Digitized by the Internet Archive

in 2018 with funding from

Kahle/Austin Foundation

https://archive.org/details/kinshipselectedrOOOOgood
 UcAn f^LlcdUj/
Penguin Education
V

Edited by Jack Goody

Penguin Modern Sociology Readings

General Editor
Tom Burns

Advisory Board
Fredrik Barth
Michel Crozier
Ralf Dahrendorf
Erving Goffman
Alvin Gouldner
Edmund Leach
David Lockwood
Gianfranco Poggi
Hans Peter Widmaier
Peter Worsley

STUDENT CC ,3
Selected Readings
Edited by Jack Goody

Penguin Books
Penguin Books Ltd, Harmondsworth,
Middlesex, England
Penguin Books Inc., 7110 Ambassador Road,
Baltimore, Md 21207, U.S.A.
Penguin Books Australia Ltd,
Ringwood, Victoria, Australia

First published 1971

This selection copyright © Jack Goody, 1971
Introduction and notes copyright © Jack Goody, 1971

Made and printed in Great Britain by

Cox & Wyman Ltd, London, Reading and Fakenham
Set in Monotype Times

This book is sold subject to the condition that

it shall not, by way of trade or otherwise, be lent,
re-sold, hired out, or otherwise circulated without
the publisher’s prior consent in any form of
binding or cover other than that in which it is
published and without a similar condition
including this condition being imposed on the
subsequent purchaser
 Contents

Introduction 9

Note on Abbreviations 15

Part One The Family 17

R. N. Adams (1960) 19
The Nature of the Family
B. Malinowski (1930) 38
The Principle of Legitimacy
B. Malinowski (1927) 42
The Family Complex in Patrilineal and Matrilineal
Societies

Part Two Incest and Sex 45

4 C. Levi-Strauss (1949) 47
The Principles of Kinship

^5 J. Goody (1956) 64
Incest and Adultery

Part Three The Developmental Cycle 83

6 M. Fortes (1958) 85
The Developmental Cycle in Domestic Groups

Part Four Joking and Avoidance 99

7 A. R. Radcliffe-Brown (1952) 101
Joking Relationships

Part Five Marriage Transactions 117

A. R. Radcliffe-Brown (1950) 119
\t Dowry and Bridewealth

\ E. Friedl (1962) 134

Dowry, Inheritance and Land-Tenure
10 J. Hajnal (1965) 140
European Marriage Patterns in Perspective

Part Six Plural Marriage 149

11 E. R. Leach (1955) 151

Polyandry, Inheritance and the Definition of Marriage

12 R. Clignet (1970) 163

Determinants of African Polygyny

Part Seven Marriage and Alliance 181

13 L. Dumont (1957) 183

The Marriage Alliance

14 D. Maybury-Lewis (1965) 199

Prescriptive Marriage Systems

Part Eight Divorce and Marriage Stability 225

15 M. Gluckman (1950) 227

Marriage Payments and Social Structure among the Lozi
and Zulu

16 J. C. Mitchell (1961) 248

Social Change and the Stability of Marriage in
Northern Rhodesia

Part Nine Kin Groups 261

17 M. Fortes (1953) 263

The Structure of Unilineal Descent Groups

18 A. I. Richards (1950) 276

Matrilineal Systems

19 R. N. Pehrson (1964) 290

Bilateral Kin Groupings

Part Ten Kin Terms 297

20 J. Goody (1970) 299

The Analysis of Kin Terms

21 A. R. Radcliffe-Brown (1930) 307

Kin Terms and Kin Behaviour
22 E. A. Hammel (1965) 317
Formal Semantic Analysis

Part Eleven Ritual and Fsctive Kinship 329

23 E. N. Goody (1970) 331
Forms of Pro-Parenthood: The Sharing and Substitution of
Parental Roles

24 S. W. Mintz and E. R. Wolf (1950) 346

Ritual Co-Parenthood (compadrazgo)

Part Twelve Changing Patterns of Kinship 363

25 W. J. Goode (1963) 365
World Changes in Family Patterns

Further Reading 384

Acknowledgements 391

Author Index 393

Subject Index 397

Introduction

In this book I have tried to set the study of kinship in as wide a

framework as possible. Social anthropologists have been primarily
concerned with studies of other cultures, while sociologists have
dwelt very much upon the part of the world in which they live.
Hence the rough division of labour between this Volume of
Readings and its parallel on the family (Anderson, 1971) is
between other cultures and Western society. This division is far
from satisfactory and not one that should be perpetuated. The
alternative split between close (familial) and distant (kin) re¬
lationships is equally objectionable, especially when it gives rise
to the idea that complex societies have the family and simpler
ones have kinship.
To try to dissolve the boundaries, I have included some sections
that refer to European kinship, especially those by Friedl, Hajnal
and Goode (Readings 9,10 and 25). But as important as extending
the scope of studies of other cultures in a geographical or socio¬
logical sense is the job of extending their theoretical and technical
range. In the early days of anthropology, large-scale, long-run
hypotheses were subjected to wide-ranging discussions; E. B.
Tylor’s essay (1889) on marriage and descent is perhaps the most
remarkable. Since that time, the range of topics considered has in
many ways grown smaller. In recent years, interest has been
concentrated on three main fields, the role of kin groups (see
Readings 6, 17 and 18), the role of marriage as alliance (see
Readings 4, 13 and 14), and the componential analysis of kin
terms. In all of these fields, stress has been placed upon ‘formal’
analysis, a tendency which can lead to a severe restriction in the
number of variables considered. Such theoretical model building
has its uses in letting us see how systems of limited possibility
might work, but it is often at a disadvantage in showing us how
they actually do work. The alternative view, which holds that these
models apply to how the actors see them work, is usually backed by

Introduction 9
insufficient evidence; ‘ideological models’ are in any case no
substitute for ‘ working models’.
In this book, then, I have included important contributions to
the main developments in the study of kinship, and especially
articles which attempted to sum up the state of play. But there are
also sections on important subjects which have had too little
attention given to them in recent years, at any rate on the general
level. In this category of neglected topic I would include the
analysis of marriage transactions (except in the limited context
of divorce), as well as the study of joking and avoidance relations,
and, indeed, that kind of depth analysis of intra-familial relation¬
ships with which Malinowski was much concerned (Fortes
(1949) provides a notable exception). I have also included sections
upon two other aspects of kinship which have never received
sufficient attention. The first is the influence of ‘demographic’
factors. The work of Hajnal (Reading 10), among others, has
emphasized how important for the social system are variables
like the age of marriage (for particular studies see Tait (1961),
Hart and Pilling (I960)); the excerpt from the work of Clignet
(Reading 12) examines the effects of widespread plural marriage
in Africa. Significantly, these two authors also analyse problems of
social change; Hajnal discusses the emergence as well as the effects
of the ‘European marriage pattern ’ (characterized by a late age of
marriage for men and women and the large proportion of spinsters
and bachelors), and Clignet looks at the part played by polygyny
in the processes of urbanization and development in contemporary
West Africa. The study of changing kinship systems has inevitably
played a smaller part in field studies, whose duration is limited to a
restricted period. But it is a necessary part of the total picture of
those institutions and one that has received an intelligent treatment
in the work of Goode (Reading 25).
The study of kinship goes back about a century to the work of
the American lawyer, Lewis Henry Morgan (1818-81), whose
major books included an account of the matrilineal society of the
Iroquois (1851), a worldwide analysis of kin terms (1871) and an
important essay on the history of the human family (1877). Mor¬
gan s work hud much influence on Marx, w'hose notes were written
up by Engels as The Origin of the Family, Private Property and the
State (1884). Apart from Morgan, the most significant figures of

10 Introduction
the nineteenth century writing on social anthropology were J. J.
Bachofen (1861), Sir Henry Maine (1861), and J. F. McLennan
(1865); the first of these claimed the primacy of matriarchy, the
second stressed the significance of early patriarchy. Maine also
insisted upon the importance of the kinship corporation, the clan
or lineage. It was McLennan, a supporter of theories of early
matriarchy, who introduced the terms exogamy (the rule of out¬
marriage) and endogamy (the rule of in-marriage) into the tech¬
nical vocabulary.
A book of this kind cannot go into these early discussions of
kinship. Indeed there is insufficient space even for what is perhaps
the most seminal paper ever written on kinship institutions: the
article of E. B. Tylor (1889) to which I have already referred.
This contribution provided the starting point for much of the
exchange theory of incest and marriage (see Levi-Strauss and
Goody (Readings 4 and 5)); in it the author describes early mar¬
riage as a ‘family transaction’ and introduces the term ‘cross¬
cousin marriage’ (see Reading 14). But it is also the first example of
the systematic application of numerical techniques to the com¬
parison of human societies, and pioneered the cross-cultural
method continued by Hobhouse, Wheeler and Ginsberg (1930) and
later developed by G. P. Murdock (1949; see also Marsh, 1967).
Tylor was fortunate in having as the chairman of the meeting at
which he gave his paper the statistician, Francis Galton, who raised
a number of points in discussion about the possible effects of
diffusion (often referred to as ‘Gabon’s problem’), which have
been given much thought in the revival of interest in cross-
cultural studies over the past decade (Naroll, 1965).
In the same essay, Tylor discussed a number of institutions
such as the levirate (by which he meant the inheritance of the
widow by the husband’s brother or other close kinsman), the
sororate (the replacement of a dead wife by her sister or close
kinswoman), the couvade (‘ the father, on the birth of his child,
makes a ceremonial pretence of being the mother’), and marriage
‘by capture’. Like other writers of the later nineteenth century he
was concerned to present long-term schemes of unilineal develop¬
ment, which turned largely upon changes in the major systems of
descent (matrilineal and patrilineal), and in the main forms of
marriage (polyandry, marriage by capture, etc.). However, his

Introduction 11
paper was a turning point in that it stressed correlational studies:
the relations of institutions existing side by side with one another.
A turning point of yet greater importance was the shift to
intensive analysis encouraged by the development of professional
fieldwork, notably by A. R. Radcliffe-Brown and B. Malinowski.
Because of their sociological bent towards functional analysis
and their experience in the field, they were able to bring a new
dimension to the study of kinship, not only in the analysis in
depth of particular societies (in which many of their students
excelled) but also in the more general fields of kinship: Radcliffe-
Brown’s paper on Australia (1930) and his introduction to
African Systems of Kinship and Marriage (1950) remain the best
general statements on a number of aspects of kinship, although
much more work has been done on studies of particular societies
and on specific topics such as cross-cousin marriage, lineage
systems and the formal analysis of kin terms.
What about the future of kinship studies? With so broad a
subject, there is an equally broad range of possibilities. We can
look forward to growing numbers of analyses in depth of particular
societies, especially as more scholars from developing countries
study their own or neighbouring cultures. The study of one’s own
culture has some disadvantages; but one clear gain is in the com¬
mand one has in matters of communication, a command that
can help to avoid some of the superficiality to which the outside
scholar is likely to fall victim. In this way, the qualitative data on
inter-personal relations should greatly improve. Secondly, the
advent of computers is already making itself felt in another way,
that is, in the formalization, categorization and measurement of
data for constructing models that attempt to simulate social
processes. Thirdly, the division between the ‘family studies’ of
‘industrial societies’ and the ‘kinship studies’ of‘pre-industrial’
ones is inevitably becoming blurred. This shift is taking place
partly as a result of recognizing the greater differentiation in the
pre-industrial sphere. For example, the lumping together of
Australian, Chinese and South Indian systems as ‘elementary
structures’ would appear to overlook the radical effects of econo¬
mic factors on kinship; there are many features of Chinese kin¬
ship that link it to the peasants of Europe rather than to the
hunters of Australia. This change in emphasis is also affected by a

12 Introduction
recognition of greater differentiation within and between in¬
dustrial societies themselves (for example, the matrifocal ten¬
dencies among the urban worldng class as contrasted with the
landed aristocracy), especially in those societies that are in the
process of ‘modernization,’ namely, the countries of the Third
World. These factors are blurring the old lines of division between
the family and kinship, between sociology and anthropology,
and hence raising more questions and leading to more adequate
answers.
In making this selection I have tried to cover a range of topics in
the field and to give preferences to articles which summarize a
number of alternative approaches as well as making a theoretical
point. In some cases I have been unable to find a suitable article
that dealt generally with the field (outside the pages of an ency¬
clopedia) ; in these cases (Readings 20 and 23), contributions have
been prepared specially for this volume. In two other cases
(Readings 15 and 22) Professors Gluckman and Hammel have
kindly made additions to their original articles. Finally, I would
like to thank Esther Goody, Sarah Cattermole and Joan Buckley
for their help with this volume.

References
Anderson, M. (ed.) (1971), Sociology of the Family, Penguin.
Bachofen, J. J. (1861), Das Mutterrecht, von Krais & Hoffman.
Fortes, M. (1949), The Web of Kinship among the Tallensi, Oxford
University Press.
Engels, F. (1884), The Origin of the Family, Private Property and the
State, Hottingen.
Hart, C. W. M., and Pilling, A. R. (1960), The Tiwi of North
Australia, Holt, Rinehart & Winston.
Hobhouse, L. T., Wheeler, G. C., and Ginsberg, M. (1930),
The Material Culture and Social Institutions of the Simpler Peoples,
London School of Economics.
McLennan, J. F. (1865), Primitive Marriage, Black.
Maine, H. (1861), Ancient Law, Murray.
Marsh, R. M. (1967), Comparative Sociology, Harcourt Brace & World.
Morgan, L. H. (1851), The League of the Iroquois, Sage & Brother.
Morgan, L. H. (1871), Systems of Consanguinity and Affinity of the
Human Family, Smithsonian Institution.
Morgan, L. H. (1877), Ancient Society, Macmillan Co.
Murdock, G. P. (1949), Social Structure, Macmillan Co.
Naroll, R. (1965), ‘Gabon’s problem: the logic of cross-cultural
analysis’, Soc. Res., vol. 32, pp. 428-51.

Introduction 13
Radcliffe-Brown, A. R. (1930), ‘Social organization of Australian
tribes’, Oceania, monogr. no. 1.
Radcliffe-Brown, A. R. (1950), ‘Introduction’, African Systems of
Kinship and Marriage, Oxford University Press.
Tait, D. (1961), ‘The Konkomba of Northern Ghana, Oxford University
Press.
Tylor, E. B. (1889X ‘On a method of investigating the development of
institutions: applied to laws of marriage and descent’, J.
anthropol. Inst., vol. 18, pp. 245-72.

14 Introduction
Note on Abbreviations

The following abbreviations occur in this book:

FS Father’s son
MS Mother’s son
FBS Father’s brother’s son
FBD Father’s brother’s daughter
MZS Mother’s sister’s son
MZD Mother’s sister’s daughter
FZS Father’s sister’s son
FZD Father’s sister’s daughter

Note on Abbreviations 15
Part One The Family

The study of the family began with a long controversy about

whether non-Western societies had such an institution at all. Its
extensive, if not universal, distribution (the answer depends upon
one’s definition) was established by the investigations of E. A.
Westermarck (1891) and by a fellow member of the London
School of Economics, Bronislaw Malinowski (1913). The nature
and necessity of the nuclear (or elementary) family, found to
have such a widespread distribution, has been subject to
extensive discussion which has recently been reviewed by Adams
(Reading 1). There follow two short excerpts from Malinowski,
one dealing with the importance of parent-child ties in the
process of socialization, and the other specifically discussing the
Freudian hypothesis about the universality of the Oedipus
Complex. From his experience among the matrilineal
Trobiands, he argues that different family systems will give rise
to different kinds of ‘nuclear complex’ (Reading 3). While
Malinowski’s treatment of kinship has contributed little to
later attempts in the direction of its formal analysis, his
approach has been important in the study of the character of
specific relationships, especially insofar as these are associated
with different systems of inheritance and descent.

References
Malinowski, B. (1913), The Family among the Australian Aborigines:
A Sociological Study, University of London Monographs in Sociology,
no. 2.
Westermarck, E. A. (1891), The History of Human Marriage,
Macmillan.
1 R. N. Adams
The Nature of the Family

Excerpts from R. N. Adams, ‘An inquiry into the nature of the family’,
in G. E. Dole and R. L. Carneiro (eds.), Essays in the Science of Culture,
Crowell, 1960, pp. 30-49.

Literature on the human family appearing during the past decade

has taken a decided swing away from the earlier simple clas-
sificatory goals of identifying lineality, locality, descent groups
and formal kin structures. The new direction, as has been noted
by many persons active in the movement, has been towards
examining the phenomenon within wider dimensions. No longer,
for example, is it possible to speak simply and securely of matri-
locality or of patrilocality without extensive and adequate analysis
of the precise configurations standing behind the activities of the
members of the particular society concerned (Fortes, 1949;
Goodenough, 1956). In a very real sense many of the formerly
analytic terms have become heuristic and descriptive.
With respect to the form of the nuclear family, however, there
has been little evidence of increased interest in fundamentals.
Concern here is as ancient as any in the field of social organization,
but treatments of it continue to be predominantly expressions of
profound convictions, buttressed by more or less convincing
logical arguments stemming from a variety of theoretical premises.
A recent example of this may be found in LaBarre’s (1954)
absorbing and provocative though unconvincing argument for the
absolute necessity and inevitability of a continuing nuclear
family. A more rigorous argument with the same conclusion but
based on different kinds of evidence is contained in Murdock
(1949). Murdock claims, on the basis of an examination of 250
societies, that there are no cases where the nuclear family is not
the fundamental unit or cell upon which all further familial and
kin elaborations are based. Both before and after Murdock’s
study, exceptions to this picture were cited, specifically the Nayar
of Malabar (Linton, 1936; Gough, 1952; Cappannari, 1953), but

R. N. Adams 19
in principle Murdock’s judgement has met with general approval.
Even an examination of the kibbutz led Spiro (1954) to conclude
that whereas the kibbutz may have eliminated the nuclear family,
it did so only through converting the entire community into a
single large gemeinschaft.
The purpose of the present essay is to question whether some
arguments in support of this general view are satisfactory, and to
do so through a review of selected cases in which the nuclear
family is manifestly only one type of basic form. This is in accord
with, but varies in focus from, the interest expressed by Levy
(1955) when he asked whether the nuclear family was ‘institu¬
tionalized’ in all societies. Levy pointed out that even though the
statuses of father, mother, spouse, sister and brother may be
present, they may not function as a nuclear family unit. He gave
as an example die case of the traditional Chinese family. In the
present paper the position is taken that social organization is
flexible enough to permit different forms of the family to exist
simultaneously. These different forms may not even take care of
the same general functional needs of the total society, and in
many cases certain of the standard nuclear family statuses (that is,
mother-wife, father-husband, unmarried children) may not
function at all. So far as present evidence indicates, there is no
question but that these statuses are present in the society; rather
it is a question of how they are filled and how they function. The
flexibility of social organization permitting the appearance of
different family forms rests on the fact that there are more ele¬
mental forms of the family than the nuclear, and that different
forms may function in relation to different aspects or character¬
istics of the total social structure.
The cases to be discussed are taken from contemporary Central
and South America. We are intentionally treating only this material
(and omitting the Nayar and similar cases) because it better
illuminates the propositions we wish to explore. Studies in Latin
America have increasingly indicated that while most contemporary
family systems of that region reckon descent bilaterally, there are
many instances where family forms other than the nuclear are
operative. The nuclear family is generally replaced in these
circumstances by a group based on what we will call the maternal
dyad, a residential unit composed of a mother and one or more

20 The Family
children. As is the case in many nuclear family residences, these
dyad households may also have a variety of other members present,
both kin and non-kin.
Our interest will focus on two dyad forms: the maternal dyad,
just described; and an adult dyad, composed of a man and woman,
which we shall clumsily call the sexual or conjugal dyad. This
dyad may be based simply on the sexual act, or may be further
sanctioned by marriage. There is a third dyad, the paternal
(composed of father and one or more children) which we will not
treat here. It is with no intent of minimizing the importance of this
dyad in the world at large that it is minimized here, but simply
because it does not appear in significant numbers apart from the
nuclear family in our data.1 The identification of the maternal
dyad, as distinct from the nuclear family, is made on the basis of
the fact that there is no husband-father regularly resident. The
cases used here are based on a distinction made between house¬
holds with a woman head and those with a man head. This identi¬
fication in terms of the sex of household heads stems from the
nature of census data from which much of the information is
derived. While having both theoretical and practical disadvantages,
it serves sufficiently well for present purposes. The presence of
woman-headed households (in these bilateral societies) is being
used here as an index to the prevalence of the maternal dyad family
form, and man-headed households as an index to the prevalence
of the nuclear family form. While some woman-headed households
are doubtless due to widowhood, the percentage of widows seldom
exceeds 5 per cent of the women in the society, and, of course,
many widows are not heads of households. While some man¬
headed households may be paternal dyads and not nuclear families,
the number is not significant in all cases where specific information
is available.
Some cases from Latin America
In his recent monograph on the community of Villa Reconcavo,
Bahia, Brazil, Harry W. Hutchinson (1957) defines an entire
1. It would perhaps be well to note at this point that not only the paternal
dyad, but many other forms both of family elements and artificial or
pseudo-kin relationships are pertinent to the discussion as it progresses. In
the interests of brevity, I am raising these principles for discussion, and am
intentionally not pursuing here all the lines of exploration they suggest.

R. N. Adams 21
social class segment of his community in terms of the fact that it is

the 290 households in the community were reported to be of this

type in the 1950 census. Although Hutchinson says (1957, p. 151)
that, ‘The composition of these households almost defies classifi¬
cation ’, he promptly notes that 55 of them (19 per cent of the total
number of families, and 61 per cent of the 90) are ‘composed of
mothers and children, with the addition in some cases of relatives
and an agregado as well as boarders’. The other households in this
class, Hutchinson describes as ‘left-overs’ from other families or
marital unions. Although Hutchinson evidently feels that these
families offer the scientist nothing but confusion, the fact that
they w'ere sufficiently distinct to move him to the extreme of
categorizing them as an entirely separate ‘social class’, and the
fact that they do manifest a considerable consistency with respect
to the presence of the dyad family indicate that they do not defy
classification.
The Services, in their report on Tobatf, Paraguay (1954),
indicate that what they call ‘ incomplete ’ families form a prominent
part of the community. Of a total of 292 families, only 133 (45-5
per cent) are complete nuclear families (with or without additional
members); of the remainder, 113 (38-8 per cent of the total) are
woman-headed households. This detailed report gives a somewhat
higher woman-headed household rate than Emma Reh’s earlier
study (1947) of the Paraguayan community of Piribebuy where she
estimated that 60 per cent of the families were complete and 33
per cent were headed by women. Since there are almost as many
woman-headed households as man-headed households in Tobatf,
there is little doubt that the maternal dyad is the basis of a highly
significant portion of the household units.
Although national statistics for Brazil and Paraguay were not
available to the writer, there is evidence from other areas that the
presence of maternal dyad families is not a matter of limited or
local significance. In Central America, 1950 census data are avail¬
able for four countries concerning the relative number of families
recorded as having women as heads of households: see page 23.
[. . .] The presence of woman heads of households in Central
America is not a confused and random situation but is definitely
associated with the Ladino population, as opposed to the Indian

22 The Family
Country Number of families Per cent of families
with woman heads
Guatemala 561,944 16*8
El Salvador 366,199 25-5
Nicaragua 175,462 26-0
Costa Rica 143,167 17-2

population, is concentrated in certain regions, and is probably

more commonly associated with town dwellers than with rural
populations (Adams, 1957).
Another region from which there has been an increasing number
of reports of dyad families is the Caribbean and the Guianas. Of
the studies that have appeared in recent years one in particular has
addressed itself to this issue and should concern us here. Raymond
T. Smith (1956) studied three Negro towns in British Guiana in
which the percentage of woman heads of households was as
follows:

Town Number of Per cent of households

households with woman heads
August Town 275 37-1
Perseverance 103 16-5
Better Hope 71 29-2

Many accounts of West Indian societies have indicated the presence

of these families (as in the work of Herskovits, Campbell, Simey,
and Henriquez) but for present purposes we will restrict ourselves
to the work of Smith.
These cases from Paraguay, Brazil, Central America and British
Guiana give ample evidence that in contemporary populations
with bilateral descent systems woman-headed households are
quite common. We infer, especially from those cases which have
been described in some detail, that this is an index to an almost
equally high incidence of families that have the maternal dyad as
their basic unit.

The universal functions approach

The problem now is to arrive at a theoretical framework that will
make these data intelligible. As literature on the family is extensive,

R. N. Adams 23
we will restrict ourselves to a limited number of theories con¬
cerning the status of the nuclear family. The writers of particular
interest to us here are Murdock, Parsons and R. T. Smith.

Murdock's multiple-function approach

Murdock’s major reasons for seeing the nuclear family as a uni¬
versal and inevitable phenomenon are that it was present in all the
societies in his original sample for Social Structure (1949), and that
logically it seemed to him that the family fulfilled a number of
functions better than any other conceivable agency. The four
functions he regards as primary (although he would doubtless
allow others for any specific society) are ‘fundamental to human
social life - fhe sexual, the economic, the reproductive, and the
educational’. Murdock is quite explicit in saying that ‘Agencies
or relationships outside of the family may, to be sure, share in
the fulfillment of any of these functions, but they never supplant
the family’ (1949, p. 10). The immediate issue that arises from
Murdock’s propositions is whether in fact other agencies have
not frequently taken over the functions that he regards as being
uniquely served by the nuclear family. In reading Murdock, one
gathers that he is referring not only to the presence of a nuclear
family in all societies, but also to its pervasiveness among house¬
hold groups in all societies. The implication is that its absence is
considered by him to be an abnormal situation. When he says that
‘no society . . . has succeeded in finding an adequate substitute
for the nuclear family, to which it might transfer these functions’,
one cannot help concluding that almost everyone in all societies
must therefore rely on the nuclear family to fulfill these functions.
The cases cited earlier make it clear that large segments of some
contemporary societies do not have functioning nuclear families,

as ‘abnormal’ or disorganized’, but are regular, viable,

4«milv units in a regular, functioning society. With respect to the
four functions listed by Murdock, we simply find that other social
agents do in fact take over the functions for extended periods;
piecisely who may do it varies from one society to another. The
educational function may be taken care of by the mother, other
relatives, chums, schools, and so on. The rationale that a male
child must have a resident father in order to learn to be a man does

24 The Family
not hold in fact. The economic function may be handled by the
mother and the children as they grow older; to this can be added
grandparents, brothers and other relatives who help either regu¬
larly or periodically. And, of course, the sexual function is handled
well by other married men, boarders, visitors, friends, and so forth.
The reproductive function does not need the father’s presence; a
midwife is more useful. While there is no denying the social
necessity of the functions that Murdock has delineated, there is
evidence that some families can achieve them without the presence
of someone identified as a ‘father-husband’.

Parsons’s dual-function approach

In a recent collection of papers Talcott Parsons (1955) has ex¬
pressed the opinion that the multiple-functions approach is not
adequate to explain the basic necessity of the nuclear family. In
its place, he offers another functional explanation. There are, he
feels, two functions, and two functions only, that are necessary
everywhere and account for the universal presence of the nuclear
family. One of these concerns, which Murdock calls the ‘edu¬
cational ’, is namely the necessity of providing socialization of the
child. The other (not on Murdock’s list, but again he probably
would not deny its potential importance) consists of the constant
development and balancing of the adult personality which is
achieved because of the constant interaction between spouses.
Parsons singles out this second function as being of particular
importance in explaining the restrengthening of the American
(USA) nuclear family today.
Since Parsons proposes these two functions as being essential
everywhere, any documented instance in which they are not
operative should be sufficient to cast doubt on his thesis. Such an
instance is provided by R. T. Smith’s detailed study of the British
Guiana Negro family. While Smith would hold that the nuclear
family does have universality in the sense that all the statuses
therein are recognized, he makes it clear in his study that some
households remain with women as heads for extended periods,
often for the greater part of the adult life of the woman concerned.
He adds, furthermore, that even when men are attached to the
household, it is precisely during this period that the ‘men spend a
considerable amount of time working away from home and they

R. N. Adams 25
do not take any significant part in the daily life of the household.
.. . There are no tasks allotted to a man in his role as husband-
father beyond seeing that the house is kept in good repair, and
providing food and clothing for his spouse and the children’
(1956, pp. 112-13). The function of socio-psychological integra¬
tion assigned by Parsons to the husband-wife relationship would
have considerable difficulty operating if the husband were absent
most of the time. The specific functions that Smith assigns to the
husband-father are economic. Parsons’s argument for the univer¬
sality of the nuclear family is basically no stronger than that of
Murdock since the functions delineated by both can be taken care
of by other agents in the society, or by other members of the
family.
The fundamental weakness in Murdock’s and Parsons’s points
of view is that they take functions that may be ‘imperatives’,
‘universal functions’, or ‘basic prerequisites’ for a society, and
try to correlate them with functions that are fulfilled by the nuclear
family. Since it is mistakenly believed that the nuclear family form
is found everywhere, that is, a universal, it must therefore be
correlated with some universal requirement of human society.
It is correct that there are social prerequisites, and that the nuclear
family has numerous functions; but to correlate the two is a
deduction that is not empirically supported.

A structural approach
Another approach to the problem of the significance of the woman¬
headed household and maternal-dyad families is taken up in
Smith’s study (1956). Following the lead of his mentor, Meyer
Fortes, Smith regards the family as something to be studied
empirically and within a temporal as well as spatial framework
(Fortes, 1949) and not a hardshelled cell that forms the building
unit of all kin-based social structure. Unlike Murdock and Parsons,
Smith has approached the family from the point of view of the
ethnographer and not the ethnologist or comparative sociologist,
and studied a society where the dyad family and woman-headed
households are normal. Much of Smith’s work is of interest, but
we will concentrate here on some major propositions referring to
the woman-headed households.
Smith reports that the woman-headed household in British

26 The Family
Guiana Negro society almost always goes through a stage during
which there is a man attached to it.2 A family starts in a nuclear
form, and later develops into the maternal dyad form when the
man leaves. Smith goes on to propose that there is a basic ‘matri-
focal ’ quality in the familial relations so that it is relatively easy for
a family to be reduced to the maternal dyad type; the husband-
wife relationship and the father-child relationship are much less
important than is the mother-child relationship. The weak
character of the husband-father role is related to a situation in the
general social structure in British Guiana. General social status
is conferred through ascribed membership in an ethnic-class
group. The specific occupation of the husband, in the lower class,
confers no prestige, and hence the children have nothing to gain
from their fathers in this matter. This is made more obvious by
comparing the lower class Negroes with members of the higher
class. In the latter, the occupation of the father is of importance
for the general social status of the entire family, and the father is
considered an indispensable part of the family. Smith correlates
the presence of the woman-headed household with a social status
system in which the father can achieve no superior status.
Smith’s work provides an important structural analysis of the
significance of the woman-headed household and shows that the
maternal dyad can and does exist effectively in spite of the theor¬
etical positions of Murdock and Parsons. Parsons, who had access
to Smith’s study prior to the preparation of his own paper, failed
to see the full implications of the Guiana material (Parsons,
1955, pp. 13 ff.). The fact that the Guiana family may include a \
man long enough to get a household institutionalized in the local
society and to undertake the procreation of children, does not
mean that the man is present to fulfill either of the functions that

2. This temporal difference was also noted in a survey of El Salvador in

terms of residence pattern: ‘even though the patterned residence at the
time of marriage or beginning to live together may be neo-local, the
subsequent departure of the man of the family leaves it a domestic establish¬
ment based on the fact that the woman lives there. It is, if you like, matri-
locality by default.’ And further: ‘The solidarity of the Salvadorean nuclear
family was reported in some places (Texistepeque and Chinameca) to be
increased after the birth of children. This does not, however, seem to hold
in all cases in view of the numerous cases in which the woman has retained
her children and the man has gone elsewhere’ (Adams, 1957, pp. 460-61).

R. N. Adams 27
Parsons tries to hold as being ‘root functions’ that ‘must be found
wherever there is a family or kinship system at all ...

The elemental family units: dyads and nucleus

In rejecting the propositions advanced by Parsons and Murdock
in favour of a structural approach, their position concerning the
elemental importance of the nuclear family is also cast into doubt.
If ‘functions’ do not explain the absence of the nuclear family in
some situations, they can hardly be called upon to support the
claim of universality for that form. No matter how fruitful this
position has been in reference to other problems in social structure,
we must seek an alternative view here.
The nuclear family comprises three sets of relationship that are
identifiable as dyads. There is the relationship based on coitus
between a man and a woman, and which may be identified as the
sexual dyad until or unless it is recognized as a marital union, in
which case it becomes a conjugal dyad. There is, second, the
maternal dyad, composed of mother and child, that presumably
begins at the time of conception but is not of great social signi¬
ficance as a dyad until parturition. And third, there is the paternal
dyad,*’' between father and child, that is identified specifically
because of the relationship established by the sexual or conjugal
dyad. Both the sexual and conjugal dyad, on the one hand, and the
maternal on the other, have clear cut correlates in biological
activity. The paternal does not. So no matter what importance it
may hold in a given society, at the present level of analysis it must
be looked upon as a dyadic relationship of a different order; it
exists not by virtue of a biological correlate, but by virtue of other
dyads. Once given these dyads (all three: the sexual-conjugal,
maternal and paternal) there are important economic functions
that may be assigned them. Infant dependency through nursing is,
after all, an economic relationship as well as a biological one. But
the economic cooperation and interdependency that may be as¬
signed beyond this level is clearly a socially defined activity with
no immediate biological correlates.
If we reject the idea that the nuclear family is the fundamental
‘atom’ in the social ‘molecule’, or the irreducible unit of human
kin organization, and take initially the two dyads with biological
correlates as two distinct components which must each be present

28 The Family
at certain times, but not necessarily always or simultaneously, we
will be approaching a view of the elements of social organization
which is less biased by contemporary social system philosophy.
If we allow that the nuclear family is not the minimum model for
the building of subsequent structures, then we can see that it is
basically, as Lowie partially suggested (1948, p. 215), an unstable
combination of two simpler elements, each of which is also un¬
stable and temporal. This allows us to look at more complex
forms without the bias of assuming the nuclear family always to be
present, and to seek excuses for its absence. There is a significance
to be attached to the nuclear as well as the dyad forms, but it is
distinctive. The conjugal or sexual dyad is particularly significant
because it is the reproductive unit of the society; the maternal
dyad is the temporal link between successive generations of adult
dyads. While theoretically the two kinds of dyads can operate
independently at all times, the society would be a sadly disjointed
affair were they to do so. Their combination into a nuclear family
provides generational relationships for all concerned. Since such
combinations can be a short-lived activity for the individuals
involved, and actually may occupy only a limited time, most
people are theoretically available most of the time to focus on the
dyadic relationships.
The reason that human societies have supported the nuclear
family in such abundance can be found at the level of social analysis.
Like all animals, human beings live not only in families, but in
larger aggregates which, following general usage, can generically
be called communities. A community cannot maintain stability
and continuity solely with such unstable and temporal forms as
dyads for elemental units. Seen from this point of view, the
nuclear family becomes one combination that, if on nothing
more than a random basis, must inevitably occur from time to
time. It is the simplest way of joining the two dyads. Since the
mother is the only adult in the maternal dyad, and the wife is the
only female in the sexual dyad, they can be jointed most readily by
identifying the wife with the mother. Once this identification is
made, the nuclear unit is created and can fulfill many potential
functions. But while its occurrence is inevitable, its continuation
is by no means inevitable because each of the dyads alone can
also fulfill some functions, and there are, in addition, presumably

R. N. Adams 29
other societal agents that can also fulfill them. The nuclear

maintains itself. For some functions and under some circumstances,

individuals may be effective agents; for others the elemental dyads
are more efficient; for yet others the nuclear family may serve, and
still others find other kinds of groups more useful. There are, in
short,
combined for continued maintenance of the
The social universal of human society are not, then, as has
been held by many students, the nuclear family and the com¬
munity, but rather the community and the two dyads. The nuclear
family is, in a sense, a structural by-product of the nature of the
dyads, but one which is almost inevitable, even if for the briefest
period. However, beyond these, the dyads may be subject to a
variety of combinations to further the continuity of the community.
The case described by Spiro (1954) as existing in the kibbutz and
the details of the woman-headed households of the British Guiana
Negroes described by Smith should not be interpreted as being
exceptions to a principle of nuclear family universality, but as
positive illustrations of how dyads may and do operate outside
of the nuclear family. [. . .]
It should not be thought that the concept of the dyad in social
structure has gone unnoticed in social anthropology. Its signifi¬
cance, however, has usually been in descriptive terms rather than
as an analytical tool. A. R. Radcliffe-Brown, certainly a pioneer in
structural studies, pointed out on a number of occasions that the
basic elements of social structure were dyadic:
I regard as a part of the social structure all social relations of person
to person. For example, the kinship structure of any society consists
of a number of such dyadic relations, as between a father and son, or a
mother’s brother and his sister’s son. In an Australian tribe the whole
social structure is based on a network of such relations of person to
person, established through genealogical connections (1952a, p. 191;
see also 1952b, pp. 52-3).
But Radcliffe-Brown’s view was somewhat different from that
proposed here, as he also held that, ‘The unit of structure from
which a kinship system is built up is the group which I call an
“elementary family”, consisting of a man and his wife and their
child or children, whether they are living together or not’ (1952b

30 The Family
p. 51). The nuclear family, as a constellation of statuses, served
as the central block although, unlike Murdock and Parsons,
Radcliffe-Brown did not hold that this unit must everywhere
exist.
While Radcliffe-Brown saw in the ‘elementary family’ three
kinds of social relationships, ‘that between parent and child, that
between children of the same parents (siblings), and that between
husband and wife as parents of the same child or children’
(1952b, p. 51), he did not expressly project these as potential
analytical units that could themselves be examined apart from the
nuclear family context and considered as distinctive building
blocks. On the other hand, Radcliffe-Brown did, in his principles
of ‘the unity of the sibling group’ and ‘the unity of the lineage’,
recognize the theoretical significance of a society’s placing em¬
phasis upon a given set of relationships that, in terms of the
present discussion, we would see as a ‘sibling dyad’ and either the
maternal or paternal dyad. He did not carry it farther at the time
of the essay in question to include the husband-wife dyad as also
being a potential centre of emphasis, nor did he distinguish between
other maternal and paternal relations.

The woman-headed household and the total society

The thesis presented by Smith concerning the reasons for the
appearance of the woman-headed household provides an analysis
that on the surface fits well into the present argument. Over a
single life cycle of Guiana Negroes the sexual or conjugal dyad
tends to come into play strongly only at limited periods - for
procreation and for support of the woman with infant. As a
woman becomes free of dependent infants, the conjugal relation
can and often does disappear or change its character. This dyad is
weak because the members are part of an ethnically distinct,
lower class community in which there is no status differentiation
possible between males, and hence, little that one man can offer a
family or son over what another can offer. According to this
analysis we would expect to find similar developments in other
similar situations. However, the data from Latin America do not
support the extension of the analysis. Three examples will indicate
the nature of the variations.
The first involves a comparison of the Guatemalan Indians and

R. N. Adams 31
neighbouring Ladinos. The former have predominantly nuclear
families while the latter have a significantly high proportion of
woman-headed households. The populations involved hold
comparable positions within the total social structure, but the
Indians in particular are similar to the Guiana Negroes in being a
lower class ethnic group within which the status of the father does
not necessarily give status to the son. There is some variation in
this matter, and a situation comparable to that of the Guiana
Negroes is to be found less among the more traditional Indians
than among the more acculturated ones. Among both Indians and
Ladinos some segments of the population work on plantations,
some live in independent villages, and some are part-time sub¬
sistence farmers and part-time labourers. Both have the same
general concept of land tenure, and both live within the same
general national context. But, it will be remembered, in the
predominantly Indian departments the percentage of households
w'ith women as heads is considerably lower than that of the
Ladino departments.
Although Indians and Ladinos live under similar conditions,
the Ladino family is much closer to the model Smith sets up for the
British Guiana Negroes than is the Mayan Indian family. The
difference lies in what Smith has referred to in the Guiana situa¬
tion as the ‘marginal nature of the husband-father role’ that
gives rise to a ‘matrifocal system of domestic relations and house¬
hold groupings’. Shifting the theoretical focus from the structure
of the family to the values associated with it is in one sense a
shorthand method of indicating that somewhere the structure, in
spite of overtly similar conditions, is different. Thus, presumably
the Indians have within the structure of their total community
certain features which stress the father-husband role, but they are
not necessarily the same features whose absence causes the weak
role in the Guiana Negro situation.
Smith reports another case in a later paper (n.d.) in which he
says that the East Indian residents of British Guiana (like the
Guatemalan Mayan Indians) have retained a strong father-
husband role in spite of the similarity to the Negroes in their
general circumstances. ‘ Quite apart from their historical derivation
the ideal patterns of [East] Indian culture and family life have
themselves become an object of value in distinguishing Indians

32 The Family
from their nearest neighbours in the ethnic status system, the
Negroes.’ If Smith is interpreting his Guiana data and I my
Guatemalan material correctly, the reasons behind women-
headed households among the Guiana Negroes are relative to the
structure of the particular society. Values associated with one
phase or aspect of the social structural system may in fact con¬
flict with or contradict values stemming from or associated with
other aspects. Thus in many Guatemalan Indian situations, where
the population works on coffee plantations, the nuclear family is
not sustained through variable social status derived from the
father, but is important economically. During the five or six
months of harvest, the wife also brings in a significant income
through picking coffee. This means that a man with a wife has
access to a larger income than one without a wife.
Societies, in which families exist, offer many faces, and the form
a specific family takes must integrate with as much of the total
system as possible. Total systems are complex and seldom com¬
pletely self-integrated, so some aspects will be more significant for
the family form of some parts of the population, while other aspects
prove to be more important for others. There is thus room for
variation in the form a family may take simply because different
families may be answering to different structural features.
The last case involves the Guiana Negroes and the Ladinos
themselves, both societies in which two distinctive forms of the
family appear within similar total structural situations. Smith
reported, and the censuses for Central America show, that within
these populations there are variations in the degree to which the
woman-headed household occurs. If Smith’s argument with
respect to the relation between the dyadic Negro household and
the total system is valid, we must then account for the presence of
some continuous nuclear families. The answer here is probably the
same as that just discussed. Within the total structure, there is
room for variation, and we must assume that in spite of the
structural features appearing to be the same, we are not identifying
those features which the different family forms are answering to.
The evidence from the present cases does not provide us with a
clear enough picture to delineate with precision why some families
go one way and some go another. It is here that we must rest our
case simply by preferring to place our confidence in the structural

T-K-B R. N. Adams 33
approach to solve the issue as over and against the ‘universal
functions’ preferred by other writers. We need to seek out facts
pertaining to a number of situations:

1. We need to delineate the types of structural aspects which can

differentially affect family forms within a single class or ethnic
societal group, or both.
2. Given this, we need then to establish the principles which will
hold for such relationships within any society.

Summary and conclusions

The preceding discussion has been exploratory, working on the
hypothesis-building level. The following summary remarks are
made in the light of the same approach.

1. The concept of ‘functions’ as being activities necessary to the

maintenance of the species, society, or individual personality is
one which is not satisfactory to explain the various forms that the
family may have in a given society. The economic, sexual, re¬
productive and educational functions as outlined by Murdock, or
the socialization and adult-personality-maintaining functions of
Parsons, may be taken care of by the nuclear family under some
circumstances and not under others. We cannot agree with Parsons
that there are ‘root functions’ everywhere associated with the
nuclear family. If there are such things, they would probably be
better identified in terms of the community and the dyads. The
search for universal functions has unfortunately become an
activity not unlike the continuing search for human instincts: it
is not that there are none, but that it is misleading unless it is
correlated with structure.
2. A theoretical analysis of the human family must not start with
the assumption that the nuclear family is a basic cell or atom, but
rather that there are two distinct dyadic relations that go into the
formation of the nuclear family as well as into other family
forms. While the concept of the nuclear family is doubtless useful
for many kinds of social analysis, the fact that it fails in analysing
some family forms means we must look further. A full under¬
standing of family form requires an analysis beginning with
dyads. With this kind of approach, it may well prove that the
nuclear family has not had the extensive ramifications which have

34 The Family
been attributed to it heretofore. By adding other dyads, we are in
a position to re-analyse kin and family structure as well as to
pursue more analytically the nature of intrafamilial and other
interpersonal relationships. It has been recognized that the nuclear
family, as found among apes and men, is essentially a very primitive
form. It is not surprising to find that man’s culture elaborates on
the dyadic possibilities of the family and produces forms intricate
and fantastic.

3. Smith’s work among the British Guiana Negroes gives us a

most important insight into the structural correlates of the
woman-headed household in that society. It leads us to the next
step, which is to seek the structural correlates which will explain
why woman-headed households sometimes appear and sometimes
do not within apparently a single structural system. One step in
this explanation is to have recourse to the theoretical position
that the other aspects of the total social structure may be working
adversely to those which are producing a nuclear or a dyadic
emphasis. The emphasis thus placed, however, must have struc¬
tural correlates, even if they are merely reflective of some structural
aspect that is about to disappear. In this case we need more
research into the exact nature of form and structure relationships,
both in a synchronic and a diachronic context.

4. The final general position to be derived from the preceding

discussion is that it is neither necessary nor valid to attempt to
find a single normal structural form for the family within a society.
That there may be only one is possible; but the assumption that
there can be only one is unfruitful. The conviction that there is
only one right way is older than social science, but it continues to
make itself felt today. Many sociologists and anthropologists
have regarded the woman-headed household as an abnormal,
incomplete or disorganized form of the family. This has contri¬
buted to the argument that the nuclear family is an indispensable,
basic, stable, family type, and that its absence must therefore
represent a breakdown. If we accept the notion, however, that the
basic relational elements of the family are dyadic, and that the
nuclear family is a more complex arrangement but one which
is probably even less significant temporally than its dyad com¬
ponents, then we are in a position to see women-headed

R. N. Adams 35
households as alternative or secondary norms rather than forms of
disorganization. The assertion that the nuclear family successfully
fulfills certain functions is perfectly valid. But the reverse assertion
that other social forms can never suitably fulfill these functions
is both empirically and theoretically invalid.
The denial of this reverse assertion is also important for our
approach to other cultural forms. The search for a fundamental
cell or building block of kin organization leads not only to a
misplaced emphasis on the nuclear family, but towards a biased
approach in the study of the entire family system. As Goodenough
(1956) has pointed out with respect to residence, there are ethno¬
graphic ways of seeing things, and there are ethnological ways of
seeing the same things. Just as the desire to discover cross-cultural
regularities has led to forcing an ethnological strait jacket on a
society’s residence rules, so it has led to misleading assumptions
concerning the identification of the nuclear family as the minimum
structural form of family organization. If we look into other
aspects of culture, it seems likely that we should assume that all
cultural forms are alternatives (in the Lintonian sense) until a
given form can be demonstrated to be universal by the ethno¬
graphers. To assume that a form, because it is a variant, is abnor¬
mal, is to evade the task before us. The first job of science is, after
all, to study what is, not what might, or could, or should be.3
3. I am indebted to William Davenport, Iwao Ishino, Raymond T. Smith,
Nancie Solien de Gonzalez, John Useem, Gertrude E. Dole and Robert
L. Carneiro for critical readings of earlier drafts of this paper.

References
Adams, R. N. (1957), ‘Culture surveys of Panama - Nicaragua - Guate¬
mala - El Salvador - Honduras’, Pcinamerican Sanitary Bureau,
Scientific Publications, no. 33.
Cappannari, S. C. (1953), ‘Marriage in Malabar’, Southwestern J.
AnthropoL, vol. 9, pp. 263-7.
Fortes, M. (1949), ‘Time and social structure: An Ashanti case study’,
in M. Fortes (ed.), Social Structure: Studies Presented to A. R.
Radcliffe-Brown, Clarendon Press, pp. 54-84.
Frazier, E. F. (1939), The Negro Family in the United States, University
of Chicago Press.
Goodenough, W. H. (1956), ‘Residence rules’. Southwestern J.
Anthropol., vol. 12, pp. 22-37.

36 The Family
Gough, E. K. (1952), ‘Changing kinship usages in the setting of
political and economic change among the Nayar of Malabar’, /. roy,
anthropol. Inst., vol. 82, pp. 71-88.
Hutchinson, H. W. (1957), Village and Plantation Life in Northeastern
Brazil, American Ethnological Society, University of Washington
Press.
LaBarre, W. (1954), The Human Animal, University of Chicago Press.
Levi-Strauss, C. (1956), ‘The family’, in H. L. Shapiro (ed.),
Man, Culture, and Society, Oxford University Press, pp. 261-85.
Levy, M. J., Jnr (1955), ‘Some questions about Parsons’ treatment of
the incest problem’, Brit J. Sociol., vol. 6, pp. 277-85.
Linton, R. (1936), The Study of Man, Appleton-Century-Crofts.
Lowie, R. H. (1948), Social Organization, Holt, Rinehart & Winston.
Murdock, G. P. (1949), Social Structure, Macmillan Co.
Parsons, T. (1955), ‘The American family: Its relations to personality
and to the social structure’, in T. Parsons and R. F. Bales, Family,
Socialization and Interaction Process, Free Press, pp. 3-33.
Radcliffe-Brown, A. R. (1952a), ‘On social structure’, in A. R.
RadclifFe-Brown, Structure and Function in Primitive Society, Cohen &
West, pp. 188-204.
Radcliffe-Brown, A. R. (1952b), ‘The study of kinship systems’,
in A. R. Radcliffe-Brown, Structure and Function in Primitive Society,
Cohen & West, pp. 49-89.
Reh, E. (1946), Paraguayan Rural Life, Institute of Inter-American
Affairs.
Service, E. R., and Service, H. S. (1954), Tobatl: Paraguayan Town,
University of Chicago Press.
Smith, R. T. (1956), The Negro Family in British Guiana: Family
Structure and Social Status in the Villages, Routledge & Kegan Paul.
Smith, R. T. (1957), ‘Family structure and plantation systems in the New
World’, paper presented at the seminar on Plantation Systems of the
New World, San Juan, Puerto Rico.
Spiro, M. E. (1954), ‘Is the family universal?’ Amer. Anthropol.,
vol. 56, pp. 839-46.

R. N. Adams 37
2 B. Malinowski
The Principle of Legitimacy

Excerpt from B. Malinowski, ‘Parenthood - the basis of social

structure’, in V. F. Calverton and S. D. Schmalhausen (eds.), The
New Generation, Allen & Unwin, 1930, pp. 113-68.

We can say that motherhood is always individual. It is never

allowed to remain a mere biological fact. Social and cultural
influences always endorse and emphasize the original individuality
of the biological fact. These influences are so strong that in the
case of adoption they may override the biological tie and substitute
a cultural one for it. But statistically speaking, the biological ties
are almost invariably merely reinforced, redetermined and re¬
molded by the cultural ones. This remolding makes motherhood
in each culture a relationship specific to that culture, different
from all other motherhoods, and correlated to the whole social
structure of the community. This means that the problem of
maternity cannot be dismissed as a zoological fact, that it should
be studied by every field-worker in his own area, and that the
theory of cultural motherhood should have been made the founda¬
tion of the general theory of kinship.
What about the father? As far as his biological role is con¬
cerned he might well be treated as a drone. His task is to impreg¬
nate the female and then to disappear. And yet in all human
societies the father is regarded by tradition as indispensable. The
woman has to be married before she is allowed legitimately to con¬
ceive. Roughly speaking, an unmarried mother is under a ban, a
fatherless child is a bastard. This is by no means only a European
or Christian prejudice; it is the attitude found amongst most
barbarous and savage peoples as well. Where the unmarried
mother is at a premium and her offspring a desirable possession,
the father is forced upon them by positive instead of negative
sanctions.
Let us put it in more precise and abstract terms. Among the
conditions which define conception as a sociologically legitimate

38 The Family
fact there is one of fundamental importance. The most important
moral and legal rule concerning the physiological side of kinship
is that no child should be brought into the world without a man -
and one man at that - assuming the role of sociological father, that
is, guardian and protector, the male link between the child and the
rest of the community.
I think that this generalization amounts to a universal sociologi¬
cal law and as such I have called it in some of my previous writings
The Principle of Legitimacy.1 The form which the principle of
legitimacy assumes varies according to the laxity or stringency
which obtains regarding prenuptial intercourse; according to the
value set upon virginity or the contempt for it; according to the
ideas held by the natives as to the mechanism of procreation;
above all, according as to whether the child is a burden or an asset
to its parents. Which means according as to whether the unmarried
mother is more attractive because of her offspring or else degraded
and ostracized on that account.
Yet through all these variations there runs the rule that the
father is indispensable for the full sociological status of the child
as well as of its mother, that the group consisting of a woman and
her offspring is sociologically incomplete and illegitimate. The
father, in other words, is necessary for the full legal status of the
family.
In order to understand the nature and importance of the prin¬
ciple of legitimacy it is necessary to discuss the two aspects of pro¬
creation which are linked together biologically and culturally,
yet linked by nature and culture so differently that many diffi¬
culties and puzzles have arisen for the anthropologists. Sex and
parenthood are obviously linked biologically. Sexual intercourse
leads at times to conception. Conception always means pregnancy
and pregnancy at times means childbirth. We see that in the
chain there are at least two possibilities of a hiatus; sexual inter¬
course by no means always leads to conception, and pregnancy can
be interrupted by abortion and thus not lead to childbirth.

1. Compare article entitled ‘Kinship’ in the Encyclopedia Britannica,

14th edn.; also Sex and Repression in Savage Society, Routledge & Kegan
Paul (1927), and The Family Among the Australian Aborigines, Schoken
(1913), ch. 6. In this latter the relevant facts are presented though the term
is not used.

B. Malinowski 39
 The moral, customary and legal rules of most human com¬
munities step in, taking advantage of the two weak links in the
chain, and in a most remarkable manner dissociate the two sides
of procreation, that is sex and parenthood. Broadly speaking, it
may be said that freedom of intercourse though not universally is
yet generally prevalent in human societies. Freedom of conception,
outside marriage is, however, never allowed, or at least in ex¬
tremely few communities and under very exceptional circum¬
stances.
Briefly to substantiate this statement: it is clear that in those
societies, primitive or civilized, where prenuptial intercourse is
regarded as immoral and illegitimate, marriage is the conditio sine
qua non of legitimate children - that is children having full social
status in the community.
In the second place, in most communities which regard pre¬
nuptial intercourse as perfectly legitimate, marriage is still re¬
garded as essential to equip the child with a full tribal position.
This is very often achieved without any punitive sanctions, by the
mere fact that as soon as pregnancy sets in a girl and her lover
have to marry. Often in fact pregnancy is a prerequisite of mar¬
riage or the final legal symptom of its conclusion.
There are tribes, again, where an unmarried mother is definitely
penalized and so are her children. What is done under such condi¬
tions by lovers who want to live together sexually and yet not
to produce children is difficult to say. Having had in my own
field-work to deal with the case in point, I was yet unable to arrive
at a satisfactory solution. Contraceptives, I am firmly con¬
vinced, do not exist in Melanesia, and abortion is not sufficiently
frequent to account for the great scarcity of illegitimate children.
As a hypothesis, I venture to submit that promiscuous intercourse,
while it lasts, reduces the fertility of woman. If this side of the
" hole question still remains a puzzle it only proves that more
research, both physiological and sociological, must be done in
oider fully to throw light upon the principle of legitimacy.
There is still one type of social mechanism through which the
principle of legitimacy operates, and that is under conditions where
a child is an asset. There an unmarried mother need not trouble
about her sociological status, because the fact of having children
only makes her the more desirable, and she speedily acquires a

40 The Family
husband. He will not trouble whether the child is the result of his
love-making or not. But whether the male is primed to assume
his paternity, or whether child and mother are penalized, the
principle of legitimacy obtains throughout mankind; the group
of mother and child is incomplete and the sociological position of
the father is regarded universally as indispensable.

B. Malinowski 41
3 B. Malinowski

The Family Complex in Patrilineal and Matrilineal

Societies

Excerpts from B. Malinowski, Sex and Repression in Savage Society,

Routledge & Kegan Paul, 1927, pp. 79-262.

The repressing and molding forces in Melanesia are twofold -

exogamy. The first is brought about by the influence of the mother’s

brother, who, in appealing to the child’s sense of honour, pride and
ambition, comes to stand to him in a relation in many respects
analogous to that of the father among us. On the other hand,
both the efforts which he demands and the rivalry between suc¬
cessor and succeeded introduce the negative elements of jealousy
and resentment. Thus an ‘ambivalent’ attitude is formed in
which veneration assumes the acknowledged dominant place,
while a repressed hatred manifests itself only indirectly.
The second taboo, the prohibition of incest, surrounds the
sister and to a lesser degree other female relatives on the maternal
side, as well as clanswomen, with a veil of sexual mystery. Of all
this class of women, the sister is the representative to whom the
taboo applies most stringently. We noted that this severing taboo,
entering the boy’s life in infancy, cuts short the incipient tenderness
towards his sister which is the natural impulse of a child. This taboo
also, since it makes even an accidental contact in sexual matters a
crime, causes the thought of the sister to be always present, as
well as consistently repressed.
Comparing the two systems of family attitudes briefly, we see
that in a patriarchal society, the infantile rivalries and the later
social functions introduce into the attitude of father and son,
besides mutual attachment, also a certain amount of resentment
and dislike. Between mother and son, on the other hand, the
premature separation in infancy leaves a deep, unsatisfied craving
which, later on, when sexual interests come in, is mixed up in
memory with the new bodily longings, and assumes often an

42 The Family
 erotic character which comes up in dreams and other fantasies.
In the Trobriands there is no friction between father and son, and
all the infantile craving of the child for its mother is allowed
gradually to spend itself in a natural, spontaneous manner. The
ambivalent attitude of veneration and dislike is felt between a man
and his mother’s brother, while the repressed sexual attitude of
incestuous temptation can be formed only towards his sister.
Applying to each society a terse, though somewhat crude formula,
we might say that in the Oedipus complex there is the repressed
desire to kill the father and marry the mother, while in the matri-
lineal society of the Trobriands the wish is to marry the sister and
to kill the maternal uncle. [. . .]
Cultural training is not merely the gradual development of
innate faculties. Besides an instruction in arts and knowledge,
this training also implies the building up of sentimental attitudes,
the inculcation of laws and customs, the development of morality.
And all this implies one element which we have found already in
the relation between child and mother, the element of taboo,
repression, of negative imperatives. Education consists in the last
instance in the building up of complex and artificial habit responses,
of the organization of emotions into sentiments.
As we know, this building up takes place through the various
manifestations of public opinion and of moral feeling, by the
constant influence of the moral pressure to which the growing
child is exposed. Above all it is determined by the influence of
that framework of tribal life which is made up of material ele¬
ments and within which the child gradually grows up, to have its
impulses molded into a number of sentiment patterns. This
process, however, requires a background of effective personal
authority, and here again the child comes to distinguish between
the female side of social life and the male side. The women who
look after him represent the nearer and more familiar influence,
domestic tenderness, the help, the rest and the solace to which the
child can always turn. The male aspect becomes gradually the
principle of force, of distance, of pursuit of ambition and of
authority. This distinction obviously develops only after the
earlier period of infancy, in which, as we have seen, the father
and the mother play a similar part. Later on, though the mother,
side by side with the father, has to train and teach the child, she

B, Malinowski 43
still continues the tradition of tenderness, while the father in
most cases has to supply at least a minimum of authority within
the family.
At a certain age, however, there comes the time at which the
male child becomes detached from the family and launches
into the world. In communities where there are initiation cere¬
monies this is done by an elaborate and special institution, in
which the new order of law and morality is expounded to the
novice, the existence of authority displayed, tribal conditions
taught and very often hammered into the body by a system of
privations and ordeals. From the sociological point of view, the
initiations consist in the weaning of the boy from the domestic
shelter and submitting him to tribal authority. In cultures where
there is no initiation the process is gradual and diffused, but
its elements are never absent. The boy is gradually allowed or
encouraged to leave the house or to work himself loose from the
household influences, he is instructed in tribal tradition and
submitted to male authority.
But the male authority is not necessarily that of the father.
[...] In societies where the authority is placed in the hands of the
maternal uncle the father can remain the domestic helpmate and
friend of his sons. The father to son sentiment can develop
simply and directly. The early infantile attitudes gradually and
continually ripen with the interests of boyhood and maturity.
The father in later life plays a role not entirely dissimilar to that
at the threshold of existence. Authority, tribal ambition, re¬
pressive elements and coercive measures are associated with
another sentiment, centring round the person of the maternal
uncle and building up along entirely different lines. [...]
There is only one way of avoiding the dangers which surround
the paternal relation and this is to associate the typical elements
which enter into the paternal relation with two different people.
This is the configuration which we find under mother-right.

44 The Family
Part Two Incest and Sex

The role of prohibitions on sex between close kin (‘the incest

taboo ’) in the development of human society was the theme of
Freud’s study Totem and Taboo (1913). Like Malinowski, Freud
attempted to relate these taboos to the requirements of family
structure, and particularly to the process of socialization.
Another line of thought, which was set in train by E. B. Tylor
in a famous essay (1889), linked incest with exogamy, that is,
with the rule of out-marriage by which lines of communication
are established and maintained between adjacent social groups.
This view was developed by Levi-Strauss in his book, The
Elementary Structures of Kinship (1949), an excerpt from which
is included in this section. The article by Goody attempts to
reconcile these two approaches and to suggest that they can each
be seen as applying to different aspects of the taboo, that is, to
parent-child and sibling incest respectively.

References
Freud, S. (1913), Totem and Taboo, Routledge & Kegan Paul.
Levi-Strauss, C. (1969), The Elementary Structures of Kinship, Eyre
& Spottiswoode.
Tylor, E. B. (1889), ‘On a method of investigating the development of
institutions: applied to laws of marriage and descent’, /. anthropol.
Inst., vol. 18, pp. 245-72.
\
4 C. Levi-Strauss
The Principles of Kinship

Excerpt from C. Levi-Strauss, The Elementary Structures of Kinship,

Eyre & Spottiswoode, 1969, pp. 478-90. First published in 1949.

It is always a system of exchange that we find at the origin of

rules of marriage, even of those of which the apparent singularity
would seem to allow only a special and arbitrary interpretation.
In the course of this work, we have seen the notion of exchange
become complicated and diversified; it has constantly appeared
to us in different forms. Sometimes exchange appears as direct
(the case of marriage with the bilateral cousin), sometimes as
indirect (and in this case it can comply with two formulas, one
continuous, the other discontinuous, corresponding to two differ¬
ent rules of marriage with the unilateral cousin). Sometimes it
functions within a total system (this is the theoretically common
characteristic of bilateral marriage and of matrilateral marriage),
and at others it instigates the formation of an unlimited number
of special systems and short cycles, unconnected among them¬
selves (and in this form it represents a permanent threat to
moiety systems, and as an inevitable weakness attacks patrilateral
systems). Sometimes exchange appears as a cash or short-term
transaction (with the exchange of sisters and daughters, and
avuncular marriage), and at other times more as a long-term
transaction (as in the case where the prohibited degrees include
first, and occasionally second, cousins). Sometimes the exchange
is explicit and at other times it is implicit (as seen in the example
of so-called marriage by purchase). Sometimes the exchange is
closed (when marriage must satisfy a special rule of alliance
between marriage classes or a special rule for the observance of
preferential degrees), while sometimes it is open (when the rule
of exogamy is merely a collection of negative stipulations, which,
beyond the prohibited degrees, leaves a free choice). Sometimes
it is secured by a sort of mortgage on reserved categories (classes

C. Levi-Strauss 47
or degrees); sometimes (as in the case of the simple prohibition
of incest, as found in our society) it rests on a wider fiduciary
guarantee, viz., the theoretical freedom to claim any woman of
the group, in return for the renunciation of certain designated
women in the family circle, a freedom ensured by the extension
of a prohibition, similar to that affecting each man in particular,
to all men in general. But no matter what form it takes, whether
direct or indirect, general or special, immediate or deferred,
explicit or implicit, closed or open, concrete or symbolic, it is
exchange, always exchange, that emerges as the fundamental
and common basis of all modalities of the institution of marriage.
If these modalities can be subsumed under the general term of
exogamy (for endogamy is not opposed to exogamy, but pre¬
supposes it), this is conditional upon the apperception, behind
the superficially negative expression of the rule of exogamy, of the
final principle which, through the prohibition of marriage within
prohibited degrees, tends to ensure the total and continuous
circulation of the group’s most important assets, its wives and
its daughters.
The functional value of exogamy, defined in its widest sense,
has been specified and brought out in the preceding chapters.
This value is in the first place negative. Exogamy provides the
only means of maintaining the group as a group, of avoiding
the indefinite fission and segmentation which the practice of
consanguineous marriages would bring about. If these consan¬
guineous marriages were resorted to persistently, or even over-
frequently, they would not take long to ‘fragment’ the social
group into a multitude of families, forming so many closed
systems or sealed monads which no pre-established harmony
could prevent from proliferating or from coming into conflict.
The rule of exogamy, applied in its simplest forms, is not entirely
sufficient to the task of warding off this mortal danger to the
group. Such is the case with dual organization. With it there is no
doubt that the risk of seeing a biological family become estab¬
lished as a closed system is definitely eliminated; the biological
group can no longer stand apart, and the bond of alliance with
another family ensures the dominance of the social over the
biological, and of the cultural over the natural. But there im¬
mediately appears another risk, that of seeing two families, or

48 Incest and Sex

rather two lineages, isolate themselves from the social continuum
to form a bi-polar system, an indefinitely self-sufficient pair,
closely united by a succession of intermarriages. The rule of
exogamy, which determines the modalities for forming such pairs,
gives them a definite social and cultural character, but this social
character is no sooner given than it is disintegrated. This is the
danger which is avoided by the more complex forms of exogamy,
such as the principle of generalized exchange, or the subdivision
of moieties into sections and subsections in which more and more
numerous local groups constitute indefinitely more complex
systems. It is thus the same with women as with the currency
the name of which they often bear, and which, according to the
admirable native saying, ‘depicts the action of the needle for
sewing roofs, which, weaving in and out, leads backwards and
forwards the same liana, holding the straw together’ (Leenhardt,
1930, pp. 48, 54). Even when there are no such procedures, dual
organization is not itself ineffective. We have seen how the inter¬
vention of preferred degrees of kinship within the moiety, e.g.
the predilection for the real cross-cousin, and even for a certain
type of real cross-cousin, as among the Kariera, provides the
means of palliating the risks of an over-automatic functioning
of the classes. As opposed to endogamy and its tendency to set
a limit to the group, and then to discriminate within the group,
exogamy represents a continuous pull towards a greater cohesion,
a more efficacious solidarity, and a more supple articulation.
This is because the value of exchange is not simply that of the
goods exchanged. Exchange - and consequently the rule of
exogamy which expresses it - has in itself a social value. It
provides the means of binding men together, and of superimpos¬
ing upon the natural links of kinship the henceforth artificial
links - artificial in the sense that they are removed from chance
encounters of the promiscuity of family life - of alliance governed
by rule. In this connection, marriage serves as model for that
artificial and temporary ‘conjugality’ between young people of
the same sex in some schools and on which Balzac makes the
profound remark that it is never superimposed upon blood ties
but replaces them: ‘It is strange, but never in my time did I
know brothers who were “Activists”. If man lives only by his
feelings, he thinks perhaps that he will make his life the poorer

C. Levi-Strauss 49
if he merges an affection of his own choosing in a natural tie.’1
On this level, certain theories of exogamy which were criticized
at the beginning of this work find a new value and significance.
If, as we have suggested, exogamy and the prohibition of incest
have a permanent functional value, co-extensive with all social
groups, surely all the widely differing interpretations which have
been given for them must contain an atom of truth? Thus the
theories of McLennan, Spencer and Lubbock have, at least,
a symbolical meaning. It will be recalled that McLennan believed
that exogamy had its origin in tribes practising female infanticide,
and which were consequently obliged to seek wives for their
sons from outside. Similarly, Spencer suggested that exogamy
began among warrior tribes who carried off women from neigh¬
bouring groups. Lubbock proposed a primitive opposition
between two forms of marriage, viz., an endogamous marriage
in which the women were regarded as the communal property
of the men of the group, and an exogamous marriage, which
reckoned captured women as the private property of their captor,
thus giving rise to modern individual marriage. The concrete
detail may be disputed, but the fundamental idea is sound, viz.,
that exogamy has a value less negative than positive, that it
asserts the social existence of other people, and that it prohibits
endogamous marriage only in order to introduce, and to prescribe,
marriage with a group other than the biological family, certainly
not because a biological danger is attached to consanguineous
marriage, but because exogamous marriage results in a social
benefit.
Consequently, exogamy should be recognized as an important
element - doubtless by far the most important element - in that
solemn collection of manifestations which, continually or period¬
ically, ensures the integration of partial units within the total
group, and demands the collaboration of outside groups. Such
are the banquets, feasts and ceremonies of various kinds which
form the web of social life. But exogamy is not merely one mani¬
festation among many others. The feasts and ceremonies are
periodic, and for the most part have limited functions. The law
1. ‘The conjugal regard that united us as boys, and which we used to
express by calling ourselves “Activists” ...’ (Balzac, 1937, vol. 10, pp.
366, 382).

50 Incest and Sex

of exogamy, by contrast, is omnipresent, acting permanently
and continually; moreover, it applies to valuables - viz., women -
valuables par excellence from both the biological and the social
points of view, without which life is impossible, or, at best, is
reduced to the worst forms of abjection. It is no exaggeration,
then, to say that exogamy is the archetype of all other mani¬
festations based upon reciprocity, and that it provides the funda¬
mental and immutable rule ensuring the existence of the group
as a group. For example, among the Maori, Best tells us:
Female children of rank, as also male children of that status, were
given in marriage to persons of important, powerful tribes, possibly of a
quite unrelated people, as a means of procuring assistance from such
tribes in time of war. In this connection we can see the application of
the following saying of older times: ‘He taura taonga e motu, he taura
tangata e kore e motu> (‘A gift connection may be severed, but not so
a human link’). Two peoples may meet in friendship and exchange gifts
and yet quarrel and fight in later times, but intermarriage connects them
in a permanent manner (Best, 1929, p. 34).
And, further on, he quotes another proverb: ‘ He hono tangata e
kore e motu, kapa he taura waka, e motu\ 4A human joining is
inseverable, but not so a canoe-painter, which can be severed’
(p. 36). The philosophy contained in these remarks is the more
significant because the Maori were by no means insensible to
the advantages of marriage within the group. If both families
quarrelled and insulted each other, they said, this would not be
serious, but merely a family affair, and war would be avoided
(Best, 1924, vol. 1, p. 447).

The prohibition of incest is less a rule prohibiting marriage with

the mother, sister or daughter, than a rule obliging the mother,
sister or daughter to be given to others. It is the supreme rule of
the gift, and it is clearly this aspect, too often unrecognized,
which allows its nature to be understood. All the errors in inter¬
preting the prohibition of incest arise from a tendency to see
marriage as a discontinuous process which derives its own limits
and possibilities from within itself in each individual case.
Thus it is that the reasons why marriage with the mother,
daughter or sister can be prevented are sought in a quality in¬
trinsic to these women. One is therefore drawn infallibly towards

C. Levi-Strauss 51
biological considerations, since it is only from a biological,
certainly not a social, point of view that motherhood, sisterhood
or daughterhood are properties of the individuals considered.
However, from a social viewpoint, these terms cannot be regarded
as defining isolated individuals, but relationships between these
individuals and everyone else. Motherhood is not only a mother’s
relationship to her children, but her relationship to other mem¬
bers of the group, not as a mother, but as a sister, wife, cousin or
simply a stranger as far as kinship is concerned. It is the same for
all family relationships, which are defined not only by the in¬
dividuals they involve, but also by all those they exclude. This
is true to the extent that observers have often been struck by the
impossibility for natives of conceiving a neutral relationship, or
more exactly, no relationship. We have the feeling - which, more¬
over, is illusory - that the absence of definite kinship gives rise
to such a state in our consciousness. But the supposition that this
might be the case in primitive thought does not stand up to
examination. Every family relationship defines a certain group
of rights and duties, while the lack of family relationship does
not define anything; it defines enmity:
If you wish to live among the Nuer you must do so on their terms,
which means that you must treat them as a kind of kinsman and they
will then treat you as a kind of kinsman. Rights, privileges and obli¬
gations are determined by kinship. Either a man is a kinsman, actually or
by fiction, or he is a person to whom you have no reciprocal obligations
and whom you treat as a potential enemy (Evans-Pritchard, 1940,
p. 183).
The Australian aboriginal group is defined in exactly the same
terms:
When a stranger comes to a camp that he has never visited before, he
does not enter the camp, but remains at some distance. A few of the
older men, after a while, approach him, and the first thing they pro¬
ceed to do is to find out who the stranger is. The commonest question
that is put to him is ‘Who is your maeli (father’s father)?’ The dis¬
cussion proceeds on genealogical lines until all parties are satisfied of the
exact relation of the stranger to each of the natives present in the camp.
When this point is reached, the stranger can be admitted to the camp,
and the different men and women are pointed out to him and their
relation to him defined ... If I am a blackfellow and meet another
blackfellow that other must be either my relative or my enemy. If he

52 Incest and Sex

is my enemy I shall take the first opportunity of killing him, for fear
he will kill me. This, before the white man came, was the aboriginal view
of one’s duty towards one’s neighbour ... (Radcliffe-Brown, 1913,
p. 151).
Through their striking parallelism, these two examples merely
confirm a universal situation:
Throughout a considerable period, and in a large number of societies,
men met in a curious frame of mind, with exaggerated fear and an
equally exaggerated generosity which appear stupid in no one’s eyes
but our own. In all the societies which immediately preceded our own
and which still surround us, and even in many usages of popular
morality, there is no middle path. There is either complete trust or
complete mistrust. One lays down one’s arms, renounces magic, and
gives everything away, from casual hospitality to one’s daughter or
one’s property (Mauss, 1925, p. 138).

There is no barbarism or, properly speaking, even archaism

in this attitude. It merely represents the systematization, pushed
to the limit, of characteristics inherent in social relationships.
No relationship can be arbitrarily isolated from all other
relationships. It is likewise impossible to remain on this or that
side of the world of relationships. The social environment should
not be conceived of as an empty framework within which beings
and things can be linked, or simply juxtaposed. It is inseparable
from the things which people it. Together they constitute a
field of gravitation in which the weights and distances form a
coordinated whole, and in which a change in any element pro¬
duces a change in the total equilibrium of the system. We have
given a partial illustration at least of this principle in our analysis
of cross-cousin marriage. However, it can be seen how its field
of application must be extended to all the rules of kinship, and
above all, to that universal and fundamental rule, the prohibition
of incest. Every kinship system (and no human society is without
one) has a total character, and it is because of this that the mother,
sister and daughter are perpetually coupled, as it were, with
elements of the system which, in relation to them, are neither
son, nor brother, nor father, because the latter are themselves
coupled with other women, or other classes of women, or
feminine elements defined by a relationship of a different order.
Because marriage is exchange, because marriage is the archetype

C. Levi-Strauss 53
of exchange, the analysis of exchange can help in the under¬
standing of the solidarity which unites the gift and the counter¬
gift, and one marriage with other marriages.
It is true that Seligman disputes that the woman is the sole or
predominant instrument of the alliance. She cites the institution
of blood brotherhood, as expressed by the henamo relationship
among the natives of New Guinea (1935, pp. 75-93). The
establishment of blood brotherhood does indeed create a bond of
alliance between individuals, but by making them brothers it
entails a prohibition on marriage with the sister. It is far from our
mind to claim that the exchange or gift of women is the only way
to establish an alliance in primitive societies. We have shown
elsewhere how, among certain native groups of Brazil, the com¬
munity could be expressed by the terms for ‘brother-in-law’
and ‘brother’. The brother-in-law is ally, collaborator and
friend; it is the term given to adult males belonging to the band
with which an alliance has been contracted. In the same band, the
potential brother-in-law, i.e. the cross-cousin, is the one with
whom, as an adolescent, one indulges in homosexual activities
which will always leave their mark in the mutually affectionate
behaviour of the adults (Levi-Strauss, 1948a). However, as well
as the brother-in-law relationship, the Nambikwara also rely
on the notion of brotherhood: ‘Savage, you are no longer my
brother! * is the cry uttered during a quarrel with a non-kinsman.
Furthermore, objects found in a series, such as hut posts, the
pipes of a Pan-pipe, etc. are said to be ‘brothers’, or are called
‘others’, in their respective relationships, a terminological
detail which is worth comparing with Montaigne’s observation
that the Brazilian Indians whom he met at Rouen called men the
‘halves’ of one another, just as we say ‘our fellow men’ (1962,
vol. 1, ch. 31). However, the whole difference between the two
types of bond can also be seen, a sufficiently clear definition being
that one of them expresses a mechanical solidarity (brother),
while the other involves an organic solidarity (brother-in-law,
or god-father). Brothers are closely related to one another, but
they are so in terms of their similarity, as are the post or the reeds
of the Pan-pipe. By contrast, brothers-in-law are solidary be¬
cause they complement each other and have a functional efficacy
for one another, whether they play the role of the opposite sex

54 Incest and Sex

 in the erotic games of childhood, or whether their masculine
alliance as adults is confirmed by each providing the other with
what he does not have - a wife - through their simultaneous
renunciation of what they both do have - a sister. The first form
of solidarity adds nothing and unites nothing; it is based upon a
cultural limit, satisfied by the reproduction of a type of connection
the model for which is provided by nature. The other brings
about an integration of the group on a new plane.
Linton’s observation on blood-brotherhood in the Marquesas
helps to place the two institutions (blood-brotherhood and inter¬
marriage) in their reciprocal perspectives. Blood-brothers are
called enoa: ‘When one was enoa with a man, one had equal
rights to his property and stood in the same relation to his
relatives as he did’ (Linton, 1945, p. 149). However, it emerges
very clearly from the context that the enoa system is merely an
individual solution acting as a substitute, while the real and
effective solution of the relations between the groups, i.e. the
collective and organic solution of intermarriages, with the conse¬
quent fusion of the tribes, is made impossible by the international
situation. Although vendettas may be in progress, the institution
of enoa, a purely individual affair, is able to ensure a minimum
of liaison and collaboration, even when marriage, which is a
group affair, cannot be contracted.
Native theory confirm's our conception even more directly.
Mead’s Arapesh informants had difficulty at first in answering
her questions on possible infringements of the marriage pro¬
hibitions. However, when they eventually did express a comment
the source of the misunderstanding was clearly revealed: they
do not conceive of the prohibition as such, i.e. in its negative
aspect; the prohibition is merely the reverse or counterpart of a
positive obligation, which alone is present and active in the
consciousness. Does a man ever sleep with his sister? The
question is absurd. Certainly not, they reply: ‘No, we don’t
sleep with our sisters. We give our sisters to other men, and other
men give us their sisters’ (1935, p. 84). The ethnographer pressed
the point, asking what they would think or say if, through some
impossibility, this eventuality managed to occur. Informants
had difficulty placing themselves in this situation, for it was
scarcely conceivable: ‘What, you would like to marry your sister!

C. Levi-Strauss 55
What is the matter with you anyway? Don’t you want a brother-
in-law? Don’t you realize that if you marry another man’s sister
and another man marries your sister, you will have at least two
brothers-in-law, while if you marry your own sister you will have
none? With whom will you hunt, with whom will you garden,
whom will you go to visit?’ (1935, p. 84).
Doubtless, this is all a little suspect, because it was provoked,
but the native aphorisms collected by Mead, and quoted as the
motto to the first part of this work, were not provoked, and their
meaning is the same. Other evidence corroborates the same thesis.
For the Chukchee, a ‘bad family’ is defined as an isolated family,
‘brotherless and cousinless’ (Bogoras, 1904-9, p. 542). Moreover,
the necessity to provoke the comment (the content of which, in
any case, is spontaneous), and the difficulty in obtaining it,
reveal the misunderstanding inherent in the problem of marriage
prohibitions. The latter are prohibitions only secondarily and
derivatively. Rather than a prohibition on a certain category of
persons, they are a prescription directed towards another cate¬
gory. In this regard, how much more penetrating is native theory
than are so many modern commentaries! There is nothing in the
sister, mother or daughter which disqualifies them as such.
Incest is socially absurd before it is morally culpable. The
incredulous exclamation from the informant: ‘So you do not
want to have a brother-in-law?’ provides the veritable golden
rule for the state of society.

There is thus no possible solution to the problem of incest with¬

in the biological family, even supposing this family to be already
in a cultural context which imposes its specific demands upon it.
The cultural context does not consist of a collection of abstract
conditions. It results from a very simple fact which expresses it
entirely, namely, that the biological family is no longer alone,
and that it must ally itself with other families in order to endure.
Malinowski supported a different idea, namely, that the pro¬
hibition of incest results from an internal contradiction, within
the biological family, between mutually incompatible feelings
such as the emotions attached to sexual relationships and
parental love, or ‘the sentiments which form naturally between
brothers and sisters’ (1934, p. lxvi). These sentiments neverthe-

56 Incest and Sex

less only become incompatible because of the cultural role which
the biological family is called upon to play. The man should
teach his children, and this social vocation, practised naturally
within the family group, is irremediably compromised if emotions
of another type develop and upset the discipline indispensable to
the maintenance of a stable order between the generations:
‘Incest would mean the upsetting of age distinctions, the mixing
up of generations, the disorganization of sentiments and a
violent exchange of roles at a time when the family is the most
important educational medium. No society could exist under
such conditions’ (Malinowski, 1927, p. 251).
It is unfortunate for this thesis that there is practically no
primitive society which does not flagrantly contradict it on every
point. The primitive family fulfils its educative function sooner
than ours, and from puberty onwards - and often even before -
it transfers the charge of adolescents to the group, with the
handing over of their preparation to bachelor houses or initiation
groups. Initiation rituals confirm this emancipation of the young
man or girl from the family cell and their definitive incorporation
within the social group. To achieve this end, these rituals rely on
precisely the processes which Malinowski cites as a possibility
solely in order to expose their mortal dangers, viz., affective
disorganization and the violent exchange of roles, sometimes
going as far as the practice, on the initiate’s very person, of most
unfamilial usages by near relatives. Finally, different types of
classificatory system are very little concerned to maintain a
clear distinction between ages and generations. However, it is
just as difficult for a Hopi child to learn to call an old man ‘my
son ’, or any other assimilation of the same order, as it would be
for one of ours (Simmons, 1942, p. 68). The supposedly disastrous
situation that Malinowski depicts in order to justify the prohibi¬
tion of incest, is, on the whole, no more than a very banal picture
of any society, envisaged from another point of view than its
own.
This naive egocentrism is so far from being new or original that
Durkheim made a decisive criticism of it years before Malin¬
owski gave it a temporary revival in popularity. Incestuous
relationships only appear contradictory to family sentiments
because we have conceived of the latter as irreducibly excluding

C. Levi-Strauss 57
the former. But if a long and ancient tradition allowed men to
marry their near relatives, our conception of marriage would be
quite different. Sexual life would not have become what it is. It
would have a less personal character, and would leave less room
for the free play of the imagination, dreams and the spontaneities
of desire. Sexual feeling would be tempered and deadened, but
by this very fact it would compare closely with domestic feelings,
with which it would have no difficulty in being reconciled. To
conclude this paraphrase with a quotation: ‘Certainly, the
question does not pose itself once it is assumed that incest is
prohibited; for the conjugal order, being henceforth outside the
domestic order, must necessarily develop in a divergent direction.
This prohibition clearly cannot be explained in terms of ideas
which obviously derive from it’ (Durkheim, 1898, p. 63).
Must we not go even further ? On the very occasion of marriage,
numerous societies practise the confusion of generations, the
mingling of ages, the reversal of roles, and the identification of
what we regard as incompatible relationships. As these customs
seem to such societies to be in perfect harmony with a prohibition
of incest, sometimes conceived of very rigorously, it can be
concluded, on the one hand, that none of these practices is
exclusive of family life, and, on the other hand, that the pro¬
hibition must be defined by different characteristics, common to
it throughout its multiple modalities. Among the Chukchee, for
example:
the age of women thus exchanged is hardly considered at all. For in¬
stance, on the Oloi River, a man named Ql’mlqai married his young
son five years old to a girl of twenty. In exchange he gave his niece,
who was twelve years of age, and she was married to a young man more
than twenty years old. The wife of the boy acted as his nurse, fed him
with her own hands and put him to sleep (Bogoras, 1904-9, p. 578).
The writer also cites the case of a woman who, married to a two-
year-old baby and having a child by ‘a marriage companion’, i.e.
an official and temporary lover, shared her attentions between
the two babies: ‘When she was nursing her own child, she also
nursed her infant husband. ... In this case the husband also
readily took the breast of his wife. When I asked for the reason of
the wife’s conduct, the Chukchee replied, “Who knows? Perhaps
it is a kind of incantation to insure the love of her young husband

58 Incest and Sex

in the future’” (Bogoras, 1904-9, p, 578). At all events, it is
certain that these apparently inconceivable unions are compatible
with a highly romantic folklore, full of devouring passions.
Prince Charmings and Sleeping Beauties, shy heroines and
triumphant loves (Bogoras, 1904-9, pp. 578-83). We know of
similar facts in South America (Means, 1931, p. 360).
However unusual these examples may appear, they are not
unique, and Egyptian-style incest probably represents only the
limit. They have their parallel among the Arapesh in New Guinea,
among whom infant betrothals are frequent, the two children
growing up as brother and sister. But this time the age difference
is on the side of the husband :
An Arapesh boy grows his wife. As a father’s claim to his child is not
that he has begotten it but rather that he has fed it, so also a man’s
claim to his wife’s attention and devotion is not that he has paid a
bride-price for her, or that she is legally his property, but that he has
actually contributed the food which has become flesh and bone of
her body (Mead, 1935, p. 80).
Here again, this type of apparently abnormal relationship
provides the psychological model for regular marriage: ‘The
whole organization of society is based upon the analogy between
children and wives as representing a group who are younger,
less responsible, than the men, and therefore to be guided.
Wives by definition stand in this child-relationship ... to all of
the older men of the clan into which they marry’ (Mead, 1935,
pp. 80-81).
Likewise, among the Tapirape of central Brazil, depopulation
has brought about a system of marriage with young girls. The
‘husband’ lives with his parents-in-law and the ‘wife’s’ mother is
responsible for woman’s work (Wagley, 1940, p. 12). The Mohave
husband carries the little girl that he has married on his shoulders,
busies himself with household duties, and generally speaking
acts both as husband and in loco parentis. The Mohave comment
upon the situation cynically, and ask, sometimes even when the
person concerned is present, whether he has married his own
daughter: ‘“Whom are you carrying around on your back? Is
that your daughter?” they ask him. When such marriages break
up, the husband often has a manic attack’ (Devereux, 1939,
p. 519).

C. Levi-Strauss 59
 I myself have been present, among the Tupi-Cawahib of the
upper Madeira, in central Brazil, at the betrothal of a man about
thirty years old with a scarcely two-year-old baby, still in its
mother’s arms. Nothing was more touching than the excitement
with which the future husband followed the childish frolics of his
fiancee. He did not tire of admiring her, and of sharing his
feelings with the onlookers. For some years his thoughts would
be filled with the prospect of setting up house. He would feel
strengthened by the certainty, growing alongside him in strength
and beauty, of one day escaping the curse of bachelorhood.
Henceforth, his budding tenderness is expressed in innocent
gifts. According to our standards, this love is torn between three
irreducible categories, viz., paternal, fraternal and marital,
but in an appropriate context it reveals no element of disquiet
or defect, endangering the future welfare of the couple, let alone
the whole social order.
We must decide against Malinowski and those of his followers
who vainly attempt to support an outmoded position (Seligman,
1931-2, pp. 250-76), in favour of those, like Fortune and
Williams, who, following Tylor, found the origin of the incest
prohibition in its positive implications (Fortune, 1932, pp.
620-2; Williams, 1936, p. 169; Tylor, 1889). As one observer
rightly puts it: ‘An incestuous couple as well as a stingy family
automatically detaches itself from the give-and-take pattern of
tribal existence; it is a foreign body - or at least an inactive one
- in the body social’ (Devereux, 1939, p. 529).
No marriage can thus be isolated from all the other marriages,
past or future, which have occurred or which will occur within
the group. Each marriage is the end of a movement which, as
soon as this point has been reached, should be reversed and
develop in a new direction. If the movement ceases, the whole
system of reciprocity will be disturbed. Since marriage is the
condition upon which reciprocity is realized, it follows that
marriage constantly ventures the existence of reciprocity. What
would happen if a wife were received without a daughter or a
sister being given ? This risk must be taken, however, if society
is to survive. To safeguard the social perpetuity of alliance, one
must compromise oneself with the chances of descent (i.e. in
short, with man’s biological substructure). However, the social

60 Incest and Sex

recognition of marriage (i.e. the transformation of the sexual
encounter, with its basis in promiscuity, into a contract, cere¬
mony or sacrament) is always an anxious venture, and we can
understand how it is that society should have attempted to
provide against the risks involved by the continual and almost
maniacal imposition of its mark. The Hehe, says Brown, practise
cross-cousin marriage, but not without hesitation, for if cross¬
cousin marriage allows the clan-line to be maintained, it risks
it in the case of a bad marriage, and informants report: ‘Thus
some forbid their children to marry a cousin’ (1934, p. 28).
The ambivalent attitude of the Hehe towards a special form of
marriage is the pre-eminent social attitude towards marriage in
any of its forms. By recognizing and sanctioning the union of the
sexes and reproduction, society influences the natural order, but
at the same time it gives the natural order its chance, and one
might say of any culture of the world what an observer has noted
of one of them: ‘Perhaps the most fundamental religious con¬
ception relates to the difference between the sexes. Each sex is
perfectly all right in its own way, but contact is fraught with
danger for both’ (Hogbin, 1935, p. 330).
Marriage is thus a dramatic encounter between nature and
culture, between alliance and kinship. ‘Who has given the bride?’
chants the Hindu hymn of marriage: ‘ To whom then is she given ?
It is love that has given her; it is to love that she has been given.
Love has given; love has received. Love has filled the ocean.
With love I accept her. Love! let her be yours’ (Banerjee, 1896,
p. 91).2 Thus, marriage is an arbitration between two loves,
parental and conjugal. Nevertheless, they are both forms of love,
and the instant the marriage takes place, considered in isolation,
the two meet and merge; ‘love has filled the ocean’. Their
meeting is doubtless merely a prelude to their substitution for one
another, the performance of a sort of chasse-croise. But to inter¬
cross they must at least momentarily be joined, and it is this
2. As to marriage considered as bordering upon incest, compare the
following, written in a completely different spirit: ‘Profound sentiment
[between husband and wife] would have seemed odd and even “ridiculous”,
in any event unbecoming; it would have been as unacceptable as an earnest
“aside” in the general current of light conversation. Each has a duty to all,
and for a couple to entertain each other is isolation; in company there
exists no right of the tete-a-tete’ (Taine, 1876, p. 133).

C. Levi-Strauss 61
which in all social thought makes marriage a sacred mystery.
At this moment, all marriage verges on incest. More than that,
it is incest, at least social incest, if it is true that incest, in the
broadest sense of the word, consists in obtaining by oneself, and
for oneself, instead of by another, and for another.
However, since one must yield to nature in order that the
species may perpetuate itself, and concomitantly for social
alliance to endure, the very least one must do is to deny it while
yielding to it, and to accompany the gesture made towards it
with one restricting it. This compromise between nature and
culture comes about in two ways, since there are two cases, one in
which nature must be introduced, since society can do every¬
thing, the other in which nature must be excluded, since it rules
from the first - before descent and its assertion of the unilineal
principle, and before alliance, with its establishment of pro¬
hibited degrees.

References
de Balzac, H. (1937), Louis Lambert, Oeuvres Completes, vol. 10, Paris.
Banerjee, G. N. (1896), The Hindu Law of Marriage and Stridhana,
Calcutta.
Best, E. (1924), ‘The Maori’, Mem. Polynesian Soc. no 5.
Best, E. (1929), ‘The Whare Kohanga (the “Nest House”) and its
lore’, Dominion Mus. Bull., no. 13, pp. 1-72.
Bog or as, W. (1904-9), ‘The Chukchee’, Mem. Amer. Mus. nat. Hist.,
no. 11, pp. 1-733.
Brown, G. G. (1934), ‘Hehe cross-cousin marriage’, in E. E. Evans-
Pritchard, et al. (eds.), Essays Presented to C. G. Seligman, Routledge &
Kegan Paul.
Devereux, G. (1939), ‘The social and cultural implications of incest
among the Mohave Indians’, Psychoanal. Q., vol. 8, pp. 510-33.
Durkheim, E. (1898), ‘La prohibition de l’inceste et ses origines’, Ann.
sociol., vol. 1, pp. 1-70.
Evans-Pritchard, E. E. (1940), The Nuer, Clarendon Press.
Fortune, R. F. (1932), Sorcerers of Dobu, Routledge & Kegan Paul.
Hogbin, H. I. (1935), ‘Native culture in Wogeo: report of field work in
New Guinea’, Oceania, vol. 5, pp. 308-37.
Leenhardt, M. (1930), ‘Notes d’ethnologie n6o-cal£donniene’,
Travaux Mem. Inst. Ethnol., vol. 8.
L£vi-Strauss, C. (1948), ‘La vie familiale et sociale des Indiens
Nambikwara’, J. Soc. Amer. vol. 37.
Linton, R. (1945), ‘Marquesan culture’, in A. Kardiner (ed.) The
Individual and his Society, Columbia University Press.
Malinowski, B. (1927), Sex and Repression in Savage Society,
Routledge & Kegan Paul.

62 Incest and Sex

Malinowski, B. (1934), ‘Introduction’, in H. I. Hogbin, Law and
Order in Polynesia, Allen & Unwin.
Mauss, M. (1925), ‘Essai sur le don’, Ann. sociol., vol. 1, pp. 30—186.
Mead, M. (1935), Sex and Temperament in Three Primitive Societies,
Apollo.
Means, P. A. (1931), Ancient Civilizations of the Andes, Gordion.
de Montaigne, M. (1962), Essais, Gamier Freres. First published 1580.
Radcliffe-Bro wn, A. R. (1913), ‘Three tribes of western Australia’,
J. roy. anthropol. Inst., vol. 43, pp. 143-70.
Seligman, B. Z. (1931-2), ‘The incest barrier: its role in social
organisation’, Br. J. Psychol., vol. 22, pp. 250-76.
Seligman, B. Z. (1935), ‘The incest taboo as a social regulation’,
Sociol. Rev., vol. 27, pp. 75-93.
Simmons, L. W. (ed.) (1942), Sun Chief, New Haven.
Taine, H. A. (1876), Les Origines de la France Contemporaine, Hachette.
Tylor, E. B. (1889), ‘On a method of investigating the development of
institutions: applied to laws of marriage and descent’, J. anthropol.
Inst., vol. 18, pp. 245-72.
Wagley, C. (1940), ‘The effects of depopulation upon social
organisation as illustrated by the Tapirape Indians’, Trans. New York
Acad. Sci., vol. 3, pp, 12-16.
Williams, F. E. (1936), Papuans of the Trans-Fly, Oxford University
Press.

C. Levi-Strauss 63
5 J. Goody
Incest and Adultery

Excerpts from J. Goody, ‘A comparative approach to incest and

adultery’, British Journal of Sociology, vol. 7, 1956, pp. 286-305.
Reprinted in J. Goody, Comparative Studies in Kinship, Routledge
& Kegan Paul, 1969, pp. 13-38.

... The continuous analysis in depth of different societies calls for

more precise conceptual discriminations than were previously
required. The terms often employed by social scientists are those
which we use as members of a particular society to refer to our
own institutions. Such concepts may turn out to be quite in¬
appropriate for the purpose of cross-cultural analysis. The
English ‘family’ is an obvious case in point. From the sociologi¬
cal point of view, the term has at least four analytically separable
meanings. A statement of the kind, ‘the family is a universal
institution among all human societies’, is meaningless without
further elaboration.
A refinement of concepts is a product of onward-going
research; it proceeds hand in hand with it. The depth analysis of
societies through long periods of residence by trained observers
is a necessary concomitant of the sharpening of concepts for
cross-cultural studies.
The particular concept in which I am interested here is that of
‘incest’. I want also to mention the related ones of adultery and
fornication, as I shall later be concerned with them as categories
of heterosexual offence. The everyday meanings given by the
Concise Oxford Dictionary are as follows:

Incest Sexual commerce of near kindred.

Adultery Voluntary sexual intercourse of married person with
one of the opposite sex, married (double adultery) or not
(single adultery).
Fornication Voluntary sexual intercourse between man (some¬
times restricted to unmarried man) and unmarried woman.

64 Incest and Sex

 These particular definitions are by no means standardized. For
instance, the Encyclopaedia Britannica (Eleventh Edition) and
Webster's Dictionary both define incest as ‘sexual intercourse
between persons so related by marriage or affinity that legal
marriage cannot take place between them’, a formula which
assumes an identical range in prohibitions on heterosexual
intercourse and prohibitions on marriage.
It is these everyday usages which have formed the basis of the
anthropological concepts. Malinowski, for example, appeared to
treat the incest taboo (the prohibition on sexual intercourse) and
exogamy (the prohibition on marriage) as being but two sides of a
single coin.
Murdock, on the other hand, adheres more closely to the
Concise Oxford Dictionary when he defines incest and adultery:

When it (heterosexual intercourse) takes place outside of marriage

between two persons of whom at least one is married to another person,
it is called adultery. If its participants are related to one another by a
real, assumed, or artificial bond of kinship which is culturally regarded
as a bar to sex relations, it is classed as incest (1949, p. 261).

Radcliffe-Brown, while retaining the criterion of kinship,

offers a more restricted definition of incest. He writes, ‘Incest is
properly speaking the sin or crime of sexual intimacy between
immediate relatives within the family, father and daughter,
mother and son, brother and sister’ (1950, p. 69).
Such extensive controversies have raged around the ‘incest
taboo’ that it may perhaps appear impertinent to raise the
question as to whether all these writers are in fact discussing
the same range of phenomena or looking for explanations of
the same set of prohibitions. But when we put the definitions
of Murdock and Radcliffe-Brown side by side it is obvious that
such doubts are not altogether misplaced. It is clear for instance
that in terms of Radcliffe-Brown’s definition, Murdock’s ‘second
factual conclusion ... that incest taboos do not apply universally
to any relative of opposite sex outside of the nuclear family’
(1949, p. 285) is tautologous. Equally, on the basis of Murdock’s
formula, it is difficult to decide whether sexual intimacy with the
father’s wives other than one’s own mother would constitute
incest or adultery, particularly in societies like the Tiv or Bedouin

T-K-C J. Goody 65
where kinship is universal. The difference between the two
definitions is this, that though both apparently see the regulations
as ‘grounded in the constitution of the nuclear family’ (1949,
p. 284), Radcliffe-Brown attempts to limit the application of the
term to the elementary family itself, while Murdock prefers to
include all kin-based prohibitions, seeing these as ‘extensions’
of the primary taboo. Murdock’s emphasis is in line with Malin¬
owski’s stress upon the elementary family and with his dogma
of ‘extension of sentiments’. Both definitions are clearly based
upon the institutions of our own society, where prohibitions on
intercourse, like prohibitions on marriage, are bilaterally organ¬
ized within limited ranges of kin. But are these necessarily ade¬
quate for the analysis of non-European societies?
In order to answer this question, let us examine the evidence
from two societies characterized by unilineal descent, one by
matrilineal, the other patrilineal descent. I have selected for this
purpose the Ashanti and Tallensi of Ghana, for which the
main sources on incest are Rattray (1929) and Fortes (1936,
1949) respectively. These societies were chosen partly because of
the high standard of the available reports and partly because of
my own familiarity with the area. The Trobriand and the Nuer
material will be used as a check upon the results obtained from
an analysis of the examples from Ghana. In each case I want to
examine both the explicit verbal categories of the actors them¬
selves and the classifications implicit in the system of sanctions.
These will be compared with the concepts employed by the
observers.

The matrilineal case

In his treatment of sexual offences among the Ashanti, Rattray
distinguished what he calls sins or tribal offences (Oman akyiwadie)
from household offences (efiesem). The former demanded the
intervention of the central authority and the execution of the
guilty party, although in some instances compensation was
allowed. The latter ‘were settled by the persons directly concerned
or were decided by argument before any Elder, without reference
to the “house-father”, who stood entirely aloof’ (1929, p. 287).
The offences falling under these two categories were discussed
separately.

66 Incest and Sex

 Among the Ashanti sexual offences can be categorized in two
ways, firstly according to the different names used by the Ashanti
themselves, and, secondly, according to the different sanctions
employed. I shall consider first the classification according to
the nature of the sanctions.
This reveals three classes of offence. In the first class falls
mogyadie, intercourse with a woman of the same clan, punishable
by death; this includes intercourse with full siblings and maternal
half-siblings and with the mother; it excludes intercourse of
father with daughter. But there is another type of offence, which
though not given the same name, is also punishable by death;
this is atwebenefie, intercourse with a member of the same patri¬
lineal sub-group, which of course includes that between father
and daughter. Terminologically this constitutes a different
category, but in respect of the nature of the sanction it must be
associated with mogyadie. Both are cases of intercourse with
members of the same descent group. The terminological dis¬
tinction indicates that it is intercourse within the matriclan
which is the major prohibition here, while that within the patri¬
lineal sub-group is subsidiary. This is consistent with the nature
of double clanship among the Ashanti.
The second class of offence consists basically of intercourse not
with members of the same descent group but with the wives of
fellow members, as well as with other classificatory wives. It also
includes some prohibitions on intercourse with affines which
might tend to confuse the social position of the wife herself.
The punishment for this class of offence varies. It is never death,
but consists of some variant of the adultery payment.
The third class of sexual offence is with wives of other men, and
the sanction here is the simple adultery payment.
The threefold typology on the basis of sanctions is an indication
of the weight placed by the society on these various offences.
The first class brings together offences relating to the structure
of descent groups, both matrilineal and patrilineal, offences
relating to the hierarchical organization and ‘ritual’ offences
relating to the cult of the Earth and to the fertility of women.
When we look at the terms used by the Ashanti themselves,
we find there is another threefold typology, if we exclude the
category baratwe, which represents a different method of classi-

J. Goody 67
fying these offences. Intercourse within the matriclan is sharply
differentiated terminologically from intercourse within the
patrician, the latter falling into the same category as intercourse
with the wives of members of the matriclan and of other social
groups. In this way it is assimilated to what I wish to call group-
wife offences to distinguish them from intra-group offences.
This is clearly related to the overwhelmingly greater importance
of the matriclans in the social system. The third category is
residual in that it consists essentially in sexual intercourse with
people other than the members or wives of members of the
descent groups, and of a few other quasi-kin groups such as
guilds and military companies. Thus the concepts of the Ashanti
themselves concerning heterosexual offences closely reflect the
system of social groups. Intercourse with a daughter falls into a
different category from intercourse with a sister, although for us
both would be classified as ‘ incest ’.
Now let us turn to Rattray’s own use of the terms ‘incest’
and ‘adultery’ to see how he meets this situation. ‘Incest’ he
uses simply to translate mogyadie, ‘eating of one’s own blood’,
that is, sexual intercourse with a matriclanswoman. He applies
the term to none of the other offences, even those also punishable
by death. The term ‘adultery’ he uses to translate all the house¬
hold offences, ‘eating a man’s wife’ (di obi yere) and atwebenefie
- ‘a vagina near to the dwelling-house’. He also uses it to trans¬
late those offences called di obi yere which fall under tribal
jurisdiction and are therefore punishable by death. This consists
of two offences only, intercourse with a chief’s wife, and the
worst type of sexual sin against the Earth, the rape of a married
woman in the bush. His difficulty arises with atwebenefie. In
his original list of offences, he translates this neutrally as ‘sexual
intercourse with certain individuals other than those related by
“blood”’ (p. 304) - i.e. females of the same matriclan. On the
following page he writes:
Atwe-bene-fie means literally (having sexual intercourse with) ‘a
vagina that is near to the dwelling-house’, and the offence, as the title
implies, consisted in committing adultery with the wives of certain
persons with whom the existing menage necessarily compelled close
social intercourse or constant physical proximity . . . (p. 305).
The term ‘adultery’ has now replaced the neutral circumlocution

68 Incest and Sex

previously used. However, when we examine the list of atwebenefie
offences we find that those included under ‘tribal sins’ are
not defined by the affinal relationship to ego. The women are
forbidden not because they are someone’s wives but because
they are female members of the same patrilineal sub-group. For
such an offence adultery seems a misleading translation.
The point at issue is this. In English usage the term ‘adultery’
is defined in relation to the marital status of one or both partici¬
pants and is in effect residual to the category ‘incest’. The term
‘incest’ is defined bilaterally, in keeping with other aspects of the
social system. Heterosexual offences among the Ashanti do not
fall into these categories, and in trying to translate these simply
by the English words ‘incest’ and ‘adultery’, Rattray was faced
with an impossible task. The English concept ‘incest’ refers to
heterosexual intercourse with persons within a particular range
of kin, whether they fall within that range by birth or by marriage.
When a male ego marries, the immediate female kin of his wife
are assimilated to his own kinship chart by becoming sisters- or
mothers-in-law. Intercourse with affines is defined as incestuous,
and placed in the same conceptual category as intercourse with
consanguineous kin.
Thus, whereas the Ashanti differentiate between intra-group
offences and group-wife offences, the European system does not
have to do this because at marriage the spouses are assimilated,
for many social purposes, into each other’s natal groups. There
is no distinction, in the context of heterosexual offences, between
group-member and group-spouse.
This interpretation is strikingly confirmed in another matri-
lineal case, that of the Trobriands. Malinowski discusses incest
in considerable detail in his book The Sexual Life of Savages
(1929). First let us ask what Malinowski means by incest. ‘Incest
within the family and breach of exogamy’, he says, is the meaning
of the Trobriand word suvasova (p. 389). As the family is bilateral,
the term suvasova should therefore cover the intercourse of a
man with his mother, his sister, or his daughter.
When we look at the Trobriand concepts themselves we find
that this is not the meaning of the word suvasova. Malinowski
himself makes this apparent in another context, although he
continues to assume an equivalence.

J. Goody 69
It must be clearly understood that, although father to daughter incest
is regarded as bad, it is not described by the word suvasova (clan
exogamy or incest), nor does any disease follow upon it; and, as we
know, the whole ideology underlying this taboo is different from that of
suvasova (p. 447).
Suvasova corresponds precisely to the Ashanti concept
mogyadie. It is the name for what I have called intra-group
offences (intercourse or marriage), and has to be distinguished
from intercourse with wives of members of the matriclan, such
as brother’s wife, which to judge from the example Malinowski
gives (p. 98) is not heavily sanctioned. The category suvasova
includes intercourse with the mother, the daughter and the
sister, the last being considered the most heinous, possibly
because this was felt to be the most likely. The worst heterosexual
offences in the Trobriands, as among the Ashanti, in each case
distinguished terminologically, are those committed with
members of the same matriclan. Malinowski repeatedly insists
that it is the brother-sister prohibition which is the basis of the
‘incest’ taboo in Trobriand society.

The patrilineal case

Let us now consider a patrilineal case, the Tallensi. According to
Fortes (1949) the Tallensi have no word for incest. There is a
term poyamboon which might be translated literally ‘matters
concerning women’. Fortes himself translates this as ‘adultery’,
but on the basis of my own experience among the LoDagaa I
would suggest that it covers a wider range of heterosexual
offences than is usually indicated by this term.
If the Tallensi have no specific word for incest, what range of
phenomena does Fortes include under this term and how does he
differentiate this from other types of offence? Looking at his
analysis, we find that incest consists in sexual relations within the
‘expanded family’, that is, the family group based upon the inner
lineage (1949, p. 111). Thus, in the absence of an indigenous
concept, Fortes has introduced what is essentially a bilateral
classification, one that includes in the same category offences
with a paternal aunt, a sister or a daughter (intra-group offences)
as well as offences with the wife of a father, brother or a son
(group-wife offences). But though he calls both of these offences

70 Incest and Sex

‘incest’ he emphasizes that they are differently thought of by the
Tallensi. For the first category of offence is merely ‘disreputable’,
while the latter is viewed with the horror usually taken as being
characteristic of incest. Outside the inner (or medial) lineage this
dichotomy becomes even more obvious, for a lover relationship
with a female lineage member is in fact permitted, while inter¬
course with the wife of a lineage or even a clan member is still
considered a wrong. Fortes claims that this latter offence is not
incest, but ‘the most reprehensible form of adultery. It does not
bear the same moral stigma as the corresponding form of incest,
nor does it carry religious penalties for the adulterer’ (1949,
p. 116). For Fortes, therefore, incest consists in sexual intercourse
with female members of the inner lineage and with the wives of
its male members, while adultery consists in intercourse with the
wives of male members outside that range as well as with wives
of non-clansmen.
There is then no Tallensi term for heterosexual offences other
than one for ‘matters concerning women’. Fortes uses the English
terms ‘incest’ and ‘adultery’ to divide up this category. The way
in which he does so is bilaterally oriented. ‘Incest’ is the offence
of sexual intercourse within the ‘expanded family’, ‘adultery’ the
offence of sexual intercourse with any married woman outside it.
An alternative method of treating this problem is to infer the
implicit classification of offences among the Tallensi from the
nature of their reaction to any breach. This in effect is what
Fortes does when he insists that ‘incest’ with a sister or daughter
falls in a different category of sexual acts from ‘incest’ with a
wife of the lineage (1949, p. 114). This standardized procedure
for the investigation of moral, ritual or legal norms gives the
following threefold division:

1. Sexual intercourse with a member of the same patrician (up

to the inner lineage only).
2. Sexual intercourse with the wife of a member of the same
patrician.
3. Sexual intercourse with the wife of a non-clansman.

I suggest that this classification has more inherent probability

for three reasons. Firstly, it appears to fit better with the Tallensi
emphasis on unilineal descent. Secondly, it corresponds to the

J. Goody 71
classification I found among the LoDagaa of the same general
area who are culturally similar to the Tallensi in very many ways.
Thirdly, it is analogous to the classification which we have found
among the Ashanti. Thus in both the matrilineal and patrilineal
cases prohibitions on sexual intercourse are grouped together,
depending upon whether they were:

1. With a member of the same descent group (intra-group sexual

prohibition).
2. With the wife of a member (group-wife prohibition).
3. With another married woman (extra-group prohibition).

I suggest that a similar typology will be found in most societies

characterized by unilineal descent, but has been obscured in
anthropological reports because of the ethnocentric bias of the
observers towards bilateral classifications. It is only possible to
rectify this in the case of the Tallensi and Ashanti because of the
excellence of the reporting and the fact that the authors have
provided us with the terms used by the actors themselves. If we
accept these three basic categories for heterosexual prohibitions
and offences in societies characterized by unilineal descent
groups, it would be reasonable to refer to the last as adultery, or
more specifically non-group adultery. But what about the other
two types of offence? Which of these should be called ‘incest’?

The classification of heterosexual offences

The whole lengthy discussion of incest has turned on the sup¬
position that it is a type of illicit sexual intercourse which is
characterized by a particular horror. In the Western European
system it is true that the entire range of offences included under
the category incest is so regarded. But in many other societies,
this is not so. Even within the minimal domestic units, hetero¬
sexual offences may be differently classified both terminologically
and with regard to the organized sanctions with which they are
met. Furthermore, they are also distinguished by diffuse sanctions,
by the reactions which they arouse in the other members of the
community. Among the Tallensi, offences between brother and
sister (intra-group offences) are merely ‘disreputable’, while
group-wife offences are met with ‘horror’. On the other hand,
and this is a point of fundamental theoretical interest, among the

72 Incest and Sex

 Ashanti the reverse is the case. It is the intra-group offences
which are dealt with by death, while the group-wife offences are
treated as a heightened form of extra-clan adultery. I would
claim that it is a mistake in either of these societies to class both
these types of offence together as ‘incest’, because they are
treated in such markedly different ways in terms of the sanctions
employed, and, among the Ashanti, in terms of the actor cate¬
gories themselves. Equally it would be difficult to classify either
the first or the second types as incest on the basis of the internal
reaction to them, as this varies so markedly in the two societies.
I suggest that the word incest be retained for the category of
offences inside the group and that it be divorced from the criterion
of ‘horror’. The group-spouse category should be associated
with adultery rather than incest, for at the core of the prohibition
lies the fact that the woman is married into the group; the taboo
depends upon her married status. If she were not married, inter¬
course with her would be neither incest nor adultery but rather
fornication, an act which may not be negatively sanctioned at
all. For the group-wife category I therefore suggest the somewhat
clumsy phrase, ‘group-wife adultery’. Let me now schematize
the threefold categorization of offences which we found among
the Ashanti and the Tallensi. The terminology I suggest seems to
me more appropriate for the cross-cultural analysis of hetero-
sexual acts outside marriage (see Table 1). [...]

Table 1 The Classification of Heterosexual Offences

Offences that are Offences with
Unmarried person Married person
Intra-group Incest Incestuous adultery
Extra-group Fornication i. Spouse of group
(group-spouse
adultery)
ii. Other married
person (non-group
adultery)

A further variable has been introduced into this table, that of

marital status. I have already explained why this is essential in

J. Goody 73
considering extra-group offences. But it may also be relevant
in the case of intercourse with a fellow-member of the group. For
instance, the LoDagaa of northern Ghana, among whom I
worked, and who are in many ways very similar to the Tallensi,
regarded intercourse with a clanswoman before her marriage,
that is, before her sexuality had been alienated to a member of
another clan, as being of very minor importance. But intercourse
with the same woman after her marriage, what I have called
‘incestuous adultery’, is more severely treated.
For the comprehensive analysis of heterosexual offences, it is
essential to introduce another variable, not shown in the table,
that of generation. Social relationships with a member of the
same or alternate generation are usually characterized by relative
equality and those between adjacent generations by super or
sub-ordination. This fact is likely to affect the severity with
which the offence is treated. It will tend to be more severely
treated where the relationship is characterized by authority, and
especially where the male offender is of junior generation, for
example, in the event of intercourse of a man with his father’s
wife. The same is true of other unequal statuses.

The incidence of horror

By breaking down the categories of incest and adultery in this
manner, it is possible to offer not only a more adequate analysis
of heterosexual offences in any one particular society, but also to
begin to examine these offences on a cross-cultural basis. I have
already called attention to the different incidence of ‘horror’
among the Tallensi and the Ashanti. In the former case it was
offences with clan wives that were considered most heinous,
whereas among the latter it was with the clan females themselves.
The category heavily sanctioned among the Ashanti was relatively
lightly treated among the Tallensi and vice versa. Why should the
Tallensi represent the ‘mirror image’ of the Ashanti in this
respect ?
I suggest the following is the explanation of this remarkable
reversal. The Tallensi are patrilineal; their classification of
offences resembles that of many other patrilineal peoples. The
category ‘wives’ is of fundamental importance to the descent
group because it is through them that the continuity of the clan is

74 Incest and Sex

obtained. Hence illegal intercourse with the wife of another
member of the group is treated most severely.
The Ashanti are matrilineal. Social reproduction, as distinct
from physiological reproduction, is obtained not through wives
but through ‘sisters’, the female members of the clan. Hence it is
interference with their sexuality that constitutes the most heinous
heterosexual offence. An interesting aspect of this explanation is
that it accounts for the differential treatment of father-daughter
and mother-son offences. In neither the patrilineal Tallensi nor in
matrilineal Ashanti does the father-daughter relationship fall into
the most heinous category, while in both societies the mother-son
relationship does. In the Tallensi the mother is the closest wife of
a clansman of senior generation, while in the Ashanti she is the
closest female clan member of senior generation. This I suggest
forms a more satisfactory explanation of the different treatment
of these offences than the usual ‘biological’ one.
To put this difference in another way, in patrilineal societies the
rights over a woman that are transferred at marriage include
rights to her reproductive capacities as well as rights to her
sexual services, whereas in matrilineal societies it is only the latter
that are transferred. Indeed among the Ashanti, a male only
acquires exclusive sexual rights by the payment of a special sum,
known as the tiri-nsa, which is not intrinsic to the ‘marriage’
itself.
The rights over the sexual services of women are customarily
vested in one man, except in the rare cases of polyandrous systems.
But the degree of this exclusiveness varies. For example, the
LoDagaa, like the Tallensi, regard intercourse with the wife of a
patriclansman as being the worst form of heterosexual offence.
Yet the junior of a pair of male twins, if unmarried, is said to have
access to the wife of his elder brother. In this case, the social
identification of the siblings is such that it overrules the individual¬
ization of rights to the sexual services of the wife. There is
always an incipient contradiction in patrilineal societies centring
around the fact that while rights to the sexual services of women
are in general acquired by individuals, rights to their procreative
capacities are to some extent vested in the clan as a whole. An
offspring of a particular union is an offspring of the entire clan.
This contradiction is differently resolved in various societies. In

J. Goody 75
Brahmin groups, for example, rights over women are so highly
individualized that a widowed woman may not marry again.
Among the Tallensi a man’s exclusive rights in a woman cease
at his death, and by the institution of widow inheritance are
taken over by another member of the same patrician. Fraternal
polyandry, or polycoity, represents the extreme case of corporate
rights over the sexual services of women, at the opposite pole as
it were to the individualization of Brahmin society. The problem
of plural access is different from, but not unrelated to, that of
plural marriage. [. ..]

Explanations of incest
Once the distinction between intra-group and group-wife
sexual offences has been understood the problems of the ‘ex¬
planation ’ of incest, and of the relationship between incest and
exogamy, can be seen in a new light. Explanations of incest fall
into three categories. Firstly, there are those framed in terms of
the internal relations of the group. These are associated with
writers who have concentrated their attention on sexual pro¬
hibitions within the elementary family: Freud, Radcliffe-Brown,
Malinowski, Brenda Seligman, Murdock, Parsons and others.
Secondly, there are those framed in terms of the external relations
of the group, which are associated principally with Tylor,
Fortune, and Levi-Strauss. In the third category fall the bio¬
logical, psychological-genetic variety. With this last I am not
concerned here, although I am aware that they find their way
into the formulations of some of the writers mentioned above.
I take the two sociological hypotheses as my starting point not
because I automatically assume that they will serve as complete
explanations, but because for heuristic purposes it seems to me
desirable to see how far one can get with these before employing
theories which from the sociologist’s standpoint are residual.
The two sociological theories are normally viewed as alternatives
and a considerable literature has accrued as to their relative
merits. Brenda Seligman has already summarized this discussion,
herself coming down on the side of internal relations. Her
argument is worth presenting not only because it gives some idea
of how the discussion has developed but also because it deals
fairly with both points of view. She writes:

76 Incest and Sex

Dr R. W. Fortune ... considers that the barrier itself is adopted not
because of its internal value to the family, but because the external
value of the marriage alliance is essential to social structure (1950,
p. 313).
She distinguishes two types of incest. ‘One is the union of
parent and child, the other is of siblings of opposite sex’ (p.
306). And she maintains that, while the marriage alliance might
account for the brother-sister taboo, it cannot possibly explain
the parent-child prohibition. Therefore, she concludes, it is the
internal value of the arrangement which is the most important
aspect of incest. ‘With the prohibition of incest within the ele¬
mentary family, the foundation of social structure is laid’
(p. 307). Thus she succeeds in categorizing heterosexual offences
on generation lines and perceives that different explanations
might be appropriate to each. But she fails to dichotomize either
in terms of group members and group wives, or in terms of the
structure of unilineal descent groups. The reason for this appears
to be her commitment to the Malinowskian stress on the ele¬
mentary family. If this is seen as the primary unit in relation to
which the incest taboo functions, then the only possible break¬
down of incest is by generation. The point elaborated in this
paper is that, in the analysis of ‘descent societies’, a further
breakdown is necessary, and exists within the actor frame of
reference either in the terms used or in the sanctions employed.
But the breakdown is made according to whether the prohibition
is on intercourse with a group member or with a group wife; and
the groups in question are in general based upon unilineal descent.
It is from this point of view that explanations of incest and exo¬
gamy must be considered.
Incest and exogamy are usually analysed as related prohibi¬
tions, the one on intercourse, the other on marriage. For example,
Evans-Pritchard in his study of the Nuer maintains that the
former is derived from the latter. Malinowski sometimes speaks
of incest and exogamy as if they were entirely complementary.
This point of view arises from a failure to make the distinction
discussed above. For while the rule prohibiting marriage inside
the group (exogamy) may be associated with the prohibition on
intercourse within the group (intra-group prohibition), it cannot
possibly be related, in any direct manner, to the prohibition on

J. Goody 77
intercourse with the wives of the group, for these women must
of necessity fall within the general category of permitted spouse.
They cannot possibly be excluded by any marriage rule.
Exogamy, then, can only be related to the prohibition on intra¬
group intercourse. But as Fortes has shown, there need be no
complete overlap even here. The Tallensi allow sexual inter¬
course with distant clansmen where they do not allow marriage.
The reason is clear. Marriage affects the alignment of relation¬
ships between groups; it has to be publicly validated by overt
transactions and it provides a precedent for similar arrangements
in the future. Sexual intercourse in itself does none of this, and
therefore when carried on in semi-secrecy requires no realign¬
ment of social relations. And indeed, as Fortes has also shown,
under certain conditions there may be advantages for the in¬
dividuals concerned if the lover is forbidden as a spouse, for then
these relationships are necessarily of limited duration. Within
groups of more restricted span, however, intercourse between
members can render other social relations difficult. This is
especially true where the relationship is characterized by super¬
ordination, as for example between members of adjacent gener¬
ations.
Although there is no inevitable overlap between the prohib¬
ition of intra-group intercourse and the prohibition of intra¬
group marriage, there is nevertheless a strong tendency for such
an overlap to occur. Exogamy is frequently phrased in terms of
kinship. . . . ‘We cannot marry our “sisters”.’ So is the intra¬
group sexual taboo. . . . ‘We cannot sleep with our “sisters”.’
It is true that the classificatory reference of the term ‘sister’
may not be the same in the two cases. This is so with the Tallensi.
In the first instance ‘sister’ refers to clan females as a whole, in
the second, to those belonging to the inner or medial lineage.
But the principle of structural congruence acts in favour of the
same referent in both cases. And indeed the prohibition on
temporary sexual relations and the prohibition on semi-perman¬
ent sexual relations are patently not unrelated.
If therefore the rule of exogamy is to be related to the external
value of the marriage alliance, as Tylor and others have suggested,
I think correctly, then the intra-group prohibition on intercourse
cannot be dissociated from it. The rejection of temporary sexu-

78 Incest and Sex

 ality within the group is in part a reflection of the rejection of
permanent sexuality, and the latter is related to the importance
of establishing inter-group relationships by the exchange of
rights in women.
Let us now turn to the prohibition on intercourse with those
who have married members of the descent group. This is spoken
of by Seligman, Fortes and many others as incest. Yet clearly
the explanations of Fortune, Levi-Strauss and others concerning
marriage alliances have no bearing at all upon this phenomenon,
because it is not intercourse with the women as such that is
forbidden, but intercourse with them as wives of group members.
Rights over their sexual services have been pre-empted by other
males with whom one has prior relationships. These women are
not necessarily consanguineal kin at all, with the exception of
ego’s mother; they are afflnes. Moreover, when the specific
relationship with the member of the descent group ceases, then
they may be legitimate sexual partners. In many cases one is in
fact obliged to marry them when their husband dies, because of
one’s relationship with the dead man. Now this type of prohibition
has nothing directly to do with marriage alliances, but rather
with the other explanation which has been put forward, namely,
the necessity of preserving the structure, not merely of the ‘family
for there would then be no need for a rule of any extensive appli¬
cation, but rather of the descent group. For where rights of sexual
access are individualized, conflict over females may be a cause
of internecine dispute, and this prohibition renders such disputes
less likely. It is indeed closely related to the taboo, found among
the Tallensi and among many other African peoples, against
more than one clansman having sexual relationships with one
woman during the same period.

Conclusions
The current sociological explanations of incest are not, then,
alternatives. Explanations in terms of external relations are
relevant to the prohibitions on intra-group intercourse, while
those in terms of internal relations are primarily relevant to the
group-wife prohibition, although they also bear upon the intra¬
group taboo. Exogamy can be related to the former, but not to
the latter.

J. Goody 79
 This paper has attempted to establish a typology of hetero¬
sexual prohibitions to facilitate both cross-cultural studies and
the depth analysis of particular societies. The typology depends
in the first place upon a distinction between women who are
considered to belong to the group and women who are married
to its male members. In the societies with which the discussion has
been mainly concerned, the reference group is the unilineal descent
group rather than the elementary family. It is impossible to
relate the concepts ‘incest’ and ‘exogamy’ when one term is
held to refer to a bilateral group, the family, and the other to a
unilineal one, the clan or lineage. It is impossible to account for
the different sanctions placed upon these acts among the patri¬
lineal Tallensi and the matrilineal Ashanti unless one introduces
the system of descent as a variable. For the ‘grisly horror of
incest’ is not a universal characteristic of all heterosexual offences
with kinswomen and the wives of kinsmen. The reactions to a
breach vary within and between societies. This is a fact which
psychologists venturing into the cross-cultural field have often
forgotten. Indeed, so concerned have they been with their own
findings that they have tended, even more than anthropologists,
to impose the categories derived from their own institutions upon
the other societies with which they have been concerned. This is
noticeable even in the type cases which psychologists have taken
from classical Greek mythology. The nature of early Greek
society makes it possible that their system of classification was
closer to the patrilineal societies of Africa than the bilateral ones
of modern Europe.
Like anthropologists, sociologists and psychologists dealing
with our own society have patently failed to realize the ethno¬
centric nature of their categories. They have tended to treat
‘incest’ as an isolate instead of examining the system of prohibi¬
tions as a whole in relation to the social structure. Thus there is a
quite disproportionate amount of literature devoted to ‘incest’
as compared to ‘adultery’, yet from the stand-point of social
problems the latter would seem to deserve the greater attention.
But the lure of the exotic has overcome the attraction of the
mundane.
The study of ‘incest’ in any society must be related not merely
to the analysis of marriage prohibitions or preferences, but also to

80 Incest and Sex

‘adultery’, so that it can be seen within the total constellation of
sexual offences within that society. And this can only be done by
accepting a breakdown of the monolithic category ‘incest’ into
concepts more closely related to the structure of the society in
question.

References
Fortes, M. (1936), ‘Kinship, incest and exogamy of the Northern
Territories of the Gold Coast’, in L. H. D. Buxton (ed.), Custom is
King, Hutchinson.
Fortes, M. (1949), The Web of Kinship among the Tallensi, Oxford
University Press.
Malinowski, B. (1929), The Sexual Life of Savages, Routledge &
Kegan Paul.
Murdock, G. P. (1949), Social Structure, Macmillan Co.
Murdock, G. P. (1955), ‘Changing emphases in social structure’.
Southwestern J. Anthropol., vol. 11, pp. 361-70.
Radcliffe-Brown, A. R. (1950), ‘Introduction’, in A. R.
Radcliffe-Brown and D. Forde (eds.), African Systems of Kinship and
Marriage, Oxford University Press.
Rattray, R. S. (1929), Ashanti Law and Constitution, Oxford University
Press.
Seligman, B. (1950), ‘Incest and exogamy: A reconsideration’, Amer.
Anthropol., vol. 52, pp. 305-16.

J. Goody 81
Part Three The Developmental Cycle

One way of approaching the study of ‘the family’, the

domestic aspect of kinship, is through the analysis of the
various types of group - productive, reproductive, residential
and consuming - around which domestic activities revolve.
Such groups are dynamic systems. Their structure may be
changing over the long term, due to external or internal forces.
But there is also a cyclical movement of growth and decline,
fission and fusion, as members marry, set up on their own,
bear children and die. In pre-industrial society, this pattern
of growth may be complicated by the fact that the groups
formed on the basis of these activities do not necessarily
overlap, so that more sensitive concepts are required for the
process of building up a comparative sociology of kinship
and the family. The analysis of the developmental cycle has
been developed by Fortes and his colleagues, who have used
it to shed light on patterns of residence, divorce and other
more general aspects of kinship. In this volume we present
Fortes’ introduction to a series of studies on this theme,
which has influenced much sociological thinking (e.g.
Stacey, 1969).

Reference
Stacey, M. (ed.) (1969), ‘Family and Household’, Comparability in
Social Research, Heinemann.
6 M. Fortes
The Developmental Cycle in Domestic Groups

Excerpts from M. Fortes, ‘Introduction’, in J. Goody (ed.). The

Developmental Cycle in Domestic Groups, Cambridge University Press,
1958, pp. 1-14.

The most promising advance in recent research on the social

structures of homogeneous societies has been the endeavour to
isolate and conceptualize the time factor. By this I do not mean
the amorphous subject matter usually labelled ‘culture change’
or ‘social change’. I mean the more fundamental and difficult
problems involved in the truism that the idea of society, the
notion of a social system or a social structure, necessarily implies
extension through a stretch of time. A social system, by definition,
has a life. It is a social system, that particular social system, only
so long as its elements and components are maintained and
adequately replaced; and the replacement process is the crucial
one because the human organism has a limited life span. Main¬
tenance and replacement are temporal phenomena. It is the
processes by which they are ensured that concern us when we
study the time factor in social structure.
These processes have biological determinants. One is the life
span of the individual; the other is the physical replacement of
every generation by the next in the succession of death and birth.
We must leave to physiology, genetics and demography the
exact study of these determinants. It is enough to remind our¬
selves that a social system will not persist if the average life span
of its members is too short for them to have offspring and to
rear them to the age when they in turn can have offspring, or,
in demographic terms, if the balance of births and deaths does
not yield a net reproduction rate of unity or more. From the
anthropological point of view, the important thing is that the
physical growth and development of the individual is embodied
in the social system through his education in the culture of his
society, and the succession of the generations through their

M. Fortes 85
incorporation in the social structure. The facts of physical
continuity and replacement are thus converted into the process
of social reproduction.
These generalities can be put in another way. For a social
system to maintain itself its two vital resources must be main¬
tained at an adequate level by continuous use and replacement.
These two resources are its human capital and its social capital,
and it is the latter that specially concerns the anthropologist.
It consists of the total body of knowledge and skill, values and
beliefs, laws and morals, embodied in the customs and institu¬
tions of a society and of the utilities made available for sup¬
porting the livelihood of its members through the application of
the cultural outfit to natural resources. The process of social
reproduction, in broad terms, includes all those institutional
mechanisms and customary activities and norms which serve to
maintain, replenish and transmit the social capital from genera¬
tion to generation.
Of course generalizations of this sort are not susceptible of
investigation by observation and experiment, nor do they lend
themselves to profitable theoretical discussion. They are useful
only as a step in the task of giving empirical content to the study
of the time factor in social structure. They lead us to ask what
are the institutional mechanisms and customary activities of
social reproduction in a particular society and how do they
operate? The nodal mechanism is well known. In all human
societies, the workshop, so to speak, of social reproduction, is
the domestic group. It is this group which must remain in
operation over a stretch of time long enough to rear offspring to
the stage of physical and social reproductivity if a society is to
maintain itself. This is a cyclical process. The domestic group
goes through a cycle of development analogous to the growth
cycle of a living organism. The group as a unit retains the same
form, but its members, and the activities which unite them, go
through a regular sequence of changes during the cycle which
culminates in the dissolution of the original unit and its replace¬
ment by one or more units of the same kind.
I shall later explain why it is useful to distinguish between the
domestic group and the family, in the strict sense. Here I am
interested in a different distinction. It is now commonly agreed

86 The Developmental Cycle

 that it is necessary, for analytical purposes, to distinguish between
the domestic field of social relations, institutions and activities,
viewed from within as an internal system, and the politico-jural
field, regarded as an external system. A significant feature of
the developmental cycle of the domestic group is that it is at one
and the same time a process within the internal field and a
movement governed by its relations to the external field.
To investigate this process in a given society we must first
establish what the domestic group is in that society. The con¬
ventional ethnographic method is to give a generalized descrip¬
tion derived from the observation of casually selected examples
and couched in terms of stereotyped persons and institutions.
This is like the amateur demography of travellers and colonial
officials in the days before rigorous census methods were intro¬
duced. To find out what the average family size was in a primitive
community, one rounded up twenty or thirty women at random
and questioned them about their children. One then divided the
total number of living children recorded by the total number of
women and so obtained an ‘average’. Such data are now re¬
garded as useless, owing partly to the faulty sampling method,
but chiefly to the failure to take into account age differences
among the women questioned. Similarly, if we wish to determine
reliably the structure and boundaries of the domestic group in a
given society, it is essential to use a reliable and representative
sample of domestic groups, and more particularly, to take into
account their ‘age-specific’ characters - that is, the stages of the
developmental cycle. A domestic group comprising only two
successive generations is at a different stage from one made up
of three generations; and so is one in which all the filial genera¬
tion are pre-adolescent as compared with one with some or all
the children at marriageable ages. The developmental factor is
intrinsic to domestic organization and to ignore it leads to
serious misinterpretation of the descriptive facts.
Residence patterns illustrate this very well. We know that they
provide a basic index of the boundaries of the internal structure
of domestic groups. But they are not a primary factor of social
structure of the order of kinship, descent, marriage and citizen¬
ship. The alignments of residence are determined by the econo¬
mic, affective, and jural relations that spring from these primary

M. Fortes 87
factors, and it is fallacious to analyse them in terms of ostensibly
discrete rules or types that come into effect at marriage. There are
numerous examples in the descriptive literature of kinship, but a
timely and particularly pertinent one is a recent paper by Good-
enough (1956).
There are, as he notes, several distinct questions involved.
First, there is the question of the normal residential composition
of the domestic group in the society. He shows how two investi¬
gators can arrive at totally discrepant conclusions about the
incidence of different ‘types’ of residence in the same com¬
munity though they use what seem to be the same census methods.
In fact the source of the apparent discrepancies is the neglect
by both investigators of the developmental dimension. [...] Resi¬
dence patterns are the crystallization, at a given time, of the
development process.
Secondly, there is quite a different problem when we consider
residential alignments from the point of view of the person
rather than from that of the domestic group as a unit. Genetical
analysis then needs to be supplemented by the numerical and
conceptual isolation of the structural and cultural variables
involved. Marriage is certainly a crucial element in determining
choice of residence by or for a person. In developmental terms,
the reason for this is because marriage leads to an actual or
incipient split in one or both of the spouses’ natal families and
domestic groups, and fission in the domestic group is always
translated into spatial representation in the residence arrange¬
ments. In analytical terms, this developmental moment is the
starting point of a redistribution of control over productive and
reproductive resources associated with a change in the jural
status of the spouses. Other things being equal, a wife will reside
with her husband if he, or whoever has jural authority over him,
has unrestricted rights over her sexual and economic services and
her reproductive powers, and children will reside with those who
have similar powers over, and the concomitant responsibilities
towards, them. Only numerical analysis can show what ‘degree
of freedom’, if any, exists.[. . .]
We can set up a paradigm distinguishing three main stages or
phases in the developmental cycle of the domestic group. First
there is a phase of expansion that lasts from the marriage of two

88 The Developmental Cycle

people until the completion of their family of procreation. The
biological limiting factor here is the duration of the wife’s
(or wives’) fertility. In structural terms it corresponds to the
period during which all the offspring of the parents are economi¬
cally, affectively and jurally dependent on them. Secondly, and
often overlapping the first phase in time (hence my preference for
the term ‘phase’ instead of ‘stage’), there is the phase of dis¬
persion or fission. This begins with the marriage of the oldest
child and continues until all the children are married. Where the
custom by which the youngest child remains to take over the
family estate is found, this commonly marks the beginning of
the final phase. This is the phase of replacement, which ends with
the death of the parents and the replacement in the social structure
of the family they founded by the families of their children, more
specifically, by the family of the father’s heir amongst the
children. [.. .]
Mutatis mutandis this paradigm can be applied to all social
systems. The birth of a couple’s first child, so frequently picked
out by special ritual observances, which initiates the phase of
expansion, and the marriage of their oldest child, which precipit¬
ates the eventual dissolution and replacement of their domestic
group, are always critical episodes in the developmental cycle.
But they are not, of course, the only critical turning points. The
initiation, retirement, or death of a member of the group may be
equally important.
In short, by the structural and cultural variables involved in the
developmental cycle I mean all the forces generated by the social
structure, and all the customs and institutions through which these
forces and the values they reflect are manifested. Biological laws
ensure that children inexorably grow up if they are not cut off by
death. Growing up requires a minimum time span, at least
fifteen years for the attainment of physiological maturity, and
often rather longer for the attainment of social maturity. The
complex and fundamental tasks of child-rearing imposed on the
domestic group by this fact generates critical forces for its cycle
of development.
The most important of these forces is the opposition between
successive generations focused in the incest taboos. This is not a
static condition. The opposition develops in intensity and may

M. Fortes 89
change in its customary forms of expression during the time that
the filial generation is growing up. It is a factor in the partial
or complete secession of offspring at marriage; for the essential
stake is the right to use and dispose of the productive and repro¬
ductive resources which every generation must gain possession
of when it reaches maturity. Among the Fulani we see very
clearly how growing up, for a boy, is projected into the social
structure through his increasing skill and responsibility in cattle
husbandry and the corresponding extension of his rights in herd
ownership, and culminates, after his marriage and achievement
of fatherhood, in the dispossession and virtual expulsion of his
father from the productive and reproductive organization of the
domestic group. In general, the allocation by gift, prestation,
inheritance and succession of rights over property, persons and
office on the one hand, and of rights over the fertility of women on
the other, is a major, if not the most significant, factor in the
developmental cycle of the domestic group.
Now the opposition between successive generations operates
primarily within the internal structure of the domestic group. But
it is legitimized and kept within bounds through being allowed
customary expression in forms sanctioned by the total society.
Marriage, inheritance, succession, and so forth, are events in the
internal system, or, to be more specific, domain of the domestic
group; but they are simultaneously events in the external domain,
where the domestic group is integrated into the total social
structure in its political, jural and ritual aspects. The interests
involved are those of society at large as well as those of the domes¬
tic group per se. This is shown in many customary forms, e.g.
in the conjunction of rules of exogamy with rules of incest in
the regulation of marriage, in the obligatory participation of
extra-domestic kin and of political authorities in funeral cere¬
monies and in decisions about inheritance and succession, in
initiation ceremonies, and so on. That is to say, it is through
political, jural and ritual institutions and customs which derive
their force from society at large that the interests of the total
social system, as opposed to those specific to the domestic
domain, are brought to bear on the latter. Classificatory kinship
institutions, unilineal descent corporations, age sets, and the
great variety of institutions and organizations through the me-

90 The Developmental Cycle

dium of which citizenship is exercised, are the structural links
between the two domains. We now have a number of excellent
studies showing how the domestic group and the unilineal descent
group are interlocked. The former is the source from which the
latter is continually replenished. This is not just a matter of
physical recruitment. There is a ‘feeding in’ process by which the
differentiation of persons in the domestic domain by generation,
filiation, and descent, is projected into the structure of the uni¬
lineal descent group to generate the modes of collocation and
segmentation so characteristic of lineage systems. It is a con¬
tinuous process that goes on as long as a lineage endures.
But there is a feature of this process that can easily be over¬
looked. It is true that fission in the domestic group can be re¬
garded as the model and starting point of segmentation in the
lineage, if we are concerned with the internal growing points of
the lineage as a temporal system. But if we look at lineage systems
from the point of view of their place in the external politico-
jural domain and consider their connection with the domestic
domain from that angle, we can see that differentiation and fission
in the domestic group are reciprocally determined by norms and
rules derived from the external domain. The classical example is
descent rules. [. . .]
In primitive societies the domain of domestic relations is
commonly organized around a nucleus consisting of a mother
and her children. Where the conjugal relationship and patri-
filiation are jurally or ritually effective in establishing a child’s
jural status, the husband-father becomes a critical link between
the matricentral cell and the domestic domain as a whole. In
this case the elementary family may be regarded as the nucleus.
This is the reproductive nucleus of the domestic domain. It
consists of two, and only two, successive generations bound
together by the primary dependence of the child on its parents
for nurture and love and of the parents on the child as the link
between them and their reproductive fulfilment. The domestic
group, on the other hand, often includes three successive genera¬
tions as well as members collaterally, or otherwise, linked with
the nucleus of the group. In this domain, kinship, descent and
other jural and affectional bonds (e.g. of adoption or slavery)
enter into the constitution of the group, whereas the nucleus is

M. Fortes 91
formed purely by the direct bonds of marriage, filiation and
siblingship. The domestic group is essentially a householding
and housekeeping unit organized to provide the material and
cultural resources needed to maintain and bring up its members.
The distinction, as I have said before, is an analytical one. The
actual composition of the nuclear family and the domestic group
may be identical, as it generally is in our own society; but the
strictly reproductive functions, in the sense given to our concept
of social reproduction, are distinguishable from the activities
concerned with the production of food and shelter and the
non-material means for ensuring continuity with society at large.
One might put it that the domestic domain is the system of social
relations through which the reproductive nucleus is integrated
with the environment and with the structure of the total society.
If we consider a person’s life cycle in the context of the domestic
group and its development, we can distinguish four major phases
in the period between his birth and his attainment of jural
adulthood. In the first he is wholly contained within the matri-
central cell. He is virtually merged with his mother, being no
more than an appendage to her in the social and affective as well
as the physiological sense. He is related to the total society only
through her. This phase may last for only the few days of post¬
partum seclusion and may be ritually terminated, or it may merge
imperceptibly into the second phase. In this the child is accepted
into the patricentral nuclear family unit and his father assumes
responsibility for him in relation to society and to spiritual
powers. Or rather, the husband-father assumes responsibility
for the mother-and-child as a unit. Presently, in the paradigmatic
case, after weaning and with the acquisition of the ability to
walk, he enters the third phase. He now moves into the domain
of the domestic group. The spatial correlate of this phase is that
the child is no longer confined to his mother’s quarters but has
the freedom of the whole dwelling house. He now comes under
the jural and ritual care of the head of the domestic group, who
may or may not be his own parent. This is the phase of childhood
proper and may last for some years. During the whole of it he has
no autonomous rights over property or productive resources,
not even his own developing skills, no independent access to
ritual institutions, and no political or jural standing in his own

92 The Developmental Cycle

right. Finally he is admitted to the politico-jural domain. This
confers on him actual or potential autonomy in the control of
some productive resources, the elements of jural independence,
rights of access to ritual powers and institutions, and some rights
and duties of citizenship, as in warfare or feud. It is common for
this phase to be legitimized by rites de passage, and to have a
spatial correlate, as with the Trobriand boy who takes up resi¬
dence with his maternal uncle. The culmination of the fourth
phase is marriage and the actual or incipient fission of the natal
domestic group.
What I am stressing in this paradigm is the changing structural
relationships that make up the framework of a person’s life-
cycle. The stages of physiological maturation that accompany
this development are of secondary significance. They are chiefly
important as signs of readiness for a shift from one phase to the
next. For each phase has its appropriate norms and activities
connected with the basic psycho-physical capacities and needs.
In the first phase a child is wholly dependent on the mother’s
breast for food and her arms for shelter and love. In the next
phase he usually eats with his mother, sleeps in her room, and
learns from her the fundamental self-oriented skills and values
involved in walking, talking, feeding and cleanliness. He is
regarded as sexually neutral and morally irresponsible. This
pattern persists through the second phase. In the third phase the
sexual division of roles and activities becomes effective. Boys are
attached to their fathers and girls to their mothers. A boy
commonly eats with his father or older brothers, sleeps with them,
and learns objectively oriented social and economic skills and
values from them. Moral responsibility is demanded of both
sexes. They have to learn to control their affective attitudes to
conform to customary norms of conduct and, in particular, they
become subject to the incest taboos. In the next phase boys and
girls eat and sleep with their like-sex age mates or peers. They
are expected to take a responsible part in the performance of
economic, military, jural and ritual duties for the benefit of the
total society. They become answerable, to a greater or lesser
degree, for moral and jural misdemeanours. Above all, they are
now permitted to enter into relationships which involve adult
sexuality for procreative ends, as opposed to childish sexuality

M. Fortes 93
for pleasurable ends. They are subject not only to the incest rules,
which belong to the domestic domain, but also to the marriage
regulations, which emanate from the politico-jural domain. Rites
de passage often serve to dramatize this fact.
Though these phases do not invariably conform to stages of
physiological growth, in relatively homogeneous social systems
there is a close parallelism between them. For in such societies
the basic educational tasks required to produce an adult person
capable of playing a full part in maintaining and transmitting the
social capital seem to be complete at about the same time as
the attainment of physical and sexual maturity and therewith
the capacity for replacing the parental generation in productive
and reproductive activities. But what I want particularly to
emphasize is that the maturation of the individual and his proper
passage through the life cycle is of paramount concern to society
at large. This is shown in the widespread occurrence of institution¬
alized procedures for legitimizing each step in it, and especially
for terminating the period of jural infancy, whether it ends with
adolescence or extends into the stage of physical adulthood.
Initiation, puberty and nubility ceremonies are the most
dramatic instances of such procedures. In these ceremonies the
domestic group’s task of social reproduction is terminated. Hav¬
ing bred, reared and educated the child, it hands over the
finished product to the total society. It is a transaction in which
the power and authority of the politico-jural order as the final
arbiter over the human and social capital of society are asserted.
It is a situation in which the distinctive interests of the domestic
group and those of the total society are liable to clash. In their
capacity as citizens, parents wish their children to be admitted
to the politico-jural domain and to have the rights of jural
adulthood conferred on them. But as parents they may fear and
resent having to relinquish their children to the superior and
impersonal powers of society at large. Their resistance may be
strengthened by the knowledge that initiation is the thin end of
the wedge that will ultimately split the family. The children for
their part, however mature they are and however much they
value admission to adulthood, may hesitate to step out of the
protective circle of the home. It may be particularly hard to
renounce the bond of primal dependency on the mother which

94 The Developmental Cycle

goes back to the first phase of the life cycle. If there is a marked
cleavage between the domestic domain and the politico-jural
domain these resistances may be institutionalized and the more
difficult to overcome. Hence society may have to use harsh and
abrupt rites to tear the new citizen away from his natal family
and to assert its right to incorporate him as an adult. Citizenship
may need a drastic reorientation of moral values and of social
and economic roles in the new recruit. Shock tactics may be the
readiest way to bring this about. Furthermore, the stamp of
legitimacy must be publicly and incontrovertibly set upon the
new rights (notably those of jural autonomy and procreative
sexuality) - and corresponding duties (notably those defending
the social order against dangers from within, such as crime, and
against the external dangers of war and feud) - which are con¬
ferred by citizenship.
I am not here concerned with the theory of initiation ceremon¬
ies and further discussion of them would be out of place. I have
referred to them merely in order to illustrate what I mean by a
movement or transaction between the two domains of social
structure we have been analysing. There are many societies in
which the movement is not legitimized by means of initiation or
other ceremonies. The reason may be that the two domains are
not, analytically speaking, separated by a decisive cleavage. In
any case, the movement does take place. There is a phase in the
life cycle when jural infancy draws to an end and jural adulthood
begins. It may be initiated, as has already been suggested, by
marriage or by the birth of a couple’s first child. Initiation
ceremonies, in the strict sense, are often regarded as the prelude
to marriage, if they do not actually end in marriage. In general,
what finally terminates jural infancy is the emergence of the
family nucleus for the new domestic group that is destined to
replace that of the parents. Initiation ceremonies are sometimes
spread over months or years, the preliminary rites serving, as it
were, to apprentice the new recruit to the politico-jural domain
and the later ones serving to make him free of that domain when
he has proved worthy. Analogously, institutions like the shift
of residence at adolescence from the father’s to the maternal
uncle’s house may be regarded as the first steps in a longer
process of jural emancipation which ends with marriage.

M. Fortes 95
 One consideration that must not be lost sight of is the recipro¬
cal relationship between the two domains. Every member of a
society is simultaneously a person in the domestic domain and
in the politico-jural domain. His status in the former receives
definition and sanction from the latter. Jural infancy is structur¬
ally located in the domestic domain, but its character is defined
by norms validated in the politico-jural domain. Take the extreme
case of an Ashanti infant which is defined as being non-human,
that is, not a potential member of society, if it dies before the
naming ceremony on the eighth day after its birth. This jural
status derives from the politico-jural domain. The parents are
obliged to accept the definition whatever their private emotions
may be.
This has a direct bearing on the internal structure of the domes¬
tic group. The differentials in this structure are in part inherent
in the procreative relationship and spring from the require¬
ments of child rearing. But their character is also decisively
regulated by politico-jural norms. The gap between successive
generations can be large or small, varying with the kind and
degree of authority and power vested in the parental generation;
solidarity can be stressed more than rivalry in the sibling group,
as in lineage systems, or vice versa, as among the Iban. These
are differences of magnitude and of precedence related to the
balance that is struck in a particular social system between the
variables that combine together in the organization of the domes¬
tic domain. They are expressed in customs, beliefs and institutions
that are the collective possessions of the whole society not the
private culture of each domestic group.
The classical illustration is the contrast in the relationships of
fathers and children in patrilineal and in matrilineal descent
systems. It is because the father is not vested with jural authority
over his son and the son has no title to the inheritance of his
father’s properties or to succession to his offices and rank, that
matrilineal fathers and sons have an affectionate, non-competitive
relationship. Conversely, it is because maternal uncles have
jurally sanctioned rights over their nephews and the latter have
jural ly sanctioned claims on their uncles that there is tension in
their relationship. And the pattern is reversed in patrilineal
systems because the locus of rights and claims is jurally reversed.

96 The Developmental Cycle

 Matrilineal fatherhood is defined as primarily a domestic re¬
lationship with only a minimal function in the politico-jural
domain. Hence its focus is the task of bringing up and educating
a child and fathers must rely on moral and affectional sanctions
to fulfil it. In the last resort society will stand behind them to
prevent trespass on their prerogative but gives them no support in
the enforcement of their will on their children. We can contrast
this with the juridical support society gives to the matrilineal
husband in enforcing his rights to the sexual services of his wife.
A patrilineal father, on the other hand, has not only the domestic
and parental roles of provider and educator. He also has rights
enforceable by juridical sanctions over and towards them and
they have corresponding claims on him. He represents the power
of society as a force within the domestic group in a way that the
matrilineal father does not. And this analysis could be carried
further if we were to take into consideration a third domain of
social structure, that of ritual institutions. I have made allusions
to this domain but it is not directly pertinent to our inquiries.
This formulation enables us to see why numerical data are
essential for the analysis of the developmental cycle of the domes¬
tic group. Each phase of the cycle can be thought of as the out¬
come of a set of * pushes ’ and 4 pulls ’, antecedent and contempor¬
aneous. They come in part from within the domestic domain and
in part from the external structure of society. Numerical data
provide a means of assessing the relative strength of these forces
and of describing their configuration at a given phase. Let us
take a case such as we find in a society like the Tallensi, with
their rigorous patrilineal descent system. During the expansion
phase of the domestic group all forces converge on supporting
the paramountcy of the father in the domestic domain. He
controls all the productive resources required to provide for his
wife and children and he is vested with jural authority over them.
Neither his wife nor his children have jural status, economic
rights or ritual standing except through him. Consistently with
this a man’s wife, and his children during their jural infancy, can
be expected to live with him, and numerical data show that they
invariably do so. In the dispersion phase, however, a son’s rights
to some measure of jural, economic and ritual independence
become operative, and he may set up his own dwelling group.

T-K-D M. Fortes 97
But whether he moves out of the parental home altogether to
farm on his own or remains residentially attached to his father’s
homestead depends on factors internal to the domestic group.
If he is the only son he will be less likely to move away than if
he has brothers and if he is the oldest son he will be more likely
to do so than if he is a younger son. Moreover the move may
take place by stages and will not be complete until he has young
children of his own. Numerical data are essential to assess the
relative weight of these factors; and it has now become an
established practice among social anthropologists to use such
data in the analysis of social structure.

Reference
Goodenough, W. H. (1956), ‘Residence rules’, Southwestern J.
Anthropol., vol. 12, pp. 22-37.

98 The Developmental Cycle

Part Four Joking and Avoidance

The fact that, in Western societies, the widespread custom of

avoidance of the wife’s mother has provided the substance of so
many music-hall jokes indicates one of its functions: the
preservation of social distance when two persons with
conflicting, interests in a third person are brought into direct
confrontation. Somewhat crudely, formalized behaviour
between persons in kin relationships can be seen as varying on
a continuum between avoidance and familiarity, with piety,
respect, joking and friendship lying in between. The subject was
discussed by E. B. Tylor (1889) and later pursued in much field
research. An important step forward in systematizing ideas
about such behaviour was taken by Radcliffe-Brown (Reading
7). To do justice to the many facets of even the standardized
behaviour between kin would require a great deal more space
than we have at our disposal; meanwhile the present essay must
stand as representative of a wide field of interest.

Reference
Tylor, E. B. (1889), ‘On a method of investigating the development of
institutions: applied to laws of marriage and descent’, /. anthropol. Inst.,
vol. 18, pp. 245-72.
7 A. R. Radcliffe-Brown
Joking Relationships1

A, R. Radcliffe-Brown, ‘On joking relationships’, Africa, vol. 13, 1940,

pp. 195-210. Reprinted in A. R. Radcliffe-Brown, Structure and Function
in Primitive Society, Cohen & West, 1952, pp. 90-104.

What is meant by the term ‘joking relationship’ is a relation

between two persons in which one is by custom permitted, and in
some instances required, to tease or make fun of the other, who in
turn is required to take no offence. It is important to distinguish
two main varieties. In one the relation is symmetrical; each of
the two persons teases or makes fun of the other. In the other
variety the relation is asymmetrical; A jokes at the expense of
B and B accepts the teasing good humouredly but without re¬
taliating; or A teases B as much as he pleases and B in return
teases A only a little. There are many varieties in the form of
this relationship in different societies. In some instances the
joking or teasin0 is only verbal, in others it includes horse¬
play; in some the joking includes elements of obscenity, in others
not.
Standardized social relationships of this kind are extremely
widespread, not only in Africa but also in Asia, Oceania and
North America. To arrive at a scientific understanding of the
phenomenon it is necessary to make a wide comparative study.
Some material for this now exists in anthropological literature,
though by no means all that could be desired, since it is un¬
fortunately still only rarely that such relationships are observed
and described as exactly as they might be.
The joking relationship is a peculiar combination of friendli¬
ness and antagonism. The behaviour is such that in any other
social context it would express and arouse hostility; but it is not
meant seriously and must not be taken seriously. There is a
pretence of hostility and a real friendliness. To put it in another

1. Professor Marcel Mauss has published a brief theoretical discussion of

the subject (1927-8). It is also dealt with by Dr F. Eggan (1937).

A. R. Radcliffe-Brown 101
way, the relationship is one of permitted disrespect. Thus any
complete theory of it must be part of, or consistent with, a
theory of the place of respect in social relations and in social
life generally. But this is a very wide and very important sociologi¬
cal problem; for it is evident that the whole maintenance of a
social order depends upon the appropriate kind and degree of
respect being shown towards certain persons, things and ideas or
symbols.
Examples of joking relationships between relatives by marriage
are very commonly found in Africa and in other parts of the
world. Thus Mademoiselle Paulme (1939) records that among the
Dogon a man stands in a joking relationship to his wife’s sisters
and their daughters. Frequently the relationship holds between
a man and both the brothers and sisters of his wife. But in some
instances there is a distinction whereby a man is on joking terms
with his wife’s younger brothers and sisters but not with those
who are older than she is. This joking with the wife’s brothers
and sisters is usually associated with a custom requiring extreme
respect, often partial or complete avoidance, between a son-in-
law and his wife’s parents.2
The kind of structural situation in which the associated
customs of joking and avoidance are found may be described as
follows. A marriage involves a readjustment of the social struc¬
ture whereby the woman’s relations with her family are greatly
modified and she enters into a new and very close relation with
her husband. The latter is at the same time brought into a special
relation with his wife’s family, to which, however, he is an out¬
sider. For the sake of brevity, though at the risk of over-simpli¬
fication, we will consider only the husband’s relation to his wife’s
family. The relation can be described as involving both attach¬
ment and separation, both social conjunction and social dis¬
junction, if I may use the terms. The man has his own definite
position in the social structure, determined for him by his birth
into a certain family, lineage or clan. The great body of his
rights and duties and the interests and activities that he shares
with others are the result of his position. Before the marriage
his wife s family are outsiders for him as he is an outsider for
2. Those who are not familiar with these widespread customs will find
descriptions in Junod (1927) and in F. Eggan (1937).

102 Joking and Avoidance

them. This constitutes a social disjunction which is not destroyed
by the marriage. The social conjunction results from the con¬
tinuance, though in altered form, of the wife’s relation to her
family, their continued interest in her and in her children. If the
wife were really bought and paid for, as ignorant persons say that
she is in Africa, there would be no place for any permanent close
relation of a man with his wife’s family. But though slaves can be
bought, wives cannot.
Social disjunction implies divergence of interests and therefore
the possibility of conflict and hostility, while conjunction requires
the avoidance of strife. How can a relation which combines the
two be given a stable, ordered form? There are two ways of
doing this. One is to maintain between two persons so related an
extreme mutual respect and a limitation of direct personal con¬
tact. This is exhibited in the very formal relations that are, in
so many societies, characteristic of the behaviour of a son-in-law
on the one side and his wife’s father and mother on the other.
In its most extreme form there is complete avoidance of any social
contact between a man and his mother-in-law.
This avoidance must not be mistaken for a sign of hostility.
One does, of course, if one is wise, avoid having too much to do
with one’s enemies, but that is quite a different matter. I once
asked an Australian native why he had to avoid his mother-in-
law, and his reply was, ‘Because she is my best friend in the world;
she has given me my wife’. The mutual respect between son-in-
law and parents-in-law is a mode of friendship. It prevents
conflict that might arise through divergence of interest.
The alternative to this relation of extreme mutual respect and
restraint is the joking relationship, one, that is, of mutual dis¬
respect and licence. Any serious hostility is prevented by the
playful antagonism of teasing, and this in its regular repetition is
a constant expression or reminder of that social disjunction which
is one of the essential components of the relation, while the social
conjunction is maintained by the friendliness that takes no
offence at insult.
The discrimination within the wife’s family between those who
have to be treated with extreme respect and those with whom it is
a duty to be disrespectful is made on the basis of generation and
sometimes of seniority within the generation. The usual respected

A. R. Radcliffe-Brown 103
relatives are those of the first ascending generation, the wife’s
mother and her sisters, the wife’s father and his brothers, some¬
times the wife’s mother’s brother. The joking relatives are those
of a person’s own generation; but very frequently a distinc¬
tion of seniority within the generation is made; a wife’s older
sister or brother may be respected while those younger will be
teased.
In certain societies a man may be said to have relatives by
marriage long before he marries and indeed as soon as he is
born into the world. This is provided by the institution of the
required or preferential marriage. We will, for the sake of brevity,
consider only one kind of such organizations. In many societies
it is regarded as preferable that a man should marry the daughter
of his mother’s brother; this is a form of the custom known as
cross-cousin marriage. Thus his female cousins of this kind, or
all those women whom by the classificatory system he classifies
as such, are potential wives for him, and their brothers are his
potential brothers-in-law. Among the Ojibwa Indians of North
America, the Chiga of Uganda, and in Fiji and New Caledonia,
as well as elsewhere, this form of marriage is found and is ac¬
companied by a joking relationship between a man and the sons
and daughters of his mother’s brother. To quote one instance
of these, the following is recorded for the Ojibwa.

When cross-cousins meet they must try to embarrass one another.

They ‘joke’ one another, making the most vulgar allegations, by their
standards as well as ours. But being ‘kind’ relations, no one can take
offence. Cross-cousins who do not joke in this way are considered
boorish, as not playing the social game (Landes, 1937, p. 103).

The joking relationship here is of fundamentally the same kind

as that already discussed. It is established before marriage and is
continued, after marriage, with the brothers- and sisters-in-law.
In some parts of Africa there are joking relationships that
have nothing to do with marriage. Mr Pedler’s note (1940)
refers to a joking relationship between two distinct tribes, the
Sukuma and the Zaramu, and in the evidence it was stated that
there was a similar relation between the Sukuma and the Zigua
and between the Ngoni and the Bemba. The woman’s evidence
suggests that this custom of rough teasing exists in the Sukuma

104 Joking and Avoidance

 tribe between persons related by marriage, as it does in so many
other African tribes.3
While a joking relationship between two tribes is apparently
rare, and certainly deserves, as Mr Pedler suggests, to be care¬
fully investigated, a similar relationship between clans has been
observed in other parts of Africa. It is described by Professor
Labouret (1929) and Mademoiselle Paulme (1939) in the articles
previously mentioned, and amongst the Tallensi it has been
studied by Dr Fortes (1945).
The two clans are not, in these instances, specially connected
by intermarriage. The relation between them is an alliance
involving real friendliness and mutual aid combined with an
appearance of hostility.
The general structural situation in these instances seems to be
as follows. The individual is a member of a certain defined group,
a clan, for example, within which his relations to others are
defined by a complex set of rights and duties, referring to all the
major aspects of social life, and supported by definite sanctions.
There may be another group outside his own which is so linked
with his as to be the field of extension of jural and moral relations
of the same general kind. Thus, in East Africa, as we learn from
Mr Pedler’s note, the Zigua and the Zaramu do not joke with
one another because a yet closer bond exists between them since
they are ndugu (brothers). But beyond the field within which
social relations are thus defined there lie other groups with which,
since they are outsiders to the individual’s own group, the
relation involves possible or actual hostility. In any fixed relations
between the members of two such groups the separateness of the
3. Incidentally it may be said that it was hardly satisfactory for the
magistrate to establish a precedent whereby the man, who was observing
what was a permitted and may even have been an obligatory custom, was
declared guilty of common assault, even with extenuating circumstances.
It seems quite possible that the man may have committed a breach of
etiquette in teasing the woman in the presence of her mother’s brother,
for in many parts of the world it is regarded as improper for two persons
in a joking relationship to tease one another (particularly if any obscenity
is involved) in the presence of certain relatives of either of them. But the
breach of etiquette would still not make it an assault. A little knowledge
of anthropology would have enabled the magistrate, by putting the appro¬
priate questions to the witnesses, to have obtained a fuller understanding
of the case and all that was involved in it.

A. R. Radcliffe-Brown 105
groups must be recognized. It is precisely this separateness which
is not merely recognized but emphasized when a joking relation¬
ship is established. The show of hostility, the perpetual disrespect,
is a continual expression of that social disjunction which is an
essential part of the whole structural situation, but over which,
without destroying or even weakening it, there is provided the
social conjunction of friendliness and mutual aid.
The theory here put forward, therefore, is that both the
joking relationship which constitutes an alliance between clans or
tribes and that between relatives by marriage are modes of
organizing a definite and stable system of social behaviour in
which conjunctive and disjunctive components, as I have called
them, are maintained and combined.
To provide the full evidence for this theory by following out
its implications and examining in detail its application to different
instances wouia take a book rather than a short article. But some
confirmation can perhaps be offered by a consideration of the
way in which respect and disrespect appear in various kinship
relations, even though nothing more can be attempted than a
very brief indication of a few significant points.
In studying a kinship system it is possible to distinguish the
different relatives by reference to the kind and degree of respect
that is paid to them (see Eggan, 1937; Mead, 1934). Although
kinship systems vary very much in their details there are certain
principles which are found to be very widespread. One of them is
that by which a person is required to show a marked respect to
relatives belonging to the generation immediately preceding his
own. In a majority of societies the father is a relative to whom
marked respect must be shown. This is so even in many so-called
matrilineal societies, i.e. those which are organized into matrilineal
clans or lineages. One can very frequently observe a tendency
to extend this attitude of respect to all relatives of the first
ascending generation and, further, to persons who are not
relatives. Thus in those tribes of East Africa that are organized
into age-sets a man is required to show special respect to all
men of his father’s age-set and to their wives.
The social function of this is obvious. The social tradition is
handed down from one generation to the next. For the tradition
to be maintained it must have authority behind it. The authority

106 Joking and Avoidance

is therefore normally recognized as possessed by members of the
preceding generation and it is they who exercise discipline. As a
result of this the relation between persons of the two generations
usually contains an element of inequality, the parents and those
of their generation being in a position of superiority over the
children who are subordinate to them. The unequal relation
between a father and his son is maintained by requiring the latter
to show respect to the former. The relation is asymmetrical.
When we turn to the relation of an individual to his grand¬
parents and their brothers and sisters we find that in the majority
of human societies relatives of the second ascending generation
are treated with very much less respect than those of the first
ascending generation, and instead of a marked inequality there is
a tendency to approximate to a friendly equality.
Considerations of space forbid any full discussion of this
feature of social structure, which is one of very great importance.
There are many instances in which the grandparents and their
grandchildren are grouped together in the social structure in
opposition to their children and parents. An important clue to
the understanding of the subject is the fact that in the flow of
social life through time, in which men are born, become mature
and die, the grandchildren replace their grandparents.
In many societies there is an actual joking relationship, usually
of a relatively mild kind, between relatives of alternate genera¬
tions. Grandchildren make fun of their grandparents and of those
who are called grandfather and grandmother by the classificatory
system of terminology, and these reply in kind.
Grandparents and grandchildren are united by kinship; they
are separated by age and by the social difference that results from
the fact that as the grandchildren are in process of entering into
full participation in the social life of the community the grand¬
parents are gradually retiring from it. Important duties towards
his relatives in his own and even more in his parents’ generation
impose upon an individual many restraints; but with those of the
second ascending generation, his grandparents and collateral rela¬
tives, there can be established, and usually is, a relationship of
simple friendliness relatively free from restraint. In this instance
also, it is suggested, the joking relationship is a method of order¬
ing a relation which combines social conjunction and disjunction.

A. R. Radcliffe-Brown 107
 This thesis could, I believe, be strongly supported if not
demonstrated by considering the details of these relationships.
There is space for only one illustrative point. A very common
form of joke in this connection is for the grandchild to pretend
that he wishes to marry the grandfather’s wife, or that he intends
to do so when his grandfather dies, or to treat her as already being
his wife. Alternatively the grandfather may pretend that the wife
of his grandchild is, or might be, his wife (see Labouret, 1931,
p. 248; Roy, 1915, pp. 352-4). The point of the joke is the
pretence at ignoring the dilference of age between the grand¬
parent and the grandchild.
In various parts of the world there are societies in which a
sister’s son teases and otherwise behaves disrespectfully towards
his mother’s brother. In these instances the joking relationship
seems generally to be asymmetrical. For example the nephew
may take his uncle’s property but not vice versa; or, as amongst
the Nama Hottentots, the nephew may take a fine beast from
his uncle’s herd and the uncle in return takes a wretched beast
from that of the nephew (Hoernle, 1925).
The kind of social structure in which this custom of privileged
disrespect to the mother’s brother occurs in its most marked
forms, for example the Thonga of South-East Africa, Fiji and
Tonga in the Pacific and the Central Siouan tribes of North
America, is characterized by emphasis on patrilineal lineage and a
marked distinction between relatives through the father and
relatives through the mother.
In a former publication (1924) I offered an interpretation of this
custom of privileged familiarity towards the mother’s brother.
Briefly it is as follows. For the continuance of a social system
children require to be cared for and to be trained. Their care
demands affectionate and unselfish devotion; their training
requires that they shall be subjected to discipline. In the societies
with which we are concerned there is something of a division of
function between the parents and other relatives on the two
sides. The control and discipline are exercised chiefly by the
father and his brothers and generally also by his sisters; these are
relatives who must be respected and obeyed. It is the mother who
is primarily responsible for the affectionate care; the mother and
her brothers and sisters are therefore relatives who can be looked

108 Joking and Avoidance

 to for assistance and indulgence. The mother’s brother is called
‘male mother’ in Tonga and in some South African tribes.
I believe that this interpretation of the special position of the
mother’s brother in these societies has been confirmed by further
field work since I wrote the article referred to. But I was quite
aware at the time it was written that the discussion and interpre¬
tation needed to be supplemented so as to bring them into line
with a general theory of the social functions of respect and dis¬
respect.
The joking relationship with the mother’s brother seems to
fit well with the general theory of such relationships here outlined.
A person’s most important duties and rights attach him to his
paternal relatives, living and dead. It is to his patrilineal lineage
or clan that he belongs. For the members of his mother’s lineage
he is an outsider, though one in whom they have a very special
and tender interest. Thus here again there is a relation in which
there is both attachment or conjunction, and separation or
disjunction, between the two persons concerned.
But let us remember that in this instance the relation is
asymmetrical.4 The nephew is disrespectful and the uncle accepts
the disrespect. There is inequality and the nephew is the superior.
This is recognized by the natives themselves. Thus in Tonga it is
said that the sister’s son is a ‘chief’ (eiki) to his mother’s brother,
and Junod (1927, p. 255) quotes a Thonga native as saying ‘The
uterine nephew is a chief! He takes any liberty he likes with his
maternal uncle’. Thus the joking relationship with the uncle
does not merely annul the usual relation between the two genera¬
tions, it reverses it. But while the superiority of the father and the
father’s sister is exhibited in the respect that is shown to them, the
nephew’s superiority to his mother’s brother takes the opposite
form of permitted disrespect.
It has been mentioned that there is a widespread tendency to
feel that a man should show respect towards, and treat as social
superiors, his relatives in the generation preceding his own, and

4. There are some societies in which the relation between a mother’s

brother and a sister’s son is approximately symmetrical, and therefore one
of equality. This seems to be so in the Western Islands of Torres Straits,
but we have no information as to any teasing or joking, though it is said
that each of the two relatives may take the property of the other.

A. R. Radcliffe-Brown 109
the custom of joking with, and at the expense of, the maternal
uncle clearly conflicts with this tendency. This conflict between
principles of behaviour helps us to understand what seems at first
sight a very extraordinary feature of the kinship terminology of
the Thonga tribe and the VaNdau tribe in South-East Africa.
Amongst the Thonga, although there is a term malume (= male
mother) for the mother’s brother, this relative is also, and perhaps
more frequently, referred to as a grandfather (kokwana) and he
refers to his sister’s son as his grandchild (ntukulu). In the
VaNdau tribe the mother’s brother and also the mother’s
brother’s son are called ‘grandfather’ (tetekulu, literally ‘great
father’) and their wives are called ‘grandmother’ (mbiya),
while the sister’s son and the father’s sister’s son are called
‘grandchild’ (muzukulu).
This apparently fantastic way of classifying relatives can be
interpreted as a sort of legal fiction whereby the male relatives
of the mother’s lineage are grouped together as all standing
towards an individual in the same general relation. Since this
relation is one of privileged familiarity on the one side, and solici¬
tude and indulgence on the other, it is conceived as being basically
the one appropriate for a grandchild and a grandfather. This is
indeed in the majority of human societies the relationship in
which this pattern of behaviour most frequently occurs. By this
legal fiction the mother’s brother ceases to belong to the first
ascending generation, of which it is felt that the members ought
to be respected.
It may be worth while to justify this interpretation by con¬
sidering another of the legal fictions of the VaNdau terminology.
In all these south-eastern Bantu tribes both the father’s sister and
the sister, particularly the elder sister, are persons who must be
treated with great respect. They are also both of them members of
a man’s own patrilineal lineage. Amongst the VaNdau the father’s
sister is called ‘female father’ (tetadji) and so also is the sister
(see Boas, 1922). Thus by the fiction of terminological classifica¬
tion the sister is placed in the father’s generation, the one that
appropriately includes persons to whom one must exhibit marked
respect.
In the south-eastern Bantu tribes there is assimilation of two
kinds of joking relatives, the grandfather and the mother’s

110 Joking and Avoidance

 brother. It may help our understanding of this to consider an
example in which the grandfather and the brother-in-law are
similarly grouped together. The Cherokee Indians of North
America, probably numbering at one time about 20,000, were
divided into seven matrilineal clans (Gilbert, 1937). A man could
not marry a woman of his own clan or of his father’s clan. Com¬
mon membership of the same clan connects him with his brothers
and his mother’s brothers. Towards his father and all his relatives
in his father’s clan of his own or his father’s generation he is
required by custom to show a marked respect. He applies the
kinship term for ‘father’ not only to his father’s brothers but
also to the sons of his father’s sisters. Here is another example of
the same kind of fiction as described above; the relatives of his
own generation whom he is required to respect and who belong
to his father’s matrilineal lineage are spoken of as though they
belonged to the generation of his parents. The body of his
immediate kindred is included in these two clans, that of his
mother and his father. To the other clans of the tribe he is in a
sense an outsider. But with two of them he is connected, namely
with the clans of his two grandfathers, his father’s father and his
mother’s father. He speaks of all the members of these two clans,
of whatever age, as ‘grandfathers’ and ‘ grandmothers’. He stands
in a joking relationship with all of them. When a man marries he
must respect his wife’s parents but jokes with her brothers and
sisters.
The interesting and critical feature is that it is regarded as
particularly appropriate that a man should marry a woman whom
he calls ‘grandmother’, i.e. a ipember of his father’s father’s
clan or his mother’s father’s clan. If this happens his wife’s
brothers and sisters, whom he continues to tease, are amongst
those whom he previously teased as his ‘grandfathers’ and
‘grandmothers’. This is analogous to the widely spread organiz¬
ation in which a man has a joking relationship with the children
of his mother’s brother and is expected to marry one of the
daughters.
It ought perhaps to be mentioned that the Cherokee also have
a one-sided joking relationship in which a man teases his father’s
sister’s husband. The same custom is found in Mota of the
Bank Islands. In both instances we have a society organized on a

A. R. Radcliffe-Brown 111
 matrilineal basis in which the mother’s brother is respected, the
father’s sister’s son is called ‘ father ’ (so that the father’s sister’s
husband is the father of a ‘father’), and there is a special term
for the father’s sister’s husband. Further observation of the
societies in which this custom occurs is required before we can
be sure of its interpretation. I do not remember that it has been
reported from any part of Africa.
What has been attempted in this paper is to define in the most
general and abstract terms the kind of structural situation in
which we may expect to find well-marked joking relationships.
We have been dealing with societies in which the basic social
structure is provided by kinship. By reason of his birth or adop¬
tion into a certain position in the social structure an individual
is connected with a large number of other persons. With some of
them he finds himself in a definite and specific jural relation, i.e.
one which can be defined in terms of rights and duties. Who
these persons will be and what will be the rights and duties depend
on the form taken by the social structure. As an example of
such a specific jural relation we may take that which normally
exists between a father and son, or an elder brother and a younger
brother. Relations of the same general type may be extended over
a considerable range to all the members of a lineage or a clan or
an age-set. Beside these specific jural relations which are defined
not only negatively but also positively, i.e. in terms of things that
must be done as well as things that must not, there are general
jural relations which are expressed almost entirely in terms of
prohibitions and which extend throughout the whole political
society. It is forbidden to kill or wound other persons or to take or
destroy their property. Besides these two classes of social relations
there is another, including many very diverse varieties, which can
perhaps be called relations of alliance or consociation. For
example, there is a form of alliance of very great importance in
many societies, in which two persons or two groups are connected
by an exchange of gifts or services (Mauss, 1927-8). Another
example is provided by the institution of blood-brotherhood
which is so widespread in Africa.
The argument of this paper has been intended to show that the
joking relationship is one special form of alliance in this sense. An
alliance by exchange of goods and services may be associated

112 Joking and Avoidance

with a joking relationship, as in the instance recorded by Pro¬
fessor Labouret (1929, p. 245). Or it may be combined with the
custom of avoidance. Thus in the Andaman Islands the parents
of a man and the parents of his wife avoid all contact with each
other and do not speak; at the same time it is the custom that
they should frequently exchange presents through the medium of
the younger married couple. But the exchange of gifts may also
exist without either joking or avoidance, as in Samoa, in the
exchange of gifts between the family of a man and the family
of the woman he marries or the very similar exchange between
a chief and his ‘talking chief’.
So also in an alliance by blood-brotherhood there may be a
joking relationship as amongst the Zande (Evans-Pritchard,
1933); and in the somewhat similar alliance formed by exchange
of names there may also be mutual teasing. But in alliances of
this kind there may be a relation of extreme respect and even of
avoidance. Thus in the Yaralde and neighbouring tribes of
South Australia two boys belonging to communities distant from
one another, and therefore more or less hostile, are brought into
an alliance by the exchange of their respective umbilical cords.
The relationship thus established is a sacred one; the two boys
may never speak to one another. But when they grow up they
enter upon a regular exchange of gifts, which provides the machin¬
ery for a sort of commerce between the two groups to which they
belong.
Thus the four modes of alliance or consociation, (1) through
intermarriage, (2) by exchange of goods or services, (3) by
blood-brotherhood or exchanges of names or sacra, and (4) by
the joking relationship, may exist separately or combined in
several different ways. The comparative study of these combin¬
ations presents a number of interesting but complex problems.
The facts recorded from West Africa by Professor Labouret and
Mademoiselle Paulme afford us valuable material. But a good
deal more intensive field research is needed before these problems
of social structure can be satisfactorily dealt with.
What I have called relations by alliance need to be compared
with true contractual relations. The latter are specific jural
relations entered into by two persons or two groups, in which
either party has definite positive obligations towards the other,

A. R. Radcliffe-Brown 113
and failure to carry out the obligations is subject to a legal
sanction. In an alliance by blood-brotherhood there are general
obligations of mutual aid, and the sanction for the carrying out
of these, as shown by Dr Evans-Pritchard,' is of a kind that can
be called magical or ritual. In the alliance by exchange of gifts
failure to fulfil the obligation to make an equivalent return for
a gift received breaks the alliance and substitutes a state of
hostility and may also cause a loss of prestige for the defaulting
party. Professor Mauss (1923-4) has argued that in this kind of
alliance also there is a magical sanction, but it is very doubtful
if such is always present, and even when it is it may often be
of secondary importance.
The joking relationship is in some ways the exact opposite of
a contractual relation. Instead of specific duties to be fulfilled
there is privileged disrespect and freedom or even licence, and the
only obligation is not to take offence at the disrespect so long as
it is kept within certain bounds defined by custom, and not to
go beyond those bounds. Any default in the relationship is like a
breach of the rules of etiquette; the person concerned is regarded
as not knowing how to behave himself.
In a true contractual relationship the two parties are conjoined
by a definite common interest in reference to which each of them
accepts specific obligations. It makes no difference that in other
matters their interests may be divergent. In the joking relation¬
ship and in some avoidance relationships, such as that between a
man and his wife’s mother, one basic determinant is that the
social structure separates them in such a way as to make many of
their interests divergent, so that conflict or hostility might result.
The alliance by extreme respect, by partial or complete avoidance,
prevents such conflict but keeps the parties conjoined. The
alliance by joking does the same thing in a different way.
All that has been, or could be, attempted in this paper is to
show the place of the joking relationship in a general comparative
study of social structure. What I have called, provisionally,
relations of consociation or alliance are distinguished from the
relations set up by common membership of a political society
which are defined in terms of general obligations, of etiquette, or
morals, or of law. They are distinguished also from the true con¬
tractual relations, defined by some specific obligation for each

114 Joking and Avoidance

 contracting party, into which the individual enters of his own
volition. They are further to be distinguished from the relations
set up by common membership of a domestic group, a lineage or
a clan, each of which has to be defined in terms of a whole set of
socially recognized rights and duties. Relations of consociation
can only exist between individuals or groups which are in some
way socially separated.
This paper deals only with formalized or standardized joking
relations. Teasing or making fun of other persons is of course a
common mode of behaviour in any human society. It tends to
occur in certain kinds of social situations. Thus I have observed
in certain classes in English-speaking countries the occurrence
of horse-play between young men and women as a preliminary to
courtship, very similar to the way in which a Cherokee Indian
jokes with his ‘grandmothers’. Certainly these unformalized
modes of behaviour need to be studied by the sociologist. For the
purpose of this paper it is sufficient to note that teasing is always
a compound of friendliness and antagonism.
The scientific explanation of the institution in the particular
form in which it occurs in a given society can only be reached by
an intensive study which enables us to see it as a particular ex¬
ample of a widespread phenomenon of a definite class. This
means that the whole social structure has to be thoroughly
examined in order that the particular form and incidence of
joking relationships can be understood as part of a consistent
system. If it be asked why that society has the structure that it
does have, the only possible answer would lie in its history. When
the history is unrecorded, as it is for the native societies of Africa,
we can only indulge in conjecture, and conjecture gives us
neither scientific nor historical knowledge.5

References
Boas, F. (1922), ‘Das Verwandtschaftssystem der VandaiT, Zeits.
Ethnolog., vols. 53-4.
Eg gan, F. (ed.) (1937), Social Anthropology of North American Tribes,
University of Chicago Press.

5. The general theory outlined in this paper is one that I have presented in
lectures at various universities since 1909 as part of the general study of the
forms of social structure. In arriving at the present formulation of it I have
been helped by discussions with Dr Meyer Fortes.

A. R. Radcliffe-Brown 115
Evans-Pritchard, E. E. (1933), ‘Zande blood-brotherhood’, Africa,
vol. 6, pp. 369-401.
Fortes, M. (1945), The Dynamics of Clanship among the Tallensi,
Oxford University Press.
Gilbert, W. H. (1937), ‘Eastern Cherokee social organisation’, in
F. Eggan (ed.), Social Anthropology of North American Tribes,
University of Chicago Press, pp. 285-338.
Hoernl£, A. W. (1925), ‘Social organisation of the Nama Hottentot’,
Amer. Anthropol., vol. 27, pp. 1-24.
Junod, H. (1927), The Life of a South African Tribe, vol. 1,
Macmillan. 2nd rev. edn.
Labouret, H. (1929), ‘La parente a plaisanteries en Afrique occidentale’,
Africa, vol. 2, pp. 244 ff.
Labouret, H. (1931), Les Tribus du Rameau Lobi, Institut d’Ethnologie,
Paris.
Landes, R. (1937), ‘The Ojibwa of Canada’, in M. Mead (ed.),
Co-operation and Competition among Primitive Peoples, McGraw-Hill,
pp. 87-126.
Mauss, M. (1923-4), ‘Essai sur le don’, Ann. sociol., vol. 1, pp. 30-186.
Mauss, M. (1927-8), Ann. Ecolepratique hautes Etudes: Section Sci. relig.
Mead, M. (1934), ‘Kinship in the Admiralty Islands’, Anthropol.
Papers Amer. Mus. nat. Hist., vol. 34, pp. 181-358.
Paulme, D. (1939), ‘Parents a plaisanteries et alliance par le sang en
Afrique occidentale’, Africa, vol. 12, pp. 433-44.
Pedler, F. J. (1940), ‘Joking relationships in East Africa’, Africa,
vol. 13, pp. 170 ff.
Radcliffe-Brown, A. R. (1924), ‘The mother’s brother in South
Africa’, South Africa J. Sci., vol. 21, pp. 542-55.
Roy, S. C. (1915), The Oraons of Chota Nagpur, Ranchi.

116 Joking and Avoidance

Part Five Marriage Transactions

The study of marriage transactions revolves around the major

distinction between bridewealth and dowry, though
bride-service, sister exchange and token gifts provide alternative
general forms. The typology is crude but the distinction is
important, as Radcliffe-Brown indicates (Reading 8). Bridewealth
is particularly characteristic of Africa, while dowry is found, in
varying forms, in all the major cultures in Europe and Asia.
While bridewealth tends to be associated with the possibility of
marrying more than one wife (polygyny), dowry is linked with
monogamy. Moreover, while bridewealth circulates wealth
throughout the society, in the exchange of rights over property
for rights over women, dowry is a kind of anticipated
inheritance whereby the bride receives her Tot’ or portion of the
familial estate at her marriage. The operation of the system of
dowry, which extends from Japan to Ireland, is well illustrated
in Friedl’s account of the Greek village of Vasilika. In many
parts of Europe, especially in Ireland, the dowry system is
linked to a late age of marriage and a low proportion of persons
who never get married. As Hajnal (Reading 10) points out in
his study of European marriage patterns, these features were
widely distributed in Western Europe.
8 A. R. Radc 1 iffe-Brown

Dowry and Bridewealth

Excerpt from A. R. Radcliffe-Brown, ‘Introduction’, in A. R.

Radcliffe-Brown and D. Forde (eds.), African Systems of Kinship and
Marriage, Oxford University Press for the International African Institute,
1950, pp. 1-85.

In order to understand the African customs relating to marriage

we have to bear in mind that a marriage is essentially a rearrange¬
ment of social structure. What is meant by social structure is any
arrangement of persons in institutionalized relationships. By a
marriage certain existing relationships, particularly, in most
societies, those of the bride to her family, are changed. New
social relations are created, not only between the husband and the
wife, and between the husband and the wife’s relatives on the one
side and between the wife and the husband’s relatives on the
other, but also, in a great many societies, between the relatives
of the husband and those of the wife, who, on the two sides, are
interested in the marriage and in the children that are expected
to result from it. Marriages, like births, deaths, or initiations at
puberty, are rearrangements of structure that are constantly re¬
curring in any society; they are moments of the continuing social
process regulated by custom; there are institutionalized ways of
dealing with such events.
We tend, unless we are anthropologists, to judge other people’s
customs by reference to our own. To understand African marriage
we must remember that the modern English idea of marriage is
recent and decidedly unusual, the product of a particular social
development. We think of a marriage as an event that concerns
primarily the man and woman who are forming a union and the
State, which gives that union its legality and alone can dissolve
it by divorce. The consent of parents is, strictly, only required for
minors. Religion still plays some part, but a religious ceremony
is not essential.

A. R. Radcliffe-Brown 119
 We may compare English marriage with the following account
of a ‘wedding’ in early England.

If people want to wed a maid or a wife and this is agreeable to her and
to her kinsmen, then it is right that the bridegroom should first swear
according to God’s right and secular law and should wage [pledge
himself] to those who are her forspeakers, that he wishes to have her
in such a way as he should hold her by God’s right as his wife - and his
kinsmen will stand pledge for him.
Then it is to be settled to whom the price for upfostering her belongs,
and for this the kinsmen should pledge themselves.
Then let the bridegroom declare what present he will make her for
granting his desire, and what he will give if she lives longer than he does.
If it is settled in this way, then it is right that she should enjoy half the
property, and all if they have a child, unless she marries another man.
All this the bridegroom must corroborate by giving a gage, and his
kinsmen stand to pledge for him.
If they are agreed in all this, then let the kinsmen of the bride accept
and wed their kinswoman to wife and to right life to him who desires
her, and let him take the pledge who rules over the wedding.
If she is taken out of the land into another lord’s land, then it is
advisable that her kinsmen get a promise that no violence will be done
to her and that if she has to pay a fine they ought to be next to help her
to pay, if she has not enough to pay herself.1

The marriage here is not any concern of the State or political

authorities; it is a compact between two bodies of persons, the
kin of the woman who agree to wed their daughter to the man,
and his kinsmen who pledge themselves that the terms of the
agreement will be carried out. The bridegroom and his kinsmen
must promise to make a payment (the ‘marriage payment’) to
her father or other legal guardian. He must also state what
present he will give to his bride for permitting the physical
consummation of the marriage; this was the so-called ‘morning-
gift’ to be paid after the bridal night. There was further an agree¬
ment as to the amount of the dowry, the portion of the husband’s
wealth of which the wife should have the use during her lifetime

1. Quoted by Vinogradoff, Outlines of Historical Jurisprudence, vol. 1,

p. 252, from Liebermann, Gesetze der Angelsachsen, vol. 1, p. 442. The
‘wedding’ was the agreement or contract entered into by the kinsfolk of
bride and bridegroom, equivalent to the Roman sponsalia, not the cere¬
mony of handing over the bride (the Roman traditio puellae).

120 Marriage Transactions

if her husband died before her. The agreement is concluded by
giving of the wed, the symbolic payment made by the bridegroom
and his kin to the woman’s kinsmen.
In modern England the pledge or gage, in the form of a
‘wedding’ ring, is given, not to the bride’s kinsmen when the
marriage arrangement is made, but to the bride herself at the
wedding ceremony. The change in custom is highly significant.
The ‘giving away’ of the bride is a survival of something which at
one time was the most important feature of the ceremonial of
marriage.
Thus in Anglo-Saxon England a marriage, the legal union of
man and wife, was a compact entered into by two bodies of kin.
As the Church steadily increased in power and in control of
social life, marriage became the concern of the Church and was
regulated by canon law. There was a new conception that in
marriage the man and woman entered into a compact with God
(or with His Church) that they would remain united till parted by
death. The marriage was under the control of the Church;
matrimonial cases were dealt with in the ecclesiastical courts.
At the end of the Middle Ages there came the struggle for
power between Church and State in which the State was, in
Protestant countries, victorious. Marriage then came under State
control. At the present day to legalize a union of man and wife
the marriage, whether there is or is not a religious ceremony
must be registered by someone licensed by the State and a fee
must be paid. It is the State that decides on what conditions the
marriage may be brought to an end by a divorce granted by a
court which is an organ of the State.
A most important factor in the development of the modern
English (and American) conception of marriage was the idea of
romantic love, a theme that was elaborated in the nineteenth
century in novel and drama and has now become the mainstay
of the cinema industry. In its early development romantic love
was conceived as not within but outside marriage, witness the
troubadours and their courts of love and Dante and Petrarch. In
the eighteenth century Adam Smith could write: ‘ Love, which was
formerly a ridiculous passion, became more grave and respect¬
able. As a proof of this it is worth our observation that no ancient
tragedy turned on love, whereas it is now more respectable and

A. R. Radcliffe-Brown 121
influences all the public entertainments.’ The idea that marriage
should be a union based on romantic love leads logically to the
view that if the husband and wife find they do not love one another
they should be permitted to dissolve the marriage. This is the
Hollywood practice, but conflicts with the control of marriage
by the Church or by the State.
Another very important factor has been the change in the
social and economic position of women during the nineteenth
and twentieth centuries. A married woman may now hold
property in her own right; she may take employment that has
no connection with her family life but takes her away from it. In
the marriage ceremony many women now refuse to promise that
they will obey their husbands.
Not only are marriage and ideas about marriage in England and
America the product of a recent, special and complex develop¬
ment, but there is good evidence that they are still changing. The
demand for greater freedom of divorce is one indication of this.
Yet it is clear that despite all this some people take twentieth-
century English marriage as a standard of ‘civilized’ marriage
with which to compare African marriage.
The African does not think of marriage as a union based on
romantic love although beauty as well as character and health
are sought in the choice of a wife. The strong affection that
normally exists after some years of successful marriage is the
product of the marriage itself conceived as a process, resulting
from living together and cooperating in many activities and
particularly in the rearing of children.
An African marriage is in certain respects similar to the early
English marriage described above. The dowry or dower does not
exist in Africa, though writers who do not know, or do not care
about, the meanings of words use the term ‘dowry’ quite
inappropriately to refer to the ‘marriage payment’.2 There is
also in Africa nothing exactly corresponding to the English
morning-gift’ regarded as a payment for accepting sexual
embraces, though it is usual for the bridegroom to give gifts to
his bride. The two other features of the early English marriage
2. Belgian and some French writers make a similar misuse of the term
dot , which is a woman s marriage portion of which the annual income is
under her husband’s control.

122 Marriage Transactions

 are normally found in African marriages. Firstly, the marriage
is not the concern of the political authorities but is established by
a compact between two bodies of persons, the kin of the man and
the kin of the woman. The marriage is an alliance between the
two bodies of kin based on their common interest in the marriage
itself and its continuance, and in the offspring of the union,
who will be, of course, kin of both the two kin-groups. The
understanding of the nature of this alliance is essential to any
understanding of African kinship systems. Secondly, in Africa
generally, as in early England, and in a great number of societies
in ancient and modern times in all parts of the world, a marriage
involves the making of a payment by the bridegroom or his kin
to the father or guardian of the bride. Africans distinguish, as
we do, between a ‘legal’ marriage and an irregular union. In
modern England a marriage is legal if it is registered by a person
licensed by the State. Only children born of such a union are
legitimate. But in Africa the State or political authority is not
concerned with a marriage. How then, are we to distinguish a
legal marriage ? The answer is that a legal marriage, by which the
children who will be born are given definite ‘legitimate’ status
in the society, requires a series of transactions and formalities in
which the two bodies of kin, those of the husband and those of the
wife, are involved. In most African marriages, as in the early
English marriage, the making of a payment of goods or services
by the bridegroom to the bride’s kin is an essential part of the
establishment of ‘legality’.
Some people regard payments of this kind as being a ‘ purchase’
of a wife in the sense in which in England today a man may
purchase a horse or a motor-car. In South Africa it was at one
time held officially that a marriage by native custom with the
payment of cattle (lobola) was ‘an immoral transaction ’ and not a
valid marriage. The Supreme Court of Kenya in 1917 decided
that ‘a so-called marriage by the native custom of wife-purchase
is not a marriage’. The idea that an African buys a wife in the
way that an English farmer buys cattle is the result of ignorance,
which may once have been excusable but is so no longer, or of
blind prejudice, which is never excusable in those responsible for
governing an African people.
A marriage in many, perhaps most, African societies involves a

A. R. Radcliffe-Brown 123
whole series of prestations3 (payments, gifts, or services), and
while the most important of these are from the husband and his
kin to the wife’s kin, there are frequently, one might say usually,
some in the other direction. One of the best accounts of the whole
procedure is that given by Father Hulstaert for the Nkundo of
the Belgian Congo (1938). The procedure begins with the present¬
ation, on the part of the future husband, of the ikulas at one time
an arrow, now two copper rings. The acceptance of this by the
woman and her kin constitutes a formal betrothal. The marriage,
i.e. the ‘ tradition ’ of the bride, may take place before any further
payment. At the marriage, gifts are made to the bride by the
parents of the bridegroom, by others of his relatives, and by the
bridegroom himself. The next step is the formal prestation of
the ndanjza. formerly a knife, to the bride’s father. It signifies that
the fiusband thereafter becomes responsible for accidents that
might befall his wife. In return there is a prestation from the
bride’s family to the husband and his family. This is part of the
nkomi, the payment that is made to the bridegroom by the wife’s
family. The marriage is not fully established until the husband
pays his father-in-law the waloj a substantial payment consisting
chiefly of objects of metal. After this the woman becomes fully
the man’s wife. When the walo is handed over the woman’s
family make a return payment {nkomi) and give a present of food
to the husband’s family. The husband must also make a special
payment to his wife’s mother and must give a considerable
number of presents to the father, mother, brothers and other
relatives of the bride. The relatives of the husband then demand
and receive presents from the wife’s family. The final payment to
be made by the husband is the bosongo, formerly a slave, now a
quantity of copper rings.
There is, of course, an immense diversity in the particulars of
prestations connected with betrothal and marriage in different
societies and in each case they have to be studied, with regard to
their meanings and functions, in relation to the society in which

3. Prestation is defined in the Oxford Dictionary as ‘the act of paying,

in money or service, what is due by law or custom’. The prestations with
which we are here concerned are all those gifts and payments of goods or
services which are required by custom in the process of establishing a valid
marriage.

124 Marriage Transactions

 they are found. For general theory, however, we have to look for
general similarities. In the first place it is necessary to recognize
that whatever economic importance some of these transactions
may have, it is their symbolic aspect that we chiefly have to
consider. This may be made clear by the English customs of the
engagement ring, the wedding ring and the wedding presents.
Though an engagement ring may have considerable value (more
than many Africans ‘ pay ’ for their wives), the giving of it is not
regarded as an economic or at least not as a business transaction.
It is symbolic.
In what follows the term ‘marriage payment’ will be used for
the major payment or payments made by the bridegroom to the
wife’s kin. Where there is a payment from the wife’s kin to the
husband (as in the Nkundo) this will be called the ‘counter¬
payment’. The rule in many African societies is that if there is a
divorce the marriage payment and the counter-payment must be
returned. There are qualifications of this; for example, in some
tribes where on divorce there are children and they belong to the
father the marriage payment may be not returnable, or returnable
only in part. Also, there are tribes in Africa in which, instead
of a payment in goods, the bridegroom must serve for his wife, by
working for her kin, just as Jacob served his mother’s brother
Laban seven years for each of the two sisters, Leah and Rachel,
his cousins, whom he married (Genesis, ch. 29). This service,
the equivalent of the marriage payment or of part thereof, is of
course not returnable if there is a divorce.
Let us return to the early English marriage. In the formulary
quoted above the marriage payment was called ‘the price of
upfostering ’ and was thus interpreted as a return to the father or
guardian of the expense of rearing a daughter. But in somewhat
earlier times the payment was differently interpreted. It was a
payment for the transfer of the woman’s mund from the father or
guardian to the husband, whereby the latter gained and the
former lost certain rights. The term for a legitimately married
wife in Old Norse law was mundi kjdbt, meaning one whose
mund has been purchased. In Sweden the transfer of mund was
not by purchase but by gift, and the expression for marriage was
giftarmal. In Roman law the marriage by coemptio, sometimes
called ‘marriage by purchase’, was not the sale of a woman but

A. R. Radcliffe-Brown 125
the legal transfer of mamis to her husband, and tnund and
manus are roughly equivalent terms. In these Roman and Teutonic
marriages the important point is that to legalize the union of a
man and a woman, so that it is really a marriage, legal power over
his daughter must be surrendered by the father and acquired by
the husband, whether the transfer be by gift or by payment.
The early English marriage was of this type.
In Africa an unmarried woman is in a position of dependence.
She lives under the control and authority of her kin, and it is
they who afford her protection. Commonly, if she is killed or
injured her guardian or her kinsfolk can claim an indemnity.
At marriage she passes to a greater or less extent, which is
often very considerable, under the control of her husband (and
his kin), and it is he (and they) who undertake to afford her
protection. (Note the ndanga payment amongst the Nkundo, by
which the bridegroom accepts responsibility for accidents that
may befall the bride.) The woman’s kin, however, retain the right
to protect her against ill treatment by her husband. If she is killed
or injured by third parties it is now the husband and his kin
who can claim an indemnity. It is this transfer of murid, to use the
Old English term, that is the central feature of the marriage
transaction.
To understand African marriage we must think of it not as an
event or a condition but as a developing process. The first step
is usually a formal betrothal, though this may have been pre¬
ceded by a period of courtship or, in some instances in some
regions, by an elopement. The betrothal is the contract or agree¬
ment between the two families. The marriage may proceed by
stages, as in the instance of the Nkundo mentioned above. A
most important stage in the development of the marriage is the
birth of the first child. It is through the children that the husband
and wife are united and the two families are also united by having
descendents in common.
We may consider African marriage in three of its most import¬
ant aspects. First, the marriage involves some modification or
partial rupture of the relations between the bride and her im¬
mediate kin. This is least marked when the future husband comes
to live with and work for his future parents-in-law while his
betrothed is still a girl not old enough for marriage. It is most

126 Marriage Transactions

 marked when, as in most African societies, the woman when she
marries leaves her family and goes to live with her husband and
his family. Her own family suffers a loss. It would be a gross
error to think of this as an economic loss.4 It is the loss of a
person who has been a member of a group, a breach of the family
solidarity. This aspect of marriage is very frequently given sym¬
bolic expression in the simulated hostility between the two
bodies of kin at the marriage ceremony, or by the pretence of
taking the bride by force (the so-called ‘capture’ of the bride).
Either the bride herself or her kin, or both, are expected to make a
show of resistance at her removal.
Customs of this kind are extremely widespread not only in
Africa but all over the world, and the only explanation that fits
the various instances is that they are the ritual or symbolic
expression of the recognition that marriage entails the breaking
of the solidarity that unites a woman to the family in which she
has been born and grown up. Ethnographical literature affords
innumerable instances. One example may be given here. In
Basutoland, or at least in some parts of it, on the day fixed for
the marriage the young men of the bridegroom’s group drive the
cattle that are to constitute the marriage payment to the home of
the bride. When they draw near, the women of the bride’s party
gather in front of the entrance to the cattle kraal. As the bride¬
groom’s party try to drive the cattle into the kraal the women, with
sticks and shouts, drive them away so that they scatter over the
veld and have to be collected together again and a new attempt
made to drive them into the kraal. This goes on for some time
until at last the cattle are successfully driven into the kraal. The
women of the group make a show of resistance at the delivery of
the cattle which will have as its consequence the loss of the bride.
The proper interpretation of these customs is that they are
symbolic expressions of the recognition of the structural change
that is brought about by the marriage.
When this aspect of marriage is considered the marriage pay¬
ment can be regarded as an indemnity or compensation given by
the bridegroom to the bride’s kin for the loss of their daughter.
4. This is the view of modern English law. If an unmarried woman is
seduced her father can recover damages for the loss of her ‘services’; as
though the only value attached to a daughter is as a servant.

A. R. Radciiffe-Brown 127
 This is, however, only one side of a many sided institution and
in some kinship systems is of minor importance. In societies in
which the marriage payment is of considerable value it is com¬
monly used to replace the daughter by obtaining a wife for some
other member of the family, usually a brother of the woman who
has been lost. A daughter is replaced by a daughter-in-law, a
sister by a wife or sister-in-law. The family is compensated for
its loss.
A second important aspect of legal marriage is that it gives the
husband and his kin certain rights in relation to his wife and the
children she bears. The rights so acquired are different in different
systems. Some of these are rights of the husband to the perform¬
ance of duties by the wife (rights in personam) and he accepts
corresponding duties towards her. He has, for example, rights to
the services of his wife in his household. But the husband usually
acquires rights in rem over his wife. If anyone kills or injures
her, or commits adultery with her, he may claim to be indemnified
for the injury to his rights.
The husband acquires his rights through an action by the wife’s
kin in which they surrender certain of the rights they have pre¬
viously had. The marriage payment may be regarded in this
aspect as a kind of ‘consideration’ by means of which the
transfer is formally and ‘legally’ made. It is the objective instru¬
ment of the ‘legal’ transaction of the transfer of rights. Once the
payment, or some specific portion of it, has been made the
bride’s family have no right to fetch their daughter back, and in
most tribes, if the union is broken by divorce at the instance of
the husband, the payment has to be returned and the woman’s
family recover the rights they surrendered.
The rights obtained by a husband and his kin are different in
some respects in different systems. The most important difference
is in the matter of rights over the children the wife bears. An
African marries because he wants children - liberorum auaeren-
mdorumgratiaA The most important part of the ‘value’ of a woman
is her child-bearing capacity. Therefore, if the woman proves to
be barren, in many tribes her kin either return the marriage
payment or provide another woman to bear children.
In a system of father-right, such as the Roman nalria potestas.
the rights of the father and his kin over the children of a marriage

128 Marriage Transactions

are so preponderant as to be nearly absolute and exclude any
rights on the part of the mother’s kin. On the other hand, in a
system of mother-right such as that formerly existing amongst
the Nayars of southern India, the father has no legal rights at all:
the children belong to the mother and her kin. This does not, of
course, exclude a relationship of affection between father and
child. Both father-right and mother-right are exceptional con¬
ditions; most societies have systems which come between these
extremes and might be called systems of joint right or divided
right. The system of division varies and there may be an approxi¬
mation either to father-right or to mother-right.
Some societies in Sumatra and other parts of the Malay
Archipelago have two kinds of marriage. If a full marriage
payment is made the children belong to the father; we may call
this a father-right marriage. But if no payment is made the child¬
ren belong to the mother and her kin, the marriage being one of
mother-right.
The same sort of thing is reported from some parts of Africa,
for example from Brass in Southern Nigeria (Talbot, 1926,
pp. 437-40). The father-right marriage, with a substantial
marriage payment, is the usual form, but if only a small payment
is made the children belong to the mother’s kin. The most
definite example is from the Nyamwezi. In the kukwa form of
marriage there is a payment (nsabg) made by the bridegroom to
the father or guardian of the bride; children of such a marriage
fall into the possession of the husband and his agnatic kin. In the
butende form of marriage there is no payment and the children
felong to the mother and her kin.
There is another aspect of marriage that must be taken into
account. In Africa a marriage is not simply a union of a man and
a woman; it is an alliance between two families or bodies of kin.
We must consider the marriage payments in this connection also.
In so-called primitive societies the exchange of valuables is a
common method of establishing or maintaining a friendly relation
between separate groups or between individuals belonging to
separate groups. Where material goods are exchanged it is
common to speak of gift-exchange. But the exchange may be of
services, particularly those of a ritual character. There are
societies in which there is an exchange of women, each group

T-K-E A. R. Radcliffe-Brown 129

(family, lineage or clan) providing a wife for a man of the
other. The rule governing transactions of this kind is that for
whatever is received a return must be made. By such exchanges,
even by a single act of exchange, two persons or two groups are
linked together in a more or less lasting relation of alliance
(Mauss, 1923^1).
There are societies in some parts of the world in which the
marriage payment and the counter-payment are equal or approxi¬
mately equal in value. We may regard this as an exchange of gifts
to establish friendship between two families, of which the son of
one is to marry a daughter of the other. The kind, and to some
extent the amount of the gifts is fixed by custom. But where the
marriage payment is considerable in amount and there is a much
smaller counter-payment, or none at all, we must interpret this as
meaning that the bride’s family is conferring a specific benefit
on the bridegroom by giving him their daughter in marriage, a
benefit that is shared by his kin, and that the marriage payment
is a return for this. The transaction can still be regarded as a form
of ‘ gift-exchange ’ and as such establishes a relation (of alliance)
between the parties.
It is characteristic of a transaction of purchase and sale that
once it has been completed it leaves behind no obligations on either
the buyer or the seller. (This does not, of course, exclude claims
based on warranty.) In an African marriage the position is very
different. For one thing the marriage payment may in certain
circumstances have to be repaid. In some tribes where the pay¬
ment consists of cattle it is the same cattle with all their increase
that should be returned. Further, in some African societies the
family that has made the marriage payment continues to have an
interest in the cattle or other goods of which it consists. The pay¬
ment received for a woman’s marriage may be used to obtain a
wife for a member of her family, usually her brother. This sets up
a number of important relations between the persons involved.

A = b B = c C = d

B and b are brother and sister, and so are C and c. A marries b and
makes a marriage payment which is used to obtain a wife (c) for
B. In various tribes the marriage payment establishes a series of

130 Marriage Transactions

special personal relations between b and B, between A and B,
between b and c, and between A and c. These are defined differently
in different tribes. We may briefly consider three varieties.
It is usual to speak of B and b as ‘linked’ brother and sister, and
B is the ‘linked’ mother’s brother of the children of A and b,
while b is the ‘linked’ father’s sister of B's children. In the
Shangana-Tonga tribes there is a very special relation between
A and his ‘great mukonwana’ c, the wife that B married with the
payment provided by A. A can claim in marriage a daughter of c,
particularly if his wife b dies and there is no younger sister to take
her place (Junod, 1927, pp. 231 ff.).
In the Lovedu the relations between the families of A, B and C
ought to be continued in the next and succeeding generations. A
son of A should marry a daughter of B and a son of B should
marry a daughter of C, There is thus established a chain of con¬
nected families. The B family (or lineage) gives brides to and re¬
ceives cattle from the A family and gives cattle to and receives
brides from C. The linked sister b is said to have ‘ built the house
for her brother ’ B, and she ‘ has a gate ’ by which she may enter the
house. She has the right to demand a daughter of the house to
come as her daughter-in-law, to marry her son and be her helper.
Thus, in this tribe, cross-cousin marriage is systematized in terms
of marriage payments, and a complex set of relations between
persons and between families is created (Krige and Krige, 1943;
J. D. Krige, 1939). In the Shangana-Tonga tribes b can demand a
daughter of c as her co-wife or ‘helper’, the wife of her husband,
not as her daughter-in-law (Earthy, 1933).
Amongst the Nkundo the relationships are given a different
form. There is a special relation of b to c, the wife of her linked
brother whose marriage was provided for by her marriage pay¬
ment. The sister b is the nkolo of c, who is her nkita. The nkolo (b)
stands in a position of superiority to the nkita (c). This relation is
continued in the succeeding generations; the children of b (the
nkolo) are in a position of superiority to the children of c. This is
connected with a peculiar ordering of relations amongst the
Nkundo by which the relation between cross-cousins is an
asymmetrical one in which they are treated as if they belonged to
different generations. The children of the father’s sister are
‘fathers ’, male and female (baise), to their cousins, the children of

A. R„ Radcliffe-Brown 131
the mother’s brother who are their ‘ children ’ (Jbana). The * children ’
must show respect to their ‘fathers’ and help them. As a conse¬
quence a man regards the son of his father’s sister’s son as his
‘brother’ and uses that term for him (Hulstaert, 1938, pp. 164 ff.).
It should now be evident that the marriage payment is a com¬
plex institution having many varieties in form and function. In
any given society it has to be interpreted by reference to the whole
system of which it is apart. Nevertheless, there are certain general
statements that seem to be well grounded. In Africa the marriage
payment, whether it be small or large, is the objective instrument
by which a ‘legal’ marriage is established. In some instances it
is a compensation or indemnity to the woman’s family for the
loss of a member. This is particularly so where the marriage pay¬
ment is considerable and is used to obtain a wife for the woman’s
brother. The payment may in some instances be regarded as part
of an exchange of a kind that is used in many parts of the world to
establish a friendly alliance between two groups. In some societies
of South Africa and the Nilotic region it is the derivation of the
cattle used in the marriage payment that fixes the social position
of the children bom of the union. Where the same cattle or other
goods are used in two or more successive marriages this is in some
tribes held to establish a special relation between the families thus
formed. Where cattle are sacred in the sense that the cattle of a
lineage are the material link between the living and their ancestors
(having been received from those ancestors and being used for
sacrifices to them), the use of cattle in marriage payments has a
significance which a transfer of other goods would not have. This
is not intended as a complete survey, which would be impossible
within the limits of this essay. It is only an indication of how this
institution, which is the procedure by which a husband acquires
those rights which characterize a legal marriage (rights that vary
in different societies), may be elaborated in different ways.

References
Earthy, E. D. (1933), Valenge Women, Oxford University Press.
Hulstaert, R. P. G. (1938), Le Mariage des Nkundo, Inst. roy. colon,
belg., M&moires, vol. 8, ch. 2.
Junod, H. (1927), The Life of a South African Tribe, Macmillan. Second
rev. edn.

132 Marriage Transactions

Krige, J. D. (1939), ‘The significance of cattle exchanges in Lovedu
social structure’, Africa, vol. 12, pp. 393-424.
Krige, E. J., and Krige, J. D. (1943), The Realm of a Rain Queen, Oxford
University Press.
Mauss, M. (1923-4), ‘Essai sur le don’, Ann. sociol., vol. 1, pp. 30-186.
Talbot, P. A. (1926), The Peoples of Southern Nigeria, vol. 3., Oxford
University Press.

A. R. Radcliffe-Brown 133
9 E. Friedl
Dowry, Inheritance and Land-Tenure

Excerpt from E. Friedl, Vasilika: A Village in Modern Greece, Holt,

Rinehart & Winston, 1962, pp. 64-8.

The discussion of the dowry as a mechanism of inheritance for

children of farmers who remain farmers may now be summarized.
The dowry is part of a system in which children receive property
through the parents of both their fathers and their mothers.
Property from two sources merges in each generation and is
redistributed in the next generation. Although patrilineally
inherited lands (those passed from a father to his sons in each
generation) have some continuity in space and have some con¬
tinuity of ownership in the male line, such lands always constitute
only one part of the total holdings of any particular elementary
family. Therefore, in Vasilika and its vicinity, the pattern of land
and money circulation through inheritance has two facets: one
portion that is patrilineally inherited straight down the
line of males; a second, distributed at marriage, that eventually
circulates among unrelated elementary families. The system can¬
not be described as one in which women inherit from women, in
spite of the legal residual control of a woman over her dower
properties, because brothers have equal rights with their sisters to
their mother’s dower lands. In Vasilika, at least, there is neither an
explicit nor an implicit pattern of giving daughters only dower
property and sons only patrilineally inherited lands.
The combination of practices including the function of the dowry
as inheritance, land as a major form of property, and village
exogamy have some further consequences for the relation of
property to groups of kinsmen in the Boeotian countryside. In spite
of patrilocal residence and some patrilineally inherited land, the
descendants of a male line are not associated with any particular
landed estates. Since no man owns or farms exactly the same hold¬
ings as his father farmed before him, nor the same holdings his
brothers have, and since he himself will expect to work different

134 Marriage Transactions

lands in different communities even in the course of his own adult
life, the permanent association of certain estates with certain
lineages is obstructed (Friedl, 1959a.) Moreover, since dowries
move down the generations and not across to a man of one’s own
generation who might use the property to marry off one of his
daughters, there is no economic advantage to brother-and-sister
exchange marriages. This situation is congruent with both the
Greek Orthodox Church’s and the Civil Code’s (Section 1357)
prohibitions on marriages between sisters and brothers-in-law and
between cousins to the third degree. This rule, as well as the bilat¬
eral reckoning of kin in Greece (Section 1356), prevents the trans¬
fer of property at marriage from resulting in either a series of equal
exchanges between two sets of kin groups or in a regular pattern of
circulation through several generations among particular sets of
such groups. The control of property establishes social links
between a man and his wife’s relatives not only in the ways we have
mentioned, but because his wife’s parents and brothers and sisters
have a legal and customary right to be consulted before the final
alienation of dower property (Section 1418). These links last only
one generation, however, so that in each generation a new network
of relations between elementary families in the neighborhood
develops, clustering around the management of dower properties.
These matters are important because they reveal some of the eco¬
nomic and social structure of Vasilika and its neighborhood that is
congruent with the type of relationships people, both kin and non¬
kin, have with each other.
One other situation occurs frequently enough in Vasilika to be
worth mentioning. When a farmer has no sons to inherit his land
and his house, but has a daughter, the father may acquire a man’s
assistance on the farm by importing a son-in-law. Such a man is
called a soghambros in Vasilika; he is a husband who moves into
his wife’s"householcfTnstead of vice versa. The importation of a
soghambros is also a solution to the problem of a young widow who
has no grown sons to run the farm left by her husband and arranges
a second marriage for herself with a man who is willing to work her
first husband’s holdings and her original dower lands. Since a
soghambros is always a manager of, and laborer on, property
belonging to others and is expected to bring no property of his own
with him, there is a slight social stigma attached to the status in

E. Friedl 135
Vasilika. The villagers say also that a soghambros is not ‘master in
his own house’. Certainly, the lot of a man who is not a master in
his own house and comes as a stranger to a village where most of
the other men have known each other all their lives is not a happy
one. But a soghambros is not really dishonored by his position;
the villagers, in this situation, as in so many others, recognize the
practical necessities which brought about the arrangement and do
not strongly condemn a man for making the best of his difficulties.
The dowry in Vasilika enters into the life of the villagers in other
ways besides inheritance at marriage. It has long served as a means
of upward social mobility for girls. Since the marriage of a daughter
is among the most important obligations of parents, the dowry
comes into the consciousness of the villagers more often as a
property requirement for marrying off their daughters than as a
means of transmitting inheritance. When, in addition, the high
value placed on upward social mobility is translated into an effort
to find urban sons-in-law for one’s daughters, the dowry emerges as
a mechanism for increasing the social prestige of the family.
Farmers are willing to give larger dowries in exchange for the
great satisfaction they derive from having town sons-in-law. In
Vasilika, the education of some sons has released land to add to the
girls’ dowries. Improved agricultural income has also enabled the
farmers to give larger dowries. In the decade ending in 1959, every
marriage of a Vasilika daughter whose father was in the upper half
of the village’s income range has been one with a man of respectable
occupation who lived in a provincial town or in Athens. The hus¬
bands are tailors, small retail store owners, or civil service workers.
One young man is a photographer, another is a gymnasium pro¬
fessor who has become the principal of his gymnasium. In 1961,
however, the son of a rather prosperous village farmer became
attracted to a daughter of another Vasilika farmer with good land
holdings. The young man asked his father to arrange the marriage
for him, and, since all the conditions were entirely suitable, the
young couple were engaged.
The trend in favor of town husbands prevails, however, among
most of those who can afford it. Once a few girls had married urban
men, the rivalry between village families led to greater efforts to
secure town husbands for the others. Consequently, there has been
an inflation in dowries which is alarming the villagers themselves.

136 Marriage Transactions

In the early 1950s, a prika worth $3000 was enough for a town
husband of no special prestige; by the late 1950s, the same kind of
man was asking for one worth $4500.
The inflation has had several consequences. Often the value of
the dowry can no longer be limited to the share of the inheritance to
which a girl is entitled. Farming sons are willing to give up some
portion of their shares so that their sisters can ‘live well’, as they
put it. The brothers gain also from the added prestige and influence
of the family. These in turn may make it possible for them to find a
girl with a larger dowry than their property qualifications might
warrant. A farmer so situated may make higher demands on the
ground that his and his wife’s children will have an urban aunt and
uncle. The town sister might provide board for her nephews and
nieces (girls are increasingly being sent to gymnasium) while they
are in school and can also be expected to help them find jobs.
Another consequence of the dowry inflation is the increasingly
late age of marriage for the village girls and the town men. It takes
farmers longer to accumulate the larger amounts of cash for the
dowry, and the prospective grooms longer to attain a position or
income at least partially commensurate. Vasilika’s girls who marry
town men are usually between twenty-five and thirty. Those whose
fathers have few land holdings, and consequently can offer only
small dowries, have been marrying farmers from other villages or
within Vasilika itself and have been in their early twenties at the
time of the engagement.
Now let us consider the disposition of the dowry which goes with
a girl marrying a town husband. First, if he accepts land as part of
the dowry, he will of necessity have it worked by his wife’s male
relatives. In time, however, often by the end of the first decade of
the marriage, the son-in-law may wish to expand his shop, or may
wish to start building a house for himself or for his daughter’s
dowry, or may hear of a good investment opportunity. He will
want money for these purposes, and so he begins, with his wife’s
consent and after consultation with her father, to arrange for the
sale of dowry lands. Since the 1950s, prospective town grooms have
been less willing to take land which they know they will eventually
want to sell, and have been asking for cash or for a house in town.
Anticipating this situation, several of Vasilika’s farmers who have
small daughters have begun to use their savings and even to sell a

E. Friedl 137
little land in order to build houses in Athens or in a provincial
town so that by the time their daughters are ready to marry, the
houses will provide the main portion of the dowry. Rents are high
and earnings relatively low in cities so that houses are a good in¬
vestment. However, since building materials must be bought, and
contractors paid, in cash, the houses may take many years to build.
The young couple may then move into the habitable shell, and the
rest of the dwelling will be completed out of the farmer’s then
current income. This kind of dowry-on-the-installment-plan
becomes part of the marriage contract. It is often a source not only
of continuing long-term discussions between a farmer and his town
son-in-law, but also a source of quarrels - what in Greece are called
fasaries.
The rate of movement of farmers’ daughters into the towns and
cities has been accelerating in the last decade, but marriages to
urban husbands are not a new phenomenon. Between 1930 and
1950, at least five of Vasilika’s women married ‘into Athens’.
This type of marriage, like those discussed above in which both
bride and groom are members of farm families, accelerates land
sales and exchanges. The network of association which develops
between Vasilika’s residents and their relatives in Athens and the
towns, is, however, perhaps the most important consequence of
the urban marriages of village women (Friedl, 1959). In addition
to the economic reasons for continuing relationships, there are
several customary patterns which increase the frequency of con¬
tacts between the two groups. Lonely village mothers are grateful
for visits from their married daughters and their grandchildren,
and the midsummer season not uncommonly finds city women
with their children back in the village for Easter, sometimes for
most of the Holy Week as well, is a well-known pattern in Greece.
Paniyiri in one’s patridhd (home village) is another occasion for
visiting. *
Journeys in the opposite direction also occur with some fre¬
quency. Village men and women, when ill, may enter hospitals in a
town in which they have sons or daughters, or brothers or sisters.
Village men visit their town brothers and bring their children;
occasionally a child spends a summer with his town aunt or uncle.
Visits to the village or to the town usually last several days. They
are made possible by still another congruent pattern of Greek

138 Marriage Transactions

culture. Neither the villagers nor their town relatives seem to have
any strong need for personal privacy. Visiting relatives are bedded
down on pallets when there are not enough regular beds, and it is
not considered either indecent or especially uncomfortable for one
or even two families to sleep in the same room. It is through this
inter-visiting process that so many urban traits of culture are
introduced to the villagers and, indeed, that some village patterns
are conserved in urban areas.
An additional consequence of the town marriages of village
girls is that rural wealth derived from land flows into the cities.
This wealth is an addition to what the productivity of land normally
contributes to urban centers in the form of food, taxes and the
export of rural produce. A portion of the farmer’s profits, through
the dowry, is being used, it would seem, directly to support a part
of the urban population. When the dowries are invested in housing
or in small commercial enterprises, many low-salaried employees,
civil service workers, or economically marginal entrepreneurs find
it possible to support themselves and their families in the town and
cities. Without the aid of dower wealth, they might not be able to
do so.

References
Friedl, E. (1959), ‘The role of kinship in the transmission of national
culture to rural villages in mainland Greece’, Amer. Anthropol.
vol. 61, pp. 30-38.
Friedl, E. (1959a), ‘Dowry and inheritance in modern Greece’,
Trans. New York Acad. Sci. vol. 22, pp. 49-54.

E. Friedl 139
10 J. Hajnal
European Marriage Patterns in Perspective

Excerpt from J. Hajnal, ‘European marriage patterns in perspective’,

in D. V. Glass and D. E. C. Eversley (eds.), Population in History, Edward
Arnold, 1965, pp. 101-43.

The main theme of this paper is not new. It is one of the main
topics of Mai thus’ Essay (1970) and indeed implicit in its very
structure (especially in the revised version of the second edition).
Malthus devoted Book I of his Essay to ‘the checks to population
in the less civilized parts of the world and in past times and Book
II to ‘the different states of modern Europe’. In Europe he traces
again and again the workings of the ‘preventive checks of moral
restraint’ which implies ‘principally delay of the marriage union’
and he contrasts the condition of Europe with that of the peoples
described in Book I.
Was Malthus right in thinking that late marriage in Europe
resulted in lower birth rates, and hence lower death rates, than
obtained among non-European populations? Whatever the
nature of the causal connection, his notions about the levels of
birth and death rates gain some support from modem research.
European birth rates, so far as we can tell, were rarely over 38
before the spread of birth control; in underdeveloped countries,
they are almost always over 40 and often over 45. So far as
mortality is concerned, the contrast is less clear cut, but there
seems no record in European experience since the eighteenth
century of conditions such as those in India or Formosa in the
initial decades of the twentieth century.
The way in which a non-European marriage pattern goes with
non-European birth and death rates may be illustrated by a
recent study by Csoscan (1959) of the parish registers for three
Hungarian villages in the eighteenth century. This population is
not in ‘Europe’ as defined for this paper. The distribution of
marriages by age of bride in 1770-1800 was quite definitely ‘non-
European’, as the following figures show:

140 Marriage Transactions

Age group Per cent distribution1 of
Bridegrooms Brides

Under 20 11 52
20-24 48 25
25-29 14 9
30-39 15 9
40 and over 12 5
Total 100 100

In the same period (1770-1800) the crude birth rate in these

Hungarian villages was 52 per 1,000 and the death rate was 43 per
1,000. This may be compared, for example, with the following
figures for the French village of Crulai given in the study by
Gautier and Henry (1958):

Period Birth rate Death rate

1675-1749 36 31
1750-89 31 28

The marriage data for Crulai are, of course, quite clearly

‘European’. The eighteenth-century Hungarian villages are thus
non-European in all three respects (age at marriage, birth rate,
death rate) in contrast with the European levels of Crulai’s vital
rates, as well as its marriage patterns.
There was a widespread conviction among eighteenth-century
authors that European conditions were fundamentally different
not only in marriage, birth and death rates, but above all in
standards of living, from those obtaining elsewhere in the world.
Europeans, a large proportion of them, not just the rich, had better
housing, better clothing, a greater variety of food, more furniture
and utensils, than people elsewhere. This uniqueness of Europe, so
evident to contemporaries, has been largely ignored in recent
discussions of economic development; all that is pre-industrial,
including eighteenth-century Europe, is often lumped together in
1. These percentages are based on 440 cases for men and 442 for women.
The number of marriages in which the age was not recorded was, surprisingly
for this period, very small. The figures are from p. 106 of Csoscdn’s paper.
Remarriages are included and they form a high proportion of the total.

J. Hajnal 141
generalizations about ‘agricultural’ or ‘peasant’ or ‘under¬
developed ’ societies.
Presumably the uniqueness of Europe in standards of living and
in death rates did not extend back beyond the seventeenth century
(except in limited regions). But if European death rates were as
high as in other parts of the world, could birth rates have been
lower? And if the European birth rate before the seventeenth
century was as high as elsewhere, does this not imply that European
women must have married young as in other populations of
high fertility? These large and vague questions need to be broken
up and investigated by careful calculations on the interrelation¬
ships between marriage, birth and death rates. Even the rela¬
tion between marriage patterns and crude birth rates is not
nearly as obvious as is often supposed and it is not independent of
mortality.
An inquiry into the origins of the European marriage pattern
will inevitably take one into fundamental issues of economic and
social history. This is so not only because of the connections just
discussed between marriages and births and deaths. There are
other links. A marriage almost by definition requires the estab¬
lishment of an economic basis for the life of the couple and their
children. The arrangements current in a society for achieving
this must fit in with the marriage pattern: they will shape it and
will be in turn influenced by it. Unmarried men and women must
be attached to households in some way, or form independent
households. The structure and size of households and the rate of
formation of new households and disappearance of old ones,
therefore, depend on the marriage pattern. In societies where
the household is the principal unit of economic production as well
as consumption, all this means that the marriage pattern is tied
in very intimately with the performance of the economy as a
whole. The emotional content of marriage, the relation between
the couple and other relatives, the methods of choosing or allo¬
cating marriage partners - all this and many other things cannot
be the same in a society where a bride is usually a girl of sixteen
and one in which she is typically a woman of twenty-four. These
things are perhaps obvious, but they have not been much explored,
at least not in histories which trace the emergence of modern
Europe. A full explanation of the background of European

142 Marriage Transactions

marriage patterns would probably lead into such topics as the
rise of capitalism and the protestant ethic.
The economic system influences the marriage pattern through
the arrangements by which the economic basis for the support of
a couple and their children is established. It is equally true that
the marriage pattern influences the economic system. The tradi¬
tional argument, that late marriage retarded population growth,
has already been mentioned but other possible effects need to be
explored. In the European pattern a person would usually have
some years of adult life before marriage; for women especially
this period would be much larger than outside Europe. It is a
period of maximum productive capacity without responsibility
for children; a period during which saving would be easy. These
savings (e.g. by means of the accumulation of household goods in
preparation of marriage) might add substantially to the demand
for goods other than the food etc. required for immediate survival.
In this respect delayed marriage may be similar to income in¬
equality in stimulating the diversion of resources to ends other
than those of minimum subsistence; but when later marriage is the
norm the total volume of demand generated might be much
larger than that which can be caused by a small class of wealthy
families in a population at subsistence level.2 Could this effect,
which was uniquely European, help to explain how the ground¬
work was laid for the uniquely European ‘take-off’ into modern
economic growth?
If late marriage brings about wealth, wealth may equally cause
late marriage. It was suggested in the eighteenth century (for ex¬
ample by Cantillon) that people married late because they insisted
on a certain standard of living (a standard varying with the social
position of the individual) as a prerequisite of marriage. More
simply, men marry late because they cannot ‘afford’ to marry
young; they have to wait until they have a livelihood, a farmer till
he acquires land, an apprentice till he finishes his apprenticeship
and so on.
It is tempting to see in this feature a key to the uniqueness of the
European marriage pattern. In Europe it has been necessary for a
2. The mere presence in the labour force of a large number of adult women
not involved in child bearing or rearing must have been a considerable
advantage to the eighteenth-century European economies.

J. Hajnal 143
man to defer marriage until he could establish an independent
livelihood adequate to support a family; in other societies the
young couple could be incorporated in a larger economic unit,
such as a joint family.3 This, presumably, is more easily achieved
and does not require such a long postponement of marriage. This
line of argument seems especially convincing if the larger economic
unit is such that extra labour is often felt to be an economic asset.
A system of large estates with large households as in Eastern
Europe might thus be conducive to a non-European marriage
pattern, while small holdings occupied by a single family and-
passed on to a single heir would result in a European pattern. If
this reasoning has substance, the uniqueness of the European
marriage pattern must be ascribed to the European ‘stem-
family’.4 (The term ‘stem-family’ was coined by Le Play in
describing the type of family organization in which land descends
to a single heir, the other sons going elsewhere.) This explanation
calls attention to a force which may have helped to bring about the
European marriage pattern, if it did not exist in the Middle
Ages. If men had to wait till land became available, presumably a
delay in the death of the holders of land resulting from declining
death rates would tend to raise the age at marriage. Whether there
was, in fact, a decline in mortality over the relevant period is a
3. The young Due de la Rochefoucauld, on his visit to England, makes a
similar point: ‘ Perhaps another reason for this [i.e. late marriage in England]
is because it is usual to set up house immediately after marriage. The young
couple never stay with their parents and they must be sensible enough to
avoid extravagance both in their conduct and in their expenditure.’
4. The extent to which generalizations can justifiably be made about the
family system in parts of Europe at particular times (let alone about the
whole of Europe or all non-European societies) is totally beyond my com¬
petence. There have been large estates and joint families in some regions
of Western Europe in the Middle Ages and beyond. Presumably it would be
possible to have a system in which each couple is in principle an independent
economic unit, but in which early marriage is made possible by arrange¬
ments to provide for the couple until they achieve complete independence.
The study by Katz (1959) mentions such arrangements among Jews in
Eastern Europe. Joint families can perhaps be regarded as fulfilling the
function of such transitional arrangements. In countries where in theory
joint families prevail the average size of household is not large. Households
consisting of several nuclear families may not remain long in this condition,
but this arrangement makes it possible for young couples to be part of a
larger unit at the beginning of marriage for some years.

144 Marriage Transactions

dubious point (the decline in question must have occurred before
the eighteenth century); but this is certainly a hypothesis that
merits study.
The connection between the death of the holder of land and its
availability for the founding of a new family is, however, rather an
indirect one. Under the mortality conditions of the Middle Ages
fathers often died while their children were very young; interim
arrangements had to be made till the son was old enough to take
over. Even if the father survived to old age, it does not follow that
a young family could not be set up on the holding until he died.
Homans (1941) in his book on thirteenth-century England
describes many instances where a father made over the land to his
son while he lived, thus permitting the latter to get married. He
also mentions instances where a father, while he lived, turned
over his holding to be shared between two sons, where a man
transferred his holding to someone other than his son, etc. To
understand the effect on the frequency of marriage and age of
marriage of a rule that a man must acquire land before marrying
we should have to know the frequency of the various arrangements
by which land was passed on. The rate at which land became
available for the founding of new families may have been con¬
trolled not so much by death as by social arrangements. It is not at
all clear a priori how a rule that a man must have a livelihood
before marrying would operate to produce just such a postpone¬
ment as is in fact observed. Even if we understood how the age at
marriage of men was determined at a given period it would still
need to be explained how women’s age at marriage was affected.
The uniqueness of the European pattern lies primarily in the high
age at marriage of women (often with a relatively small difference
between the age of husband and wife), rather than in a high age at
marriage for men.
There is no space for further speculation in the causes or
consequences of the European marriage pattern. The primary con¬
cern of this account has been the mere existence of the pattern.
This aspect should be kept distinct from the search for explanations.
It has been shown (1) that the distinctively European pattern
can be traced back with fair confidence as far as the seventeenth
century in the general population; (2) that its origins lie somewhere
about the sixteenth century in several of the special upper class

J. Hajnal 145
groups available for study and in none of these groups was the
pattern European before the sixteenth century; (3) the little frag¬
mentary evidence which exists for the Middle Ages suggests
a non-European pattern, as do scraps of information for the
ancient world.
Some at least of the data presented have probably been mis¬
interpreted. In dealing with sources of a type of which one has no
experience coming from remote periods of whose historical back¬
ground one is ignorant, one is very likely to make mistakes. In an
effort to survey so great a variety of materials, some of them could
only be looked at superficially. Even if individual pieces of inform¬
ation have been soundly interpreted, there remains the problem
how far generalizations can properly be based on isolated demo¬
graphic facts. This is a basic problem of much of historical demo¬
graphy. We wish to draw conclusions about the demography of
large groups. The terms in which questions are posed (like the
distinction between European and non-European marriage
patterns) are based on modern statistics for whole countries; but
the historical data often relate to small groups such as one village.
To what extent are conclusions from such data to larger units justi¬
fied? How far are statistics of particular groups likely to deviate
from those of larger populations of which they form a part (aside
from sample fluctuations)? Are data likely to be systematically
misleading because they do not relate to a closed population?
In spite of these and other difficulties, there seem to be good
prospects of obtaining substantial further information on the
origins and spread of the European marriage pattern. The
distinction between European and non-European patterns is
substantial so that no very refined measuring instrument is
required for its detection. There is probably a good deal of material
for the seventeenth and even the sixteenth century. The parish
registers offer a large mine of information waiting to be exploited.
If it were indeed to prove the case that in the Middle Ages the
marriage pattern of Europe was entirely ‘non-European’, traces
of the transition should be visible in some of the early parish regis¬
ter materials. Even for the Middle Ages there seems hope that
various types of records (for example manorial extents5) if carefully
5. The recent study by Hallam (1958), which came to my attention after this
paper was written, suggests that promising material may be awaiting analysis.

146 Marriage Transactions

handled may yield useful information. If the recent rate of output
of studies in historical statistics is maintained, and if those who
engage in such work keep their eyes open for information on
marriage, the mystery of the origins of European marriage pat¬
terns may be cleared up.

References
CsoscAn, J. (1959), ‘Harom Pest megyei falu nSpesedese a XVIII szazad
masodik feleben’, Tortenelmi Statisztikai Kozlemenyek, vol. 3, nos.
1-2, pp. 58-107. ‘Demographic changes in three villages in the
county of Pest in the second half of the 18th century’, Publications
of Historical Statistics, Budapest.
Gautier, £., and Henry, L. (1958), La Population de Crulai,
Paroisse Normande, Institut National d’fitudes Demographiques:
Travaux et Documents, Cahier no. 33.
Hall am, H. E. (1958), ‘Some thirteenth century censuses’, Econ.
Hist. Rev., vol. 10, no. 3, pp. 340-61.
Homans, G. C. (1941), English Villagers of the Thirteenth Century, Harvard
University Press.
Katz, J. (1959), ‘Family kinship and marriage among Ashkenazim in the
sixteenth to eighteenth centuries’, Jewish J. Sociol., vol. 1, no. 1,
pp. 3-22.
Malthus, T. R. (1970), An Essay on the Principle of Population, Penguin.
First published 1798.

J. Hajnal 147
Part Six Plural Marriage

Plural marriage may take a variety of forms, depending on

whether it is the husband (polygyny) or wife (polyandry) who
has more than one spouse, and depending too upon whether the
marriage is plural in a simultaneous or in a consecutive sense.
Frequent divorce establishes a kind of serial polygyny. But here
we are concerned with the two main kinds of plural marriage,
namely, polyandry, which was virtually limited to a small
area of India and Tibet, and polygyny, which is most widely
practised in Africa (in bridewealth and non-bridewealth
societies alike). In Reading 11 Leach reconsiders some of the
main characteristics of polyandrous marriages and relates these
to the type of inheritance, as Friedl does with dowry
(see Reading 9). In Reading 12 Clignet reports on some
general features of African polygyny.
11 E. R. Leach
Polyandry, Inheritance and the Definition of Marriage

E. R. Leach, ‘Polyandry, inheritance and the definition of marriage’,

Man, vol. 55, 1955, pp. 182-6. Reprinted in E. R. Leach, Rethinking
Anthropology, Athlone Press, pp. 105-13.

Although polyandry has been an important topic of anthropolo¬

gical discussion for almost a century the definition of the concept
remains strikingly unsatisfactory.1 This is sufficiently indicated by
the fact that Fischer (1952) maintains that adelphic polyandry,
regarded as a form of polygamy, is non-existent, while HRH
Prince Peter of Greece and Denmark (1955), ignoring Fischer,
continues to discuss adelphic polyandry as a species of polygamy.
At first sight the issue seems a simple one with the logic all on
Fischer’s side. The Notes and Queries (1951) definition of marriage
is: * Marriage is a union between a man and a woman such that
children bom to the woman are recognized legitimate offspring
of both partners \ Now certainly, in many cases of polyandry, the
legal status of the children is similar to that described by Caesar
for the ancient Britons (Fischer, 1952, p. 114): ‘Wives are shared
between groups of ten or twelve men, especially between brothers
and between fathers and sons; but the offspring of these unions
are counted as the children of those to whom the maid was con¬
ducted first.’ This clearly is not a condition of polygamy; the
children have only one legal father and the woman has only one
legal husband. The other ‘husbands’ have tolerated sexual access
to the woman, but she is not married to them in terms of the Notes
and Queries definition. The situation is one of plural mating, or, as
Fischer would call it, ‘polykoity’.
More specifically, Fischer argues that we should reserve the
concept of polygamy for situations in which the polygamous
spouse goes through a succession of marriage rites with different
partners. In adelphic polyandry ‘the woman does not contract
1. This paper is based in part upon fieldwork carried out in Ceylon in
1954.

E. R. Leach 151
different successive marriages. There is no reason for this, since
the social position of her children is guaranteed completely by the
fact that she is married’ (Fischer, 1952, p. 114).
Fischer agrees that an institution of polyandrous polygamy is a
possibility. For example a woman might be mated to several men
in such a way that each of them in turn assumed the role of social
father in respect to her successive children. This very approxi¬
mately seems to be the state of affairs among the Todas, and
Fischer concedes that it ‘approaches very closely to that of poly¬
gamy’. The institution of secondary marriage as described by
Smith (1953) is also polyandrous polygamy in Fischer’s sense. In
both these cases every child has one, and only one, clearly defined
social father.
But is it really so certain that the role of social father cannot be
vested simultaneously in several different individuals? Is it not
possible that in some societies social fatherhood is not an attribute
of individuals at all but of a collective corporation which may
include several brothers or even fathers and sons ?
When Radcliffe-Brown (1941) argued that adelphic polyandry is
to be ‘interpreted in the light of the structural principle of the
solidarity of the sibling group ’, he presumably had in mind that
social fatherhood might sometimes be vested in a collective cor¬
poration of this kind, and Prince Peter sought to demonstrate
that this is in fact the case. Does this mean that the notion of
group marriage is once again respectable?
There is certainly one well attested case of ‘ corporate polyandry ’
of this kind, namely that of the Iravas (Aiyappan, 1945, pp. 98-
103). Although Aiyappan states that on the occasion of a marriage
‘ the common practice is for the eldest brother alone to go to the
bride’s house to fetch her’, it is plain, from the further details that
he gives, that the eldest brother is here acting as representative
of the group of brothers considered as a corporation. Even so, it is
not entirely clear what rights this corporation possesses. It is
Aiyappan’s thesis that all marital rights are completely merged in
the corporation - that the sexual rights of the individual husbands
and the property rights of the individual children are alike
indistinguishable. Nevertheless one would welcome more detailed
evidence on these points.
There is another way of looking at the problem. Instead of

152 Plural Marriage

arguing pedantically about whether adelphic polyandry does or
does not constitute plural marriage, let us consider whether a
definition of marriage solely in terms of legitimacy (Notes and
Queries, p. Ill; Fischer, p. 108) is altogether adequate. There are
other definitions of marriage with respectable backing, e.g. ‘a
physical, legal, and moral union between a man and a woman in
complete community of life for the establishment of a family*
(Ranasinha, 1950, p. 192). Is the Notes and Queries definition any
less question-begging than this? What, for example, does the
phrase ‘legitimate offspring’ really connote?
Prince Peter, in the lecture under discussion, seemed to assume
that, of the various forms of heterosexual mating recognized and
tolerated in any society, there is always one which may properly
be described as ‘marriage’ in the anthropological sense. Yet if we
adhere rigidly to our Notes and Queries definition this is not the
case.
Thus traditionally among the matrilineal Nayar of South India
(Gough, 1952; 1955) a woman had a ritual husband in her enangar
lineage and also various ‘recognized lovers’ (sambandham), who
lacked ritual status; but all of these men were excluded from any
legal rights in respect to the woman’s children. There was here
then no marriage in the strict sense of the term but only a ‘ relation¬
ship of perpetual affinity’ between linked lineages (Gough, 1955).
The woman’s children, however they might be begotten, were
simply recruits to the woman’s own matrilineage.
Yet as Gough has shown, even in this system, certain elements of
a normal marriage institution are present.
The notion of fatherhood is lacking. The child uses a term of
address meaning ‘lord’ or ‘leader’ towards all its mother’s lovers,
but the use of this term does not carry with it any connotation of
paternity, either legal or biological. On the other hand the notion
of affinity is present, as evidenced by the fact that a woman must
observe pollution at her ritual husband’s death (Gough, 1955).
Both Gough (1952) and Prince Peter have described the Nayar
as having a system of polyandrous marriage. I do not wish to
insist that this is a misnomer, but we need to be clear that if we
agree that the Nayar practise polyandrous marriage then we are
using the term ‘marriage’ in a sense different from that employed
by Fischer and by Notes and Queries.

E. R. Leach 153
 My personal view is that the Notes and Queries definition of
marriage is too limited and that it is desirable to include under the
category ‘marriage’ several distinguishable sub-types of insti¬
tution.
The institutions commonly classed as marriage are concerned
with the allocation of a number of distinguishable classes of rights.
In particular a marriage may serve:
1. To establish the legal father of a woman’s children.
2. To establish the legal mother of a man’s children.
3. To give the husband a monopoly in the wife’s sexuality.2
4. To give the wife a monopoly in the husband’s sexuality.
5. To give the husband partial or monopolistic rights to the
wife’s domestic and other labour services.
6. To give the wife partial or monopolistic rights to the husband’s
labour services.
7. To give the husband partial or total rights over property
belonging or potentially accruing to the wife.
8. To give the wife partial or total rights over property belonging
or potentially accruing to the husband.
9. To establish a joint fund of property - a partnership - for the
benefit of the children of the marriage.
10. To establish a socially significant ‘relationship of affinity’
between the husband and his wife’s brothers.
One might perhaps considerably extend this list, but the point I
would make is that in no single society can marriage serve to
establish all these types of right simultaneously; nor is there any
one of these rights which is invariably established by marriage in
every known society. We need to recognize then that the institut¬
ions commonly described as marriage do not all have the same legal
and social concomitants.
If we attempt a typology of marriage on these lines it is at once
obvious that the nature of the marriage institution is partially
correlated with principles of descent and rules of residence. Thus
in a society structured into patrilineal patrilocal lineages we com¬
monly find that right 1 is far and away the most important element,
2.1 use the term ‘monopoly’ advisedly. I consider that this right 3 is to be
regarded as a monopoly control over the disposal of the wife’s sexuality
rather than an exclusive right to the use thereof. In discussing adelphic
polyandry this distinction is important.

154 Plural Marriage

whereas among the matrilineal matrilocal Nayar, as we have seen,
right 10 is the only marriage characteristic that is present at all.
Or again, in the matrilineal virilocal structure of the Trobriands,
right 7 assumes prior, though not altogether unique, importance
in the form of urigubu (Malinowski, 1932, pp. 69-75).
Although the early writers on polyandry (e.g. McLennan, 1865)
supposed that it was an institution closely associated with matri-
liny, Prince Peter has pointed out that the best-established cases of
adelphic polyandry occur in societies which express patrilineal
ideals. This was true of the Kandyan Sinhalese (D’Oyly, 1929);
it is true of the patrilineal Iravas of Madras (Aiyappan, 1945) and
of the Tibetans (Bell, 1928, p. 88). But it is also the case that the
patriliny in these societies is of an ambiguous and rather uncertain
type. The position in each case is that while the people concerned
profess a preference for patrilocal marriage and the inheritance
of landed property through males only, matrilocal marriage and
inheritance through females is not at all uncommon (Aiyappan,
1945; Li An-Che, 1947; D’Oyly, 1929, p. 110). Moreover although
women who marry patrilocally surrender their claims on their
own ancestral land, they receive a dowry of movable goods in
lieu.
This aspect of adelphic polyandry, namely that it is intimately
associated with an institution of dowry, has previously received
inadequate attention. In patrilineal systems of the more extreme
type all significant property rights are vested in males so that, from
the inheritance point of view, marriage does no more than establish
the rights of a woman’s sons in her husband’s property (right 1
above). Fission of the patrimonial inheritance group does of
course occur, and when it occurs it is very likely that individual
marriages will be cited (retrospectively) as a justification for such a
split; the model given by Fortes (1945, p. 199) is typical in this
respect. Yet, in such cases, marriage, as such, does not create an
independent partible estate.
But when property in land and saleable valuables is vested in
women as well as in men, a very different state of affairs prevails;
for each marriage then establishes a distinct parcel of property
rights and the children of any one marriage have, of necessity, a
different total inheritance potential from the children of any other
marriage.

E. R. Leach 155
 Systems of inheritance in which both men and women have
property endowment are very general in Southern India, Ceylon
and throughout South-East Asia. Such systems are found in
association with patrilineal, matrilineal and cognatic descent
structures. The general pattern is that the nuclear family, as a
unit, possesses three categories of property, namely the entailed
inheritance of the father, the entailed inheritance of the mother
and the ‘acquired property’ - that is, the property owned jointly
by the parents by virtue of their operations as a business part¬
nership during the period of the marriage. The children of the
marriage are heirs to all three categories of property, but the
categories are not merged.
Now it is quite obvious that an inheritance principle whereby
women as well as men can be endowed with property conflicts
with the ideal that landed property should be maintained intact
in the hands of the male heirs. Yet it is a fact that there are many
societies which manage to maintain both principles simultan¬
eously. There are a variety of customary behaviours which can
best be understood if they are regarded as partial solutions to the
dilemma that arises from maintaining these contradictory ideals.
Let us be clear what the dilemma is. On the one hand there is the
ideal that the patrimonial inheritance ought to be maintained
intact. Full brothers and the sons of full brothers ought to remain
together in the ancestral home and work the ancestral land. On
the other hand, since the wives of these men, when they join the
household, bring with them property which will be inherited by
their own children but not by their husbands’ nephews and nieces,
each new marriage creates a separate block of property interests
which is in conflict with the ideal of maintaining the economic
solidarity of male siblings.
One way out of the difficulty was that adopted in the Jaffna
Tamil code of Thesawalamai (Tambiah, 1954, p. 36): the sons
inherited the hereditary property of their father, and the acquired
property of both spouses was inherited by the sons and the un¬
dowered daughters. The dowries to the daughters were given out
of the mother’s dowry. Systems of double unilineal descent such as
that described by Forde for the Yako operate in a somewhat com¬
parable way (Forde, 1950, p. 306), though the distinction here is
between property passed to men through men (the patrilineal in-

156 Plural Marriage

heritance of land rights) and property passed to men through
women (the matrilineal inheritance of movables).
The Moslem preference for patrilineage endogamy likewise
resolves the conflict between female rights of inheritance and a
patrilineal principle of descent. A declared preference for recipro¬
cal or patrilateral cross-cousin marriage may sometimes have
similar implications. Indeed, marriage preferences of this latter
type seem to be more or less confined to societies in which a
substantial proportion of the inheritance rights are transmitted
through women (cf. Homans and Schneider, 1955).
Adelphic polyandry, I would suggest, is to be understood as yet
another variation on the same theme. If two brothers share one
wife so that the only heirs of the brothers are the children bom of
that wife, then, from an economic point of view, the marriage will
tend to cement the solidarity of the sibling pair rather than tear it
apart, whereas, if the two brothers have separate wives, their
children will have separate economic interests, and maintenance
of the patrimonial inheritance in one piece is likely to prove im¬
possible. If the ethnographical evidence is to be believed, poly-
androus institutions, where they occur, are deemed highly vir¬
tuous and tend to eliminate rather than heighten sexual jealousies
(Aiyappan, 1937).
In the lecture under discussion, Prince Peter referred repeatedly
to contemporary polyandry among the Kandyan Sinhalese. It
seems important that we should be clear what the word ‘poly¬
andry’ means in this case. Sinhalese law does not recognize the
existence of polyandrous marriage and it is not possible for any
individual to maintain in a law court that he or she is ‘ the recog¬
nized offspring’ of two different fathers, nor can a woman bear
‘legitimate offspring’ to two different husbands, without an inter¬
mediate registration of divorce. Thus, strictly speaking, polyandry
in Ceylon is not a variety of marriage, if marriage be narrowly
defined. On the other hand it is certainly the case that there are
parts of Ceylon where two brothers often share a common domes¬
tic household with one ‘ wife ’, these arrangements being permanent,
amicable and socially respectable.3
3. It is difficult to accept Prince Peter’s claim that in the Ratnapura
District of Ceylon polyandry is so common as to be the norm. The Census
(1946, vol. I, pt 2) includes figures for ‘customary marriages’ as well as

E. R. Leach 157
 Polyandrous households of this type contrast rather strikingly
with the more normal pattern in which two or more brothers live
together in a single compound each with his separate ‘wife’.
This latter situation is characterized by marked restraint between
the brothers and even complete avoidance between a man and
his ‘sister-in-law’.
The ‘wives’ in such cases may or may not be married according
to Sinhalese law. In a high proportion of cases they are not so
married. In law the children of these unions are then illegitimate.
The children, however, have birth certificates and these certificates
give the name not only of the mother but also of the acknowledged
father, a circumstance which provides the child with a potential
claim to a share of the heritable property of each of its parents.* * * 4
The child therefore, although not the legitimate offspring of
both its parents, is nevertheless a legitimate heir of both its parents.
If then the principle of legitimacy be here defined in terms of
property rights rather than descent it seems quite proper that
Sinhalese customary unions should be regarded as marriages.
Is it then possible in this case to have a polyandrous marriage ?
registered marriages’. The Census enumerators were required to enter as
‘married’ anyone who ‘claimed to be married according to custom or
repute’ and there seems no reason why they should have excluded ‘poly¬
androus husbands’. However, in all districts, the overall total of‘married’
males is roughly equal to the overall total of‘married’ females, which does
not suggest that the frequency of polyandry can be numerically significant.
For Ceylon as a whole the Census gave 389,846 women as ‘married by
custom’ and 843,493 as ‘legally married by registration’. While this is
evidence that the strict definition of legitimate marriage is unrealistic, it
does not follow that the anthropologist must accept the Census enumerators’
notions of what constitutes customary marriage.
4. The Report of the Kandyan Law Commission (1935, paragraphs 199-210)
recommends that all children born of non-registered marriages shall be
deemed illegitimate and shall be excluded from any share in the entailed
property of the father. The Report recognizes that this conflicts with the
customary law of the pre-British period which did not restrict entailed
(paraveni) property to the offspring of formal marriages. Ranasinha (1950,
vol. I, pt l,[p. 192) ignores this Report and asserts that the highest authorities
have held that ‘registration was not essential to the validity of a marriage
in Ceylon, and the marriage relation could be presumed on adequate
evidence of cohabitation and repute’. Certainly in many parts of Ceylon
today the children of non-registered ‘marriages’ are treated as having full
inheritance rights in their father’s property, but whether this right could
now be sustained in a Court of Law I am uncertain.

158 Plural Marriage

Legally, no. Since a birth certificate certainly cannot show more
than one father, no child possesses the basis for establishing a
legal claim to the property of a polyandrous corporation. All the
same, it seems probable that in polyandrous households the
children do ordinarily inherit jointly the undivided property of the
two fathers and that Sinhalese custom recognizes their right to do
so. Provided that we are not too pedantic about what we mean by
‘legitimate ’ it does appear that we are dealing here with something
that an anthropologist can properly call polyandrous marriage.
Even so the issue is by no means clear-cut.
Aiyappan (1945, p. 103), in commenting on the refusal of an
English judge to admit the possibility of a woman being simul¬
taneously married to two brothers at the same time, treats the issue
as being simply one of a conflict between English law and cus¬
tomary Irava law. But so far as the Sinhalese are concerned the
issue is not so simple.
The classical formulation of the former Sinhalese law regarding
polyandry appears in Sawers’s Digest (see D’Oyly, 1929, p. 129):

Polygamy as well as polyandry is allowed without limitation as to the

number of wives or husbands - but the wife cannot take a second
associated husband without the consent of the first - though the hus¬
band can take a second wife into the same house as his first wife
without her consent. The wife, however, has the power of refusing to
admit a second associated husband at the request of her first husband,
even should he be the brother of the first. And should the proposed
second associated husband not be a brother of the first, the consent of
the wife’s family to the double connection is required.

It is clear that two separate rights are here distinguished. First,

there is the right in the wife’s sexuality which marriage serves to
vest partly, but not completely, in the person of the first husband.
The sexual rights of the other husbands are exercised, not by
virtue of the marriage, but through the individual consent of the
first husband and the joint wife. On the other hand, the ritual of
patrilocal marriage - the essence of which is that a man conducts
his bride from her father’s house to his own {Report, 1935,
paragraph 168) - serves to establish a relation of affinity between
the wife’s family as a whole and the husband’s family as a whole.
The wife’s family have no interest in what sexual arrangements

E. R. Leach 159
pertain unless it is proposed to extend the rights of sexual access
beyond the limits of the husband’s sibling group.
It is notable that, in this formulation of Sawers, the rights of the
children are not mentioned; the ritual procedures of Sinhalese
marriage are not concerned with the rights of the potential children.
The marriage rite disposes of the woman’s sexuality to her first
husband; it also has the effect of making a public pronouncement
that the woman has been properly endowered so that she has no
further claims on her parental property. The status of children
arises from quite a different source.
In Sinhalese customary law it was (and is) the rule that if a man
and a woman are publicly known to have cohabited together and
the woman bears a child, then the woman has a claim on the man
for the support of the child (D’Oyly, 1929, p. 84). In ordinary rural
practice, all of a man’s acknowledged children are equally his heirs
whether or not he has at any time gone through a ritual of marriage
with the children’s mother. Likewise all of a woman’s children have
equal claims on her inheritance.
My conclusion is that in the Sinhalese case, and very probably
in other analogous cases, we are dealing with two different
institutions both of which resemble marriage as ordinarily under¬
stood, but which need to be carefully distinguished. Neither
institution corresponds precisely to the ideal type of marriage as
defined in Notes and Queries.
On the one hand we have a formal and legal arrangement, by
which, so far as Ceylon is concerned, a woman can only be married
to one man at a time. ‘ Marriage’ in this sense establishes a relation¬
ship of affinity between the family of the bride and the family of the
first husband, and it gives the disposal of the bride’s sexuality to
the first husband, subject to the bride’s personal consent. On the
other hand we have another institution of ‘marriage’, which is
entered into quite informally but which nevertheless, by virtue of
its public recognition, serves to provide the children with claims
upon the patrimonial property of the men with whom the woman
cohabits and publicly resides. This second form of ‘marriage’, al¬
though it establishes the inheritance rights of the children, does not
establish their permanent status as member of a corporate descent
group, and Sinhalese children, as they grow up, have wide choice as
to where they finally align themselves for the purposes of affiliation.

160 Plural Marriage

 If we accept this second institution as a form of‘marriage’, then
polyandry in Ceylon is a form of polygamy. If we confine the term
‘ marriage ’ to the first institution, polyandry in Ceylon is a form of
polykoity. These niceties of definition are worth making because it
is important that anthropologists should distinguish the various
classes of right that are involved in marriage institutions.
Of greater importance is my hypothesis that adelphic polyandry
is consistently associated with systems in which women as well
as men are the bearers of property rights. Polyandry exists in
Ceylon because, in a society where both men and women inherit
property, polyandrous arrangements serve, both in theory and
practice, to reduce the potential hostility between sibling brothers.

References
Ai yapp an, A. (1937), ‘Polyandry and sexual jealousy’, Man, art. no. 130.
Aiyappan, A. (1945), ‘Iravas and Culture Change’, Bull. Madras
Govt. Mus., General Section, vol. 5, no. 1.
Bell, Sir C. (1928), The People of Tibet, Oxford University Press.
D’Oyly, J. (1929), A Sketch of the Constitution of the Kandyan
Kingdom {Ceylon), Colombo.
Fischer, H. (1952), ‘Polyandry’, Int. Arch. Ethnog., vol. 46, pp. 106-15.
Forde, D. (1950), ‘Double descent among the Yako’, in A. R. RadclifFe,
Brown and D. Forde (eds), African Systems of Kinship and Marriage,
Oxford University Press.
Fortes, M. (1945), The Dynamics of Clanship among the Tallensi,
Oxford University Press.
Gough, K. (1952), ‘Changing kinship usages in the setting of political
and economic change among the Nayars of Malabar’, J. roy. anthropol.
Inst., vol. 82, pp. 71-88.
Gough, K. (1955), ‘The traditional lineage and kinship systems of the
Nayar’, unpublished manuscript in the Haddon Library, Cambridge.
Li An-Che (1947), ‘Dege: A study of Tibetan population’. South¬
western J. Anthropol., vol. 3, pp. 279-93.
Malinowski, B. (1932), The Sexual Life of Savages, Routledge &
Kegan Paul.
McLennan, J. F. (1865), Primitive Marriage, Black.
Notes and Queries in Anthropology, 1951, Royal Anthropological
Institute, 6th edn.
Peter, H. R. H. Prince of Greece and Denmark (1955), ‘Polyandry
and the kinship group’, lecture delivered to the Royal Anthropological
Institute, 8 December. Summarized in Man, 1955, no. 198.
Radcliffe-Brown, A. R. (1941), ‘The study of kinship systems’,
J. roy. anthropol. Inst., vol. 71, pp. 1-18.

T-K-F E. R. Leach 161

Ranasinha, A. G. (1946), Census of Ceylon, vol. 1, pt 1, Colombo.
Report of the Kandyan Law Commission (1935), Sessional Paper 24,
Colombo (Govt. Press).
Smith, M. G. (1953), ‘Secondary marriage in northern Nigeria’, Africa,
vol. 23, pp. 298-333.
Tambiah, H. W. (1954), The Laws and Customs of the Tamils of Ceylon,
Colombo.

162 Plural Marriage

12 R. Ciignet
Determinants of African Polygyny

Excerpt from R. Ciignet, Many Wives, Many Powers, Northwestern

University Press, 1970, pp. 16-34.

Our purpose here is to demonstrate that there are variations in

(1) the distribution of plural marriage among African cultures,
(2) the principles underlying the recruitment of additional co¬
wives, (3) the social, economic and political characteristics of
African societies which practise plural marriage, and (4) the
individual motivations which underlie this particular form of
familial arrangement. In a last section, we will show how social
change affects these various dimensions.

The distribution of polygyny in Africa

The distribution of polygyny in a given population presents two
characteristics relevant to our analysis: its incidence and its
intensity (Dorjahn, 1959). Incidence is measured by the number of
married women per hundred married males or, alternatively, by
the number of males married to more than one wife. For the first
measure to be useful, however, the communities investigated
should be demographically stable. The first indicator of incidence
is misleading in urban centers, where the relative excess of men
over women and the uneven incidence of singles among the male
and female segments of the population prevents distributions of
married men and women from being really matched. Intensity,
on the other hand, refers to the number of married women per
hundred polygynous men or, alternatively, to the number of
polygynous males with more than two wives.
An examination of the Human Relations Area Files gives us
a first approximation of the significance of plural marriage in
Africa.1 Approximately three-fourths of the 136 peoples included
1. For all criteria used in the first section of this Reading, see
Murdock (1957), quoted in Moore (1961).

R. Ciignet 163
in these files are characterized by general polygyny - that is, by
both an incidence of polygynous families in excess of 20 per cent,
and an absence of special restrictions on the recruitment of ad¬
ditional co-wives. Of the entire world sample, Africa has the lowest
percentage of monogamous cultures; hence, we can see the im¬
portance of an analysis of the dominant behavior patterns in
African polygynous institutions.
For all sub-Saharan countries, the mean number of wives per
hundred married males is 150, but the variance is extremely large
(Moore, 1961). Maximal ratios are found in Ghana and Sierra
Leone, and the lowest percentages in the southeastern parts of

Table 1 Incidence and Intensity of Polygyny in

selected African Countries
Countries Date Married Married
women per women per
100 married 100 polygynous
men men

Basutoland 1936 114 218

Bechuanaland 1946 112 201
Swasiland 1936 168 307
South Africa 1921 112 216
Northern Rhodesia 1947 122 211
Tanganyika 1934 117 219
Kenya 1951 236 283
Sudan 1945 199 253
Congo 1947-48 131-44 224-73
Portuguese Guinea 1950 159 245
French Equatorial
Africa 1952 145 ?
French Cameroon 1942 152 ?
•

Togo 1931 131-57 ?

Upper Volta 1951 190 233
Nigeria 1950 155 237
Gold Coast 1909-39 190-218 ?
1945 160 252
Sierra Leone 1945 304 325
Gambia 1951 184 276

Derived from Dorjahn (1954, pp. 134-6)

164 Plural Marriage

the continent. Intensity has an equally large variance and ranges
from a minimal level of 201 for Bechuanaland to a maximal value
of 325 for Sierra Leone, as shown in Table 1.
More important, these two dimensions have remained stable
throughout time. The figures obtained from adjacent territories at
different points in time are comparable. In the specific case of the
former Gold Coast, the incidence of polygyny has apparently
increased throughout time.
Finally, Table 1 indicates that there is no clear-cut relationship
between incidence and intensity of polygyny. It would be tempting
to assume a priori that the number of women per polygynous
family declines as the institution spreads among a greater variety
of social levels (Dorjahn, 1954, ch. I).2 This is obviously not the
case.

Principles of recruitment of co-wives

Both the incidence and the intensity of plural marriage should
reflect the principles underlying the recruitment of additional
co-wives. Polygyny should be more frequent wherever there are
few boundaries in the definition of the field of eligible additional
spouses. Thus, in some societies, polygyny is at least partially due
to levirate, or ‘inheritance of widows’ - that is, the transmission
to the inheritor of all the rights that the deceased testator had in
his wife or wives. In contrast, the field of eligible co-wives may be
narrowly defined; in some societies, for example, sororalpolygyny,
in which a man marries two or more sisters, is the only form of
plural marriage socially allowed. This type of plural marriage
tends to prevent the multiplication of polygynous arrangements. It
also tends to erase the incompatibilities between the functions of
polygyny, more specifically, the division of the loyalties of each
co-wife between her affinal group and her family of origin. The
common background of co-wives, as it exists in this case, should
help to lower potential tensions between familial actors. This
leads us to predict that variations in the number and the form of
restrictions imposed on plural marriage will be accompanied by

2. In fact, it has been demonstrated that there was a positive relationship

(r = 0-85) between incidence and intensity of plural marriage among the
twenty-six districts of the Congo. See Brass et al. (1968, p. 215).

R. Clignet 165
similar variations in the types of relationships prevailing both
within and between polygynous families.
Only 8 per cent of the African societies included in the Human
Relations Area Files practice sororal polygyny exclusively and they
are concentrated in the cluster of Bantu subcultures.3

Characteristics of polygynous societies

Since the incidence and intensity of polygyny are higher in some
parts of Africa than in others, it is necessary to determine whether
African cultures that are characterized by similar distributions of
plural marriage also present certain common traits in their
economic, political and social organizations. The absence of such
regularities should lead us to suspect that variations in type of
marriage reflect a number of factors, and that, accordingly, the
organization of polygynous families is not stable throughout time
or space. Conversely, the identification of certain recurrent
patterns in the distributions of plural marriage across African
cultures would not necessarily exclude the possibility of observing
marked disparities in the functioning of these families.
Interestingly enough, although monogamous familes are
characteristic of modern, highly industrialized nations, they also
tend to be frequently found in African social systems whose
subsistence patterns are predominantly gathering and hunting.
In these societies, the proportion of resources available to an
individual declines as the population density increases.4 Associ¬
ated with a high density, polygyny would lead to starvation. On
the other hand, general polygyny tends to prevail in societies
where subsistence is based upon the cultivation of roots and tubers
and upon arboriculture. This type of economic organization
encourages the maintenance of large-sized households and also

3. Twenty-two out of forty Bantu cultures are characterized by general

polygyny; in the southern cluster of these people, five out of ten cultures
indulge in sororal polygyny.
4. See Zelditch, Jr. (1964). For a more exhaustive discussion of the
correlates of marriage type, see Osmond (1965), who demonstrates on a
world-wide basis that the relationship between monogamy and societal
complexity is curvilinear in nature; see also Sawyer and LeVine
(1966). It is quite clear that the correlations between polygyny and other
variables remain quite low when the investigation deals with a world-wide
sample.

166 Plural Marriage

constitutes a necessary condition to the emergence of any form of
social differentiation. It is to this point that we will turn our
attention first.
The existence of politically or socially stratified divisions in
the population, associated with the transmission of land or other
resources, which ensures that a man’s children will remain grouped
around his own place of residence, favors the emergence of
polygynous households. In fact, the relationship between the
incidence of polygyny and social stratification is curvilinear in
nature. Rare among societies characterized by a lack of social
stratification or, alternatively, by a large number of social levels,
plural marriage is most frequent among societies divided into age
grades or in which a hereditary aristocracy is separated from the
bulk of the population (Table 2).
A stabilized mode of agricultural production, however, is not
only conducive to the emergence of certain forms of social

Table 2 Relationship between Social Stratification and Marriage

Practices among selected African Cultures (percentages)

Type of Monogamy Limited General

stratification polygynya polygynya

Complex
stratification 22 8 17
Hereditary
aristocracy 0 25 27
Wealth distinction 11 8 10
Age grades 11 13 14
No social
stratification 56 42 28
Unascertained 0 4 4

Total 100 100 100

N 9 24 103

a Limited polygyny means that the incidence is below 20 per cent,

whereas general polygyny refers to all societies where this threshold is
exceeded. Variations in the form of polygyny have been disregarded.
Computed from Murdock (1957). The Horn, Ethiopia and Moslem
Sudan have been added to African samples.

R. Clignet 167
stratification, but it also tends to encourage the maintenance of
large households and to accentuate the division of labor among
individual actors of the family group. In contrast to monogamous
cultures, polygynous social systems are characterized by principles
of division of labor leading women to carry out a substantial part
of the agricultural activities of the household (Table 3). In effect,
the incidence of polygyny is positively related to the productive
value of the female members of a group. It might have been assumed

Table 3 Relationship between Division of Agricultural Labor

and Plural Marriage (percentages)
Division of labor Monogamy Limited General
in agricultural work polygyny polygyny

Division of labor
equal along sex
11 50 49
lines
Female share of the A
u 12 21
work greater
Male share of the
56 4 20
work greater
No agricultural
activity 33 17 8
Work performed by
0 0 1
slaves
Not ascertained 0 17 1

Total 100 100 100

N 9 24 103

that plural marriage would decline with the development of cash

economies and that successful individuals would be anxious to
invest their surpluses of cash income in assets more readily
negotiable than women, but this does not appear to have happened.
In societies where subsistence depends upon agricultural pro¬
duction and where such production depends heavily on the man¬
power available, particularly so on the productive value of women,
each family group tends to use its surplus income to increase its
labor force-and, more specifically, the number of its polygynous

168 Plural Marriage

 units as well as the number of its co-wives. In short, polygyny is
most tenacious in cultures where economic rights to women can
be acquired and have a high significance (Bohannan 1963, p.
109).
The productive value of a woman in turn influences the bride-
price, that is, the goods and services which either the bridegroom or
his family must pay to the bride’s family as a compensation for the
emotional, social and economic loss resulting from her marriage.
Since polygyny is most frequent among cultures which invest
high values in their female members, it is not surprising that the
institution of brideprice is more often found in polygynous than
in monogamous African societies.5
Finally, the incidence of plural marriage depends on the rules of
residence followed by each culture. Where 84 per cent of African
peoples with widespread polygyny are patrilocal, this is true of
only two-thirds of the societies with a limited rate of polygyny,
and of 54 per cent of the cultures which practise monogamy.
Influenced by the productive value of women, polygyny is also
associated with a predominance of male orientations because
patrilocal rules of residence tend to reinforce the loyalties and
obligations of a husband toward his own family.
Having determined some relationships between the incidence of
polygyny and some features most typical of the African cultures
under study, it would be useful to examine the extent to which
these features are associated with one or another form of poly¬
gyny. Unfortunately, the number of African cultures practising
sororal polygyny is too small to enable us to do so. Analysis of a
world-wide sample of matrilineal descent groups, however, has
led to the assumption that sororal polygyny is most likely to be
found in societies with matrilocal residence combined with a
weak descent group that is not permanently attached to stable,
scarce cultivation sites. Under these circumstances, ties between
sisters and their offspring are easily maintained, the interests of
co-resident husbands are more easily harmonized, and the com-

5. Of African monogamous societies, 22 per cent are characterized by an

absence of any significant matrimonial compensation in marriage. This
percentage drops to 16-6 among cultures with a limited incidence of poly¬
gyny. Regardless of the form of polygyny that they practise, none of the
remaining societies presents this feature.

R. Clignet 169
mitment of these husbands to their respective conjugal groups is
increased (see Gough, 1961, pp. 623-4).
Thus far, we have determined the limits within which the in¬
cidence and the intensity of plural marriage vary. We have also
examined those traits in the economic and social organization
of a given culture which tend to facilitate or impede plural mar¬
riage. More specifically, we have isolated two functions of poly¬
gyny. On the one hand, plural marriage is associated with social
differentiation and makes significant distinctions between certain
categories of ‘ have ’ and ‘ have not ’ of the particular ethnic groups
investigated. On the other hand, this institution may also con¬
stitute a basic mechanism indispensable for economic survival.
In the first context, additional co-wives are mainly liabilities, but
in the second they are additional assets. It remains of course to
determine the extent to which these two types are mutually ex¬
clusive.
However, the most necessary condition to the emergence of
polygynous families in a given society is, of course, an imbal¬
ance in its sex ratio; it may be said that polygyny constitutes a
most rational solution to the problem of absorbing an actual
surplus of females. Our initial task, therefore, is to assess the
sex ratio of the populations of African countries.
A high sex ratio may result from an insufficient gross or net
reproduction rate;6 it may reflect discrepancies in the distribu¬
tion of mortality rates by sex at various points in the age pyramid.
For example, male fetuses may be more vulnerable to miscarriage
than female; or males may be more susceptible to death in child¬
hood or in early or late adulthood. As a matter of fact, the sparse
information available on these points in various parts of the world
leads us to believe that miscarriage occurs less frequently to male
than to female fetuses. While it is a fact that the number of male
children born is consistently greater than the number of female
births, this unbalanced character of the sex ratio tends to disappear
later on since male children are inclined to be more vulnerable to
fatal diseases than females (Dorjahn, 1954, pp. 310-70). In addi¬
tion, the hunting, military and commercial enterprises of males also
6. For definitions of these terms, see Winch (1964), pp. 186-7.
These rates answer the question of how many female children the average
woman in a hypothetical cohort of females will bear.

170 Plural Marriage

increase their exposure to death. To rebut this, many scientists
argue that only women are involved in the most hazardous of all
activities - childbearing - and that this offsets any increased male
exposure to death (Bohannan, 1963, p. 109).
Data available on the overall population of African countries
indicate that the sex ratio tends to be roughly equal (Table 4). For
the Gold Coast, Table 4 establishes that, between the turn of the
century and 1948, while polygyny increased slightly, the rate of
female births declined. In short, plural marriage cannot be ex¬
plained by disparities in the natural distribution of male and
female populations. Such disparities, however, might still be the
product of cultural or social factors. For example, massive
urbanization of males of a given rural society will skew the dis¬
tribution of that segment of the population which remains at
home. Migrations facilitate polygyny, not only for the stable male
elements of the rural society but also for the mobile members of
the populations who, with an increased cash income, are able to
Table 4 Sex Ratio in selected African Countries

Country Date Number of

females per 100
males
Southern Rhodesia 1948 101-0
Tanganyika 1948 108-4
Kenya 1948 102-6
Uganda 1948 100-0
Portuguese Guinea 1950 103-0
French Equatorial Africa 1948 104-0
Ruanda Urundi 1951 114-4
Gold Coast 1891 118-0
Gold Coast 1948 98-0
Derived from Dorjahn (1954, p. 267).

support two households and two wives - one in the city and one in
the village of origin.7
If male migrations can facilitate polygyny, so also can female
migration, most especially the importing of marriageable women.
7. Africans are not the only ones to follow this practice. The film The
Captain's Paradise explains the de facto polygyny of some sailors and the
difficulties which result from this state of affairs.

R. Clignet 171
For example, in a tribal or inter-village war, the women of a
weaker society may be captured and taken home by victorious
males, who thus artificially augment the female population of their
own village. On the other hand, marked inequalities in the amount
of brideprice required by neighboring societies may foster vari¬
ations in the number of eligible women available to men in both
groups. The males of the wealthier society will be able to invade
the field of potential brides of the poorer society. Similarly, the
differential rates of mobility of African males and females are
also likely to affect the sex ratio of the marriageable populations
derived from distinctive subparts of a given territory or region.
For example, the adult population of Upper Volta probably
comprises a larger proportion of women than the eastern zones
of the Ivory Coast, which attract a considerable number of male
migrants. The influence of such factors on polygyny depends,
however, on the degree of ethnocentrism prevalent in the societies
in question. The extent of this ethnocentrism varies naturally
with the various patterns of interaction investigated. Evidence
concerning patterns of interethnic marriage and the distribution
of the relevant motivations remains unfortunately scarce. In
Dakar, it has been established that in 1954 the incidence of
interethnic marriages increased with socioeconomic status and,
more important, was higher among polygynous than monogamous
families (Mercier, 1955, p. 553). It is difficult, however, to derive
long-term generalizations from this evidence, which is limited
both in time and space.
It might be interesting to speculate that polygyny reflects a
difference in the incidence of marriage for each sex. A large
percentage of celibate males certainly increases the number of
available marriageable women and thus increases polygyny.
But it has been authoritatively estimated that the proportion of
adult single males in African populations does not exceed 28
per cent, while that of unmarried adult females is still lower (22
per cent) (Dorjahn, 1954, pp. 269-309). Neither the low percentage
among males nor the limited difference in the occurrence of male
and female celibacy can account for polygynous marriages.
A more satisfactory explanation lies in the age characteristics
of conjugal partners. An excess of marriageable women can be the
result of the difference in mean age at first marriage in both sexes,

172 Plural Marriage

as well as of the specific mean ages of the people who marry.
Assuming that (1) the incidence of singles is equally low among
both sexes, (2) there is little variation in the average age at mar¬
riage of male and female populations, and (3) the size of various
age groups is equally uneven for the two sexes, it is possible to
evaluate the surplus of marriageable women. Thus, Dorjahn has
established that if, for example, the first-marriage age of women is
sixteen and that of men twenty-five, there will be a surplus of
26 per cent of marriageable women. If the first-marriage age of
women is twenty-four and that of men thirty-three, there will be a
33 per cent surplus of marriageable females. Yet inequalities in
the number of marriageable men and women may also result
from differences in the relative number and duration of their res¬
pective matrimonial experiences. The imposition of differing
limitations upon male and female divorcees or widows concerning
their remarriage is likely to affect the overall number of potential
partners.8
In summary, inequalities in the distribution of polygyny
throughout Africa mainly reflect the variety of cultural norms
pertaining to age at marriage and to the respective matrimonial
\

experiences of male and female actors.

✓

Polygyny and individual motivations

African informants frequently indicate that polygyny constitutes
the most efficacious way of maintaining a high birth rate in their
societies. This proposition has been critically examined by various
observers and social scientists; but the discussion has sometimes
been obscured by religious and political considerations which
generally lead to a formal condemnation of plural marriage. Thus
the belief that modernization of a country necessitates a constant
increase of population is often accompanied by the opinion that
modernization is incompatible with the maintenance of poly-
8. (Dorjahn, 1954, p. 299). The additional comments presented here
result from private communications with R. Cohen. In this context, we
must note a highly significant correlation (r = 0-80) among Congo districts
between number of married women per 100 married men and difference in
mean age at marriage between the sexes. There is, alternatively, a negative
correlation (r = — 0-45) between this indicator of polygyny and the propor¬
tion of widowed and divorced among women fifteen to forty-five years old.
See Brass et al. (1968, pp. 220-21).

R. Clignet 173
gynous families. The organization of such families, it is deemed,
does not meet the requirements of an industrial system and also
has negative effects on the desired growth of the population. It is
possible to demonstrate, for example, that fertility of mono-
gamously married women is one-third higher than that of women
in polygynous families.9
Sterility of his senior co-wife often leads a man to acquire an
additional wife. This necessarily affects the distribution of children
born in polygynous households. Further, since polygyny pre¬
supposes a certain order in sexual relationships between a man and
his spouses, it can be argued that the frequency of sexual inter¬
course per woman decreases, leading to a corresponding decline
of fertility.10 In addition, all African societies are subject to various
sexual taboos; and it is obvious that such taboos are more likely
to be enforced with greater frequency and for longer periods of
time by women who are polygynously married. Given the fact
that there is a significant relationship between polygyny and the
severity of post-partum taboos, the nature of this relationship
remains to be explored. It might be argued that the severity of this
taboo leads married males to contract plural marriage. Altern¬
atively, however, plural marriage may lead to the formulation of
universal prescriptions regarding the appropriate behavior of
new mothers.
The effects of all these factors are convergent and account for
the lower number of children born to polygynous wives. Per¬
ceived by all Africans as the most effective way to insure a numer¬
ous posterity, is polygyny nevertheless destined to maintain
fertility below a certain threshold? A proper answer to this
question depends on the level of analysis selected. A decline in the
number of children born of a given category of women does not
necessarily imply a corresponding decline in the number of
children of a male individual. In fact, a decline in fertility may be
compensated for by an increase in the number of potentially
child-bearing women.
9. For a general discussion of the factors affecting fertility, see Blake
and Davis (1956, p. 211-14). More specifically on the problem of fer¬
tility and plural marriage, see Whiting (1964, pp. 511 ff.).
10. See Muhsam (1956). Yet distinctions should be made between
senior and junior co-wives in this respect, and it is, for example, necessary
to control the ages of the women analysed.

174 Plural Marriage

 A second justification of polygyny offered by Africans concerns
the complexity of female roles in the society. The obligations of a
woman to her husband and his family do not cancel her duties to
the members of her own family. For example, in many African
cultures a married woman is required to visit her kin group at
regular intervals and on special occasions, such as funerals. These
prescribed absences lead to the often-voiced complaint of African
husbands that to have one wife is to have none, intimating that
wives spend more time in their households of origin than in their
own.11 Furthermore, this situation might impose on husbands the
performance of domestic chores, such as cooking and fetching
water and wood, which are considered female tasks and therefore
incompatible with male dignity. Regardless of these strains, females
are expected to fill plural roles and, as in many societies, the
significance of a wife for the hearth can be contrasted with the
importance of a wife for the heart.
Besides economic and symbolic properties attached to female
roles, Africans cite other social and psychological factors which
influence the size of a family group. If there is no difference
between having only one wife and having none, as African
husbands say, then males with two wives only are not much
better off. The positive effects of an additional wife are too often
mitigated by the husband’s obligation to mediate in the conflicts
and jealousies which arise between the women. Since three is
conducive to the formation of coalitions, four is generally con¬
sidered the optimal number of co-wives for a man to have (Joseph,
1913, p. 595).

Social change and polygyny

Having established that, on the whole, both the incidence and
intensity of polygyny have not varied throughout time, can we
then infer that there has been no alteration in the motivations
which support it ? Let us examine the effects of social change.
Social change implies differential access of the male population
to modern residential, educational and occupational structures
and is accompanied by a modification of the criteria on which the
social hierarchy is based. Education, occupation and experience
11. See also Serere proverb, quoted by Monteil (1964): ‘With one wife, a
man has only one eye.’

R. Clignet 175
in urban centers increasingly differentiate the positions occupied
by males within the social structure. Yet changes in the degree of
social stratification and in the routes leading to higher status do
not necessarily imply corresponding shifts in the symbolic qualities
attached to the various positions in the status system. Indeed, with
one notable exception, many observers of the contemporary
African scene agree that the incidence of polygyny in cities tends
to increase with the length of time spent there and with the higher
levels of occupation achieved.12
Social change implies a restructuring of ideology. The persis¬
tence of polygyny has been reinforced by the diffusion of Islam,
which on this very point converged with the principles of social
organization prevalent in traditional African cultures. It should
be pointed out, however, that Islamic teaching is not always
equated with male domination over women. Koranic laws provide
a legal status for v/omen which is often refused them by traditional
African customs (Monteil, 1964). Likewise, in spite of its mis¬
sionary success, Catholicism has failed to eliminate this type of
family deemed incompatible with Christian principles.
Even though, during most of the colonial period, efforts of the
Christian missionaries to end polygyny have been supported by
administrative authority, the institution has survived. In French
territories few women have taken advantage of the right to protest
their traditional matrimonial status. Theoretically, French
citizenship was reserved to African individuals who were mono-
gamously married. Yet this provision has too often conflicted with
other sections of the law which accorded citizenship to the most
educated and most economically successfully segments of the
population, within which many polygynists were to be found. In
addition, in the early 1950s the French government decided to
extend to African workers in both public and private sectors certain
fringe benefits already existing in metropolitan France, among
which was an allotment of money to individuals with many
12. For a discussion of the diffusion of polygyny in Senegalese urban
areas, see, for instance, Masse (1955). See also Mercier (1960) and
Thore (1964). For the diffusion of polygyny in the Belgian Congo,
see Clement (1955). For Dahomey, see Tardits (1958), pp. 47-9 and
63-5). For English-speaking Africa, see Little (1959, pp. 72-3). For
Sierra Leone, see Gamble (1963, pp. 75-84). Against these views, see
Banton (1957, pp. 207ff.). See also Bird (1963, pp. 59-74).

176 Plural Marriage

children. The purpose of the decision was both to stimulate a
natural increase in the population and to satisfy the demands of
local politicians and union leaders to be treated as ‘Black French¬
men’. This lagniappe had the unexpected effect of reinforcing
polygyny among the already privileged wage-earners in urban
areas. Introduction of this legislation in Africa is an illustration of
the inconsistencies which characterize a colonial policy based on
‘assimilationist’ principles that appear to favor at the same time
both the emergence of new patterns of action and the persistence of
traditional ways of life.13
Of all manifestations of social change, schooling of the female
population is the only one which has had a negative effect on poly¬
gyny and has, thus, tended to contribute to the disruption of the
overall functioning of traditional family structures. Not only have
educated women refused to belong to polygynous families, even as
senior co-wives, but they have frequently been in a position to
force males to change their domestic attitudes and behavior. For
example, educated women have repudiated mealtime etiquette
which obliges them to eat, in another room, leftovers from meals
taken by men. They have challenged the principle on which the
payment of a brideprice is based, and they have become increas¬
ingly eager to marry mates of their own choosing.14 Hence, there is
a certain amount of ambivalence on the part of both males and
older females toward educated women. Male domination is too
ancient a phenomenon to be easily eroded, and elder women tend
to perceive formal education as a devious way of dissociating one¬
self from traditional duties. The first generation of educated
women is therefore necessarily marginal, and many of them seem
doomed to remain single or to marry at a late age.

Summary and conclusions

In a polygynous culture, co-wives may be perceived as direct
(through their work) and indirect (through their offspring)
sources of increased income and prestige. In other words, the
functions ascribed to polygyny may be instrumental in character.

13. On the general effects of laws on changes in familial organization, see

Baker and Bird (1959).
14. For a description of the African educated woman, see Desanti,
(1962).

R. Ciignet 177
Yet, although they augment the family’s social and economic
resources, additional co-wives are also a privilege reserved to
individuals who initially hold higher than average positions. In
other words, the functions ascribed to plural marriage may also
be symbolic and closely related to social differentiation.
The recruitment of co-wives, the size of the surplus of marriage¬
able women, the instrumental or symbolic nature of polygyny
and the variety of motivations experienced by individual actors -
all these factors should affect the functioning of polygynous
families and introduce variations both within and among dis¬
tinctive cultures in the style of interaction of a senior co-wife
with her husband, her family of origin, the other spouses of her
husband and her own children.

References
Baker, T., and Bird, M. (1959), ‘Urbanization and the position of
women’, Sociol. Rev., vol. 7, pp. 110 ff.
B anton M. (1957), A Study of Tribal Life in Freetown, Oxford University
Press.
Bird, M. (1963), ‘Urbanization, family, and marriage in Western Nigeria’,
in Urbanization in African Social Change, Proc. Inaugural Seminar
in Centre of African Studies, University of Edinburgh, pp. 59-74.
Blake, J., and Davis, K. (1956), ‘Social structure and fertility: an
analytic framework’, Econ. Devel. Cultural Change, vol. 4, pp 211-35.
Bohannan, D. (1963), Social Anthropology, Holt, Rinehart & Winston.
Brass, W. et al. (eds.) (1968), The Demography of Tropical Africa,
Princeton University Press.
Clement, P. (1955), ‘Social patterns of urban life’ in D. Forde (ed.),
Social Implications of Industrialization and Urbanization in Africa South
of the Sahara, UNESCO, pp. 393-469.
Desanti, D. (1962), ‘Quand l’Africain revient d’Europe’, in Le Dossier
Afrique, Marabout University, pp. 156-63.
Dorjahn, V. (1954) The Demographic Aspects of African Polygvmy,
doctoral thesis, Northwestern University.
Dorjahn, V. (1959), ‘The factor of polygamy in African demography’,
in M. Herskovits and W. Bascom (eds.), Continuity and Change in
African Cultures, University of Chicago Press, pp. 87-112.
Gamble, D. (1963), ‘Family organization in new towns in Sierra
Leone’, in Urbanization in African Social Change,
Proc. Inaugural Seminar in Centre of African Studies,
University of Edinburgh, pp. 75-84.
Gough, K. (1961), ‘Preferential marriage forms’, in D. Schneider and
K. Gough (eds.), Matrilineal Kinship, University of California Press.

178 Plural Marriage

Joseph, G. (1913), ‘La condition de la femme en Cote d’Ivoire’, Bull,
soc. Anthropol. de Paris, vol. 4 pp. 585-9.
Kobben, A. (1956), ‘Le planteur noir’, Etudes eburneennes, vol. 5,
Centre I FAN.
Little, K. (1959), ‘Some patterns of marriage and domesticity in West
Africa’, Sociol. Rev., vol. 7, pp. 65-81.
Masse, L. (1955), ‘Preliminary results of demographic surveys in the
urban centres of Senegal’, in D. Forde (ed.), Social Implications
of Industrialization and Urbanization in Africa South of the Sahara,
UNESCO, pp. 523-35.
Mercier, P. (1955), ‘An experiment investigation into occupational and
social categories in Dakar’, in D. Forde (ed.), Social Implications of
Industrialization and Urbanization in Africa South of the Sahara,
UNESCO, pp. 510-23.
Mercier, P. (1960), ‘Etude du mariage et enquete urbaine’, Cahiers
d’etudes africaines, vol. 1, pp. 28-43.
Moore, F. (ed.) (1961), Readings in Cross-Cultural Methodology,
Human Relations Area Files Press, pp. 193-216.
Monteil, V. (1964), VIslam Noir, Le Seuil.
Muhsam, H. V. (1956), ‘Fertility of polygynous marriages’, Pop.
Stud., vol. 10, pp. 3-16.
Murdock, G. P. (1957), ‘World ethnographic sample’, Amer.
Anthropol., vol. 69, pp. 664-87.
Osmond, M. (1965), ‘Correlates of types of marriage’, Soc. Forces,
vol. 44, pp. 8-16.
Sawyer, J., and LeVine, R. (1966), ‘Cultural dimensions: factor
analysis of the World Ethnographic Sample’, Amer. Anthropol., vol. 68,
pp. 708-31.
Tardits, C. (1958), Les Nouvelles Generations Africaines entre leurs
Traditions et TOccident, Mouton.
Thore, L. (1964), ‘Monogamie et polygamie en Afrique noire’. Rev.
action pop., pp. 807-21.
Whiting, J. (1964), ‘Effects of climate on certain cultural practices’,
in W. Goodenough (ed.), Explorations in Cultural Anthropology,
McGraw-Hill.
Winch, R. (1964), The Modern Family, Holt, Rinehart & Winston.
Zelditch, M. Jr (1964), ‘Cross-cultural analysis of family structures’,
in H. Christensen (ed.), Handbook of Marriage and the Family, Rand
McNally.

R. Clignet 179
Part Seven Marriage and Alliance

The marriage of cross-cousins (that is, children of parents’

siblings of opposite sex) has been a major focus of inquiry since
the topic was studied by J. G. Frazer and E. B. Tylor. It was the
subject of Levi-Strauss’s work, The Elementary Structures of
Kinship (1969), a study much influenced in its general orientation
by Durkheim and Mauss. In it he examined the various types of
marriage exchange found in Western Asia and in Australia. The
approach behind this study, with its emphasis upon the
distinction between the ties created by marriage (alliance,
affinity) and those created by birth (descent, filiation), was
applied to South Indian societies by Louis Dumont in his
monograph, Une Sous-caste de TInde du Sud (1957), and in a
general paper, ‘Hierarchy and marriage alliance in South Indian
kinship’, from which Reading 11 is taken. The discussion of
cross-cousin marriage has been mainly concerned with those
systems that can be described as ‘prescriptive’, where the
category of spouse (e.g. mother’s brother’s daughter) is
unambiguously laid down. In his paper on ‘Prescriptive
Marriage Systems’, Maybury-Lewis summarizes the recent
discussions on this subject, pointing out that we must see
cross-cousin marriage as one type of a more general category of
systems of prescriptive marriage, which are characterized by the
marriage alliance.

References
Dumont, L. (1957), Une Sous-Caste de Vlnde du Sud: Organisation Sociale
et Religion des Pramalai Kallar, Mouton.
Le vi-Strauss, C. (1969), The Elementary Structures of Kinship,
Eyre & Spottiswoode. First published 1949.
13 L. Dumont
The Marriage Alliance

Excerpt from L. Dumont, ‘ Hierarchy and marriage alliance in South

Indian kinship’, Occasional Paper no. 12, 1957, Royal Anthropological
Institute.

To introduce an institution which is shared by all groups referred

to here and, I believe, by many others, some criticism of current
anthropological categories is first necessary.

Marriage regulations and affinity

A (positive) marriage regulation like ‘a man should marry his
mother’s brother’s daughter’ might be a native statement in¬
dicating whom a given individual should marry. But, in anthrop¬
ological thought, it takes on a slightly different meaning. There it
appears as a rule for deriving a man’s marriage from a relationship
excluding any idea of marriage or affinity, i.e. from a relationship
of consanguinity. It is implied that consanguinity is pre-existent
to the marriage, since I must be born before I marry. The marriage
regulation is in fact used as a tool for deducing a secondary
category (a certain marriage) from a primary category (a certain
relationship of consanguinity). After marriage is so introduced, it
brings with it relationships of a secondary kind (affinal relation¬
ships) which are never considered as full kinship relationships,
because they are individual and above all temporary - they disap¬
pear with the married person and are not transmitted to his or her
descendants but under the form of a consanguinity relationship.
I submit that all this is wrong and needs revision for the follow¬
ing reasons: (1) it rests only on undue generalization of our com-
monsense categories, and does not do justice to the facts because
in our societies marriage is an individual affair, not positively
regulated; (2) it is contradictory for, as I shall show, the very
existence of the marriage regulation implies that affinity is trans¬
mitted from one generation to the next just as consanguinity ties
are. We have thus to give a proper definition of marriage regulation

L. Dumont 183
on the one hand and to widen our concept of affinity on the other.
First, it is almost unnecessary to recall that marriage cannot in
general be considered as a secondary product of other institutions
such as descent, which are then taken as being primary; there is
rather an interrelation in the complete make-up. Still less it is
possible to reduce the content of the marriage regulation to the
codification of an individual affair, which marriage is not. Conse¬
quently, the regulation should not be considered as consisting of
a relation between consanguineous ties and affinity, but as a
feature of affinity itself. It is possible to do so, by pointing out that
the regulation determines one’s marriage by reference to one’s
ascendants ’ marriages: in a patrilineal, patrilocal society, marrying
the matrilateral cross-cousin means reproducing the marriage of
one’s father, while in the patrilateral formula one reproduces one’s
grandfather’s marriage, and so on. In general, the regulation
determines a cycle of repetition of a marriage of a certain sort. If
we say that ‘one marries one’s cross-cousin’, we merely state a
condition to be observed in order to maintain a certain pattern of
intermarriage.
In other words, the regulation causes marriage to be transmitted
much as membership in the descent group is transmitted. With it,
marriage acquires a diachronic dimension, it becomes an institut¬
ion enduring from generation to generation, which I therefore
call ‘marriage alliance’, or simply ‘alliance’.
In the matter of affinity, we generally admit too readily that,
while the relationship between a man and his brother-in-law is
affinal, the relationship between their sons (cross-cousins) has
no longer any affinal content, but is a mere consanguineous re¬
lationship. This is certainly not so in South India, where to call E
and A cross-cousins- as in Figure 1 (b) - instead of ‘ sons of affines ’ -
as in Figure 1 (a) - is quite deceptive. Being sons of affines, they
are ipso facto affines, at least in a virtual or rather a general sense,
before or without becoming so individually, as when E marries
A’s sister. We are now out of the vicious circle and we can look
at it with amusement: ‘marrying a cross-cousin’,1 is nothing but
1. Briffault (1927, vol. 1, pp. 563 ff.) rightly uses the Tamil term for
‘cross-cousin’, machuna (for maccuNaN) to stress the affinal content of the
category; he speaks of a ‘marriage agreement between two groups’.
Aiyappan (1944, p. 68) classifies the sister’s son among the bandhukkal or
affines, but contradicts it in a footnote.

184 Marriage and Alliance

marrying an affine, i.e. the person who is the closest affine by virtue
of the transmission of affinity ties from one generation to the next.
I submit that, in societies where there are (positive) marriage
regulations: (1) marriage should be considered as a part of a
marriage alliance institution running through generations; (2) the
concept of affinity should be extended so as to include not only

(a) / \ (b)
A 8 " 6 “ 1
/

IM

IM
A A A
A E A
Figure t Cross-cousins or affines
(a) A as an 'affine' of E
(b) A as a 'cross-cousin' of E

immediate, individual relationships (affines in the ordinary sense)

but also the people who inherit such a relationship from their
parents, those who share it as siblings of the individual affines,
etc.; (3) there is likely to be an affinal content in terms which are
generally considered to connote consanguinity or ‘genealogical’
relationships (such as ‘mother’s brother’ etc.). This is obviously
so when there are no special terms for affines, for otherwise we
should have to admit that in such cases affinity is not expressed at
all.

Terminological dichotomy: kin and affines

Structure of terminology
All our groups share with many others a structurally identical
terminology which in its broad features has been recorded from
all four written Dravidian languages. Here I shall summarize a
separate study (Dumont, 1953a; see also Radcliffe-Brown, 1953;
Dumont 19536).
The two sexes should be taken separately. With certain ex¬
ceptions there is one term for all males in the grandfather’s
generation, but two terms in the father’s generation. The latter
terms, generally translated as ‘father’ and ‘mother’s brother’,

L. Dumont 185
denote two classes, the members of which are respectively brothers-
in-law to one another. Or, if we call ‘alliance relationship’ this
relationship between two persons of the same sex, and represent
it by A[=] A, standing for A =6a as well as for aZ>— A etc.,
the relation between the two classes is A[=] A. This is true also
in Ego’s generation (for older and younger relatives), whereas
the distinction does not fully operate in Ego’s son’s generation,
where a mere prefix is used, and disappears in the grandson’s
generation. Terms for males are recapitulated in Figure 2 (a).

(a)
generations terms

grandfather A

father A[=]A

>e A[-]A
ego’ -
<e a[=]a

son 2A

grandson A

Figure 2 Structure of the system of kinship terms.

(a) terms for males only, five generations; (b) terms for both sexes, five generations.
The superposition of signs shows identity of terms, apart from word-endings

Among females, the ‘mother’ and the ‘father’s sister’ may be

distinguished exactly as above. Now if we remark (1) that the terms
for grandmother and grand-daughter are not distinct, except for
the ending, from those for grandfather and grandson, and that the
same root is used for all in the son’s generation; and (2) that the
principle of distinction is the same for males and for females, we
can represent the whole by a symmetrical scheme in which the
identity of terms is expressed by superpositions of signs - see
Figure 2 (b). One sees that the distinction of sex and the alliance
distinction go together, and that the system might be called ‘bi¬
furcate-merging’in a new sense, that is, bifurcate in the central and

186 Marriage and Alliance

merging in the extreme generations. One sees also how simple and
regular the system looks once one ceases to remove alliance arti¬
ficially from the content of kinship terms. It consists in distin¬
guishing, in three generations or age groups, two kinds of relatives
of each sex: those related to Ego by a link excluding alliance, or
‘kin link’ (Figure 2 (a)), and those related to the first by alliance
(Figure 2 (b)). From the basic structure of the system we have on
one side the ‘ fathers ’, on the other the ‘ fathers’ affines ’, and on
one side the ‘mothers’, on the other the ‘mothers’ affines’; and
nowhere such beings as mother’s brother or father’s sister, who
are just particular cases of fathers’ affines and mothers’ affines.
Obviously, too, the system implies a marriage regulation, namely
that one marries an affine in one’s generation, the nearest of these
in terms of individual relationship being a cross-cousin.

Definitions
The whole of the kinship terminology is split into two halves, ‘kin’
terms and ‘ alliance ’ or ‘ affinity ’ terms. By thus stating that kinship
= kin + affinity, we escape an ambiguity found in anthropological
writings, where ‘kin’ or ‘kinship’ is sometimes opposed to
affinity and sometimes taken as embracing it. We prefer to speak of
kin and affines rather than of parallel and cross-relatives. We
should not however forget that these are only categories abstracted
by us from the form of the terminological system. To avoid any
confusion when actual kinship configurations are studied, we
should designate them as ‘terminological kin’ and ‘terminological
affines’. Moreover, as the latter expression has been obtained by
extending the meaning of the word ‘affine’, we should distinguish
between (a) immediate or synchronic affines, i.e. affines in the
ordinary sense, in-laws, and (b) genealogical or diachronic
affines, who inherit, so to speak, an affinal tie which originated in an
upper generation (e.g. mother’s brother). When the marriage
regulation is observed, the two categories (a) and (b) merge, and
a person is at the same time a genealogical and an immediate
affine, what might be called a perfect affine.

How the terminology is applied

The system as analysed hitherto is no more than an abstract frame
of reference, no doubt pointing to alliance as a fundamental

L. Dumont 187
institution, but one to which each social group will give a particular
concrete form according to its particular institutions. (We may
imagine, for instance, that in a matrilineal matrilocal society the
kin link is with the mother’s brother while the father is an affine;
then their ordinary positions in the system would be reversed,
without the structure being altered.) Among our groups, the actual
relationship with the father’s sister may vary, but there will always
be found as a common background the fact that she is different
from a ‘ mother ’, first of all in the sense that Ego may marry her
daughter even if she is not the preferred mate, while he may not
marry a ‘mother’s’ daughter.
Again, within one and the same terminological class the dis¬
tinction of particular relatives, whether expressed or not in the
language, may differ from one group to another. The broad
opposition between kin and affines suggested by the terminology
will itself be unequally realized for different relatives: among the
terminological kin, a part is singled out as really or fully kin,
and the same happens among terminological affines. The choice
varies, but all the different choices fit into the general terminological
frame, no doubt because they fit into a common alliance pattern.
Another general difference is found in the degree to which re¬
lationships are extended from the groups of siblings to extensive
socio-political groups. A comparison will show how the termin¬
ological categories are given different shapes by different in¬
stitutions.
The Pramalai Kallar are patrilineal and patrilocal. In one
locality there is as a rule only one or a few patrilineages. Hence:
1. The category of ‘ brothers ’ is split into two: on the one hand all
my paternal ‘brothers’ (sons of my father’s brothers etc.) are
members of my local descent group, on the other hand my
maternal ‘ brothers ’ (sons of my mother’s sisters etc.) are spread
over different groups and places. On the paternal side each indi¬
vidual link is made to endure through generations by becoming an
element among all other similar elements in the continuous fabric
of the local group, clearly defined as against others by the exo-
gamic rule. On the maternal side, each individual relationship,
being isolated - see however (2) below - is liable to be rapidly
forgotten. While the paternal half is stressed to the point of
becoming almost equivalent to the whole, the maternal half

188 Marriage and Alliance

appears, except in special cases, as temporary, subsidiary, almost
conventional.
2. The opposition between kin and affines takes on a spatial aspect;
there are kin places and affinal places, and as one marries mostly
in the neighbourhood, this might be represented ideally in the
form of concentric circles, a territorial unit made up of kin, A,
being surrounded by affinal places, B. As these in their turn
intermarry with other places, there arises a third circle, C, made
up of people who are affines to B, and hence ‘brothers’ to A, as
the affine of my affine is kin to me. (In these C places will be found
some of the maternal brothers mentioned above.) The matter is
of course not so simple in fact, but on the whole, when seen from
one point, there is a picture of the division of the not-too-far-
removed localities into the two fundamental categories. It will be
readily grasped that this apparent dichotomy in space results from
the working of the organization and has nothing to do with a
systematic division or a dual organization of the society. Neverthe¬
less it represents a maximum in the extension to groups of the
basic terminological categories.
3. Let us now compare the position of the mother’s brother with
that of the father’s sister. The mother’s brother is not only termin-
ologically opposed to the father, who is here kin par excellence,
but he also lives in an affinal place. He is an affine pure and simple,
in fact the closest, at least until one marries. On the contrary, the
father’s sister, born in one’s local descent group, becomes only
with her marriage a member of an affinal group, just as the mother,
born in an affinal group, has become kin first as mother, at the
same time thrusting, so to speak, the father’s sister into the affinal
category. The terminology here directs us to look at the father’s
sister as already married and as mother of affinal cousins. Never¬
theless she is at the same time to some extent kin, and it follows that
she is less clearly and unambiguously an affine than the mother’s
brother. If we then suppose, as will be confirmed, that affinal
relatives are in charge of ceremonial functions, we may expect
the mother’s brother to precede the father’s sister in those functions.
The picture is quite different for the matrilineal, patrilocal
Kondaiyam Kottai Maravar. With them descent and locality
work in opposite directions, with the result that individual

L. Dumont 189
kinship relationships are not backed by corresponding relation¬
ships between groups. Here the sons of two brothers on the one
hand and of two sisters on the other are recognized as ‘brothers’
in two different ways and the two kinds of relationship are stressed
in quite different conditions, the first in a context of locality and the
second in a context of alliance or of special ceremonial circum¬
stances. On one side, a remote relationship between patrilocal
brothers does not in general exclude alliance, so that the category
in the long run is stripped of any kin content, being mainly a
matter of socio-economic neighbourliness. On the other side, it is
between sisters (and not brothers) and their descendants that the
matrilineal kin relationship endures. The descent group has no
tangible reality. What is stressed here is a matriline scattered in
different localities, shifting from place to place and from house to
house in each generation. In every locality a number of matrilineal
exogamous units are represented, and a man may marry into any
of them, except his own. In one’s own village the terminological
categories are fully realized only for three kinds of people: a
smaller or larger circle of patrilocal brothers, a number of matri¬
lineal brothers, and the affines of the first two. At the same time,
a great number of people are undifferentiated: they may be at the
same time brothers in a loose, merely local sense, and virtual
affines, and it is only the nexus of individual alliances and their
classificatory extensions which decides the question.
The opposition between father and mother’s brother is seen
here in different ways. From the matrilineal point of view the
situation would be reversed, the mother’s brother could be con¬
sidered as kin and the father as an affine, but nevertheless the
mother’s brother’s children will be terminological affines. We see
that it is the mother’s brother who receives the ambiguous charac¬
ter which attaches to the father’s sister among Pramalai Kallar. In
contradistinction to them, the foremost affine here is the father’s
sister, because locality is not exclusive of alliance and because
matrilineality stresses the kin link with the mother. This will be
confirmed later, when we study the ceremonial functions.
If the two preceding examples are compared, the difference in
the affinal value of the mother’s brother and of the father’s sister
can perhaps be summed up by saying that when paternal features
(authority, locality) are present, the foremost affinal relative in

190 Marriage and Alliance

the upper generation is the affine of the lineally-stressed parent,
i.e. the mother’s brother in patrilineality and the father’s sister in
matrilineality. This is only another expression of harmony and
disharmony (Figure 3). It is hoped that this brief comparison has
shown how we may speak of a common underlying alliance
pattern which, when combined with different institutions, assumes
different concrete forms.

(a) (b)

<7 = <7 A

Pramalai Kallar Kondaiyam Kottai Maravar

Figure 3 Stress on one affine in relation to descent: Pramalai Kallar and

Kondaiyam Kottai Maravar

Inheritance and gifts

The most conspicuous feature of alliance as an enduring marriage
institution that defines and links the two kinds of relatives consists
in ceremonial gifts and functions. This perspective can be in¬
directly justified. If ceremonial gifts are essentially affinal and if
they are important, it should follow that, in societies with male
predominance, property is transmitted from one generation to the
next under two forms: by inheritance in the male line and also by
gifts to in-laws, namely from father-in-law to son-in-law. This is
precisely what happens. In the groups with which we are im¬
mediately concerned, apart from the Nangudi Vellalar among
whom female property is important, daughters have no formal
share in their father’s property, but they are entitled to mainten¬
ance and to the expenses necessary for their marriage and estab¬
lishment.
Moreover, this is a case for generalization. The same rule, if it
is not absolute and universal, has a widespread validity in Indian
customary law, where it makes itself felt even when it is contra¬
dicted (Jolly, 1896, p. 83; Mayne, 1938, paragraphs 436, 421, 431,
488, etc.). The marriage expenses should not be taken as including
only the cost of the necessary feasting and display, but also that of

L. Dumont 191
gifts to the in-laws on the occasion of marriage itself and later on as
well. If the details vary, the broad institution is general, at least in
the Tamil country, even among well-to-do people. This double
transmission of property confirms the opposition between kin and
alliance. It indicates that a review of ceremonial gifts must begin
with marriage.

Marriage gifts
What is the most salient feature of the marriage ceremonies among
the groups referred to? Sacramental acts like uniting hands or
circumambulating the fire are not found. The tying of a string,
with or without the well-known marriage badge or tali, round the
bride’s neck has certainly a sacramental value, especially for the
bride. But it is not witnessed by all relatives, because the cere¬
monies take place partly in the bride’s and partly in the bride¬
groom’s house, and only a few people go from one to the other.
This explains why the tying of the tali was sometimes repeated
(Pramalai Kallar). A common meeting of the relatives of both
sides is conspicuously absent. As a sign of union between the two
families, I think we may say that it is replaced by the long series of
alternate shiftings of the couple from one place to the other and
back and again, which takes place from the marriage onwards and
is accompanied by gifts in one direction and increased gifts in
return. This chain of gifts, or ‘prestations’ and ‘counter-presta¬
tions’2 symbolizes the alliance tie and is the most important
feature of marriage ceremonies from the point of view of the
relation between the two families.
The Pramalai Kallar state with particular emphasis that ‘gifts
sent to the bride’s house return increased twofold or threefold’.
Among them, the man’s family (which I shall designate as M)
gives first a sum of money, parigam, to the woman’s family
(designated as F) which has to spend at least twice as much for the
bride’s jewels. Then with the ceremony proper begins the series of
visits to and stays with F, the couple being every time accompanied
by a number of baskets (sir), containing foodstuffs and other

2. Prestation, ‘the action of paying, in money or service, what is due by

law or custom, or feudally; a payment or the performance of a service so
imposed or exacted, also, the performance of something promised’ (Shorter
Oxford Dictionary).

192 Marriage and Alliance

 articles for consumption, from M to F and, increased, back from
F to M. Prestations from F dominate more and more as time goes
on until finally - it may be two or three years after the marriage
ceremony - the young couple establish a separate household near
M, and receive the necessary pots and pans from F without any
return gift. This is the ‘gir of going apart’.
Such are the main prestations, which I call external prestations
in order to distinguish them from the following. During the
marriage ceremony, in both houses, money is collected among the
the bridegroom’s relatives on the one hand and the bride’s relatives
on the other. This is called moy; its effect is to make the relatives
contribute to the expenses of the family; it may be called an internal
prestation. These two kinds of prestations are found in most
other groups, internal prestations being likewise called moy,
whereas there is no general term for external prestations, some¬
times called guru} if they consist of gifts in money and gir if in
kind. Linguistically, while moy indicates a mere collection (‘ crowd ’,
‘multitude’), guru} connotes a circular movement, a rolling up,
perhaps a circular accumulation. Among the Nangudi all pres¬
tations are external, and people say that the moy has been replaced
by collections where the two groups of relatives are mixed (iNam,
a solemn word for ‘gift’). The moy is not found in connection
with marriage in Mudukkulattur, where it is known on other
ceremonial occasions, i.e. girl’s puberty and funerals. In Paganeri,
internal prestations comprise a contribution in rice brought by
all taking part in the feast (which may well be more widespread)
and also a collection of money similar to the moy, bearing the
name of revei or regei, ‘list’, and accompanied by small gifts of
thanks in return. Among the Arupangu, the two moy are associ¬
ated with a series of small external gifts.
In considering the external prestations, it is necessary to single
out the matrilineal and matrilocal Nangudi who do not make
reciprocal gifts. The pattern is definitely different: M’s prestations
are very slight on a ceremonial level; there is no parigam, no
gift of tali or sari; the idea of competition is absent; the cost of the
feast is shared afterwards between the two families. The parents
of the bride make a point of providing everything except a food¬
stuff allowance, the same in all cases, which has to be regularly
delivered by M to the new household. Moreover, the emphasis is

T-K-G L. Dumont 193

here on the dowry, strictly the wife’s property. In Paganeri also
the reciprocity in gifts is weaker, but there a part of the usual
prestations from M is found.
Otherwise the comparison shows that, while there are all possible
variations for each element in particular, the whole is more uni¬
form than its parts, and still more so is the form of small cycles
inside the whole cycle. Leaving aside once for all the Nangudi
Vellalar, we see that the tali is everywhere paid by M. The gift of
one or two saris to the bride by M is lacking among the Pramalai,
so is the parigam in Paganeri and among the Arupangu, while it
is present in Mudukkulattur, and present but small among the
Ambalakkarar. As a counterpart, the importance of jewels and
dowry varies. Land is given among the Ambalakkarar and in
Paganeri. Jewels are important in those two groups, but on the
contrary their value is hardly mentioned at all in Mudukkulattur.
This is obviously related to the economic situation, for the Maravar
of Mudukkulattur are very poor and the Arupangu occupy an
intermediary position between them and Paganeri or the Nangudi.
The masculine gir brought for the ceremony is found everywhere,
Nangudi excepted. Among the Arupangu, it is like that of the
Pramalai, with one sari added. In Mudukkulattur, where it is
preceded by another one for the betrothal, it includes rice. This
also is seen, with more rice, in Paganeri and among the Ambalak¬
karar. In return the gift is multiplied thrice in Mudukkulattur,
while in Paganeri the increase is marked in a different way, by the
additions of pots and pans. The return gift is lacking among the
Ambalakkarar and the Arupangu. In the latter group, it is prob¬
ably only delayed on the one hand (pir of the first visit to F) and,
on the other, there is another form of reciprocity, that of gurul,
as will be seen below.
Regarding subsequent gir gifts, the difference between the
groups bears on the choice of the most important dates. The
household equipment will be offered, here on the Pongal festival
(in January), there on the first visit to F; the gifts of the month of
adi will be more or less important, etc. The gurul or external gifts
in money are found among the Arupangu (1) from F to M as a
return for clothes, (2) from F to M and back (individualized, as in
Paganeri, where the gurul is a small gift from the wife’s father to
the husband and from the husband’s mother to the wife).

194 Marriage and Alliance

 The common mechanism of the gifts appears clearly if one iso¬
lates small cycles based on reciprocity. There are three types. In
the first of these, there is an exact reciprocity, as in Paganeri
when clothes are given by M to F and then equivalent clothes by
F to M; among the Arupangu the clothes presented by M to F
are compensated for by a reverse gift in money (gurut) which must
be at least equal to their value. In a second type, the initial gift is
not only reciprocated, but multiplied in return, as among the
Pramalai. A third type has a reduced reciprocity, marked sometimes
by a mere symbolical counter-gift. This is true among the Pramalai
and in Paganeri for the moy; among the Arupangu the gurul which
has just been mentioned is in its turn followed by a symbolical
return. It can also be shown how one given object receives a
particular ceremonial value from its situation in the whole. This
is so with rice among the Pramalai, where it does not occur in
internal prestations nor in the masculine, but only in the feminine
gir as a sign of their substantial importance, i.e. of the ‘increase’
which characterizes them. On the contrary, rice is to be found
everywhere in Paganeri and, to express the pre-eminence of the
feminine gir, one resorts to another element, namely the pots and
pans which elsewhere appear only later.
The foregoing comparison will have shown how much stress is
laid on the chain of prestations in all its details, and also in what
sense it may be said to be common in spite of all variations. It is
clearly impossible to single out one of the marriage ‘payments’,
theparigam, and to call it ‘bride-price’.3 It represents, at least in
the examples cited here, the contribution of the husband’s family
to the buying of jewels which will be worn by the wife but normally
become the property of the household, as can be ascertained from
their treatment in case of divorce. That terms like ‘bride-price’
are inaccurate here is also obvious if one considers that, on the
whole, and to varying degrees, it is the wife’s family that gives
more. The parigam appears rather as a kind of earnest-money
which is destined to come back increased. It would be almost

3. For a contrary view see Srinivas (1942, ch. 2), but the prestations are
not analysed. Mousset and Dupuis (1928) wisely translate parigam by the
French arrhes. An exchange of gifts similar to those found here, but with
mercantile features, has been described among the Nattukkottai Chettiar
(Thurston and Rangachari, 1909, vol. 5, pp. 265 ff.).

L. Dumont 195
equally misleading to reduce the whole to * dowry ’, in cases when
something of the kind actually appears. These are rather extreme
cases among the rich and when patrilateral marriage promises a
return in the next generation. The transfer of meaning of the
classical word for dowry (Sanskrit stridhana, ‘wife’s property’)
among the Pramalai Kallar is characteristic, since they call
gridaNam, gldaNam all gifts due by the wife’s family, including
the future gifts of her brother to his sister’s children. Obviously
the meaning of the protracted exchange of gifts with which we are
dealing, if it includes the final result in terms of plus and minus,
goes far beyond this. On the whole, the final result is a gift which
accompanies the gift of the girl. A relevant question here would
be to ask why a transaction which finally amounts to a gift has to
take the form of an exchange. I would say that it corresponds to
the individual marriage’s (i.e. gift’s) being conceived of as a part
of the whole nexus of intermarriages and their consequences as seen
from the point of view of a single family (i.e. exchange). When the
Pramalai Kallar state that ‘gifts sent to the bride’s house return
increased’, this is roughly true of one individual present, but it is
still truer of the whole series, or rather it is true in the sense that it
accounts for each exchange as seen in the light of the whole. There
is certainty about increase, because increase is the law of the whole
cycle. One knows very well that masculine gifts will decrease as
time goes on, while feminine gifts will increase; the latter are
substantial, the former initiatory and provocative. Generosity
lies on the girl’s side, but it has to be set in motion; a pledge to
protracted, manifold, and mainly unilateral gifts is obtained by a
formal exchange.
Indeed, this may be taken as a formula of Kallar marriage if one
accepts the view that these prestations - and not the ‘ritual’
elements on which attention has been mainly focused - constitute
the main part of marriage ceremonies. In favour of this view, the
first argument is that prestations in fact do not stop at the point
we have somewhat arbitrarily chosen. Those which follow may be
called ‘ alliance prestations ’ and I shall trace them in all ceremonial
circumstances of the individual’s life. Marriage does not consist
only in the consecration of conjugal union and the establishment
of a new family, for this family is as inseparable from the alliance
prestations as it is from the local lineage affiliation.

196 Marriage and Alliance

 The alliance prestations are of two kinds: some are symmetrical
or reversible and some are asymmetrical or oriented. If, after the
marriage has been celebrated, a death occurs in the bride’s
family, the bridegroom’s family, together with the other affines,
will bring food presents. These I call ‘reversible’ gifts because,
the bride’s family would do the same if a death occurred in the
bridegroom’s. This reversible relationship we find reflected in the
transport of gifts which accompany the young couple both ways
after marriage. It is the most general and undifferentiated ex¬
pression of alliance in gifts. The birth of a child in the new family
will create a different situation which has no counterpart in the
bride’s family. This is an ‘oriented’ situation, where the Kallar
will stress the gifts and functions of the maternal uncle. It should be
added that one and the same ceremonial occasion calls for the two
kinds of responses and gifts from different people. Whereas
many people come and give the ordinary, reversible, affinal gift,
one particular relative, who may be the maternal uncle, is singled
out with particular, oriented gifts and functions: the oriented
relationship stands against a background of reversible relation¬
ships. Both kinds of relationships are initiated (or renewed) in
marriage, and this corresponds to the double aspect of exchange
and gift in the marriage prestations: ‘gifts sent to the bride’s
house return increased ’.
That affinal ceremonial prestations in general constitute some¬
thing like the core of family ceremonies is shown not only by their
description, but also by the fact that it is possible for the people to
dissociate what might be called the mere rite and the accompani¬
ment of prestations which overshadow it. This is true of marriage
among the Maravar (Thurston and Rangachari, 1909, pp. 37-8)
and of funerals, on which occasions the prestations may be post¬
poned. It is true also of circumcision, as two striking instances will
show. Among the Pramalai Kallar, the circumcision of a boy is a
source of income for the family. Therefore, two brothers are never
circumcised together when their age would permit it. Further, old
parents who are said to be anxious to see the circumcision while
they are alive may have the ceremony performed several years in
advance, but the operation itself will take place later and without
any ceremony. The Ambalakkarar had for girls a ceremony
parallel to the circumcision of boys, but with no technical

L. Dumont 197
counterpart. This had nothing to do with the common girls’
puberty ceremony, although the two have been sometimes con¬
fused (Francis, 1914, p. 94). An informant states that this ceremony
originated because parents who had no sons but only daughters
wished to celebrate it as well as the others.

References
Aiyappan, A. (1944), ‘Iravas and culture change’, Bull. Madras
Govt. Mus., General Section, vol. 5, no. 1.
Briffault, R. (1927), The Mothers, vol. 1, Allen & Unwin.
Dumont, L. (1953a), ‘The Dravidian kinship terminology as an
expression of marriage’, Man, vol. 53, art. 54.
Dumont, L. (1953b), ‘Dravidian kinship terminology’, Man, vol. 53,
art. 224.
Francis, W. (1914), ‘Madura’, Madras District Gazetteers, vol. 1,
Madras Government Press.
Jolly, J. (1896), ‘Recht und Sitte (einschliesslich der einheimischen
Literatur)’, Grundriss Indo-Arischen Philolog. Altertumskunde, vol. 11,
no. 8.
Mayne, J. D. (1938), A Treatise on Hindu Law and Usage, 10th edn.,
Higginbotham, Madras.
Mousset, A., and Dupuis, B. (1928), Dictionnaire tamoul-francais,
Pondichery, Societe des Missions Etrangeres.
Radcliffe-Brown, A. R. (1953), ‘Dravidian kinship terminology’,
Man, vol. 53, art. 169.
Srinivas, M. N. (1942), Marriage and Family in Mysore, Bombay
New Book Co.
Thurston, E., and Rangachari, K. (1909), Castes and Tribes of
Southern India, Madras Government Press.

198 Marriage and Alliance

14 D. Maybury-Lewis
Prescriptive Marriage Systems1
D. Maybury-Lewis, ‘Prescriptive marriage systems’. Southwestern
Journal of Anthropology, vol. 21, 1965, pp. 207-30.

This paper attempts to clarify some basic issues in the study of

prescriptive marriage systems. It might seem presumptuous of me
to offer a clarification of points which have already been thoroughly
and vigorously discussed. I do so only because it seems to me that
the discussions have lately, and inevitably, become enmeshed in
their own dialectic of critique and rejoinder and are therefore no
longer followed by many anthropologists whose attitudes, ex¬
pressed informally, vary from ‘A plague on all their houses’ to
‘What is the importance of it all anyway?’ I believe that the issues
are important and that a summary of them at this stage would
serve a useful purpose, if only perhaps that of avoiding future
disputes based on misunderstandings. At the same time I am, with
a group of colleagues and students from Harvard and Rio de
Janeiro, engaged on the study of a number of central Brazilian
societies possessing two-section systems and/or exogamous
moieties. The issues with which I deal here are of direct relevance
to that enquiry.
The systems which I shall discuss are those which Needham has
termed ‘prescriptive’. In so doing I say nothing about the desir¬
ability of treating prescriptive and preferential marriage systems
together or separately. I maintain only that an analytical distinc¬
tion must be made between them. The justification for this distinc¬
tion follows from the discussion of what in fact is meant by a
prescriptive marriage system.
Previous discussions have dealt with three types of prescriptive
marriage system:

1. I would like to express my gratitude to Dr Rodney Needham of Oxford

University, whose critical comments on a draft of this paper have been most
helpful.

D. Maybury-Lewis 199
1. Prescriptive matrilateral cross-cousin marriage (also known as
MBD2 marriage).
2. Prescriptive patrilateral cross-cousin marriage (also known as
FZD marriage).
3. Prescriptive bilateral cross-cousin marriage (also known as
marriage with the M B D/F Z D).
This terminology dates at least from the time of Rivers (1914)
and has been the source of much confusion. It has long been
realized that the terms are unsatisfactory or even misleading (see,
for example, Dumont, 1953, 1957a; Loffier, 1964). Yet anthro¬
pologists have continued to use them with the mental proviso,
sometimes (though not always) made explicit, that when they
wrote, for example, of marriage with the MBD, they really meant
something else. It is therefore important at the outset to give a
summary statement of what this ‘something else’ is.
Let us first take note of what it is not. A system of prescriptive
cross-cousin marriage does not, in the usage adopted here, denote
a system where marriage is prescribed with an individual in a
specific genealogical relationship to Ego. Nor is it a system where
Ego must marry a certain cross-cousin ‘real or classificatory’.
The notion of ‘marrying a cross-cousin’ is an analytical one,
introduced by anthropologists and purporting to describe what
happens in such systems. It is important to remember, however,
that the rule in societies which practise ‘cross-cousin marriage’
is not phrased in terms of cross-cousins at all. Thus, as Needham
pointed out (1962, p. 9), the Batak do not prescribe marriage with
the MBD but rather with a woman of the category boru ni
tulang, one of the specifications of which is MBD. Dumont has
argued, lucidly and convincingly, that ‘cross-cousin marriage’
is a misnomer (1957a, pp. 24-5) and that ‘. . . “marrying a cross¬
cousin ” is nothing but marrying an affine.’ Ideally then we should
couch a discussion of prescriptive marriage systems in terms
which avoid genealogical specifications in general and the notion
of cross-cousin in particular. Indeed this is what Needham has
been doing where he speaks of matrilateral cross-cousin marriage
2. I shall throughout use the notation where relationship terms are
represented by their initial letter, with the exception of Z = sister to
distinguish it from S = son.

200 Marriage and Alliance

systems as systems of asymmetric alliance. Sooner or later, how¬
ever, the genealogical specification comes back to plague the
discussion. It will be necessary for me to make use of genealogical
specifications, but I shall try to avoid misunderstanding by setting
out in each instance, and at the risk of repetition, exactly what the
function of the genealogical referent is. Finally, like the \/-l
I shall elide it altogether in my final formulation.
In general, then, prescriptive marriage systems are here taken
as being those in which there is a rule of marriage with a prescribed
category of relative. The implications of such prescriptions are
discussed in the following sections which deal with each type of
prescription in turn.

Prescriptive matrilateral cross-cousin marriage

A minimal characterization of such a system is that a man must
marry a woman whom he addresses by a relationship term which
denotes a category of relatives that includes his MBD and ex¬
cludes his FZD. This rule has certain consequences. The work of

Figure 1 A formal model of asymmetric alliance

Dutch scholars (e.g. Van Wouden, 1935)3 and the classic study
of prescriptive marriage systems by Levi-Strauss (1949) demon¬
strated that one result of it could be the establishment of enduring

3. I wish to acknowledge my debt to Dr Rodney Needham, who not only

instructed me in the study of prescriptive marriage systems but also intro¬
duced me to the work of the Dutch anthropologists.

D. Maybury-Lewis 201
affinal relationships between descent groups. Group A gives
women to group B, group B to group C, and so on until finally
one group gives women to A and thereby closes the cycle (as in
Figure 1).
Prescriptive matrilateral cross-cousin marriage was therefore
held to produce cycles of marriage transactions, or connubium
as the Dutch called it.
Such a model of the system contains the implicit assumption that
a given male Ego in a patrilineal society would always take a
woman from the descent group of his MBD even though he need
not marry the MBD herself. Subsequent studies by Leach (1954)
and Needham (1958a)4 served to modify this thesis in at least one
important respect. They showed that in certain patrilineal soc¬
ieties a man could marry in conformity with a rule of prescriptive
matrilateral cross-cousin marriage even though he did not take a
wife from the descent group of his MBD, provided that the wife
he did take came from a descent group which (1) was not already
thought of as taking women from Ego’s group, and (2) was classi¬
fied by virtue of the marriage as a wife-giving group vis-a-vis
Ego’s group. Such a marriage in effect created rather than con¬
tinued an affinal relationship between descent groups.5 Once it
had been contracted, Ego and the members of his descent group
addressed the members of his wife’s group (and vice versa) by the
relationship terms appropriate between members of wife-giving
and wife-taking groups. In such systems the relationship category
from which Ego took a wife, or into which he placed his wife at
marriage, also included the specification MBD. It was in this
sense that the system could be referred to as one of matrilateral
cross-cousin marriage or MBD marriage. The formal require¬
ments of the system are, however, imperfectly translated in terms
of MBD marriage, but depend instead on the distinction between
wife-giving and wife-taking groups and the rule that a spouse
4. Also by Dumont (1957a and b), considered later in connection with
bilateral systems.
5. It is not essential that descent groups should be the wife-giving and
wife-taking units, although they frequently are. In certain societies each
marriage establishes mutually exclusive categories of wife-givers and wife-
takers which do not correspond to descent groups. I couch my discussion
here in terms of descent groups so that the exposition may be as simple and
as clear as possible.

202 Marriage and Alliance

may not be taken from a group which is already in a wife-taking
relationship to Ego’s.
It could be argued that such societies cannot usefully be con¬
sidered as instances of matrilateral cross-cousin marriage. Coult
did in fact use such an argument in an exchange with Leach
concerning the Kachin. He wrote:
the usual notion of cross-cousin marriage entails that a person marry a
woman who is related to himself in one or another of a limited number
of ways. If, for example, matrilateral cross-cousin marriage is practised
among patrilineal descent groups, then this means that a man will
marry his MBD, or MBSD, or MBSSD, or MFBSD, or MFBSSD,
etc. Marriage with the MBD is regarded as true cross-cousin marriage;
marriage with any of the remaining kin types as marriage with the
classificatory cross-cousin (1963, p. 162).
He therefore insisted that if his theory of cross-cousin marriage
(Coult, 1962a) did not apply to the Kachin, it was because the
Kachin did not practise cross-cousin marriage (Coult, 1963,
p. 163).
There is more to this divergence of views than mere disagree¬
ment over the use of a term. It is important to realize that Coult
and Leach are writing about two quite different types of inquiry.
Before we consider these, however, a preliminary contradiction in
Coult’s statement must be dealt with. The theory of cross-cousin
marriage to which he refers above (Coult, 1962a) is an attempt
to discover a formula which will explain why, in societies where
men may marry their cross-cousins genealogically defined, they
sometimes prefer the MBD to the FZD, sometimes the FZD to
the MBD, and sometimes marry both indiscriminately. His
paper thus derives explicitly from Homans and Schneider (1955).
He apparently considers it irrelevant, for the purposes of his argu¬
ment, whether these marriages are prescribed or preferred, since
he tests his hypotheses by referring to Murdock’s table showing
the relationship of preferential marriage to descent (1957, p.
687), where all such systems are categorized as ‘preferred’.
However, Murdock’s table includes the Kachin (Jinghpaw),
which had previously been listed (Murdock, 1957, p. 680) as
having patrilineal descent and matrilateral cross-cousin marriage
preferred asymmetrically. Coult therefore accepts the Kachin as
an instance of a society practising cross-cousin marriage when in

D. Maybury-Lewis 203
1962 he uses Murdock’s table to test his hypotheses but denies that
they do when in 1963 he is in argument with Leach.
A clue to this contradiction may lie in the last sentence of Coult
(1963a). Commenting on Leach’s assertion that ‘A Kachin for
example must marry a nam, that is a girl junior to Ego who is a
member of any mayu (wife-giving) lineage with respect to Ego’
(1963, p. 77), Coult replies that ‘In the final analysis no theory
can be a match for the awesome and far-ranging memories of
individual ethnographers’ (1963a, p. 163). The rejoinder, with its
insinuation that Leach ‘remembered’ the evidence in order to
controvert Coult, is in poor taste. It is also poor scholarship.
The evidence which Leach is supposed to have remembered is set
forth in Political Systems of Highland Burma (Leach, 1954, p.
74) and should certainly have been familiar to any anthropologist
who seriously proposed to discuss the characterization of the
Kachin.6
The implications of Coult’s rejoinder transcend the Kachin,
however. As I have already indicated, he is advocating a different
type of inquiry from that which Leach pursued. If cross-cousin
marriage systems are held to be only those where a man must
marry his actual or classificatory cross-cousin, genealogically
defined, then the study of such systems becomes an etic investiga¬
tion which seeks to make statements about how particular genea¬
logical specifications become the foci of marriage rules and what
the consequences of this are.
Leach’s study of the Kachin (1954) is, on the other hand, an
emic investigation which seeks, among other things, to examine the
social consequences of a marriage rule such that a bride must
always belong to a specific relationship category. He therefore
analyses Kachin categories in general and the categories of their
relationship system in particular and in so doing elucidates the
defining criteria of that category which must contain a man’s
wife/M BD. This approach is also common to the work of
Dumont and Needham, who have emphasized that the study of

6. This appears to be another melancholy example of the type of mistake

made by anthropologists who work with cross-cultural compilations
without referring to the sources. The World Ethnographic Sample is certainly
valuable, but it should be used as a starting point for comparative work,
not as an isolated universe of data.

204 Marriage and Alliance i.

 systems of social classification is an integral part of their investi¬
gation of prescriptive marriage.
Both Coult’s approach and what I have here dubbed as Leach’s
approach are, of course, perfectly legitimate. It is only possible to
choose between them by seeing how they account for particular
cases or classes of cases. In dealing with prescriptive marriage
systems Coult’s approach suffers from a major disadvantage.
We have already seen that societies do not in fact prescribe marriage
with a cross-cousin. They prescribe marriage with categories such
as boru ni tulang (Batak), nam (Kachin), and so on. An analysis
based on genealogical specifications will therefore hold only for
these societies where, e.g. boru ni tulang = MBD. We can state
this as a general proposition by letting x be the vernacular term
for any relationship category which includes the specification
MBD and excludes FZD. Let y similarly symbolize the term for
a category including FZD and excluding MBD. Let z be the term
for a category which includes both MBD and FZD. A genealogi¬
cal approach will only hold for prescriptive marriage systems
where x = MBD, y — FZD, and z = MBD/FZD. A category
approach deals with x, y and z and therefore holds not only in the
cases where x = MBD,y = FZD and z = MBD/FZD but also
in the cases where x > MBDj > FZD and z > MBD/FZD.
It could nevertheless still be held that a genealogical theory was
an adequate theory of ‘prescriptive marriage systems’ if it could
be shown that all cases where MBD was part of x could be derived
from x = MBD and so on. This was perhaps an implicit assump¬
tion of Homans and Schneider’s Marriage, Authority and Final
Causes (1955). To my knowledge, however, it has not yet been
shown that this assumption is correct. On the contrary, I have
argued elsewhere (Maybury-Lewis, in preparation) that we have
no grounds for assuming that relationship terms refer to categories
which are invariably derived (or ‘extended’) from genealogical
specifications contained in them. Instead I suggest that we are not
yet in a position to make statements about the general content of
such categories. They may on occasion be genealogically derived,
but it has also been shown that frequently they are not.
To sum up, then, the approach to prescriptive marriage systems
here outlined takes as its point of departure the rule in certain
societies that a man’s wife must belong to a specific relationship

D. Maybury-Lewis 205
category. Prescriptive matrilateral cross-cousin marriage has been
so called because one of the specifications in that category is
MBD. The formal requirements of the system are that, if a male
Ego’s descent group is in a wife-taking relationship with another
one, it must:
(a) refrain from giving women in marriage to that other (wife¬
giving) group except after a certain conventional interval;7
(b) use to members of that group the relationship terms which are
applied to wife-giving groups.
These terms will also be the ones applied to the members of any
group which contains a woman who is MBD to any man of Ego’s
descent group.
In such a system the rule of marriage serves to define the relations
between descent groups.8 If a society is patrilineal, then the affinal
ties of its constituent descent groups are maintained by the
marriages of successive generations of males. Figure 2 shows the
various marriage choices open to a male Ego in a patrilineal society
with a rule of prescriptive matrilateral cross-cousin marriage.
He may take a woman from his MB’s descent group (P), thereby
repeating his F’s marriage and possibly (though not necessarily)
marrying his actual MBD. He may take a woman from a descent
group which has in the past given women to his own (Q) or he may
take a woman from some other descent group (R) which hence¬
forward become wife-givers. The only descent groups he may not
take from are those such as Y and Z, which are taking women from
his own.
In a matrilineal society the affinal ties of descent groups are
maintained by the marriages of successive generations of females.
If such a society had a rule of prescriptive matrilateral cross¬
cousin marriage, then women would always place their husbands
in the category of FZS. The marriage choices open to a female
Ego in such a society are analogous to those open to a male Ego in
a patrilineal society, the husband invariably being classed in the
same category as the FZD. In fact this type of society appears to

7. Wife-givers may then become wife-takers and vice versa, but they
must be distinguished at any one time. See Needham (1960, p. 501) for a
discussion of this point.
8. Except where it serves, as noted before, to distinguish categories of
wife-givers and wife-takers.

206 Marriage and Alliance

be exceedingly rare.91 do not propose here to attempt to explain
this. I merely point out that in matrilineal, as in patrilineal
societies, prescriptive matrilateral cross-cousin marriage precisely
defines the relation between affinally related descent groups or
categories. This is not the case in a system of patrilateral cross¬
cousin marriage, to which we now turn.

Figure 2 Types of marriage choices open to a male Ego with asymmetric alliance

Prescriptive patrilateral cross-cousin marriage

The minimal characteristic of such a system is that a man must
marry a woman whom he addresses by a relationship term which
denotes a category of relatives that includes his FZD and ex¬
cludes his MBD. The consequences of such a rule are different
from the matrilateral case, as has been amply demonstrated by
Levi-Strauss (1949). One possible outcome would be as in Figure
3. It will be seen from that figure that a descent group A which
gives women to another descent group B in one generation will
receive women from B in the next. Levi-Strauss referred to this as
discontinuous exchange and contrasted it with the generalized
exchange effected by matrilateral cross-cousin marriage. He
suggested that discontinuous exchange effected a series of short
marriage cycles, whereas generalized exchange brought about a
single unifying cycle in the society which practised it (Levi-
Strauss, 1949, p. 562). Homans and Schneider pointed out, how¬
ever, that patrilateral cross-cousin marriage systems require a
cycle of descent groups linked by marriage in the same way as with
9. Needham has argued that the Siriono are such a society (1961). If he
is correct, then they are the only known instance. I myself do not find his
argument convincing, since I do not accept his inference from the structure
of relationship terminology to the ordering of the marriage system.

D. Maybury-Lewis 207
matrilateral cross-cousin marriage (1955, p. 13). The difference
between the two lies not in the length of the cycle but in the fact
that a matrilateral prescription ensures a unidirectional flow of
marriages from group to group in successive generations, whereas
the patrilateral one results in a change of direction of this flow with
each generation.10
As a result, the patrilateral prescription does not by itself
define the relationship between descent groups. It can be seen from

Figure 3 A formal model of prescriptive patrilateral cross-cousin marriage

Figure 3 that A and C are both wife-giving and wife-taking

groups with respect to B. Consider the effect of this on the marriage
choices open to a male Ego in a patrilateral society with this form
of marriage prescription. He may, as we have already seen in the
matrilateral case, marry a woman of a descent group which is
neither wife-giver nor wife-taker to his own. In this case his mar¬
riage would create an affinal link between groups and would not
contradict any previous relationship. But if he takes a wife from
a descent group which already has an affinal link with his own,
then there is some ambiguity. The group from which he proposes
to take his wife is both wife-giving and wife-taking with respect
to his own. In a prescriptive marriage system he could not take a
wife from a wife-taking group. Such a marriage would be regarded
10. I have sometimes found it helpful to conceive of the difference in
terms of an electrical circuit with direct current in the first (matrilateral)
case, alternating current in the second (patrilateral) case. The length of the
circuit remains constant, but the type of flow changes.

208 Marriage and Alliance

as incestuous. There must, therefore, be some institutional
distinction between those women of his affinal descent group who
were in the prescribed category and those who were forbidden to
him. It would appear from Figure 3 that an age-set system or
section system which effected a rigid demarcation between the
generations in each descent group would make such a system
workable. In this hypothetical situation a man would know which
descent groups were givers and which takers vis-a-vis his own in
a particular generation.
Nevertheless Needham has consistently maintained that a
system of prescriptive patrilateral cross-cousin marriage cannot
exist in theory and does not exist in fact (1958b; 1960; 1962;
1963a). A number of writers have expressed a contrary view
(Livingstone, 1959; Coult, 1962b; Lane, 1962; and Salisbury,
1964). Let us consider their arguments first.
Coult suggests that ‘... the only necessary requirement is that
a person marry his father’s sister’s daughter, or a classificatory
father’s sister’s daughter, and the system could work perfectly,11
(1962b, p. 330). But could it? A system where every man was
obliged to marry his actual FZD would be unworkable for simple
demographic reasons. If, on the other hand, he may marry his
classificatory FZD, then how is this category to be defined?
One obvious possibility, already mentioned, is that it be defined
in terms of a section system. Indeed this is the suggestion which has
been advanced by most writers arguing for the feasibility of
patrilateral cross-cousin marriage. Livingstone (1959) proposed
that a variation on the eight-section system of the Arunta might
be the solution. Hammel (1960) suggested that formally both a
six-section and an eight-section system could provide an institut¬
ional matrix for this type of marriage prescription. Lane (1962)
likewise proposed a form of eight-section system, and Coult
(1962b) has argued, like Hammel, that both six and eight-section
systems would satisfy the formal requirements of a patrilateral
prescription.
These suggestions do not controvert Needham’s contention, for
he was not arguing that such a system was inconceivable, which
would be absurd, nor even that its formal properties could not be
11. This simplistic assertion in itself is enough to make one wonder
whether Coult has considered the matter very deeply.

T-K-H D. Maybury-Lewis 209

described. He was claiming that the rules entailed by the system
were unworkable. This is still a formal argument in the sense that
he did not state that the rules did not or would not work in particu¬
lar instances but that they could not work as a general principle.
The issue then is not whether it is possible to suggest a formal
model for a patrilateral prescriptive marriage system but whether
such a model could be translated into actuality in a given society
and by what rules.
Needham’s central objection to the workability of the suggested
models may be paraphrased as follows. The systems are cumber¬
some, but even if they were found to exist they would produce
marriage with a bilateral cross-cousin (Needham, 1960, pp.
210-12; 1963, p. 203). This seemingly straightforward proposition
is in fact rather complex, and its logical status similarly so.12 Nor
have Needham’s attempts to demonstrate it, by showing that in a
society with a patrilateral prescription FZD = MMBDD and is
therefore a bilateral relative, served to clarify matters. On the
contrary they appear to have caused some bewilderment (Coult,
1962, p. 328; Lane, 1962, p. 469-70). The difficulty stems, I believe,
from the sudden introduction of genealogical terminology into a
discussion of categories. If I may presume to try to restate Need¬
ham’s argument without the genealogical reference, it appears to
rest on the principle that a patrilateral prescriptive system effects
exchanges of women between descent groups such that if group A
gives women to group B in one generation, group B gives women to
group A in the next. If we look at marriages from the point of view
of the descent group, then it can be seen that they are not unilateral.
To the men of group A (Figure 3) the women of B are matrilateral
in one generation and patrilateral in the next. The system thus
becomes a variation of bilateral cross-cousin marriage. If I have
understood this argument correctly, it maintains that prescriptive
patrilateral cross-cousin marriage must be a form of prescriptive
bilateral cross-cousin marriage and cannot therefore be distin¬
guished as a form of prescription in its own right.
It follows from this that even if a society were found which has

12. It could be taken as a statement to the effect that prescriptive patri¬

lateral cross-cousin marriage cannot exist because, if it did, it would be
prescriptive bilateral cross-cousin marriage. Such a proposition is clearly
absurd. I shall rephrase it in the following paragraph.

210 Marriage and Alliance

an eight-section system such as the one postulated by Livingstone
(1959, p. 370) with FZD marriage, Needham would argue that it
was not a case of a distinct type of prescriptive marriage system but
rather a form of bilateral cross-cousin marriage.13
It should by now be clear that an argument as to ‘whether a
prescriptive FZD marriage system can exist ’ is too loosely phrased
to be useful. It seems to me that if Livingstone wished to call his
hypothetical construct a system of prescriptive FZD marriage,
he would be entitled to do so, and I can see no reason why such a
system could not work. But the name used to designate the system
is not the point at issue. Needham is in fact arguing that such a
prescription cannot be distinguised in terms of lateralityixom other
systems.
This is where the conventional glosses MBD and FZD are
particularly misleading. What we refer to as a prescriptive MBD
marriage system is in fact a system where there is a unilateral trans¬
fer of spouses from one descent group to another. What we refer
to as prescriptive MBD/FZD marriage is a system where there is
a bilateral exchange of spouses between descent groups. In terms
of laterality there are no other possibilities open, and prescriptive
FZD marriage thus becomes a sub-class of prescriptive MBD/
FZD marriage.
It was for this reason that Needham considered the possibility
of a six-section system as a matrix for patrilateral prescriptive
marriage (1962, p. 112). It was not, as Coult supposed (1962, p. 331),
because there are three lines in the conventional model of a patri¬
lateral prescriptive system, but because such a system had to be
asymmetric in order to be distinguished from bilateral prescrip¬
tion. Two, four, and eight-section systems effect a bilateral
prescription and thus do not serve.
Lane (1962) nevertheless accepts the implied challenge in Need¬
ham’s claim that there can be no system of prescriptive patrilateral
cross-cousin marriage and argues that on the contrary there can
and that the Pende have it. Similarly Salisbury, though he does not
specifically mention Needham, presents the Siane as another
society practising patrilateral cross-cousin marriage and feels
obliged to restate his position ‘in view of some theorists’ failure
13. I would agree with Needham here as will become clear later in my
discussion of the Siane case and of bilateral cross-cousin marriage systems.

D. May bury-Lewis 211

to consider the ethnographical evidence when they assert that an
obligatory patrilateral cross-cousin-marriage rule is impossible*
(1964, pp. 168-9).
There are some problems in the Pende material, however, which
Lane does not mention.14 The Pende are reportedly divided into
matrilineal clans, and a man will try to take a wife from his father’s
matriclan ‘ so that he may return to it the semen which was lost in
engendering him* (Lane, 1962, p. 489).
Now, since the Pende are matrilineal, it is the women who pro¬
vide the continuity of the descent groups. We should therefore
consider Pende marriage arrangements from their point of view.
A man tries to take a woman from his father’s clan, and he is
forbidden to marry his MBD. It therefore follows that a woman is
sought by a man from her MBW’s clan and is forbidden to marry
her FZ S. If the marriage rule is prescriptive, then she is faced with
the problem we have already mentioned: how is she to know which
of her MBW’s clan are marriageable? Those of her mother’s
generation were presumably FZS to her mother and therefore
forbidden. Yet we are not told how the Pende make this distinc¬
tion, which is precisely the problem with a patrilateral prescrip¬
tion.
We are not told either what the boundaries of the kinship
categories are which Lane cites. Do they include all members of
Pende society, so that if a man may not marry his tata (some of the
specifications of which are MBD, S, D) he must marry his isoni
(one of the specifications of which is FZD)? Or do these terms
refer to certain genealogical specifications ? In the latter case how
do the Pende operate a prescriptive marriage system in terms of
genealogical relationships, which may be vacant for any given
Ego?
Furthermore, Lane notes the provisions which the Pende make
for cases where ‘the preferred marriage with the FZD cannot be
arranged’ (1962, p. 485). This would seem to indicate then that
the Pende have an ideology of FZ D marriage but that this is not a
prescription, for, as we have seen, it is characteristic of a prescrip¬
tive system that all marriages are treated as if they fall into the
correct category.
14. I have unfortunately been unable to consult de Sousberghe’s report
(1955), which is unavailable here.

212 Marriage and Alliance

 Finally, and perhaps conclusively, de Sousberghe, the ethno¬
grapher on whose report Lane based his contention, has in con¬
nection with the issue of prescriptive marriage among the Pende
allowed himself to be quoted to the following effect: 4 Marriage
with the patrilateral cross-cousin among the Pende is not pre¬
scribed: it is merely preferential’ (Needham, 1963b, p. 58).
Salisbury specifically states, on the other hand, that the rule of
marriage for the Siane ‘should not be called “prescriptive” as
nothing is “prescribed”’ (1964, p. 169). Yet he appears to be in
two minds as to whether or not the rule is obligatory. Thus he
states that ‘there is definitely no obligatory marriage rule in
Siane’ (1956, p. 646) and later ‘The marriage rule is obligatory as
all men, when they marry, must be and are (with modem exceptions
of those marrying foreign women) marrying a “father’s sister’s
daughter”’ (1964, p. 169). I shall not dwell on these fine distinc¬
tions. Instead I shall try to describe what appear to be the salient
features of the Siane material in so far as they bear on the issue of
patrilateral cross-cousin marriage.
The Siane are patrilineal, and they forbid the marriage of a man
with a woman of his mother’s clan. Nor may a man marry his
true father’s sister’s child (Salisbury, 1956, p. 646). Instead he
must marry a classificatory father’s sister’s daughter, since ‘all
the children of the other15 clans are the children of father’s sisters’
husbands and Ego can, if he so desires, call them novonefo or
“my cross-cousin”’ (Salisbury, 1956, p. 647).
Now the term novonefo is presumably the same as nofonefo,16
which Salisbury shows elsewhere (1962:19) to apply to at least the
following: males and females of a male Ego’s FZ clan in the same
generation as Ego and in the next descending generation, and males
and females of a male Ego’s MB’s clan in the same generation as
Ego and the next descending generation. MBD is not therefore
terminologically distinguished from FZD.17

15. Other than mother’s clan.

16. Salisbury gives the translation ‘cross-cousin’ for novonefo (1956, p.
647), nofonefo (1962, p. 19), and hovorafo (1964, p. 169).
17. Although there is an alternate term komonefo, which may be applied
to FZD but not to MBD. One of the specifications of this term is ZD.
Salisbury gives its translation as ‘My sister’s son’. This is presumably a
slip for ‘ My sister’s child’.

D. Maybury-Lewis 213
 It would seem then that the Siane have a category of bilateral
cross-cousins from which a man must take a wife. Within the
category there are certain prohibited unions; i.e. a man may not
marry a cross-cousin of his M B’s clan or a cross-cousin who is his
actual FZD. This is not prescriptive patrilateral cross-cousin
marriage in the sense that I have been using the term in this paper.
A Siane man must marry in the category novonefo, but this is not a
unilateral category in the sense that it may be contrasted with
another category of prohibited women on the opposite side. Indeed
the whole notion of unilaterality is only meaningful if two sides
are contrasted. It is precisely the difficulty of doing this that
renders a system of prescriptive patrilateral cross-cousin marriage
so anomalous.
I would find it more useful to maintain that the Siane practised
a form of bilateral cross-cousin marriage. They have an ideology
of symmetric exchanges of women between clans (Salisbury,
1956, pp. 641, 642), they do not distinguish terminologically
between the MBD and the FZD, and a man marries into a
category which on the evidence Salisbury presents is bilateral
rather than unilateral. Salisbury has presumably called this
system patrilateral cross-cousin marriage because of a certain
asymmetric feature of it. He discovered that there was a tendency
for women in Siane country to pass from the south and west
towards the north and east, while valuables moved in the opposite
direction (1956, p. 646).
This discovery and Salisbury’s discussion of it is extremely
interesting, but it should not be permitted to confuse the issue of
characterizing Siane marriage rules. When we refer to unilateral
cross-cousin marriage as an asymmetric system, we are referring to
an asymmetry in the rule, not in its social consequences. The
consequences are perhaps likely to be asymmetric but they need
not necessarily be so. Similarly it is possible for a society to have a
symmetric marriage rule and for the application of the rule to
result in asymmetry.
If Salisbury, and Livingstone who recently came out in support
of his position (1964), wish to call the Siane a system of prescriptive
patrilateral cross-cousin marriage, they are of course at liberty to
do so. It seems to me that there is no profit in arguing about
whether it is or is not. It is more important to understand in what

214 Marriage and Alliance

sense Salisbury and Livingstone use the term and the reasons why
others might reject their classification.
Prescriptive bilateral cross-cousin marriage
I have tried to show in the preceding section why, in my view, a
discussion of prescriptive patrilateral cross-cousin marriage
inevitably leads to a discussion of prescriptive bilateral cross¬
cousin marriage. By corollary with my previous characterizations
I would suggest that a working definition of a bilateral prescriptive
system is that a man must marry a woman whom he addresses by a
relationship term denoting a category of relatives which includes
the joint specification MBD/FZD.
The genealogical referent is here more than ever misleading. It
may cause little difficulty in the consideration of systems such as
those described by Dumont for South India (1957a; 1957b),
where patrilineal descent groups enter into affinal relations with
each other, predicated on the exchange of women. The relationship
terminologies of these societies are of the type known as two-
section systems. Ego thus applies the terms for ‘own section’
(kinsmen) to his own descent group and the terms for ‘other
section’ (affines) to any group related to his own by marriage.
Such specifications as MB and FZH therefore fall into a single
category, that of male affine of the first ascending generation.
Similarly MBDandFZD are subsumed under a single term which
denotes female affine of own generation.
The inadequacy of the genealogical definition is brought out
more clearly when we consider a society divided into exogamous
moieties. Such a society may have a two-section system of re¬
lationship terminology, in which case the terms for own section/
kinsmen are applied within Ego’s moiety and the terms for other
section/affines to the opposite moiety. But it need not have a two-
section system (Maybury-Lewis, 1960, p. 210; Keesing, 1964,
p. 297; Needham, 1964, p. 303). Alternatively it may have an
explicitly asymmetrical relationship terminology of the Crow or
Omaha type18 adapted nevertheless to a two-section system.
18. I do not agree with Needham (1964, p. 312, n. 28) that either he or I
was mistaken concerning the existence of cases of symmetric alliance with
a Crow or Omaha relationship terminology. The instances I cited in my
paper (1960, p. 210) seem to me, with the exception of the Eastern Timbira,
to be still valid. I discuss the Sherente case in the next paragraph.

D. Maybury-Lewis 215
 The Sherente of Central Brazil are such a society. They had
patrilineal exogamous moieties, each subdivided into four
patricians, in conjunction with an Omaha type terminology
(Nimuendaju, 1942). I studied them in 1955-6 and 1963 and dis¬
covered that the moiety system had fallen into desuetude. The
terminology was still used, however, and applied as follows.19 A
male Ego uses one set of terms for members of those clans which
would originally have been in his own moiety and another for
members of those clans which would have belonged to the op¬
posite moiety. For example, men of Ego’s ‘own moiety’ in the
first ascending generation are referred to by a single term, some of
the specifications of which are F and FB. There are two terms
(distinguishing between those older and younger than Ego) for all
members of Ego’s generation in Ego’s ‘own moiety ’ irrespective of
sex. There is another term for all members of Ego’s ‘own moiety’
irrespective of sex in the first descending generation.
In the ‘opposite moiety’ there must be some distinctions, since
the terminology is of the Omaha type. All women of Ego’s
mother’s patrician are referred to by a single term (i-natk£)y some
of the specifications of which are M, MZ, MBD. All men of the
same patrician are similarly classed together as ndkliekwa, and
some of the specifications of that term are MB, MBS. All women
of the ‘ opposite moiety ’ not in Ego’s mother’s clan are classed in a
single category Q-tbe) if they are of first ascending generation or in
another category (kremzu) if they are of Ego’s or the first descend¬
ing generation.
The rule of marriage was until recently that a man had to marry a
kremzu.20 Sherente would not normally express it that way. They
would say instead that they married Wairi or Doiy the names of the
moieties. But they will also explain that a man may not marry a
girl who is i-natkif (from his mother’s clan). He must marry a
kremzu, which in this context refers to any woman of the opposite
moiety not in his mother’s clan.
I would not follow Salisbury and call this a system of patrilateral
cross-cousin marriage. In a sense it is, in so far as the actual

19. I give the broad outlines of the terminology without discussing its
particular features.
20. Sherente say that this is still the rule, but go on to add that nowadays
nobody pays any attention to it.

216 Marriage and Alliance

MBD is prohibited and the actual FZD falls into the prescribed
category. In fact it gives a poor understanding of the kremzu
category if we label it as ‘patrilateral’. After all, Ego’s mother
refers to all members of her ‘ own moiety * in her own generation by
a sibling term. Therefore male members of the Ego’s ‘opposite
moiety’ not in his mother’s clan are classificatory MB as well as
FZH. Similarly the specification MBD (in a classificatory sense)
is as correct as the specification FZD for the category kremzu.
Formally, then, the Sherente system could be said to prescribe
marriage with a bilateral cross-cousin, although a sector of that
category was prohibited.
We encounter similar difficulties when we progress from two-
section systems to four and eight-section systems. Here it must be
borne in mind that just as there is a difference between a two-sec¬
tion system and a system of exogamous moieties, so a four-section
system does not entail marriage rules of the Kariera type, nor an
eight-section system rules of the Arunta type. I refer specifically
to the Arunta here as an example, not a prototype of an eight-
section system.
Spencer and Gillen showed how the eight-section system had the
effect of allocating the patrilineal Arunta groups into four cate¬
gories (1927, p. 64). The model of the system thus required four
categories, but each category actually embraced a number of
local groups (there were seventy-three in all). This analysis is an
important early demonstration, for which Spencer and Gillen are
not often given credit nowadays, of an issue which has been taken
up by other writers (e.g. Levi-Strauss, 1949; Leach, 1951; Romney
and Epling, 1958) and which is fundamental to the understanding
of prescriptive marriage systems.
Seen from the point of view of a male Arunta these four cate¬
gories may be distinguished as follows. Category 1 included his
own local group, and the men in these groups were addressed by
terms which included such specifications as F, B, S. Category 2
were groups unlike his own (defined by membership in different
marriage-sections) but nevertheless of his own moiety. The men
in them were addressed by terms which included such specifica¬
tions as M M B, M M B S. Ego might not marry into either of these.
Category 3 included groups of the opposite moiety, and the men
in them were addressed by terms which included such specific-

D. Maybury-Lewis 217
ations as MB/FZH and MBS/FZS. Yet Ego could not marry
women of these groups, which included his actual MBD/FZD
and other girls classified as such. He had instead to marry a girl
of category 4 whom he addressed by a term, one of the specifica¬
tions of which was MMBDD.
Levi-Strauss has amply demonstrated that such a system was a
development of a Kariera type four-section system where the
bilateral cross-cousin category was further subdivided and mar¬
riage permitted with only half of its incumbents (1949, pp. 210-15).
Yet according to our working definition of prescriptive bilateral
cross-cousin marriage, the prescribed spouse must fall in the same
category as the MBZ/FZD. Are we then to argue that a two-
section system and a Kariera type four-section system may be
classed as bilateral cross-cousin marriage, but an Arunta type
eight-section system may not? This would be a reductio ad
absurdum. Indeed the point of my argument is that if we per¬
petually orient our discussions of prescriptive marriage systems
by means of genealogical referents, then sooner or later we are led
to absurdities.

Conclusions
Let me therefore try and rephrase my conclusions without geneal¬
ogical specifications. It is characteristic of prescriptive marriage
systems that they prescribe marriage for a given Ego within a
certain relationship category. All marriages which take place in
such societies are therefore treated as being marriages within the
prescribed category. There are two main varieties of such systems.
The first, which I call symmetric alliance, divides Ego’s conceptual
universe into two parts. There are groups like Ego’s and groups
unlike Ego’s. In a system of exogamous moieties there may be
only one group like Ego’s and one unlike Ego’s, but this is not
essential. Ego’s wife must be classed in a certain category of the
‘unlike’ part of this dyadic model. Such a model can be described
in terms of exchange between these two types of groups.
The second variety of prescriptive marriage system is the one I
call, following Needham, asymmetric alliance. In such a system
Ego’s conceptual universe is divided into three parts. There are
groups like Ego’s and two classes of group unlike Ego’s. These
two ‘unlike’ classes are defined in terms of each other as opposites.

218 Marriage and Alliance

Ego’s wife must be subsumed under a certain category of one
‘unlike’ class in the model. Let us call that class Ui. Let the other
‘unlike’ class be U2. If the descent groups in this society are
patrilineal, then certain marriages, either Ego’s or that of previous
men in his group, will bring certain other groups into a Ui
relationship with Ego’s group. Ego’s group will be U2 to such other
groups. Such a model may be described in terms of a unilateral
passage of women from Ui groups to groups like Ego’s to U2
groups.
This is in fact what writers have meant when they wrote of the
radical distinction between wife-givers (Ui) and wife-takers (U2),
and argued that it was only in terms of this distinction that asym¬
metric alliance systems could be understood. I have tried to show
that it is only in terms of this distinction that such systems can
even be properly described. Much of the argument concerning
‘prescriptive marriage systems’ thus derives from the fact that
different writers have understood ‘cross-cousinmarriage systems’
to mean different things. The formulation proposed here derives
from the work of Levi-Strauss, Leach, Dumont and Needham. I
am not suggesting that it should be universally adopted or even
that it is the only way of tackling the problems which I have
raised. I hope only that this statement of it is sufficiently explicit
for the approach to be understood even by those who have no
wish to use it.
It can be seen why there must be a distinction maintained
between prescriptive and preferential marriage systems. A
prescriptive marriage rule entails the social consequences already
mentioned: the division of Ego’s conceptual universe in a de¬
termined fashion, irrespective of the percentage of people who
marry their actual MBD or into their MB’s descent group.21
These percentages may be of considerable interest in the handling
of certain problems. They do not affect the prescription as here
defined or its minimum social entailments.
On the other hand a preference for marriage with a certain
relative, however defined, does not have such entailments. Here
the percentages of people who actually marry according to the
preference are more significant. A stated preference which is
21. It was partly a failure to appreciate this point which led Ackerman to
his erroneous analysis of the Purum material (1964).

D. Maybury-Lewis 219
rarely acted upon is unlikely to produce social alignments pre¬
dicated upon a 100 per cent compliance with it. The extent to
which a preference for a certain type of marriage will influence the
institutions of a society thus depends in large measure on the
number of people who marry according to it. Nor is the argument
that a society may order its institutions as if there were 100 per
cent compliance with a marriage preference acceptable. Such a
preference would be a prescription in terms of my definition. I
suppose there must be a theoretical limiting case where a society
may be said, for example, to prefer marriage with a woman of
MB’s clan and where 100 per cent of all marriages conformed to
this preference. The institutions of that society might then be
congruent with this fact, and it would be a nice point as to whether
it was a case of preferential or prescriptive marriage. I would still
argue that it would be preferential, on the grounds that a sudden
change in people’s marrying habits would presumably alter any
institutional alignments based on statistical trends, whereas in a
prescriptive system this would be irrelevant.
One final problem remains to be dealt with: the problem of
‘choice’. Needham has written that if a system is prescriptive
‘. .. the emphasis is on the very lack of choice; the category of
type of person to be married is precisely determined, and this
marriage is obligatory’ (1962, p. 9). Schneider has recently taken
him to task, for using the word ‘choice’ in this context and from
his use of the word has derived an argument which purports to
show that prescriptive systems cannot be distinguished in this
way from preferential ones. Now the word ‘choice’ may or may
not be a good one to use here, yet it seems to me that Needham’s
meaning is quite clear. He does not mean that in a prescriptive
marriage system Ego is told which individual to marry, but, as he
says, that the category of person to be married is precisely deter¬
mined. Schneider notes, as we have done, that this category
cannot be determined by genealogical referents (1965, pp. 65-6)
and thus poses the problem with which this paper has attempted to
deal: i.e., how then is the prescribed category determined? In
other words, what are the characteristics of a prescriptive marriage
system ?
Schneider concludes that the gloss for this category has to be
‘marriageable woman’. He then asks how such a category could

220 Marriage and Alliance

be defined in terms of choice or lack of it. After all, the opposite of
‘marriageable woman ’ is ‘ unmarrigeable woman’. He continues:

Can we say that a prescriptive system is one in which ego is obliged to

marry a woman whom he is permitted to marry and a preferential
system one in which ego is permitted to marry a woman who is
prohibited ? This seems sheer nonsense, but it is to such sheer nonsense
that one is led if one starts with a structural problem and tries to define
it in terms of individual action (Needham, 1962) on a choice versus
no choice basis (Schneider, 1965, p. 66).

It seems to me that this argument is fallacious on a number of

counts. In the first place, as Schneider himself points out, every
known system prescribes marriage with a marriageable woman.
His gloss which purports to characterize prescriptive marriage
systems is thus vacuous. To argue that prescriptive marriage
systems cannot be distinguished on the basis of it is to knock
down a straw man. In fact Needham’s formulation does not lead
to such a characterization. It could be taken as a statement of my
contention in this paper that the distinction between a prescrip¬
tive and preferential marriage system is that in the former all
marriages are treated as being of the prescribed category where¬
as in the latter not all marriages are treated as being of the
preferred category.
It seems to me that these propositions may be phrased in terms of
individual action if the phraser feels there is any advantage to be
gained by so,.doing. Unlike Schneider I do not feel that the level
of structural statements and the level of individual actions are
mutually exclusive. On the contrary, I would argue that structural
propositions are intimately linked with propositions about indi¬
vidual actions and if so desired may be expressed in terms of them.
I do not feel that Needham’s formulation is the most felicitous one,
but I do not feel either that it leads to the sort of tautology which
Schneider suggests. It is clear that in prescriptive marriage systems
there is only one type of marriage by definition. In preferential
marriage systems there are a number of possibilities. We may
argue about how best to describe this state of affairs, but a re¬
pudiation of the word ‘choice’ cannot be made into a grounds for
assuming that there is no valid distinction to be made or that Need¬
ham had not made it.

D. Maybury-Lewis 221
 The implications of these arguments transcend what Murphy
described - I hope and assume humorously - as ‘whether one can
be (or should be) forced to marry his patrilateral cross-cousin’
(1963, p. 18). As Schneider indicated (1965), there are fundamental
theoretical and methodological issues at stake. For example, the
approach to the study of prescriptive marriage systems outlined
here lays great emphasis on understanding through the analysis
of social rules. It can be claimed, and the contention is implicit
in many of the arguments which have been advanced against this
approach, that such an emphasis is unwarranted and that an
investigation along behaviourist lines would yield better results.
Similarly we have seen that at least two radically different types
of enquiry, which I have dubbed respectively as emic and etic, have
been undertaken by students of prescriptive marriage systems.
Much of the controversy which has surrounded the interpretation
of these systems stems from the fact that the implications of this
difference do not appear to have been fully appreciated. Here are
two central issues in anthropology which have been brought out
in the debate over prescriptive marriage systems and for which
indeed the debate is perhaps a crucial instance.
Furthermore the discussions have resulted in a substantial re¬
thinking of the function of anthropological terminology. Concepts
such as social structure, descent and alliance are being continually
revised or at least re-argued in the light of the work which Levi-
Strauss, Leach, Dumont and Needham have been doing. This is
not ‘merely’ a matter of definition. The actual terms used are in
a sense immaterial. It is the assumptions which lead to this or that
use of them which have been called into question, the theoretical
biases of the terminology which are being explored. In fact
the whole question of models in social anthropology is really being
debated, their use, their implications and whether and under what
circumstances they have the explanatory power which Braithwaite
indicated as a property of scientific models (1953, p. 108). The
crux of the matter is the nature of explanation in social anthro¬
pology or, in other words, the nature of the subject itself, for a
discipline is characterized by the sorts of questions it asks and the
sorts of answers it is prepared to accept. It seems to me that these
issues are broad enough, and I am therefore frequently surprised to
learn that some anthropologists consider the debate concerning the

222 Marriage and Alliance

analysis of prescriptive marriage systems to be a dispute over
narrow technicalities. To paraphrase Dr Johnson, a man who is
tired of issues such as these is tired of social anthropology.

References
Ackerman, C. (1964), ‘Structure and Statistics: the Purum case’,
Amer. Anthropol. vol. 66, pp. 53-6.
Braithwaite, R. B. (1953), Scientific Explanation, Cambridge
University Press.
Coult, A. D. (1962a), ‘The determinants of differential cross-cousin
marriage’, Man, vol. 62, art. 47.
Coult, A. D. (1962b), ‘An analysis of Needham’s critique of the
Homans and Schneider theory’, Southwestern J. Anthropol., vol. 18,
pp. 317-35.
Coult, A. D. (1963), ‘The determinants of differential cross-cousin
marriage’, (letter), Man, vol. 63, art. 199.
D umont, L. (1953), ‘The Dravidian kinship terminology as an
expression of marriage’, Man, vol. 53, art. 54.
D umont, L. (1957a), Hierarchy and Marriage Alliance in South Indian
Kinship, Occ. Papers: roy. anthropol. Inst., no. 12.
Dumont, L. (1957b), Une Sous-caste de VInde du Sud: Organisation
Sociale et Religion des Pramalai Kallar, Mouton.
Hammel, E. A. (1960), ‘Some models for the analysis of marriage-
section systems’, Oceania, vol. 31, pp. 14-30.
Homans, G. C., and Schneider, D. M. (1955), Marriage, Authority and
Final Causes, Free Press.
Keesing, R. (1964), ‘Mota kinship terminology and marriage: a
re-examination’, J. Polynesian Soc. vol. 73, pp. 294-301.
Lane, R. B. (1962), ‘Patrilateral cross-cousin marriage: structural
analysis and ethnographic cases’, Ethnol., vol. 1, pp. 467-99.
Leach, E. R. (1951), ‘The structural implications of matrilateral
cross-cousin marriage’, J. roy. anthropol. Inst., vol. 81, pp. 23-55.
Leach, E. R. (1954), Political Systems of Highland Burma, Bell &
Sons.
Leach, E. R. (1963), ‘The determinants of differential cross-cousin
marriage’, (letter), Man, vol. 63, art. 87.
L£vi-Strauss, C. (1949), Les Structures Elementaires de la
Parente, Presses Universitaires de France.
Livingstone, F. B. (1959), ‘A formal analysis of prescriptive marriage
systems among the Australian aborigines ’, Southwestern J. Anthropol.,
vol. 15, 361-72.
Livingstone, F. B. (1964), ‘Prescriptive patrilateral cross-cousin
marriage’, (letter), Man, vol. 64, art. 59.
Loffler, L. G. (1964), ‘Prescriptive matrilateral cross-cousin marriage
in asymmetric alliance systems: a fallacy’. Southwestern J. Anthropol.,
vol. 20, pp. 218-27.

D. Maybury-Lewis 223
Maybury-Lewis, D. (1960), ‘Parallel descent and the Apinaye
anomaly’. Southwestern J. Anthropol., vol. 16, pp. 191-216.
Maybury-Lewis, D., ‘Relationship systems’, n.d.
Murdock, G. P. (1957), ‘World ethnographic sample’, Amer. Anthropol,
vol. 59, pp. 664-87.
Murphy, R. F. (1963), ‘On Zen Marxism: filiation and alliance’,
Man, vol. 63, art. 21.
Needham, R. (1958a), ‘A structural analysis of Purum society’, Amer.
Anthropol., vol. 60, pp. 75-101.
Needham, R. (1958b), ‘The formal analysis of prescriptive patrilateral
cross-cousin marriage’. Southwestern J. Anthropol. vol. 14, pp. 199-219.
Needham, R. (1960), ‘Structure and change in asymmetric alliance:
comments on Livingstone’s further analysis of Purum society’, Amer.
Anthropol.. vol. 62, pp. 499-503.
Needham, R. (1961), ‘An analytical note on the structure of Siriono
society’, Southwestern J. Anthropol., vol. 17, pp. 231-55.
Needham, R. (1962), Stucture and Sentiment, University of Chicago
Press.
Needham, R. (1963a), ‘Some disputed points in the study of prescriptive
alliance’. Southwestern J. Anthropol., vol. 19, pp. 186-207.
Needham, R. (1963b), ‘Prescriptive alliance and the Pende’, (letter),
Man, vol. 63, art. 62.
Needham, R. (1964), ‘The Mota problem and its lessons’, J. Polynesian
Soc., vol. 73, pp. 302-13.
Nimuendajt5, C. (1942), The Serente, Frederick Webb Hodge
Anniversary Publication, no. 4.
Rivers, W. H. R. (1914), Kinship and Social Organization, Constable.
Romney, A. K., and Epling, P. J. (1958), ‘A simplified model of
Kariera kinship’, Amer. Anthropol., vol. 60, pp. 59-74.
Salisbury, R. F. (1956), ‘Asymmetrical marriage systems’, Amer.
Anthropol., vol. 58, pp. 639-55.
Salisbury, R. F. (1962), From Stone to Steel, Melbourne University
Press.
Salisbury, R. F. (1964), ‘New Guinea highland models and descent
theory’, Man, vol. 64, art. 213.
Schneider, D. M. (1965), ‘Some muddles in the models, or how the
system really works ’, in M. Banton (ed.), The Relevance of Models for
Social Anthropology, Association of Social Anthropologists
Monographs, no. 1, pp. 25-79.
Spencer, B. and Gillen, F. J. (1927), The Arunta, Macmillan.
De Sousberghe, L. (1953), Structures de Parente et d'Alliance
d'aprds les Formules Pende. M6moires, Academie Royale des Sciences
Coloniales: Classe des Sciences Morales et Politiques, vol. 4, no. 1.
van Wouden, F. A. E. (1935), Sociale Structuurtypen in de Groote
Oost, J. Ginsberg.

224 Marriage and Alliance

Part Eight Divorce and Marriage Stability

The study of divorce has been important in kinship studies,

firstly because of its bearing on the position of women and the
nature of the kin groups to which they belong (or between
which they are exchanged) and, secondly, because increasing
divorce has sometimes been taken as an index of the effects of
social change. The direction of recent studies of divorce in
pre-industrial societies has been set by Gluckman’s comparison
of the Lozi of Zambia and the Zulu of South Africa. In Reading
15 he shows the link between frequent divorce and the bilateral
kin groups of the Lozi, and rare divorce and the ‘strong’
patrilineal descent groups of the Zulu. His work has been
elaborated and criticized by subsequent writers. Mitchell
developed the argument in his study of divorce in the changing
situation of urban Africa, where he points to the new types of
conjugal union now emerging (Reading 16).
l

<
15 M. Gluckman
Marriage Payments and Social Structure
among the Lozi and Zulu

Excerpt from M. Gluckman, ‘Kinship and marriage among the Lozi

of Northern Rhodesia and the Zulu of Natal’, in A. R. RadclifFe-Brown
and D. Forde (eds.), African Systems of Kinship and Marriage, Oxford
University Press for the International African Institute, 1950, pp. 166-203.
Revised for this edition by the author with a Postscript.

The evidence from the Lozi and Zulu indicates that we may have to
distinguish different types of marriage payments. There are
several aspects of marriage. It breaks or modifies certain existing
social relationships and creates new social relationships: the union
of a man and his wife, and an alliance between the kinsfolk of the
two spouses. It unites men and women to produce children to
occupy specific positions in the kinship system, since normally it is
through the marriage of their parents that children acquire their
kinsfolk on both sides. But clearly the structure of the kinship
system primarily determines the consequences of marriage in the
affiliation of children and therefore the attributes of the marriage
payment.
Marriage in most societies transfers to the husband a certain
common minimum of rights: almost always an exclusive right to
the wife’s sexual services, or to the lending of them, certain powers
over her person, rights to the produce of her economic activity
balanced by economic obligations to her, and a prima facie right to
be pater to her children. The extent and duration of these rights
varies greatly. We have seen that Lozi marriage payment is low,
considering their wealth in cattle, and that even then it dates from
a recent enactment. It does not transfer the woman’s fertility
altogether to her husband, let alone to her husband’s kindred.
Comparative data show that in general marriage payments fall
with the decreasing dominance of patrilineal descent until very
little is given in purely matrilineal societies. In Central Africa a
son-in-law may give some years of service, but it is in uxorilocal
marriage. The matrilineal Ovambo are rich cattle-owners, but a
husband gives for his wife only a present to her mother, and it is
killed for the wedding feast. That is, when children are not domin-

M. Gluckman 227
antly joined to the husband’s line he tends to give but little for his
wife, whose economic and sexual services alone are transferred to
him. In almost every African tribe he does give something to get
these rights and prima facie rights in relation to her children.
On the other hand, in patrilineal societies of the Zulu type the
marriage payment permanently transfers the woman’s procreative
capacity to her husband’s lineage. Therefore, relative to the society’s
wealth, the payment tends to be large.
I suggest that it may no longer be wise to name the common
institution of transferring goods by a single term (marriage pay¬
ment, bridewealth, bride-price). This leads to disputes about
whether the attributes ascribed to it, e.g. by Evans-Pritchard in
East Africa, apply to the transfer of goods among the Yako
or Lozi or some other tribe. The Lozi institution, on the surface a
similar transfer of property, is not the Nuer or the Zulu institution
when we come to examine their structural relations. For the
common element, the rights transferred in all societies, we may
follow Radcliffe-Brown’s use of the term marriage payment, but
it may be necessary to distinguish at least marriage payment Type
A, Type B, etc. I have analysed two types, and there are likely to
be more. Certainly my crude categories will not cover every
variation, and further analysis requires to be done in many tribes.
I consider that the data we have indicate that the basis of the
variation in the complexes is probably the different affiliation of the
children. Zulu society is chiefly distinguished from Lozi by its
structure of unilineal agnatic groups, which are exogamous, and
which are associated with a marriage rule by which the giving of
cattle to a woman’s father transfers her fertility for all her lifetime
to the agnatic group of her husband, who may indeed have been
dead before she was married to his name, or who may be a woman
or an impotent man. If the wife goes off in adultery, the children
are her husband’s; if he dies, she continues to bear for him;
if she is barren, or dies before bearing, a younger sister should
replace her. The outstanding fact is the extreme endurance of the
husband’s rights and their passing on his death to his agnatic heirs.
Legal marriage and the domestic unit it establishes are thus very
stable, though there may be frequent adultery. The legal emphasis
is the same among the Nuer, even though they have frequent
changes in the constitution of households. Marriage is stable in that

228 Divorce and Marriage Stability

wherever the woman is her husband accompanies her, even after
his death, to be pater of her children, for whoever their genitors
are, they belong to the man or group which gave cattle for her. The
husband or his heirs may let his wife go and not reclaim the mar¬
riage payment, since they get the children even though they lose
her other services as wife. In these tribes marriage payment thus
binds the woman’s reproductive capacity to the perpetuation of the
extended agnatic lineage. Each such group loses its daughters, but
insists on its rights to the fertility of its daughters-in-law.
Moreover, rights of inheritance vest in children exclusively
from the agnatic lineage of their pater. They have no rights in
their mother’s family unless they are unredeemed illegitimate
children, when they rank as if they were members of her agnatic
lineage. Therefore economic and other interests pull the children
almost entirely to the home of their pater and his agnatic lineage.
The Lozi have no extended unilineal kin groups. Among them
the corporate groups which endure in time are the cognates in
family villages or in sections of a royal village, based on the
mounds which dot the plain and which are centres of surrounding
gardens, fishing-sites and pastures. Lozi’s relatives in the patri-
lateral and matrilateral lines may be scattered at many places far
over the surface of the great flood-plain. Because their productive
activities are varied, and many fall in distant places in the same
month, they require cooperation and help from many people.
Kinship provides the framework for getting this help, but since
neighbours in the various parts where they have economic resources
are linked to them by patrilineal, matrilineal and mixed ties, they
emphasize relationship in all of their lines of descent. For them it
is not important to fix an individual’s relationship to a single line,
but to emphasize his links with many lines; therefore marriage does
not tie a woman’s procreative power to one line. She produces for
many lines. Among the Zulu economic and other interests coincide
with the pull of agnatic lineage ties; among the Lozi, with their
limited resources centred in restricted dwelling-mounds, economic
interests may pull a man to settle with his mother’s kindred, in
either of her lines of descent. As a woman’s productive capacity is
not tied to a single lineage group, the child goes to its genitor, not
to the legal husband of the woman. This removes the main buttress
of marriage; since the children can shift their allegiance and

M. Gluckman 229
emphasize that relationship which most pleases or profits them, the
family as a whole is an unstable association. I am not here referring
only to the instability of households: this is also marked among the
Nuer, but among them lineage ties always draw the family to¬
gether about the woman’s husband. Among the Lozi, too, a large
proportion of children grow up elsewhere as foster children and
not in the villages where they will ultimately settle. There is no
dominant pressure of interests and law to induce them to return to
their father’s home, though most do. They may go to the home of
any of their near ancestors. Adulterers can claim their children,
and adultery is very frequent. Men divorce their wives easily and
at personal will; women are always straining to be released.
Marriage, as well as domestic association, is unstable. I have
suggested that the reason for this is to be sought in Lozi kinship
structure, which is of a pattern common in Central Africa, but
which here is directly related to their modes of production and
settlement in their physical environment.
We see then that in these societies the types of marriage and the
forms of the family, with inheritance of property, rights of children
in father’s and mother’s lines and the rights of these lines to claim
them, laws of betrothal, destination of widows, rates of divorce,
status of wives in their husband’s homes, are all consistent with
certain types of kinship system. In some societies the household
group which is usually designated as the family may be unstable.
The structural stability of the society rests in the extended kinship
lines, one of which may be the nuclear framework on which local
communities are organized as corporate groups. These reach
back into the history of the tribe through all changes of personnel,
while the domestic family itself is always an ephemeral, and often
an unstable, association.1
I have presented my argument more strongly than I myself feel is
justified at present, in order to make clear the type of data and
analysis I consider likely to be most fruitful. I should have preferred
to do further comparative research before publishing it, had it not
been for the opportunity of presenting it in this symposium. I am
fully aware of the difficulties of establishing the validity of the
hypothesis, but even if it is wrong it may be useful. Some of the
1. See references on p. 166, n.l of my original article, especially
Evans-Pritchard (1938, pp. 64-5).

230 Divorce and Marriage Stabliity

difficulties are inherent in sociological analysis, since in this
there are always complicating variables. Others arise from the
vague and embracing use of categories and concepts (of which I too
am guilty) such as patrilineal, lineage, marriage, divorce, etc.
When is a marriage complete, and when can we class the separation
of a cohabiting couple as divorce ?2 Are the Bena patrilineal and the
Ila matrilineal, and have they lineages ? The literature is generally
confused and imprecise, and the posing of problems may help to
clarify descriptions.
I am myself uncertain whether it is the stability of people’s
attachment to specific areas, or patriliny or father-right itself, or
the agnatic lineage, or all of these together, which, whatever the
other variables are, tend to be associated with a strong marriage
tie. That any or all of these are significant seems definite to me.
Evans-Pritchard has stated that the Azande, organized on
father-right but without the agnatic lineage, had rare divorce in
the past. He has not reported on the problem fully, but it might
be that the State power prevented women from leaving their
husbands if they wanted to, i.e. there was instituted authority to
compel them to remain in marriage. When that authority was
restricted by British occupation divorce became rife.3 On the other
hand, despite European support for women against men in
Zululand, something there, I suggest the agnatic lineage, has
maintained the stability of marriage; while in Loziland the mar¬
riage tie has become looser and looser, even though tribal ethics
approve of firm unions, while not condemning divorce. However,
as we have seen, the code of the Zulu age-set regiments, backed by
the State, punished severely unchastity and infidelity: perhaps
2. For example, on the difficulty of establishing when there is a marriage
and when divorce, see Richards (1940). It is difficult to tell what Fortes
means by divorce (1937). He speaks of a high divorce rate but divorce from
a ‘proper’ marriage is obviously a most serious ritual and social step
(Fortes, 1945). In Dahomey (Herskovits, 1938) the divorce rate varies
according to which of the thirteen types of marriage is contracted, but it
seems to vary according to my postulate. Bohannan (1949) appears to
confirm the assessment in my essay.
3. I would now (1971) stress that at the time when Evans-Pritchard
studied the Azande of the Anglo-Egyptian Sudan they had been moved
out of their indigenous pattern of settlement and made to live in differ¬
ent kinds of neighbourly relationships along watersheds to escape sleeping-
sickness.

M. Gluckman 231
these codes are still the main sanctions on marriage despite all
structural changes. Yet the ban on prenuptial conception does not
seem to have retained similar force. Illegitimate births are frequent
while divorces are rare. According to old records and modern
field-workers this seems to be the position among all Southern
Bantu tribes who had tribal initiation into age-regiments (Kuper,
1947; Read, 1938; Casalis, 1859, pp. 231-2; Ellenberger and
Macgregor, 1912, p. 273; Ashton, 1952; Harries, 1929, p. 36;
Stayt, 1931, p. 152; Schapera, 1940, p. 294), save for the Thonga
of Portuguese East Africa. In his book Junod (1927) implies that
divorce is frequent, though this seems to contradict an earlier
article on another group of Thonga, and Clerc states explicitly of
the related Ronga that divorce is impossible once the wife has
borne a child: ‘she is forever bound to her husband’s village’
(1938, p. 89) (see also Junod, 1905, pp. 258-9; Harries, 1929).

Postscript, January 1971

The above attempt to correlate the contrasting patterns of mar¬
riage among the Lozi and the Zulu, and to relate these to other
institutions, in the background of a wider comparative summary
for other peoples, was written in 1947 and published in 1950. One
of the hypotheses I advanced, at least, inspired much further
research and both supporting and critical argument.1 It is clear
that that part of my argument which attempted to relate rates of
divorce to the kinship structure, requires now, not surprisingly,
twenty-four years after it was written, both clarification and
emendation. But parts of the core of my analysis still stand; and in
presenting a summary statement of my views in the light of over
two decades of later research, I emphasize the mode in which I
first formulated the hypotheses in order to eliminate what has been
considerable misunderstanding of my argument. I do so to make
clear on what lines I think future research might profitably
proceed.
The section reproduced in this book of Readings is the final
summary of an argument contrasting the patterns of marriage
rules among the Lozi and the Zulu, supplemented for the Zulu by
the very similar pattern among the Nuer far, to the north, as
1. I cite in the text of the original essay and in footnotes, earlier state¬
ments of parts of the hypothesis.

232 Divorce and Marriage Stability

reported by Evans-Pritchard. I had studied the Zulu before
studying the Lozi: and immediately I began work among the
Lozi I was almost bewildered by the ‘instability’ of their marital
associations, with very frequent divorce, either at the will of the
husband or by suit in court of the wife on many grounds, when
compared with the Zulu, among whom I did not record a single
divorce. The historical records and the genealogies I collected
showed that the Zulu had always had at most a very low rate of
divorce, while similar sources of information on the Lozi showed
that divorce had always been frequent among them. Hence it
seemed to me that, following the arguments advanced by Durkheim
about suicide, marriage and birth rates, these widely contrasting
rates of divorce would have to be accounted for by reference to
their respective social milieux, and that they could not be accounted
for by the circumstances leading to the break-up of each individual
marriage. Moreover, I judged that within the general pattern of
high specialization and segregation of sexual and conjugal roles,
which marks subsistence or tribal societies,2 the relationships
between Zulu and Lozi spouses respectively were not sufficiently
different to account for the very great difference in divorce rates.
I stress this point because D. M. Schneider pointed out in a
comment on work on this theme that ‘instability’ of marriage
can refer to several different types of phenomena, and I had
thought I had made it clear that I was speaking of the breaking up
of marriage as a legal or jural relationship.
As stated in the summary section reproduced above, I was at
work on a full comparative study of the problem, centring on the
Lozi and the Zulu, but covering data on all the peoples reported
on in books and articles available to me in Livingstone and Johan¬
nesburg, when the invitation to participate in the symposium on
African Systems of Kinship and Marriage (published eventually in
1950) led me to present an abstract of my analysis in a preliminary
essay. As my researches in the ethnographic literature, as well
as my own attempt to formulate hypotheses, had shown me how
2. See Gluckman (1955, ch. 3). For an account of this high segregation
of roles in at least enclaves in industrial society, and a most illuminating
hypothesis on factors leading to high segregation of conjugal roles, see
Bott (1971), in which she discusses later work on her hypothesis (1957);
and I comment on it in my ‘ Preface ’ to this edition.

M. Gluckman 233
imprecise were many of the statements on the rate of divorce, and
how unclear were reports on whether a widow remained married
under the true levirate to her dead husband while cohabiting with
one of his kinsmen, or was married by his kinsman in a new
marital union, and so forth, I suspended work on the fuller com¬
parison. This I did in the hope that my strong statement of the
general hypothesis set out above might encourage research which
would produce better data on the whole problem of divorce rates
in tribal societies, as well as on the other institutions associated with
marriage, and lead to the formulating of hypotheses with firmer
foundations. In this respect my essay has been very fruitful.
A comparison of Zulu and Lozi, with no divorces and with a
large number respectively, was effective, as would be a testing of a
new fertilizer producing radical improvements in yields on two
pieces of ground, and not on a checkerboard of plots. But clearly
for more refined work and for the comparing of divorce rates in diff¬
erent societies, it is essential to have more refined measurements.
My essay immediately drew from my colleague, J. A. Barnes, sugges¬
tions for measuring divorce rates in tribal societies which he later
refined (Barnes, 1949; 1967). Many of the problems discussed by
Barnes arise from sampling difficulties. Others are caused by the
fact that any still extant marriage may be terminated either by
divorce or by death of a spouse; and so forth. All I have space to
bring out here is that Barnes stressed the importance of trying to
investigate at what points in its life a marriage in a particular
society is most likely to break up, and he proposed that the best
way of presenting the rates of divorce is in the form of life-tables
of predictions of the chances of marriages surviving or not.
Clearly in the end this type of analysis is crucial. For example, we
are told that among the patrilineal Tallensi marriages are at high
risk in their first year, and thereafter are likely to survive. On the
other hand in many matrilineal, and in some bilateral, societies, a
woman after living with a husband (or series of husbands) after
her menopause may leave the husband to live with her brother,
with whom her children, particularly her sons, may have settled.
This is, for example, well described by Esther Goody for the Gonja
of Ghana, whose marital institutions are very similar to those of the
Lozi, where divorce may occur at any time during the marriage
(Goody, 1962)3. Thus, obviously, a full study would have to take

234 Divorce and Marriage Stability

into account the variable points at which divorce is likely to occur.
Some studies have followed Barnes’s formulae, but on the whole
we have still to discuss divorce rates stated in general terms.
Later work has concentrated on the problem of divorce rates
and not on the very general difference between Zulu-type and
Lozi-type rules of marriages, in terms of which I formulated my
whole hypothesis. I consider that failure to deal with the whole
hypothesis has led to certain major logical errors in hypotheses
formulated by most of my commentators. We may anticipate that
my argument could be wrong or right in a certain progression:
1. It may be right in its main outlines, but I may have formulated it
incorrectly (I believe I did).
2. It may be that either I, or my commentators, have confused the
internal consistency of marriage rules within a pattern, by explaining
the occurrence of a pattern by one of the rules logically contained
therein, and not by an independent principle having explanatory
power. (I consider this is what Ackerman (1963), Fallers (1957)
and Leach (1957), cited later, have done, and what I myself to
some extent did.)
3. It may be that the stability of marriage, including a low divorce
rate, as against instability including a high divorce rate, are
related to some external variable, quite different from the one on
which I laid emphasis.
4. It may be that the stability of marriage and rate of divorce are
controlled by so many variables that we shall never work out
interconnections, or the varying stability may be quite haphazard:
this conclusion we can only accept at the end of a considerable
amount of research, since accepting it now would entail ceasing
inquiry.
5. It may be that when we apply hypotheses to a range of societies
the hypotheses will cover most, but not all. Then we have to look
for special variables in the exceptional cases; for as Durkheim
said, ‘... to invalidate the proof, it is not sufficient to show that it
is contradicated by a few particular applications of the method of
3. Since the Gonja are so distant from the Lozi, and yet their marriage
rules are so similar, it is a pity that Goody did not refer to this general
similarity or point up certain differences in her article. I lack space to do so.

M. Gluckman 235
agreement or difference ... certainly we must not abandon
hastily the results of a methodically conducted demonstration’
(1938, p. 131).
Bearing these considerations in mind, what I demonstrated was
that there was a clear internal consistency between the patterns of
rules of marriage of Zulu-type societies, and Lozi-type bilateral
(or better, omnilateral) societies and matrilineal societies, of the
form shown in Table 1.
Table 1
Zulu Lozi
Very rare divorce Frequent divorce
Marriage endures beyond death No endurance of marriage
of spouse, in true levirate,a beyond death
or sororate (deceased still
married to relict, but pro¬
husband or pro-wife)
Marriage contractable by dead No such marriage contractable
man, or by woman to woman,
or by group to woman
Sororal polygyny encouraged Marriage to near kin of spouse
forbidden
Adulterine children to wife’s Adulterine children to their
husband genitor
If goods available, high Even if goods available, high
marriage payment marriage payment is unusual
a Levirate: the institution by which a widow continues to produce child¬
ren to her dead husband.

The work done since my comparison was published, either in

new ethnographic studies, or arising from comparative studies of
records not available to me, shows that my preliminary data were
inadequate when they indicated to me a general correlation of
patriliny itself, or what I tentatively called father-right, with
stable marriage, levirate, low divorce rate, etc., and of the opposite
situation with bilateral or omnilateral, and matrilineal societies.
I know of only a few reports on matrilineal societies with very
stable marriages, and I think in all of these special variables can
be found to account for what is clearly an exceptional situation
statistically (but see Jacobsen, 1967). This is confirmed by Acker-

236 Divorce and Marriage Stability

man’s conclusion, from a survey of more than 150 societies
in the Human Relations Area Files, that ‘. . . in matrilineal
societies marital instability is widespread . . (1963, p. 19).
But the later work clearly shows that there are two very different
types both of patrilineal societies and of bilateral societies, i.e. as
against my straightforward formulation, there are some patrilineal
societies which have a highish rate of divorce (Mitchell, 1963),
and some bilateral societies with a low rate of divorce (and asso¬
ciated institutions). I begin my discussion of explanations advanced
to explain these differences with the last, the two types of bilateral
societies. The thesis I start with is that advanced by Ackerman
(1963).
Ackerman made a laudable attempt to go beyond my own
modest effort to explain patterns of marriage in tribal societies
alone: he set out to try to bring together the work on tribal
marriage with work done on the occurrence of particular divorces
in modem American societies. The general thesis on which he
bases his analysis is that it has been shown in America that the
crucially important factors in determining whether or not a
marriage will survive or break up are: (a) homogeneity or hetero¬
geneity in the background and interests of the spouses; (b)
homogeneity or heterogeneity in the network of affective affilia¬
tions surrounding the married pair; and (c) how far a family shares
traits with the families of close friends. Wherever these are posi¬
tive, they tend to stabilize marriage. Ackerman sums up:
... it is clear that when the affiliations of both spouses are conjunctive,
that is, overlapping or identical, the behaviour and expectations of both
spouses are affected by the same norm or value sets. The affiliations
of the spouses can, on the other hand, be disjunctive: that is, each
spouse may maintain membership in different collectivities. In such
a situation, the behaviour and expectations of the spouses are affected
by different norms and value sets.4
4. In terms of Bott’s hypothesis (1971) referred to on p. 231, what she
calls a close-knit network around the family - particularly where the net¬
works of the two spouses overlap - tends to be associated with high seg¬
regation of conjugal roles; and this situation according to the works cited
by Ackerman should produce stability in marriage. The two sets of findings
require to be brought together: unfortunately I have not space in this
Postcript to attempt it. Ackerman does not cite Bott’s hypothesis, which
was first published in 1954.

M. Gluckman 237
To test whether these findings could be applied in general terms
to tribal societies (since there is no information on the situation of
particular marriages in tribal societies), Ackerman selected some
150 tribes. For those like the Lozi, who are bilateral (i.e. where
succession and inheritance can pass in any line of descent), he
examined whether people could or could not marry close relatives;
and he found indeed statistically that where there could be marriage
between close relatives, the divorce rate tended to be described
as low, but where such marriages were prohibited (as among the
Lozi), the divorce rate was described as high. Hence he argued that
‘as the network of conjunctive affiliations “tightens” around the
spouses, marital stability increases’.5
Unfortunately other work has prevented me looking into the
structure of the two types of bilateral societies, classified in terms
of whether they have or have not stable marriage. But what we
can say is that Ackerman’s findings demonstrate that under my
logical possibility (2) (p. 235), there are two types of marriage
patterns in bilateral milieux: where persons marry close relatives,
the divorce rate is low; where, as among the Lozi, such marriages
are prohibited, the divorce rate is high. Ackerman might have
cited here from my own article, that if two young Lozi who are
related insist on marrying against their elders’ wishes, the latter
curse the marriage: ‘You have chosen to marry, only death may
now divorce you; you may not separate from each other’ (p.
173 of my original article). I argued that these marriages were
cursed in this way because they began to weave new ties of kin¬
ship and affinity in a neighbourhood, or between previously
distantly related people, and hence such marriages were full of
important social implications.
Unhappily Ackerman found that for the societies recognizing
one main line of descent, there was ‘no association . . . between
the assessed divorce rate and any kind or combination of kinds of
endogamy in lineal societies ’. But he suggests that if we go below the
surface, we would find that the endogamous lineal community is
not strictly analogous to the endogamous bilateral community,
since organization on a unilineal basis introduces a new criterion
of conjunctive affiliations. He postulates two mechanisms: (a)
5. For lack of space Ackerman’s argument has to be stated in very bare
terms.

238 Divorce and Marriage Stability

the affiliations of each spouse may be extended to include those of
the other; and (b) the prior affiliations of one spouse may be
severed and she/he may be incorporated into those of the other.
He then tests this by taking the levirate and he finds that ‘the
levirate is a sufficient condition for a low divorce rate’ (p. 18).
But in my formulation both the levirate and a low divorce rate
were aspects of durable marriage: they are not, in my opinion,
independent variables, and hence one cannot be the cause of the
other. They show consistency in the pattern of marriage rules, and
this pattern should be referred to some external variable. It is note¬
worthy that all the societies shown by Ackerman as having the
true levirate are patrilineal, though some patrilineal societies
have no levirate and a low divorce rate.
Ackerman’s argument is that the levirate shows the radical
severing of the wife from her group and hence her incorporation
into the affiliations of her husband. Here he explicitly endorses
a proposition of L. F. Fallers (1957).6 The Soga of Uganda are
patrilineal but Fallers’ figures show that they have a high rate of
divorce. (Curiously, when Fallers compared the rate of divorces
among Soga in two different areas, he found that the rate was higher
in that area where patrilineal lineages were more concentrated.)
Hence he suggests (p. 116) that for the Soga my formulation
would have to be reversed: the ‘stronger’ the patrilineal descent
groups, the less stable is marriage. He concludes: ‘It is, on the
face of it, unlikely that Gluckman is simply wrong, for he presents
convincing evidence that in some societies the relationship which
he suggests does in fact hold. Rather, one suspects, there are some
societies in which patrilineal institutions act to stabilize marriage
and others in which patrilineal institutions, perhaps of a rather
different sort, have the opposite effect.’ Fallers then draws atten¬
tion to the way in which marriage among the Zulu and the Nuer
(as typical respectively of Southern Bantu and of many North¬
eastern African peoples) transfer all ‘a woman’s child-bearing
properties to her husband’s lineage’ so that once the marriage
payment is handed over, all children go to the husband, whosoever
the genitor.7 With this, he says, go the institutions listed by me
6. Fallers’ argument is very compressed in my presentation.
7. Shown most strikingly in a Kipsigis rule that if a woman who has
borne children to her husband is divorced and then remarried with marriage
payment, the children of the first marriage go to the second husband.

M. Gluckman 239
(p. 236), such as the true levirate, institutions not found among the
Soga save for approval of sororal polygyny. Among the Soga he
argues that the patrilineal lineages act to pull the spouses apart
because the wife remains a member of her natal lineage and
is not absorbed into her husband’s lineage as among Zulu and
Nuer.
Before I consider to what external factors Fallers relates this
difference, I note that Leach (1957)8 reported a similar distinction
among three Burma patrilineal peoples, who are culturally very
similar. These three peoples (Ordinary Jingphaw, Gauri Kachins
and Lakher) differ from the African peoples discussed in that they
have what Leach calls ‘class hypogamy’, with girls marrying from
lineages of higher status into those of lower status. The lineages
here, like those of Soga, do not form a segmentary system. I set out
the difference in marriage rules in Table 2.

Table 2

Jingphaw Gauri Lakher

Formal religious Divorce not common No religious element

marriage, indis¬ but possible, most in marriage, divorce
soluble unless easily by provision easy and apparently
lineage sister of lineage sister frequent
replaces wife to replace wife, but
marriage payment
returnable

Widow inheritance, Widow inheritance, Widows can remarry

possibly leviratic or marriage
payment partially
returned

Marriage payment Marriage payment Marriage payment

high, transfers wife high, transfers only high, very
and children children not wife complicated,
transfers children
only, not wife

Leach (1957) argues that since there is no difference in the system

8. Presumably in press at the same time as Fallers’s article.

240 Divorce and Marriage Stability

 of descent, the differences are to be found in the institution of
marriage itself. Starting with the assumption, which I accept, that
it is ‘arguable that it is in the general nature of kinship that a
sibling link is “intrinsically” more durable than a marriage tie’
(p. 120), he states that in the Jingphaw system the institutions
operate so that ‘ the fragility is located in the sibling link between
the bride and her lineage brothers; in the Lakher case the fragility
is in the marriage relationship itself. ... It may be observed that
the fact that the Lakher and Gauri are more sharply stratified -
more class conscious - than the “Ordinary Jingphaw”, also fits
with the pattern . .. described. Jingphaw aristocrats “sell”
their daughters outright; Gauri and Lakher disdain to do so,
they merely permit their inferiors to have sexual access conditional
on the long continued payment of tribute fees’, and in return the
inferiors get the children. Leach then argues, as Fallers does, that
one cannot explain the differences in terms of patriliny, since in
these easy-divorce patrilineal systems the patrilineal tie of a woman
to her lineage is asserted. He is therefore equally, and rightly,
critical of my ascribing the difference to ‘strong father-right’.
He proceeds to propose that the factors controlling the differences
may have to be sought in a difference between two types of unilineal
descent systems: those where ‘ the ongoing structure is defined by
descent alone and marriage serves merely to create “a complex
scheme of individuation” within that structure . . . [and] In
contrast . . . the category of those societies in which unilineal
descent is linked with a strongly defined rule of “preferred
marriage”.’ Here one would have to look at the economic and
political significance of the chain of debts established by marriages,
and running from generation to generation.9
Fallers summarized his argument most generally by postulating
that ‘common corporate group memberships [of the spouses]
tend to reinforce the marriage bond, different corporate group
memberships to work against it’ (p. 121). Here is a formulation
which seems consistent with Ackerman’s (though Ackerman
followed on Fallers and he generalized more widely). It seems also
consistent with Leach’s formulation, in so far as it deals with the

9. This argument is of course compressed, but I think it fairly presents the

gist of Leach’s paper.

T-K-I M. Gluckman 241

pattern of marriage rules. But Fallers seeks for the difference, not
in the balance of the lineal and the affinal ties in the ongoing struc¬
ture, but in the role of lineal groups in the political system. He
considers that among the Nuer and Zulu, lineages are the building
blocks of the political structure, with patriliny directly determin¬
ing citizenship; in contrast among the Soga though some import¬
ant political roles are patrilineally inherited, others quite as
important are acquired through citizenship. Hence in the Zulu-
type system, he argues, since ‘patrilineal descent is crucial to
political status and, since her children are thus tied to her hus¬
band’s lineage, [the wife] is also, through them, firmly tied to it.*
She becomes ‘absorbed’ into her husband’s lineage.
I consider that on the whole Leach, as he himself says, is defining
variables contained within the patterns of marriage rules, and I
am not attracted by Fallers’ search in the political structure for the
external factor in lineage organization which determines the
absorption or not of the wife into the husband’s lineage. I do not
think that we can see either Zulu or Nuer polities as built out of
lineages directly, since territorial units in both are occupied by
many persons not related to one another by agnatic ties. While
clearly marriages of high political importance may thereby be
stabilized (see Marx, 1967, pp. 147 ff.), I still feel strongly that the
clue to the general stability of marriage is to be sought, as I sought
it, in the system of kinship relationships. Unfortunately in some of
my phrasings I wrote as if the kinship system might be summarized
simply in terms of descent rules and the rights of ascendants: as
patriliny or matriliny, and father-right. But even the summary
final section reproduced here indicates that in the text of my
argument I examined the kinship relationships of Zulu and Lozi
as these operated through links to property and to the transmission
of property. Clearly it is not possession of lineage property alone
which can stabilize marriage, as is shown by the high rate of divorce
among Ashanti and other matrilineal peoples who are strongly
attached to valuable areas of arable land. What I emphasized as
between the Zulu and Lozi was that among the Zulu a man trans¬
mits property to his sons through his wife or wives: this mode of
transmission, which I named the ‘house-property complex’,
appears most clearly where a man has more than one wife. Through¬
out the Southern Bantu, up to the Limpopo River (i.e. stopping

242 Divorce and Marriage Stability

 short of the Shona), and again in North-eastern Africa, in the
house-property complex a man allocates land and/or cattle to
each wife, and when he dies that wife retains right in those cattle
and/or that land which pass to her own sons, as against her
husband’s sons by his other wives. In short, each wife becomes the
nucleus of a set of patrimonial property for her sons, and they have
claims on the marriage payments of their own sisters. As each
wife forms thus a patrimonial nucleus she is very firmly attached
to her husband: if he dies, she and that property remain nucleated
for her sons and her future children, who are begotten by a pro¬
husband in the name of her dead husband; i.e. there is the true
levirate - if the wife dies, the patrimonial nucleus endures, and
she may be replaced in the sororate. The very low rate of divorce
and the endurance of marriages beyond death in tribes with this
system of descent and transmission of property thus keeps clear
the structure of patrimonial estates in relation to one another.
On the other hand I found that among the Lozi, when a man
dies, the land which his wives had worked is pooled after his
death and passes back into the control of his heir, who may be a
brother, son, uterine nephew, aut alius. And similarly it appears to
me that among the Soga, the Shona and a number of other patri¬
lineal tribes reported to have a highish or high (Soga) rate of
divorce and no levirate, succession and inheritance are adelphic
and not filial - i.e. they may go to brothers before passing to the
filial generation. In these systems wives are not, then, nuclei of
patrimonial estates, and are not as fully absorbed in the husband’s
lineage - perhaps in consequence.10
My early analysis laid stress on the affiliation of children, which,
in agnatic systems, I thought pulled their mother to their father, i.e.
the woman’s husband could claim her children. Hence I spoke of

10. In my Marett Lectures for 1965-66 I suggested that this is why in

house-property systems wives are suspected of witchcraft, since struggles
over position and property within agnatic groups focuses on them, as
against the situation in patrilineal societies where there is adelphic succession
and men are the focus of such struggles. See Gluckman (1971) and also
the statement by J. Middleton and E. H. Winter (1963, p. 161) that in East
African tribes, among those who ascribe misfortune to witchcraft, it is
where each wife is the centre of a separate estate within her husband’s
property, in house-property systems, that women are accused of this
inherent evil.

M. Gluckman 243
‘father-right’ to cover this claim. But as I noted in a letter quoted
by Leach in his article (1957), I had by 1957 decided that this term
was incorrect: I had begun to realize that it would be better to
characterize Zulu-type systems as dominated by ‘wife/son-right’,
rather than by ‘father-right’, by filial inheritance rather than by
patriliny. It would be interesting to see how this works outside of
Africa: certainly the ‘ Ordinary Jingphaw ’, insofar as I can identify
them in Leach (1954, p. 165), appear to have a ranking of wives
akin to the Zulu.
I argue thus that the situation of virtually no divorce and of the
true levirate and sororate will be found with house-property filial
inheritance, but there will be some divorce at least in patrilineal
systems with adelphic succession and inheritance. Mitchell found
that in a small number of African societies where divorce rates had
been calculated on Barnes’s formulae, on the whole, save for the
Soga, even adelphic succession systems had more stable marriage
than matrilineal systems. He suggested (1963) that where the
rights over a bride are transferred to a group of men, rather than
to her husband alone, marriage will be stabilized. I believe this to
be so, but consider that this rule itself appears most strongly in
house-property systems, though as Mitchell says it may act to
stabilize marriage in matrilineal systems, within their overall
relatively high instability of marriage. If there is anything in my
argument, it may have to be validated by a closer scrutiny of the
relationship between women and property nucleations within
matrilineal societies. I would also like, given time, to attempt to
investigate whether similar variations in the disposition of
property rights underlie the two types of bilateral societies dis¬
tinguished by Ackerman: i.e. to investigate whether differences in
property arrangements underlie the occurrence of endogamy or
exogamy in bilateral systems.
I am encouraged by Jacobson (1967) to feel that my hunch here
is correct. She examines the relation between marriage payments
and marital stability in all the tribes of the Congo-Gabon region,
an area of great cultural admixture. She finds that independently of
the usual classification of societies as matrilineal, bilateral or
patrilineal, a careful study of attitudes to divorce shows that the
very many tribes of the region manifest two attitudes to marital
stability: the first, a relative indifference; the second, a strongly

244 Divorce and Marriage Stability

 positive conception of marital stability as such.11 Relative indiffer¬
ence involves an elaborate specification of ‘grounds’ on which
there can be divorce, ascription of guilt and liability to economic
sanctions, absence of institutionalized obligatory mediation in
divorce cases. High valuation of stability is marked by absence of
specification of grounds, divorce being granted only when parties
can no longer live together, there are no specific economic sanc¬
tions save return of marriage payment independently of blame,
and repeated and compulsory attempts at mediation to save the
marriage.
The differing attitudes appear in proverbs: in one set of peoples
these stress the importance of details of marriage for its durability,
while in the other they stress the overlooking of similar trifles.
The rate of divorce correlates positively with the two attitudes.
Jacobson’s tables show the levirate as occurring among almost all
the tribes considered, but she does not discuss whether she means
true levirate or the possibility of widow-inheritance or remarriage
of the widow within the husband’s kin.
With what institutions does the distribution of these attitudes
correlate? Jacobson found that in tribes with ‘strict matrilineal
descent and in tribes with strict patrilineal descent ’ alone there is
the attitude linked with high divorce rates. But where the mother’s
agnatic group is also important, and in matrilineal societies
where the father’s group is important, marriage is valued highly
and the divorce rate is low. She argues that as we find peoples
which tend to fall between the two, the rate becomes intermediate.
Her hypothesis is that where ego can satisfy all his wants within
one group, there can be little interest of the groups in the stability
of the marriage. Where there are complementary descent groups
or complementary kin groups, the children will constitute a focus
of common interests and create an interest in marital stability,
i.e. where ego ‘is dependent on both his father’s and his mother’s
group to provide for the sum total of his political, economic,
social, legal, religious and affective rights and needs, seeing that
the two groups complement each other in these respects.’ This

11.1 reported that the Lozi and other tribes with high rates of divorce
spoke as if they valued stable marriages, and they do; Jacobson, as will be
seen, is concerned with the rules by which valuations of stability are in
practice achieved.

M. Gluckman 245
suggests to me that it may be in relation to rights to property
of children that factors stabilizing marriage in tribal societies
will be found. But though it is clear that rates of stability of
marriage in tribal societies are influenced by institutional milieux
and patterns of marriage rules even more obviously show internal
consistencies, more probably the degree of stability depends on
more than one type of factor.

References
Ackerman, A. (1963), ‘Affiliations: structural determinants of
differential divorce rates’, Amer. J. Sociol., vol. 69, pp. 13-20.
Ashton, E. H. (1952), The Basuto, Oxford University Press.
Barnes, J. A. (1949), ‘Measures of divorce frequency in simple societies’,
J. roy. anthrop. Inst., vol. 79, pp. 37-62.
Barnes, J. A. (1967), ‘The frequency of divorce’, in A. L. Epstein
(ed.). The Craft of Social Anthropology, Tavistock, pp. 47-100.
Bohannan, L. (1949), ‘Dahomean marriage: a revaluation’, Africa,
vol. 19, pp. 273-83.
Bott, E. (1971), Family and Social Network, Tavistock, 2nd edn.
Casalis, J. E. (1859), Les Bassotos, Societe des Missions Evangeliques.
Clerc, A. (1938), ‘The marriage laws of the Ronga tribe’, Bantu
Studies, vol. 12, pp. 75-104.
Durkheim, E. (1938), The Rules of Sociological Method, Free Press.
Translated by S. A. Solvay and J. H. Mueller from Les Regies de la
Methode Sociologique, 1895.
Ellenberger, D. F., and Macgregor, A. (1912), History of the
Basuto, Caxton Publishing Co.
Evans-Pritchard, E. E. (1938), Some Aspects of Marriage and Family
among the Nuer. Republished 1945 as Rhodes-Livingstone Paper,
no. 11.
Faller, L. A. (1957), ‘Some determinants of marriage stability in
Busoga: a reformulation of Gluckman’s hypothesis’, Africa, vol. 27,
pp. 106-23.
Fortes, M. (1937), Marriage Law among the Tallensi, Accra, Govern¬
ment Printing Office.
Fortes, M. (1945), The Dynamics of Clanship among the Tallensi, Oxford
University Press.
Gluckman, M. et al. (1953-4), ‘Bridewealth and the stability of
marriage’, correspondence in Man, arts. 75, 122, 223, 279; 96, 97, 153.
Gluckman, M. (1955), Custom and Conflict in Africa, Blackwell.
Gluckman, M. (1971), ‘Moral crises: magical and secular solutions’,
in M. Gluckman (ed.). The Allocation of Responsibility, Manchester
University Press.

246 Divorce and Marriage Stability

Goody, E. (1962), ‘Conjugal separation and divorce among the Gonja
of northern Ghana’, in M. Fortes (ed.), Marriage in Tribal Societies,
Cambridge University Press, pp. 14-54.
Harries, C. L. (1929), The Laws and Customs of the BaPedi and
Cognate Tribes, Hortors.
Herskovits, M. (1938), Dahomey, An Ancient West African Kingdom,
J. J. Augustin.
Jacobson, M. (1967), ‘Marriage and Money’, Studia Ethnographica
Upsaliensia. vol. 28.
Junod, H. (1905), ‘The Bathonga of the Transvaal’, Addresses & Papers:
Brit. South African Ass. Advancement Sci., vol. 3, pp. 222-62.
Junod, H. (1927), The Life of a South African Tribe, vol. 1, Macmillan.
Kuper, H. (1947), An African Aristocracy, Oxford University Press.
Kuper, H. (1950), ‘Kinship among the Swazi’, in A. R. Radcliffe-
Brown, and D. Forde (eds.), African Systems of Kinship and Marriage,
Oxford University Press.
Leach, E. R. (1954), Political Systems of Highland Burma, Athlone
Press.
Leach, E. R. (1957), ‘Aspects of bridewealth and marriage stability
among Kachin and Lakher’, Man, vol. 57. Reprinted in Rethinking
Anthropology, 1961, Athlone Presss, pp. 114-23.
Marx, E. (1967), Bedouin of the Negev, Manchester University Press.
Middleton, J., and Winter, E. H. (eds.), (1963), Witchcraft and
Sorcery in East Africa, Routledge & Kegan Paul.
Mitchell, J. C. (1963), ‘Marriage stability and social structure in
Bantu Africa’, Pro. internat. Union for the study of population, vol.
2, pp. 255-63.
Read, M. (1938), ‘Moral code of the Ngoni’, Africa, vol. 11, pp. 1-24.
Richards, A. I. (1940), Bemba Marriage and Modern Economic
Conditions, Rhodes-Livingstone Paper, no. 4.
Schapera, I. (1940), Married Life in an African Tribe, Faber & Faber.
Stayt, H. (1931), The Bavenda, Oxford University Press.

M. Gluckman 247
16 J. C. Mitchell
Social Change and the Stability of Marriage
in Northern Rhodesia

J. C. Mitchell, ‘Social change and the stability of marriage in Northern

Rhodesia’, in A. Southall (ed.), Social Change in Modern Africa,
Oxford University Press, pp. 316-29.

Measures of marriage stability

It is possible using appropriate statistical devices, to predict
how long a given marriage will last. These statistical devices are
analogous to those used in the life table and with suitable material
a marriage duration table can be prepared (Barnes, 1949). In these
tables divorce is analogous to death and the tables are constructed
from an analysis of the periods marriages have survived before
being dissolved by divorce or death, in the same way that life
tables are constructed from the analysis of the ages at which
people die. In this Reading I am not concerned with the termination
of marriage by the death of one of the spouses, so that when I
discuss the duration of marriage I mean the duration of marriage
before it is ended by divorce or legal separation.
Failing measures of this sort, which are the only true com¬
parative statistics we can work with, we must use the divorce
ratios of the sort we have suggested elsewhere (Mitchell and
Barnes, 1950), but we must appreciate that these measures may
be seriously disturbed by differences in the average age of first
marriage or in the age distribution of the population.
Eventually when the standard of collecting divorce statistics
has advanced sufficiently we will be able to compare the stability
of marriage from one society to another, as Barnes has started
to do. He has been able to say, on the basis of his analysis, that
the average Ngoni marriage will last for 15-5 years while among
the Americans it will last 20-4 years.
This figure, like the life expectancy, is a general average arrived
at purely empirically. It is obvious that the causes of death are
numerous but the overall operation of these causes results in an
expectancy of life which does not vary much from one year to

248 Divorce and Marriage Stability

another in the same community. The life expectancy, however,
differs widely from one society to another. Demographers look
beyond the life curve itself and are able to isolate from the myriad
causes of death several significant factors and use these to explain
the differences between societies. The high infantile mortality
rate, for example, in non-literate societies, is instrumental in
reducing the general life expectancy.
We may approach the analysis of the stability of marriage in
the same way. There is no doubt that whether the marriage breaks
up or not depends on a variety of factors, economic, social and
personal, and that these factors operating simultaneously give rise
to an overall divorce rate which does not differ much within the
same society from one year to the next. The multiplicity of
factors, therefore, operate adventitiously and not systematically,
in the same way that random factors affect the trajectory of a
missile to give rise to a cone of fire and not to a constant error.
But from one society to another certain major determinants of the
divorce rate can be isolated. In other words we may adopt the
same type of reasoning as used by Durkheim in his analysis of
suicide. A variety of factors he argues affect the incidence of
suicide but the rate is determined in different countries, and at
different times in the same country, by a few major sociological
variables.

Kinship as the major determinant of marriage stability

Therefore, in trying to assess the probability of divorce in any
particular marriage we should take into account a wide range of
relevant factors, but if we are interested in trying to understand,
say, the trend in divorce rates in general, then we need to be able
to assess the influence of the major determinants, and, knowing
these trends so deduce the trend in the divorce rate.
There have been many attempts to do this for Western European
society but these studies are not directly relevant to the situa¬
tion in Africa. Here we need to take account of the fact that
tribal societies are organized on a different basis, and indeed we
should make this the starting point of our analysis.
Marriage is essentially a relationship between two persons
which fits into a larger pattern of social relationships. Marriage,
therefore, operates within a social system and being itself part

J. C. Mitchell 249
of it we should expect that the social system would determine
its nature. In tribal societies in Africa, as in other parts of the
world, the kinship system is a dominant feature of the social
structure so that it is reasonable to suppose, as Gluckman (1950)
has done, that the major determinants of marriage stability in
tribal societies lie in the kinship system.
Basically Gluckman’s argument is that in societies which are
organized on the basis of corporate patrilineages marriage be¬
comes the institutionalized means whereby new members are
recruited to the patrilineages. Given rules of incest and exogamy,
the patrilineal groups can survive only by securing rights over
the reproductive powers of women from other groups. This
they do by marriage, involving as it does the payment of bride¬
wealth and the performance of appropriate ceremonies. These
rights over the reproductive powers of the women are trans¬
ferred permanently to the husband’s group so that the marriage
cannot easily be dissolved. On the specific reason why the rights
should be transferred permanently Gluckman is not clear, but I
would suggest that it is related to the prolonged infancy of the
human child and the fact that in general the child needs its mother
for several years before it can safely be separated from her.
Before this stage is reached the woman is usually pregnant again
and therefore will not easily be released by her husband’s group.
In societies in which there are corporate matrilineages the
woman produces children for her own descent group. Her re¬
productive powers are therefore not transferred and the dura¬
tion of the marriage does not affect the welfare of the child, nor
the rights of its matrilineal kinsmen over it. The duration of
marriage here is therefore determined mainly by the personalities
of the spouses and divorces usually are relatively frequent.
In bilateral societies kinship alone cannot be the basis of
corporate kin groups, and therefore the possession of rights over
a woman’s reproductive powers is not a live issue. Marriage
therefore, as in matrilineal societies, is brittle and of short duration.
In general these hypotheses have been upheld and we are able
to use them in interpreting the trends in societies where the
kinship systems are changing. In order to do this we need to
emphasize some general features of kinship systems.
Marriage, we have argued, will assume different forms within

250 Divorce and Marriage Stability

different kinship systems. We may approach the analysis of the
role marriage plays in different kinship systems by considering
the sorts of rights and obligations which the formal act of mar¬
riage confers upon the contracting parties. Bohannan (1949)
on the basis of her analysis of Dahomean marriage has suggested
that we can approach the problem from the point of view of the
rights a man or his kinsmen acquire over a woman through
marriage. One set of rights are those held in her as a wife. These
are the rights to her services both domestic and sexual and they
may be termed uxorial rights. The other set of rights are the rights
in the woman as a mother and refer particularly to the ownership
of the child. These are the genetricial rights.
Now it follows that where there are corporate agnatic descent
groups the patrilineage as a whole acquires the genetricial rights
in the woman while the uxorial rights are usually held by one of
their number. In this situation, as Gluckman has argued, divorce
is rare and difficult. In matrilineal societies the uxorial rights are
acquired by a man who is not a member of the matrilineal descent
group while the genetricial rights in the woman are permanently
held by the woman’s mother’s brother, or her own brothers. As
Gluckman points out the uxorial rights may easily be transferred
from husband to husband without affecting the genetricial rights
in her so that divorce here is easy and frequent.
In bilateral societies recruitment to corporate groups is not by
kinship alone, so that the genetricial rights in women are not
relevant. Here uxorial rights as in matrilineal societies are easily
transferred from husband to husband and divorce is consequently
easy and frequent.
In order to appreciate fully the significance of genetricial rights
as rights over the reproductive powers of the woman, we should
consider the rights and obligations which the possession of these
rights confers over the children born to the woman. In a society
where corporate groups exist the possession of genetricial rights
over a woman implies that her children will belong to the group
which possesses these rights. The children, therefore, may expect
the succour and protection from the other members of the group
and they in turn must be loyal to their fellows. As Bohannan
points out, the acquisition of genetricial rights confers jural
authority over the child. The word ‘jural ’ here presumably means

J. C. Mitchell 251
that the sanctions behind the implementation of the obligations
of the child to its group lie in force either through the mechanism
of the courts or through the feud. In other words genetricial
rights, being linked particularly to corporate relationships,
operate in a political field.
Where corporate groups exist, the genetricial rights are held,
as I have pointed out, by the group as a whole. But some gene¬
tricial rights may also be held by an individual as, for example,
among the Bemba of Northern Rhodesia. The Bemba are a people
amongst whom succession to office and membership of the non¬
corporate association of clanship is determined by matrilineal
descent. The most important political unit is the village. Villages
are built up on the basis of matrilocal extended families (Richards,
1950, p. 227), i.e. they are made up of the village headman and his
married daughters. There may also be several of the headman’s
sisters and their daughters in the village. The Bemba practise
shifting cultivation of a slash-and-burn type in rather poor soils
and the size to which each village may grow is limited by this
ecological factor. As the trees become exhausted in its vicinity
each village must shift its locale approximately once in every
five years and this is the time, as Richards points out, at which
villages are particularly liable to fragment. The economy of
villages is based upon the cultivation of eleusine in ash patches
prepared from the burning of branches lopped from the trees
by the young men of the village. The labour of the young men
is crucial in the ecological balance between the Bemba and their
somewhat inhospitable agricultural environment because pollard¬
ing the trees as against cutting them down completely, allows a
much more rapid regeneration of the trees, and thus ultimately a
higher carrying capacity of the land. The prosperity of a Bemba
man, therefore, in a land where the margin between subsistence and
starvation is narrow, is counted in food, and the labour of young
men is essential to the production of food. This labour is provided
by the husbands of the man’s daughters. A man, therefore, is
heavily dependent on his daughters’ husbands for economic
security, or as Richards (1956, p. 46) graphically puts it: ‘Wealth
consists of the power to command service.’
A Bemba man’s dependence on his daughters’ husbands is
limited not to economic relationships only: it extends also into

252 Divorce and Marriage Stability

the social and political field. Within the social field, prestige in
Bemba society is associated with the economic independence of
the man. As Richards points out (1956, p. 42) a man can eventually
acquire status and prestige as head of his own extended family
with sons-in-law working for him whether he moves to one of
his own relatives or stays where he is. But if he wishes to increase
his renown he must try to become a village headman. Among the
Bemba as amongst other Central Bantu, the ambition of every
man is to become a village headman. In order to do this he must
have the right to remove his wife and her children from her
village. With his daughters’ husbands then, provided that he has
the permission of the chief, he will be able to set up his own
village.
Genetricial rights amongst the Bemba therefore centre on a
man’s jural authority over his daughters and their husbands.
In Bemba marriage three payments are made (Richards, 1940).
The first, a trifling amount of say 12^p, called nsalamu, confers
uxorial rights upon the man. The second payment, called chisungu,
is made during the girls’ initiation and is made once only in respect
of each girl. We can look upon it as a ritual payment which
releases the mystical sexual powers of the girl. The third payment,
mpango, formerly made up of barkcloth, hoes or arrows, conferred
genetricial rights upon the husband. It was through this payment
that the man established his rights to remove his daughter from
his wife’s village and thereby established his rights to the services
of his sons-in-law. As Richards (1956, p. 44) puts it: ‘The mpango
gave the father those limited rights he secured over his children
although it did not, in this matrilineal society, give him complete
control over his wife’s reproductive power as does the cattle
lobola among the Southern Bantu.’
But these rights the Bemba man held personally: his brothers,
for example, could not enforce them on his behalf, and from this
point of view they differ from the genetricial rights acquired by
a man in a patrilineal society with corporate kin groups. Divorce
was easy and apparently frequent among the Bemba during the
early years of marriage (Richards, 1940, p. 101) but less frequent
and more difficult after children had been born (Wilson, 1941, p.
42). It was when the children were born that the man’s ambitions
became capable of being realized and it was to his advantage

J. C. Mitchell 253
 therefore not to seek a divorce at this stage: he could only build
a village with the assistance of his daughters’ husbands. The
genetricial rights a Bemba man acquired over his wife, therefore,
only became effective when his daughters were old enough to
form the nucleus of a village, and the divorce apparently became
more difficult and less frequent as this stage was reached.
This argument links the stability of marriage with a man’s
realization of the genetricial rights he acquires in his wife, whether
this is in a society organized in corporate patrilineages or in a
matrilineal society in which a man may personally acquire
genetricial rights over his wife. In societies organized in corporate
matrilineages such as among the Yao and Chewa a husband never
acquires genetricial rights over his wife and therefore marriage is
never stabilized by this factor. In bilateral societies also, as
amongst the Lozi, genetricial rights are not involved so that we
would expect the divorce rate to be higher than say amongst the
Bemba. The rights which are disputed amongst the Yao and the
Lozi are uxorial rights - particularly adultery - and not the rights
over children. Among the Bemba, while adultery cases clearly
figure prominently, the rights of a man to the portion of the
marriage payment made for his daughter and the rights to the
children are also involved. Note for example Richards’s record
(1940, p. 57) of the case where a woman screamed: ‘We shall take
the baby. ... That is no father that! He paid us no mpango.’
Amongst bilateral peoples, in fact, the kinship system may
operate to increase the instability of marriage. Gluckman (1950,
p. 201) in fact describes how the Lozi system encourages the wide
ramification of cognatic and affinal ties and how a person uses
these ties to exploit the resources at its command. He writes:
Because their productive activities are varied and many fall in distant
places in the same month, they require cooperation and help from many
people. Kinship provides the framework for getting this help, but since
neighbours in the various parts where they have economic resources
are linked to them by patrilineal, matrilineal and mixed ties, they
emphasize his links with many lines: therefore marriage does not tie a
woman’s procreative power to one line. She produces for many lines.

It follows from this that from the point of view of any one
individual the less his kinship links coincide with another’s

254 Divorce and Marriage Stability

 /
the better his chances of operating these links to his personal
advantage.
In other words in bilateral systems the emphasis is on filiation
while in matrilineal or patrilineal societies the emphasis is more
likely to be on corporate relations. Where the emphasis is on
filiation we expect to find high divorce rates.
To summarize we should expect:
1. In patrilineal societies with corporate groups, i.e. genetricial
rights acquired by groups - marriage duration is long.
2. In patrilineal or matrilineal societies where genetricial rights
are acquired by individuals - marriage duration is medium.
3. In matrilineal societies with corporate groups, i.e. where
genetricial rights are never transferred or bilateral societies where
genetricial rights are not relevant - marriage duration is short.
In order to test this hypothesis we need detailed information on
the duration of marriage such as to enable us to compute expec¬
tancies of marriage duration and this we cannot yet do.

Marriage stability and modern conditions

In broad terms we have tried to relate marriage stability in tribal
areas to social structure: in particular, following Gluckman’s
lead, we have tried to relate it to the kinship system and the rights
which marriage confers within these kinship systems. Therefore
when we turn to modem conditions we must try to relate changes
in marriage stability to changes in the kinship system whether
through time (historically) or from place to place (situationally).
Richards’s (1940) study of Bemba marriage and Barnes’s (1951)
study of Ngoni marriage are studies of the historical sort. I intend,
however, to try to consider marriage stability not in historical
but in situational change, i.e. to look at marriage stability where
the kinship system is altered fundamentally by the social situation
in which it is set. The most striking examples we have are in the
newly created industrial towns.
In these towns we may isolate two main types of related vari¬
ables, i.e. the demographic and the sociological. Demographic
factors influencing marriage would be such factors as the dis¬
proportion of sexes which Wilson (1941) emphasized so much.
Another is the tribal heterogeneity of the towns and a third is the

J. C. Mitchell 255
relatively high mobility of the town dwellers. All of these demo¬
graphic characteristics have their influence on marriage stability
as such, but the social structure, itself influenced by demographic
variables, also affects marriage stability. Fortes (1953, p. 24) has
pointed out that there is evidence that ‘unilineal descent groups
break down when a modem economic framework with occupa¬
tional differentiation linked to a wide range of specialized skills,
to productive capital and to monetary media of exchange, is
introduced.’ The unilineal descent groups under these conditions
tend to become reorganized in nuclear families and the descent
tends to be reckoned bilaterally (Gough, 1950, p. 85). This is
particularly true in the industrial towns where the demographic
and social conditions militate strongly against unilineal corporate
group activity. As Fortes (1953, p. 36) remarks: ‘A lineage cannot
easily act as a corporate group if its members can never get to¬
gether for the conduct of their affairs.’ In these towns where
the emphasis is on heterogeneity and on mobility the conditions
prevent the daily participation of unilineal kinsmen in joint
activities. There are no cattle, there is no land to focus the activities
of say a patrilineage, and the working hours and conditions
preclude their participating in joint ritual and ceremonies for
protracted periods. The emphasis in town is on individual success
in a competitive environment and this emphasis cuts across the
ideology of corporate descent groups.
Kinship instead remains important in the form of filiation - as
an ever-ramifying network of personal relationship which links
any given person to others in haphazard fashion across the town¬
ship. Kinsmen, who in tribal conditions played an unimportant
part in the life of a person, assume a new importance in town
where the overwhelming majority of contacts are with strangers.
A person may call on distant kinsmen for help in town whom he
would not consider approaching in his tribal home (Wilson,
1941, p. 53; Epstein, 1953, p. 11; Mitchell, 1957, p. 27). Though
kinship is still important in the new industrial towns it possesses
qualities different from kinship in tribal areas. Wilson (1941, p. 51)
has drawn attention to an example of the way in which the re¬
lationships between kinsmen must inevitably alter in the urban
situation. He points out that in the tribal areas the economic bond
between Bemba men and their daughter’s husband was par-

256 Divorce and Marriage Stability

ticularly important. We have seen that the Bemba men acquire
rights to the services of their daughters’ husbands through the
mpango payments. He writes:
the wife’s parents were the skilled foremen, the young couple were the
relatively skilled labourers of those times, and both couples alike
benefited from their cooperation. ... The domestic group formed by
the old couple and a number of daughters and sons-in-law was the
main factory of Bemba wealth and their living depended upon it.
But in towns, of course, this essential link between a man and his
daughters’ husbands cannot exist because ‘the main productive
relationship in which the urban Bemba are involved is that
between European capital and skill on the one hand, and African
unskilled labour on the other; it is on getting a job, not on getting
married that a young man’s living now depends.’ We must
therefore look at marriage and marriage stability in the matrix of
urban social relationships of which it is a part.
In this situation genetricial rights in a woman fade in favour of
uxorial rights. The rights by means of which kin groups are able
to recruit new members are not significant where these groups
cannot exist nor are the rights viable where the essential economic
and political bond between a Bemba man and his daughters’
husbands cannot exist. Instead where men are involved in wage-
earning all day and individual households, as against joint house¬
holds of the tribal type, are the pattern, the personal domestic
services of the women achieve new importance. The emphasis
on the personal services of the women is increased by what
Wilson calls ‘disequilibrium’ or ‘disproportion’ (1941, p. 40), i.e.
an unbalanced sex ratio. The adult sex ratio in the towns of
Northern Rhodesia does not depart from equality as much as it
does say in Southern Rhodesia or South Africa. On the Copper-
belt, for example, there are 142 adult men per 100 adult women.
Yet this is enough to give rise to a certain amount of competition
among the men for the sexual and domestic services of the women.
It is not surprising, therefore, that the question of the custody
of children or of a man’s rights in them are seldom raised in court.
Instead the emphasis is rather on adultery or from the woman’s
point of view, on her husband’s neglect of her. The uxorial rights
appear to be most significant in urban areas, and this accords with
the different forms that kinship in towns has taken there.

J. C. Mitchell 257
 As we may expect in towns the significance of the marriage
payment differs from that in tribal areas. The initiating gift
(nsalamu) and the ‘virginity’ (chisungu) payments are sub¬
stantially the same but the mpango payment understandably
appears to have lost its tribal function of fixing a man’s right to
the services of his daughters’ husbands. Instead the mpango pay¬
ment has become, as I have expressed it before as ‘a sort of
insurance policy against the possible dissolution of marriage’
(Mitchell, 1957, p. 25). An aggrieved husband is able to sue for
the return of his mpango payment if his uxorial rights have been
transgressed. Those kinsmen who accept a marriage payment on
the Copperbelt, therefore, do so in the full knowledge that they
may be called upon to refund the amount if a divorce takes place.
They will therefore become guardians of the husband’s uxorial
rights and will bring pressure to bear upon their kinswoman, the
wife, if she deviates from the norms of accepted behaviour. This
is the point made by a Lala man, quoted in an earlier paper
(Mitchell, 1957, p. 25), who paid six times as much in marriage
payment as he would have done in his rural home. He said his
wife would not be as ‘proud and cheeky’ as she would have been
had he only paid the usual amount. It is significant that three of
four kinsmen who accepted the marriage payments were in fact
living on the Copperbelt. They were themselves part of the social
system in which the norms operated and were able to take a
personal interest in the marriage.
We are, also, thus able to interpret the fact that even tribes
which traditionally make no marriage payments, such as the Bisa
and the Yao, fall into line on the Copperbelt and make substantial
mpango payments. In the urban situation uxorial rights are as
significant to them as other tribes and they endeavour to protect
these rights both with their wives’ kinsmen and in the courts by
making the formal payments (Mitchell, 1957, p. 27).

Marriage stability in towns

We are now able to turn to the question of marriage stability in
towns. Wilson was convinced that ‘the instability of marriage
is greater in town than in the country’ (1941, p. 41). This he
relates to the lack of traditional restraints, in town, the idleness
of the women making them ‘all the more ready for sexual adven-

258 Divorce and Marriage Stability

turns’, the disproportion of men and women and to the number of
inter-tribal marriages.
When Wilson wrote it was not generally realized that marriage
was unstable among the matrilineal and bilateral peoples of
Northern Rhodesia even in the rural areas. It seems likely that
this had been so even before the advent of the Europeans. Richards
had just published her figures (1940, p. 120) on Bemba divorce
in the rural areas but Wilson was not able to collect comparable
figures in Broken Hill. Since 1946 we have been able to make
considerable advances in collecting the quantitative material on
marriage stability and we have been able to clarify the methods of
using this data to best advantage (Barnes, 1949; Mitchell and
Barnes, 1950). The collection of data still lags a long way behind
the sophistication of analysis Barnes has proposed but we are
still in a better position to get at least a first estimate of the relative
frequency of divorce in tribal and in urban areas. I have published
some information (1957, p. 10) which suggests that urban marriages
may be more stable than tribal in Northern Rhodesia. The figures,
however, are not strictly comparable because of the possible effects
of differing age structures in the samples. More refined work is
necessary.
Nevertheless on the basis of our general analysis we might well
have predicted this. Our argument, following Gluckman, has been
that the main stabilizing influence in marriage is the legal transfer
of genetricial rights to the husband. We would expect, therefore,
that in the towns where corporate groups cannot operate as they
do in tribal areas marriage duration will be shorter and the divorce
rate will be higher. On the same reasoning since the legal transfer
of rights over the services of the daughters’ husbands is not
significant in town we would expect the probability of divorce
among urban Bemba couples to increase in the later years of
marriage as against tribal couples. The overall duration of
marriage may possibly be slightly shorter and the overall divorce
rate will then be larger.
We would expect the duration of marriage in town to be
longer, however, than among the strongly matrilineal people and
the bilateral. Here the very assaults on the stability of marriage
which Wilson emphasized have generated their own stabilizing
reaction in the form of substantial marriage payments which

J. C. Mitchell 259
protect the uxorial rights of the men. The result will be increasing
reluctance to face divorce and a consequent lowering of the
divorce rates.
These predictions are derivations from the hypothesis we have
built up on the basis of viewing marriage as an element in a total
social system in which certain clear-cut rights and obligations
are involved. The next step is to proceed to test these by the use
of refined quantitative techniques. I am able to do this for the
urban side on the basis of social survey material from the Copper-
belt but, except for material on the Ngoni, we are not yet in a
position to compare this with measures in the rural areas.

References
Barnes, J. A. (1949), ‘Measures of divorce frequency in simple societies’,
/. roy. anthropol. Inst., vol. 79, pp. 37-62.
Barnes, J. A. (1951), Marriage in a Changing Society, Rhodes-
Livingstone Paper, no. 20.
Bohannan, L. (1949), ‘Dahomean marriage: a revaluation’, Africa,
vol. 19, pp. 273-87.
Epstein, A. L. (1953), ‘The role of African courts in the urban
communities of the Northern Rhodesian Copperbelt’,
Rhodes-Livingstone J., no. 13, pp. 1-17.
Fortes, M. (1953), ‘The structure of unilineal descent groups’, Arner.
Anthropol., vol. 55, pp. 17-41.
Gluckman, M. (1950), ‘Kinship and marriage among the Lozi of
Northern Rhodesia and the Zulu of Natal’, in A. R. Radcliffe-Brown
and D. Forde (eds.), African Systems of Kinship and Marriage, Oxford
University Press.
Gough, E. K. (1950), ‘Changing kinship usage in the setting of political
and economic change among the Nayars of Malabar’, J. roy.
anthropol. Inst., vol. 82, pp. 71-87.
Mitchell, J. C. (1957), ‘Aspects of African marriage on the Copperbelt
of Northern Rhodesia’, Rhodes-Livingstone J., no. 22, pp. 1-30.
Mitchell, J. C., and Barnes, J. A. (1950), The Lamba Village: A
Report of a Social Survey, Commun. School African Stud., University
of Cape Town, no. 24.
Richards, A. I. (1940), Bemba Marriage and Present Economic
Conditions, Rhodes-Livingstone Paper, no. 4.
Richards, A. I. (1950), ‘Some types of family structure among the
Central Bantu’, in A. R. Radcliffe-Brown and D. Forde (eds.),
African Systems of Kinship and Marriage, Oxford University Press.
Richards, A. I. (1956), Chisungu: A Girls’ Initiation Ceremony among
the Bemba of Northern Rhodesia, Faber & Faber.
Wilson, G. (1941), An Essay on the Economics of Detribalization in
Northern Rhodesia, Rhodes-Livingstone Paper, no. 6.

260 Divorce and Marriage Stability

Part Nine Kin Groups

In most pre-industrial societies, kinship is important not merely

as a way of defining a range of relatives but as providing a
system of exclusive groups, or overlapping groupings, within
and between which many significant aspects of social life are
organized. These latter may include not only the transmission
of property but also arrangements for dealing with marriage,
defence, offence, religious action, and so on. Such groups may
be unilineal descent groups (clans and lineages), bilateral
kindreds of the kind reported for Anglo-Saxon England, or a
range of intermediary types found most commonly in
Polynesia. Unilineal descent groups may be recruited by
patrilineal or matrilineal descent; in some societies (double
descent systems), both types of descent group are found
side by side.
The distribution, classification and characteristics of these
groups, groupings and ranges of kin have been subject to much
discussion and controversy.
Fortes’ paper on the structure of unilineal descent groups
attempted to summarize developments in the study of clans and
lineages, groups that were widespread in Africa, where they
often played a particularly important role in societies that lacked
complex political institutions (i.e. ‘acephalous’ or ‘segmentary’
societies) (Reading 17).
Richards’ analysis of the ‘matrilineal puzzle’ in Africa, based
upon her own fieldwork among the Bemba, upon that of her
teacher, Malinowski, among the Trobriands, and upon a
comparative study of the Central Bantu, is a landmark in
clarifying some of the main features of a system which has been
so misunderstood by writers in many fields (Reading 18).
 The kinship systems of societies without unilineal descent
groups, western and non-western, have received much greater
emphasis in recent years especially in the work of Firth,
Freeman, Goodenough and Davenport. Pehrson’s article
(Reading 19) attempts to sum up some of the general
characteristics.

262 Kin Groups

17 M. Fortes

The Structure of Unilineal Descent Groups

Excerpts from M. Fortes, ‘The structure of unilineal descent groups’,

American Anthropologist, \o\. 55, 1953, pp. 17-41. Reprinted in M. Fortes,
Time and Social Structure and Other Essays, Athlone Press, 1970,
pp. 67-95.

What is the main methodological contribution of the study of

unilineal descent groups ? In my view it is the approach from the
angle of political organization to what are traditionally thought of
as kinship groups and institutions that has been specially fruitful.
By regarding lineages and statuses from the point of view of the
total social system and not from that of an hypothetical Ego we
realize that consanguinity and affinity, real or putative, are not
sufficient in themselves to bring about these structural arrange¬
ments. We see that descent is fundamentally a jural concept as
Radcliffe-Brown argued in one of his most important papers
(1935); we see its significance, as the connecting link between the
external, that is political and legal, aspect of what we have called
unilineal descent groups, and the internal or domestic aspect. It is
in the latter context that kinship carries maximum weight, first, as
the source of title to membership of the groups or to specific jural
status, with all that this means in rights over and towards persons
and property; and second as the basis of the social relations among
the persons who are identified with one another in the corporate
group. In theory, membership of a corporate legal or political
group need not stem from kinship, as Weber has made clear. In
primitive society, however, if it is not based on kinship it seems
generally to presume some formal procedure of incorporation with
ritual initiation. So-called secret societies in West Africa seem to
be corporate organizations of this nature. Why descent rather than
locality or some other principle forms the basis of these corporate
groups is a question that needs more study. It will be remembered
that Radcliffe-Brown (1935) related succession rules to the need
for unequivocal discrimination of rights in rem and in personam.
Perhaps it is most closely connected with the fact that rights over

M. Fortes 263
the reproductive powers of women are easily regulated by a
descent group system. But I believe that something deeper than
this is involved; for in a homogeneous society there is nothing
which could so precisely and incontrovertibly fix one’s place in
society as one’s parentage.
Looking at it from without, we ignore the internal structure of
the unilineal group. But African lineages are not monolithic units;
and knowledge of their internal differentiation has been much
advanced by the researches I have mentioned. The dynamic
character of lineage structure can be seen most easily in the balance
that is reached between its external relations and its internal
structure. Ideally, in most lineage-based societies the lineage tends
to be thought of as a perpetual unit, expanding like a balloon but
never growing new parts. In fact, of course, as Forde (1938) and
Evans-Pritchard (1940) have so clearly shown, fission and accre¬
tion are processes inherent in lineage structure. However, it is a
common experience to find an informant who refuses to admit that
his lineage or even his branch of a greater lineage did not at one
time exist. Myth and legend, believed, naturally, to be true history,
are quickly cited to prove the contrary. But investigation shows
that the stretch of time, or rather of duration, with which perpetuity
is equated varies according to the count of generations needed to
conceptualize the internal structure of the lineage and link it on to
an absolute, usually mythological origin for the whole social sys¬
tem in a first founder.
This is connected with the fact than an African lineage is never,
according to our present knowledge, internally undifferentiated.
It is always segmented and is in process of continuous further
segmentation at any given time. Among some peoples (e.g. the
Tallensi and probably the Ibo) the internal segmentation of a
lineage is quite rigorous and the process of further segmentation
has an almost mechanical precision. The general rule is that every
segment is, in form, a replica of every other segment and of the
whole lineage. But the segments are, as a rule, hierarchically
organized by fixed steps of greater and greater inclusiveness, each
step being defined by genealogical reference. It is perhaps hardly
necessary to mention that when we talk of lineage structure we are
really concerned, from a particular analytical angle, with the
organization of jural, economic and ritual activities. The point

264 Kin Groups

 here is that lineage segmentation corresponds to gradation in the
institutional norms and activities in which the total lineage organi¬
zation is actualized. So we find that the greater the time depth that
is attributed to the lineage system as a whole, the more elaborate is
its internal segmentation. As I have already mentioned, lineage
systems in Africa, when most elaborate, seem to have a maximal
time depth of around fourteen putative generations. More com¬
mon, though, is a count of five or six generations of named ances¬
tors between living adults and a quasi-mythological founder. We
can as yet only guess at the conditions that lie behind these limits of
genealogical depth in lineage structure. The facts themselves are
nevertheless of great comparative interest. These genealogies
obviously do not represent a true record of all the ancestors of a
group. To explain this by the limitations and fallibility of oral
tradition is merely to evade the problem. In structural terms the
answer seems to lie in the spread or span (Fortes, 1945) of internal
segmentation of the lineage, and this apparently has inherent
limits. As I interpret the evidence we have, these limits are set by
the condition of stability in the social structure which it is one of
the chief functions of lineage systems to maintain. The segmentary
spread found in a given lineage system is that which makes for
maximum stability; and in a stable social system it is kept at a
particular spread by continual internal adjustments which are
conceptualized by clipping, patching and telescoping genealogies
to fit. Just what the optimum spread of lineage segmentation in a
particular society tends to be depends presumably on extra-lineage
factors of political and economic organization of the kind referred
to by Forde (1947).
It is when we consider the lineage from within that kinship
becomes decisive. For lineage segmentation follows a model laid
down in the parental family. It is indeed generally thought of as the
perpetuation, through the rule of the jural unity of the descent
line and of the sibling group (cf. Radcliffe-Brown, 1950), of the
social relations that constitute the parental family. So we find a
lineage segment conceptualized as a sibling group in symmetrical
relationship with segments of a like order. It will be a paternal
sibling group where descent is patrilineal and a maternal one where
it is matrilineal. Progressive orders of inclusiveness are formulated
as a succession of generations; and the actual process of segmenta-

M. Fortes 265
tion is seen as the equivalent of the division between siblings in the
parental family. With this goes the use of kinship terminology and
the application of kinship norms in the regulation of intra-lineage
affairs.
As a corporate group, a lineage exhibits a structure of authority,
and it is obvious from what I have said why this is aligned with the
generation ladder. We find, as a general rule, that not only the
lineage but also every segment of it has a head, by succession or
election, who manages its affairs with the advice of his co-members.
He may not have legal sanction by means of which to enforce his
authority in internal affairs; but he holds his position by consent of
all his fellow members, and he is backed by moral sanctions com¬
monly couched in religious concepts. He is the trustee for the whole
group of the property and other productive resources vested in it.
He has a decisive jural role also in the disposal of rights over the
fertility of the women in the group. He is likely to be the representa¬
tive of the whole group in political and legal relations with other
groups, with political authorities and in communal ritual. The
effect may be to make him put the interests of his lineage above
those of the community if there is conflict with the latter. This is
quite clearly recognized by some peoples. Among the Ashanti for
instance, every chiefship is vested in a matrilineal lineage. But once
a chief has been installed his constitutional position is defined as
holding an office that belongs to the whole community not to any
one lineage. The man is, ideally, so merged in the office that he
virtually ceases to be a member of his lineage, which always has an
independent head for its corporate affairs (cf. Busia, 1951).
Thus lineage segmentation as a process in time links the lineage
with the parental family; for it is through the family that the lineage
(and therefore the society) is replenished by successive generations;
and it is on the basis of the ties and cleavages between husband and
wife, between polygynous wives, between siblings, and between
generations that growth and segmentation take place in the lineage.
Study of this process has added much to our understanding of well
known aspects of family and kinship structure.
I suppose that we all now take it for granted that filiation - by
contrast with descent - is universally bilateral. But we have also
been taught, perhaps most graphically by Malinowski, that this
does not imply equality of social weighting for the two sides of

266 Kin Groups

 kin connection. Correctly stated, the rule should read that filiation
is always complementary, unless the husband in a matrilineal
society (like the Nayar) or the wife in a patrilineal society, as per¬
haps in ancient Rome, is given no parental status or is legally
severed from his or her kin. The latter is the usual situation of a
slave spouse in Africa.
Complementary filiation appears to be the principal mechanism
by which segmentation in the lineage is brought about. This is
very clear in patrilineal descent groups, and has been found to hold
for societies as far apart as the Tallensi in West Africa and the
Gusii in East Africa. What is a single lineage in relation to a male
founder is divided into segments of a lower order by reference to
their respective female founders on the model of the division of a
polygynous family into separate matricentral ‘houses’. In matri¬
lineal lineage systems, however, the position is different. Segmen¬
tation does not follow the lines of different paternal origin, for
obvious reasons; it follows the lines of differentiation between
sisters. There is a connection between this and the weakness in
law and in sentiment of the marriage tie in matrilineal societies,
though it is usual for political and legal power to be vested in men
as Kroeber (1938) and others have remarked. More study of this
problem is needed.
Since the bilateral family is the focal element in the web of
kinship, complementary filiation provides the essential link between
a sibling group and the kin of the parent who does not determine
descent. So a sibling group is not merely differentiated within a
lineage but is further distinguished by reference to its kin ties
outside the corporate unit. This structural device allows of degrees
of individuation depending on the extent to which filiation on the
non-corporate side is elaborated. The Tiv, for example, recognize
five degrees of matrilateral filiation by which a sibling group is
linked with lineages other than its own. These and other ties of a
similar nature arising out of marriage exchanges result in a complex
scheme of individuation for distinguishing both sibling groups and
persons within a single lineage (L. Bohannan, 1951). This, of
course, is not unique and has long been recognized, as everyone
familiar with Australian kinship systems knows. Its more general
significance can be brought out however by an example. A Tiv
may claim to be living with a particular group of relatives for

M. Fortes 267
purely personal reasons of convenience or affection. Investigation
shows that he has in fact made a choice of where to live within a
strictly limited range of non-lineage kin. What purports to be a
voluntary act freely motivated in fact presupposes a structural
scheme of individuation. This is one of the instances which show
how it is possible and feasible to move from the structural frame
of reference to another, here that of the social psychologist, without
confusing data and aims.
Most far-reaching in its effects on lineage structure is the use of
the rule of complementary filiation to build double unilineal
systems and some striking instances of this are found in Africa.
One of the most developed systems of this type is that of the Yako;
and Forde’s excellent analysis of how this works (Forde, 1950)
shows that it is much more than a device for classifying kin. It is
a principle of social organization that enters into all social relations
and is expressed in all important institutions. There is the division
of property, for instance, into the kind that is tied to the patrilineal
lineage and the kind that passes to matrilineal kin. The division is
between fixed and, in theory, perpetual productive resources, in
this case farm land, with which go residence rights on the one hand,
and on the other movable and consumable property like livestock
and cash. There is a similar polarity in religious cult and in the
political office and authority linked with cult, the legally somewhat
weaker matrilineal line being ritually somewhat stronger than the
patrilineal line. This balance between ritual and secular control is
extended to the fertility of the women. An analogous double
descent system has been described for some Nuba Hill tribes by
Nadel (1950) and its occurrence among the Herero is now classical
in ethnology. The arrangement works the other way round, too, in
Africa, as among the Ashanti, though in their case the balance is
far more heavily weighted on the side of the matrilineal lineage than
on that of the jurally inferior and non-corporate paternal line.
These and other instances lead to the generalization that comple¬
mentary filiation is not merely a constant element in the pattern of
family relationships but comes into action at all levels of social
structure in African societies. It appears that there is a tendency
for interests, rights and loyalties to be divided on broadly comple¬
mentary lines, into those that have the sanction of law or other
public institutions for the enforcement of good conduct, and those

268 Kin Groups

 that rely on religion, morality, conscience and sentiment for due
observance. Where corporate descent groups exist the former
seem to be generally tied to the descent group, the latter to the
complementary line of filiation.
If we ask where this principle of social structure springs from
we must look to the tensions inherent in the structure of the
parental family. These tensions are the result of the direction given
to individual lives by the total social structure but they also provide
the models for the working of that structure. We now have plenty
of evidence to show how the tensions that seem normally to arise
between spouses, between successive generations and between
siblings find expression in custom and belief. In a homogeneous
society they are apt to be generalized over wide areas of the social
structure. They then evoke controls like the Nyakyusa separation
of successive generations of males in age villages that are built into
the total social structure by the device of handing over political
power to each successive generation as it reaches maturity
(Wilson, 1951). Or this problem may be dealt with on the level of
ritual and moral symbolism by separating parent and first bom
child of the same sex by taboos that eliminate open rivalry, as
among the Tallensi, the Nuer, the Hausa and other peoples.
Thus by viewing the descent group as a continuing process
through time we see how it binds the parental family, its growing
point, by a series of steps into the widest framework of social
structure. This enables us to visualize a social system as an inte¬
grated unity at a given time and over a stretch of time in relation
to the process of social reproduction and in a more rigorous way
than does a global concept of culture.
I do want to make clear, though, that we do not think of a
lineage as being just a collection of people held together by the
accident of birth. A descent group is an arrangement of persons
that serves the attainment of legitimate social and personal ends.
These include the gaining of a livelihood, the setting up of a
family and the preservation of health and well-being as among the
most important. I have several times remarked on the connection
generally found between lineage structure and the ownership of
the most valued productive property of the society, whether it be
land or cattle or even the monopoly of a craft like blacksmithing.
It is of great interest, for instance, to find Dr Richards attributing

M. Fortes 269
the absence of a lineage organization among the Bemba to their
lack of heritable right in land or livestock (Richards, 1950). A
similar connection is found between lineage organization and the
control over reproductive resources and relations as is evident
from the common occurrence of exogamy as a criterion of lineage
differentiation. And since citizenship is derived from lineage
membership and legal status depends on it, political and religious
office of necessity vests in lineages. We must expect to find and we
do find that the most important religious and magical concepts
and institutions of a lineage-based society are tied into the lineage
structure, serving both as the necessary symbolical representation
of the social system and as its regulating values. This is a compli¬
cated subject about which much more needs to be known. Cults of
gods and of ancestors, beliefs of a totemic nature and purely
magical customs and practices, some or all are associated with
lineage organization among the peoples previously quoted. What
appears to happen is that every significant structural differentiation
has its specific ritual symbolism, so that one can, as it were, read
off from the scheme of ritual differentiation the pattern of struc¬
tural differentiation and the configuration of norms of conduct that
go with it. There is, to put it simply, a segmentation of ritual
allegiance corresponding to the segmentation of genealogical
grouping. Locality, filiation, descent, individuation, are thus
symbolized.
Reference to locality reminds us of Kroeber’s careful argument
(1938) in favour of the priority of the local relationships of resi¬
dence over those of descent in determining the line that is legally
superior. A lineage cannot easily act as a corporate group if its
members can never get together for the conduct of their affairs.
It is not surprising therefore to find that the lineage in African
societies is generally locally anchored; but it is not necessarily
territorially compact or exclusive. A compact nucleus may be
enough to act as the local centre for a group that is widely dis¬
persed. I think it would be agreed that lineage and locality are
independently variable and how they interact depends on other
factors in the social structure. As I interpret the evidence, local
ties are of secondary significance, pace Kroeber, for local ties do
not appear to give rise to structural bonds in and of themselves.
There must be common political or kinship or economic or ritual

270 Kin Groups

interests for structural bonds to emerge. Again spatial dispersion
does not immediately put an end to lineage ties or to the ramifying
kin ties found in cognatic systems like that of the Lozi. For legal
status, property, office and cult act centripetally to hold dispersed
lineages together and to bind scattered kindred. This is important
in the dynamic pattern of lineage organization for it contains
within itself the springs of disintegration, at the corporate level in
the rule of segmentation, at the individual level in the rule of
complementary filiation.
As I have suggested before, it seems that corporate descent
groups can exist only in more or less homogeneous societies. Just
what we mean by a homogeneous society is still rather vague
though we all use the term lavishly. The working definition I make
use of is that a homogeneous society is ideally one in which any
person in the sense given to this term by Radcliffe-Brown in his
recent (1950) essay, can be substituted for any other person of the
same category without bringing about changes in the social
structure. This implies that any two persons of the same category
have the same body of customary usages and beliefs. I relate this
tentative definition to the rule of sibling equivalence, so that I
would say that, considered with respect to their achievable life
histories, in a homogeneous society all men are brothers and all
women sisters.
Societies based on unilineal descent groups are not the best in
which to see what the notion of social substitutability means. For
that it is better to consider societies in which descent still takes
primacy over all other criteria of association and classification of
persons in the regulation of social life but does not serve as the
constitutive principle of corporate group organization. Central
Africa provides some admirable instances (cf. Richards, 1950;
Gluckman, 1951). Among the Bemba, the Tonga, the Lozi and
many of their neighbours, as I have already remarked, the social
structure must be thought of as a system of interconnected
politico-legal statuses symbolized and sanctioned by ritual and not
as a collection of people organized in self-perpetuating descent
units. The stability of the society over time is preserved by per¬
petuating the status system. Thus when a person dies his status is
kept alive by being taken up by an heir; and this heir is selected on
the basis of descent rules. At any given time an individual may be

M. Fortes 271
the holder of a cluster of statuses; but these may be distributed
among several persons on his death in a manner analogous to the
widespread African custom by which a man’s inherited estate goes
to his lineage heir and his self-acquired property to his personal
heir. Ideally, therefore, the network of statuses remains stable and
perpetual though their holders come and go. Ritual symbols define
and sanction the key positions in the system. What it represents, in
fact, is the generalization throughout a whole society of the notion
of the corporation sole as tied to descent but not to a corporate
group. Descent and filiation have the function of selecting indi¬
viduals for social positions and roles - in other words, for the
exercise of particular rights and obligations - just as in cross¬
cousin marriage they serve to select ego’s spouse.
The concept of the ‘person’ as an assemblage of statuses has
been the starting point of some interesting enquiries. A general¬
ization of long standing is that a married person always has two
mutually antagonistic kinship statuses, that of spouse and parent
in one family context and that of child and sibling in another (cf.
Warner, 1958). This is very conspicuous in an exogamous lineage
system; and the tensions resulting from this condition, connected
as they are with the rule of complementary filiation, have wide
consequences. A common rule of social structure reflected in
avoidance customs is that these two statuses must not be con¬
founded. Furthermore, each status can be regarded as a compound
of separable rights and obligations. Thus a problem that has to be
solved in every matrilineal society is how to reconcile the rights
over a woman’s procreative powers (rights in genetricem as Laura
Bohannan (1949) has called them) which remain vested in her
brother or her lineage, with those over her domestic and sexual
services (rights in uxorem, cf. L. Bohannan, loc. cit.) which pass to
her husband. Among the Yao of Nyasaland, as Dr Clyde Mitchell
has shown (1956), this problem underlies the process of lineage
segmentation. Brothers struggle against one another (or sisters’
sons against mothers’ brothers) for the control of their sisters’
procreative powers and this leads to fission in the minimal lineage.
It is of great significance that such a split is commonly precipitated
by accusations of witchcraft against the brother from whose con¬
trol the sisters are withdrawn. By contrast, where rights over a
woman’s child-bearing powers are held by her husband’s patri-

272 Kin Groups

 lineal lineage the conflicts related to this critical interest occur
between the wives of a lineage segment; and among the Zulu and
Xhosa speaking tribes of South Africa these lead to witchcraft
accusations between co-wives (cf. Hunter, 1961). As Laura
Bohannan’s paper shows, many widespread customs and institu¬
tions connected with marriage and parenthood, such as the levirate
and the sororate, wife-taking by women, exchange marriage as
practised by the Tiv, and ghost marriage as found among the Nuer
(Evans-Pritchard, 1951) have structural significance not hitherto
appreciated if they are regarded from the point of view I have
indicated.
But one thing must be emphasized. This method of analysis
does not explain why in one society certain kinds of interpersonal
conflict are socially projected in witchcraft beliefs whereas in
another they may be projected in terms of a belief in punitive
spirits. It makes clear why a funeral ceremony is necessary and
why it is organized in a particular way in the interest of main¬
taining a stable and coherent social system. It does not explain
why the ritual performed in the funeral ceremonies of one people
uses materials, ideas and dramatizations of a different kind from
those used by another people. In short, it brings us nearer than we
were thirty years ago to understanding the machinery by which
norms are made effective, not only in a particular primitive society
but in a type of primitive society. It does not explain how the norms
come to be what they in fact are in a particular society.
In this connection, however, it is worth drawing attention to
certain norms that have long been recognized to have a critical
value in social organization. Marriage regulations, incest prohibi¬
tions and the laws of homicide and warfare are the most important.
Analysis of lineage structure has revealed an aspect of these norms
which is of great theoretical interest. It is now fairly evident that
these are not absolute rules of conduct which men are apt to break
through an outburst of unruly instinct or rebellious self-assertion,
as has commonly been thought. They are relatively obligatory in
accordance with the structural relations of the parties. The Beduin
of Cyrenaica regard homicide within the minimal agnatic lineage,
even under extreme provocation, as a grave sin, whereas slaying a
member of a different tribal segment is an admirable deed of
valour. The Tallensi consider sex relations with a near sister of the

T-K-K M. Fortes 273

same lineage as incest but tacitly ignore the act if the parties are
very distant lineage kin. Among the Tiv, the Nuer, the Gusii and
other tribes the lineage range within which the rule of exogamy
holds is variable and can be changed by a ceremony that makes
formally prohibited marriages legitimate and so brings marriage
prohibitions into line with changes in the segmentary structure of
the lineage. In this way previously exogamous units are split into
intermarrying units. In all the societies mentioned, and others as
well, an act of self-help that leads to negotiations if the parties
belong to closely related lineages might lead to war if they are
members of independent - though not necessarily geographically
far apart - lineages. Such observations are indications of the flexi¬
bility of primitive social structures. They give a clue to the way in
which internal adjustments are made from time to time in those
structures, either in response to changing pressures from without
or through the momentum of their own development. They sug¬
gest how such societies can remain stable in the long run without
being rigid. But this verges on speculation. [.. .]
What I wish to convey by the example of current studies of uni¬
lineal descent group structure is that we have, in my belief, got to
a point where a number of connected generalizations of wide
validity can be made about this type of social group. This is an
advance I associate with the structural frame of reference. I wish
to suggest that this frame of reference gives us procedures of
investigation and analysis by which a social system can be appre¬
hended as a unity made of parts and processes that are linked to
one another by a limited number of principles of wide validity in
homogeneous and relatively stable societies. It has enabled us to
set up hypotheses about the nature of these principles that have
the merit of being related directly to the ethnographic material
now so abundantly at hand and of being susceptible of testing by
further field observation. It cannot be denied, I think, that we have
here a number of positive contributions of real importance to
social science.

References
Bohannan, L. (1949), ‘Dahomean marriage: a revaluation’, Africa,
vol. 19, pp. 273-87.
Bohannan, L. (1951), A Comparative Study of Social Differentiation in
Primitive Society, D. Phil, thesis, University of Oxford.

274 Kin Groups

Busia, K. A. (1951), The Position of the Chief in the Modern Political
System of Ashanti, Oxford University Press.
Evans-Pritchard, E. E. (1940), The Nuer, Clarendon Press.
Evans-Pritchard,(E. E. (1951), Kinship and Marriage among the Nuer,
Clarendon Press.
Forde, C. D. (1938), ‘Fission and accretion in the patrilineal clans of
a semi-Bantu community’, J. roy. anthrop. Inst., vol. 68, pp. 311-38.
Forde, C. D. (1947), ‘The anthropological approach in social science’.
The Advancement of Science, vol. 4, pp. 213—24.
Forde, C. D. (1950), ‘Double descent among the Yako’, in A. R.
Radcliffe-Brown and C. D. Forde (eds.), African Systems of Kinship and
Marriage, Oxford University Press.
Fortes, M. (1945), The Dynamics of Clanship among the Tallensi,
Oxford University Press.
Gluckman, M. (1951), ‘The Lozi of Barotseland in North Western
Rhodesia’, in E. Colson and M. Gluckman (eds.), Seven Tribes of
British Central Africa, Oxford University Press.
EIunter, M. (1961), Reaction to Conquest, Oxford University Press.
Kroeber, A. L. (1938), ‘Basic and secondary patterns of social structure
/. roy. anthrop. Inst., vol. 68, pp. 299-309.
Mayer, P. P. (1949), ‘The lineage principle in Gusii society’, Interna¬
tional African Institute, Memorandum 24.
Mitchell, J. C. (1956), The Yao Village, Manchester University Press.
Nadel, S. F. (1950), ‘Dual descent in the Nuba Hills’, in A. R. Radcliffe-
Brown and C, D. Forde (eds.), African Systems of Kinship and Marriage,
Oxiord University Press.
Peters, E. L. (1951), The Sociology of the Beduin of Cyrenaica, D. Phil,
thesis, University of Oxford.
Radcliffe-Brown, A. R. (1935), ‘Patrilineal and matrilineal
succession’, Iowa Law Rev., vol. 20, pp. 286—303.
Radcliffe-Brown, A. R. (1950), ‘Introduction’, African Systems of
Kinship and Marriage, Oxford University Press.
Richards, A. I. (1950), ‘Some types of family structure among the
Central Bantu’, in A. R. Radcliffe-Brown and C. D. Forde (eds.), African
Systems of Kinship and Marriage, Oxford University Press.
Warner, W. L. (1958), A Black Civilization, Harper & Row.
Wilson, M. (1951), ‘Nyakyusa age villages’, J. roy. anthrop. Inst., vol.
79, pp. 21-5.

M. Fortes 275
18 A. I. Richards
Matrilineal Systems

Excerpts from A. I. Richards, ‘Some types of family structure amongst

the Central Bantu’, in A. R. Radcliffe-Brown and D. Forde (eds.),
African Systems of Kinship and Marriage, Oxford University Press
for the International African Institute, 1950, pp. 207-51.

Characteristics of matrilineal kinship organizations in

Central Africa
Most of the Bantu peoples of Central Africa reckon descent in the
matrilineal rather than the patrilineal line, and many of them
practise some form of what is usually known as matrilocal mar¬
riage. In fact, it is the matrilineal character of their kinship
organization which distinguishes them so clearly from the Bantu
of East and South Africa, and for this reason the territory stretch¬
ing from the west and central districts of the Belgian Congo to the
north-eastern plateau of Northern Rhodesia and the highlands of
Nyasaland is sometimes referred to as the ‘matrilineal belt’.1
Within this group of ‘matrilineal’ tribes there is a remarkable
degree of uniformity as to the principles governing descent and
succession and the various ideologies by which people explain
their adherence to the mother’s rather than to the father’s line,
and stress their community of interests with their maternal
relatives.2 Blood is believed to be passed through the woman and
not through the man. The metaphors of kinship stress the ties
between people ‘born from the same womb’ or ‘suckled at the
1. There are, of course, a number of patrilineal or mainly patrilineal
tribes in the Belgian Congo, including most of the Luba (as distinguished
from the Luba-Hemba), the Songe and the Nkundo.
2. Matrilineal is used here in inverted commas because it is generally
recognized that no society is entirely matrilineal or patrilineal as regards
descent, inheritance, succession and authority, but that the family system
provides a balance of interests and rights between the two sides of the family
with a predominant emphasis on one side or the other, and it is in this
sense that I shall use the term in this article. See Richards (1934) for a
description of Bemba kinship from this point of view.

276 Kin Groups

same breast’, and in some tribes the physical role of the father is
believed to be limited to the quickening of the foetus already
formed in the uterus. The duty of the woman to produce children
for her lineage is emphasized, and descent is traced from an original
ancestress or a series of ancestresses known as ‘mothers’ of the
lineage or clan, and also in some cases from the brothers of these
founding ancestresses. The ancestral cult centres round the wor¬
ship of matrilineal rather than patrilineal ancestors, although
spirits of the father’s line are sometimes the subject of subsidiary
rites.
A child belongs to his mother’s clan or lineage, and succession to
office follows the common matrilineal rule, that is to say, authority
passes to the dead man’s brothers or to his sisters’ sons, or to the
sons of his maternal nieces. Among some of the Central Bantu
women succeed to the titles of royal ancestresses, or hold positions
as chieftainesses, with special ritual functions.
But with these principles of descent and succession the similarity
ends. The Central African people differ in a rather striking way as
to their family structure, and in particular as to the various forms
of domestic and local grouping based on the family.
These variations in the family structure depend largely on the
nature of the marriage contract and the extent to which the hus¬
band is able to gain control over his wife, who belongs, by virtue of
matrilineal descent, to the lineage and clan of her mother and of
her brothers and sisters; and also on the extent to which he manages
to achieve a position of authority over the children she bears. In
matrilineal societies the man’s control over his wife and her children
can never be complete, except in the case of a union with a slave
woman, but he can gain considerable power over his wife’s labour,
her property and her child-bearing powers, as well as rights over
his children’s work and their marriages, by virtue of the service or
payments he makes to his father- or brother-in-law. Moreover, the
ways in which domestic authority is divided between a man and
the head of his wife’s kinship group are surprisingly varied. In
some cases there is a formal allocation of rights and privileges
between father and mother’s brother in return for service and pay¬
ments. In other cases the balance is less well defined, and every
marriage produces what can only be described as a constant pull-
father-pull-mother’s-brother, in which the personality, wealth

A. I. Richards 277
and social status of the two individuals or their respective kins¬
men give the advantage to one side or the other, and a number of
alternative solutions are reached within the same tribe.
In this balance of privileges and duties between the patrikin and
the matrikin the crucial point is obviously the husband’s right to
determine the residence of the bride. If she and her children live in
the same homestead or village as his kinsmen, his domestic
authority is likely to be greater than where they remain with the
wife’s relatives.
Throughout this area, at any rate, the rule of residence at
marriage seems to me to be the most important index of the
husband’s status. It also provides the most convenient basis for the
classification of these different forms of matrilineal family. In
Central Africa we find every gradation from the marriage in
which the husband has the right to remove his bride immediately
to his own village, to varieties of trial marriage, temporary unions
or customs by which the husband takes up more or less permanent
residence in the wife’s group.
For this reason I have found the old terms ‘matrilocal’ and
‘patrilocal’ of little use for comparative purposes. Many writers
have pointed out that the words are in themselves confusing, since
there are two parties to a marriage and what is ‘residence with the
mother’ for the one is ‘residence with the mother-in-law’ for the
other. If the terms are to be retained for purposes of classification
it would be necessary to adopt a convention by which, for instance,
they were always used with relation to one sex. Firth (1936, p. 596)
and Adam (1948, p. 12) suggest as an alternative the use of the
words ‘virilocal’ and ‘uxorilocal’ to meet this difficulty,3 and
I shall use these terms from time to time.
Another difficulty with regard to the Central African area is the
variety of forms of marriage relationship which could reasonably
be included under the title ‘matrilocal’. In societies practising
matrilineal descent a man may live with his wife’s people because

3. N. W. Thomas, who suggested the terms patrilocal and matrilocal,

put them forward ‘not as being specifically appropriate but as being
parallel to patrilineal and matrilineal, denoting descent in the female or
male line respectively’ (1906, p. 108), thus showing that he realized from the
start the ambiguity of the terms.

278 Kin Groups

 the marriage is on trial; because he is fulfilling a marriage contract
by service instead of the payment of goods; or because he means to
settle permanently in his wife’s group.4 He may have sex access to
his wife at night and work in her fields, but act as head of his
sister’s house and spend a large part of his day with the latter. This
is the practice of the matrilineal Menangkabau of the Padang
highlands of Sumatra and the kindred peoples of Negri Sembilan
(Verkerk-Pistorius, 1871; Cole, 1945).5 A similar position holds
good among the Hopi of Arizona. Alternatively a husband may
live at his wife’s village but often be away visiting his own relatives,
as amongst the Yao; or he may spend alternate years in his own
and his wife’s village, as in Dobu Island. In any of these cases there
will be some years, some months, or some hours in which the
marriage may reasonably be called matrilocal. The term gives no
indication of the length of time a man spends in his wife’s village,
or the degree to which he is incorporated with her matrikin, or
isolated from his own family.6 For this reason I have found it
better to use the phrases ‘marriage with immediate right of bride
removal’ or ‘marriage with delayed right of bride removal’ or
‘marriage without bride removal’ in describing the family systems
of the Central Bantu.
The terms ‘matrilocal’ or ‘patrilocal’ are similarly lacking in
precision when used as a means of classifying types of family and
domestic unit. According to the terminology adopted in this book
[i.e. African Systems of Kinship and Marriage], a parental family is
a household or homestead composed of a man, his wife or wives,
and their children. Such a unit naturally grows in size in societies
where it is customary for either the sons or the daughters to live
with their parents after marriage. This wider group, usually based
on the principle of unilineal descent and composed of members

4. Rivers points out (1908), that marriage by service ‘passes insensibly

into the matrilocal form of marriage’ and trial marriages ‘shade insensibly
into trials before marriage on the one hand and into ease and frequency of
divorce on the other’.
5. Quoted also by Taylor (1896, p. 96) where he describes the ‘Chassez
Croisez’ which takes place at dusk when each man leaves his sister’s house
where he has been by day, to join his wife and children at night.
6. R. Linton stresses the importance of the propinquity of the two
settlements in determining the character of the descent system (1936, p.
168).

A. I. Richards 279
of three generations, is here described as an extended family.7
Members of extended families of this kind usually live in a separate
kraal or homestead or they may form the nucleus or core of a
village or a section of a village. They tend to cooperate in economic
affairs, to exercise some common rights over property, to accept
their genealogical senior as a common authority and to practise
some joint ritual. Where it is a common practice for a man who
has married sons to separate off from his father’s homestead to
start a new community, and where the land situation makes this
possible, a three-generation extended family of this kind becomes
the normal pattern of residence as it is, for instance, amongst the
Zulu. Where, however, the villages are more permanent and the
splitting off of new extended families is not so easy, the residential
pattern is naturally much less uniform. The eldest brother of a
group of siblings may succeed to the position of his father in a
patrilineal society or to that of his mother’s brother in a matrilineal
society, and various other changes of this kind may take place.
Thus in the more or less permanent villages there may be two or
three extended families or remains of extended families. There may
also be two or three alternative types of marriage in the same
community, and a number of principles of residence. Such larger
residential units, usually based on a nuclear extended family but
with a number of additions, I have called local kin groups, speci¬
fying whether they are predominantly matrilineal or patrilineal in
composition.
In such extended families or local kin groups extension takes
place on the basis of certain nuclear or pivotal relationships. For
instance, according to the rules of residence at marriage, the
children of one sex marry out of the homestead, whereas those of
the other remain within it. Rules of succession and inheritance
similarly determine the incidence of authority and economic
privilege within the local community. Property rights and author¬
ity may go from father to son, from mother to daughter, from
brother to sister’s son, and so forth. I find it convenient to classify
the different types of extended family by means of these nuclear

7.1 have found it useful to employ the term ‘grand-family’ suggested to

me by R. Firth to indicate a three-generation family descended in the
direct line, patrilineal or matrilineal, when it is necessary to distinguish
between this and other forms of extended family.

280 Kin Groups

 I

relationships, even though the categories constantly overlap. Thus

I use the term ‘father-son extended family ’ for a Zulu homestead;
or the ‘mother-son extended family’ for a local unit which is a
common pattern amongst the Swazi, where the woman leaves her
husband’s kraal at the time of her son’s marriage and lives with
him in a position of authority to the end of her life. Parent-daughter
extended families of one kind or another exist where uxorilocal
marriage is practised. These include the mother-daughter family
of the Hopi, where property in land or houses is passed through the
woman; the sororal family, where a group of married sisters and
their daughters live together, usually under the care of an elder
brother; or the sibling family found amongst the Nayar of Mala¬
bar, where a man lived with his sisters and the latter were visited
by their husbands at night; or a father-daughter family, where
married daughters live with their parents, but their father is very
much the head of the group and may determine residence instead
of the mother. Other types of extended family include the matri-
lineal fraternal family, where a group of brothers live with their
wives and their sisters’ sons.
It is obvious that in the case of residential units such as these the
nuclear relationships vary from one type of extended family to the
other; that is to say, one homestead is based on the close relation¬
ship of a group of the men of the family, as in the case of the patri¬
lineal father-son family or the matrilineal fraternal family; while
another is founded on the kinship of women, as in the mother-
daughter family of the Hopi. The interests of brother and brother,
brother and sister, or mother and daughter may be identified by
one marriage system or the other, while the extended families so
formed may attract to themselves additional households of kins¬
folk or slaves, according to fixed rule or more casual association.

Conclusions
Among the Central Bantu the matrilineal system makes for certain
elements of conflict for which some kind of solution has to be
found. The problem in all such matrilineal societies is similar. It is
the difficulty of combining recognition of descent through the
woman with the rule of exogamous marriage. Descent is reckoned
through the mother, but by the rule of exogamy a woman who has
to produce children for her matrikin must marry a man from

A. 1 Richards 281
another group. If she leaves her own group to join that of her
husband, her matrikin have to contrive in some way or other to
keep control of the children, who are legally identified with them.
The brothers must divide authority with the husband who is living
elsewhere. If, on the other hand, the woman remains with her
parents and her husband joins her there, she and her children
remain under the control of her family, but her brothers are lost to
the group since they marry brides elsewhere and they are separated
from the village where they have rights of succession.
There is the further difficulty that in most societies authority
over a household, or a group of households, is usually in the hands
of men, not women, as are also the most important political offices.
Thus any form of uxorilocal marriage means that an individual of
the dominant sex is, initially at any rate, in a position of subjection
in his spouse’s village, and this is a situation which he tends to find
irksome and tries to escape from.
There are, of course, a number of solutions to the matrilineal
puzzle. The first of these may be described as the matriarchal
solution, in that property, and particularly houses and lands, pass
through the woman as well as the line of descent. The eldest brother
usually acts as manager of the estate. This is achieved either by the
institution of the visiting husband or by that of the visiting brother.
The women of the taravad or matrilineal joint family of the Nayar
of Malabar live with their brother or brothers and are visited by
their husbands at night. In the case of the Menangkabau the men
were members of their sisters’ joint household {rumah) and spent
much of the day there, but they returned to their wives’ households
at night (Cole, 1945, pp. 253,266). Among the Hopi the situation is
slightly different, in that a group of married women live together
and own land and houses jointly, but each husband acts as manager
of his wife’s land and chief worker on it, while the brother acts as
spokesman and ritual officiant for his sisters and as host in their
houses.8
Such solutions only seem to me to be possible in big settlements
with permanent housing, where a man can walk easily from his
own household to that of his sisters and perform his two functions
without clash. This is the case amongst the Nayar and something
similar occurs in the large towns of Ashanti Province in the Gold
8. Verbal communication from Mrs Aitken; see also Forde (1931).

282 Kin Groups

 Coast where a woman visits her husband at night. But Central
Africa is much less closely settled. In Northern Rhodesia shifting
cultivation is practised, and villages not only move, but are more
sparsely distributed, often at a distance of seventeen to thirty miles
apart. In these circumstances either the wife or the husband may
have to live some distance away from the group to which they are
legally affiliated. The more densely occupied Yao area allows
husbands to be away visiting their own relatives more easily and it
appears that they often do so.9
The second solution is that of the fraternal extended family with
sisters and children ‘loaned away’. This is the solution adopted by
the Trobriand islanders and the western Congo people. Here viri-
local marriage makes it possible for a group of brothers to live
together and to exercise full male authority over the community,
while their sisters are loaned out to men in other communities and
the children of the matrilineage are reclaimed at puberty.
The third solution might be said to be that of the borrowed
husband. These are the various forms of uxorilocal marriage
described in this area. The conflict of interests in these societies is
probably the most extreme, since all the men of a community
cannot at the same time act as mother’s brothers with authority
over their own local descent group and also as husbands living in
their own wives’ villages, as the rule of uxorilocal marriage de¬
mands. Amongst the Cewa and Yao it seems that the majority of
marriages are uxorilocal but that marriages are easily broken. A
man who cannot stand the situation in his wife’s village leaves and
goes elsewhere. This might in fact be described as the solution
of the detachable husband. Added to this certain senior men have a
right from the start to practise virilocal marriage since they are
selected as heads of villages or inherit such positions. They thus
escape from the difficult position in which their younger brothers
are placed. In other words, there is here a stress on primogeniture
with virilocal marriage as one of the privileges of the first born.
The Bemba deal with this difficulty by the transfer or removal

9. Daryll Forde suggests to me that the necessary factor is not the size of
the settlement but its compact and endogamous character. A man can act
simultaneously as husband and manager-brother where he has married
within the same community. Both the size and the endogamous character
of the local unit seem to me to be relevant factors.

A. I. Richards 283
marriage, by which the young man is dependent on his father-in-
law in youth but gains authority as the head of a father-daughter
grand-family in his old age; this is a solution which is possible in
areas where land is plentiful and the setting up of new homesteads
and villages is easy, and where the political system allows of the
constant creation of new units.
Still another way out of the difficulty is the solution of the
selected heir, which allows the head of the matrilineage to choose
one or more of his sisters’ children to succeed him and to transfer
these boys to his own village while the remaining children are
allowed to stop with their father. This seems to be the practice
among the Mayombe, according to Van Reeth. Kirchoff quotes
the similar custom of the Tlingit by which one or two sons of a
woman marry their cross-cousins and go to live with their mother’s
brother, whereas the rest of the children marry virilocally. The
Nuba often adopt one or two sisters’ sons to be brought up as their
own sons in the same way (Kirchoff, 1932; Nadel, 1948, p. 31).10
In every case there are constitutionally recognized alleviations
for the socially childless father in these matrilineal societies.
Slavery allows him to gain control over the children of slave wives
who thus join his clan, and Torday (1925, p. 103) speaks of Kongo
fathers choosing slave wives for preference for this reason. Cross¬
cousin marriages may give the father control over some of his
grandchildren that he cannot maintain over his children, or bring
into his village members of his clan. By the mother’s brother’s
daughter marriage of the Yao the son of one of the married sisters
belonging to the sororal family marries the daughter of the
manager-brother and therefore contracts what is virtually a viri-
local marriage.
Polygyny and the marriage of widows also allow men to contract
at least some virilocal marriages in most of the Central African
tribes, and, whatever the rule of descent, a man who has succeeded
in becoming the head of a polygynous household is in fact a
patriarch over his wives and young children. Lastly men of wealth
and distinction are able to reverse the usual rules of residence.

10. This list does not of course exhaust the logical possibilities. The
solution of alternating residence among the Dobu has been mentioned
already. There is also the separate men’s and women’s houses of the Ga
of the Gold Coast.

284 Kin Groups

 Authority over the chiidren of a marriage consists of a series of
rights and privileges, and many kinship systems allow for the
exercise of different types of authority by different members of the
patrikin or matrikin. In Central Africa there may be a fairly com¬
plete severance between domestic authority exerted over the
children brought up in their father’s homestead and rights over
their persons and the marriages they contract. These latter may be
exerted by a man who lives some distance away. Domestic author¬
ity is a treasured asset in an area where local communities grow up
round the core of the extended family and the head of such a
family is actually or potentially the head of a village, and hence of
the unit that is the basis of the whole political structure. On the
other hand, this is an area in which slavery was formerly a very
prominent institution, and, therefore, whatever hold the father
may have gained over his children, the power of the mother’s
brother to sell them into slavery must have remained as a potential
threat to be constantly feared. Even amongst the Ila, where the
father is a patriarch ruling over a large polygynous homestead
very similar to that found amongst the patriarchal Bantu, Smith
and Dale (1920) describe the woman’s brother as having ‘power of
life and death’ over her children. The right to sell into slavery is
apparently the power described.
A balance of rights and duties between the patrikin and the
matrikin tends to produce secondary forms of descent. While it
would be true to say that in every matrilineal society a man must
know his father’s clan and get some privileges from his patrikin,
there are tribes in which both patrilineal and matrilineal relatives
are organized in definite descent groups with names and corporate
functions as amongst the Yako. In Central Africa the existence of
such double descent groups does not seem to be proved, but there
is a secondary reckoning of descent in those tribes in which the
patrilineal-matrilineal balance is most even. Amongst the Kongo
the father’s matrikin are known by a separate name and a man has
sentimental ties with the men of his father’s village in which he
was brought up, and these informal ties seem to have led to the
suggestion that the Kongo practise dual descent.
Ancestor worship also reflects the duality of the reckoning of
descent. The Bemba commoner can inherit a guardian spirit from
his father’s side as well as his mother’s, and will pray to his maternal

A. I. Richards 285
grandfather as well as to his maternal uncle, although this would
not be the case in the royal clan. The Ila pray to the ancestors both
of the father and of the mother, and Kongo ceremonies honouring
the father’s ancestors have also been mentioned.
In all these forms of family structure the crucial point, I have
suggested, is the question of residence at marriage, and the deter¬
mining factor in this regard is the marriage payment or the type of
goods and services which the bridegroom gives. In Central Africa
the tribes which give large amounts of goods or money in bride¬
wealth, as do the Mayombe and the Ila, seem invariably to practise
virilocal marriage; whereas those who give service or token pay¬
ments, like the Bemba or Bisa, do not. The importance of the
marriage payment is shown by the fact that even amongst the
western Congo peoples, where the avunculate is most pronounced,
a man who gives an unusually large sum to his bride’s family is
able to gain permanent possession of his wife and to keep her
children with him instead of returning them to their maternal
uncle at puberty. The passage of cattle as chiko amongst the
matrilineal Ila is an equally important determinant. The chiko
gives the father permanent possession of his children and enables
him to keep his married sons with him to form the basic residential
unit, the mukwashi. The division of cattle among his sons and his
daughters sets up what seem to be a number of three or four-
generation patrilineages as well as similar groups united through
rights of inheritance of cattle from their mother. In Northern
Rhodesia there are no high marriage payments and a man cannot
remove his wife and children to his village immediately. He may
earn the permission to do so through service or joint residence, but
he will never acquire complete possession of his wife and children.
As regards the stability of marriage in these different types of
family, we have hints and impressions of various kinds but very
little accurate quantitative data. The payment of bridewealth is
assumed to encourage a stable marriage since the wife’s family is
pledged to return cattle or goods received in this way if she breaks
her contract. If this assumption is correct, divorce should be rare
in the western Congo, where the marriage payments are consider¬
able, but I have no figures to show whether this actually is the case
or not. The divorce rate should also be low amongst the Ila, but as
a matter of fact Smith and Dale say that it is frequent. It seems clear

286 Kin Groups

 to me that the size of the bridewealth and the way in which it is
contributed are determining factors but by no means the only ones.
Where the avunculate is strong the mother’s brother usually has
the power to take his sister’s daughter away from the husband to
whom she has been ‘loaned ’ and give her to another man, and this
must be reckoned as at least one of the causes of instability in
marriage. It occurs amongst the Mayombe, the Ua and the Bemba,
and probably elsewhere. Moreover, where the residential units are
based on ties of kinship through the woman rather than the man,
the link between mother and daughter, or sister and sister, may
become so strong that it can threaten the marriage of one of these
women. Divorces sometimes occur in Bemba society at the time of
the conversion of a marriage, that is to say when the husband has
won the right to take his wife back to his own village. The woman
may refuse to go with him because she wants to stay with her
mother and sisters.
There are also positive inducements to permanent marriage.
Bemba give token presents not bridewealth, but a man stands to
gain by a stable union because only by this means is he able to
build up the basic domestic unit which puts him in a position of
authority over several households and enables him to set up house
where he wishes and ultimately to build his own village if he
pleases. A stable marriage for a Bemba is the first step on the road
to a headmanship. It is interesting to correlate different forms of
family structure with greater or less stability in marriage in this
way, but conclusive answers will have to wait until we have even
some rough approximation to the differential divorce rates in these
and contiguous areas.
Various economic and social determinants of family structure
have been suggested in these pages, though fuller details of the
economic organization of all the selected tribes would be required
to test the hypotheses made. The presence or absence of inheritable
possessions in the form of land, trees, cattle, or money have been
correlated with the type of marriage, and the position of the father
as against that of the mother’s brother. It has also been associated
with the corporate nature of a unit like the western Congo matri-
lineage, as against the dispersed descent groups of the peoples of
Northern Rhodesia. It would no doubt be possible to start from
the same set of facts and reach very different conclusions. Is it, for

A. I. Richards 287
instance, the shortage of land and the value of the permanent palm-
trees that make members of a matrilineage willing to live together
in one closely organized group ? Or is it the system of corporate
matrilineages which makes them anxious to develop their land
together instead of doing so alone, and encourages them to accept
the rule of their genealogical head ? No doubt there is something to
be said from both points of view. Where the type of marriage makes
it possible for a group of brothers to live together, an economic
situation such as that of the western Congo increases their
economic interdependence and counteracts centrifugal tendencies.
Where, on the other hand, the family structure provides no basis
for a group of cooperating men, and where the land supply is so
ample that it is possible for men to set up new settlements easily,
they do so and the matrilineage loses its corporate function. De¬
tailed studies of contiguous people living in rather similar economic
conditions are needed before we can generalize usefully on such
questions. At present the variables are too numerous to permit
more than likely guesses.
This article merely attempts to make some crude comparisons
between four types of family structure associated with the rule of
matrilineal descent in Central Africa, and to suggest some possible
correlations between marriage type, residential grouping, economic
and political organization.

References
Adam, L. (1948), ‘“Virilocal” and “ uxorilocal”’, Man, vol. 48, art. 13.
Cole, F. C. (1945), The Peoples of Malaysia, Van Nostrand.
Firth, R. (1936), We the Tikopia, Allen & Unwin.
Forde, D. (1931), ‘Hopi agriculture and land ownership’, J. roy.
anthropol. Inst., vol. 61, pp. 357-99.
Kirchhoff, P. (1932), ‘Kinship organisation’, Africa, vol. 2,
pp. 184-91.
Linton, R. (1936), The Study of Man, Appleton-Century-Crofts.
Nadel, S. F. (1948), The Nuba, Oxford University Press.
Richards, A. I. (1934), ‘Mother right in central Africa’, in E. E. Evans-
Pritchard et al., Essays Presented to C. G. Seligman, Routledge & Kegan
Paul.
Rivers, W. H. R. (1908), ‘Marriage’, in J. Hastings (ed.). Encyclopedia of
Religion and Ethics, T. & T. Clark.
Smith, E. W, and Dale, A.M. (1920), The Ila-Speaking Peoples of
Northern Rhodesia, Macmillan.

288 Kin Groups

Taylor, A. (1896), ‘Matriarchal family system’, The Nineteenth Century,
vol. 40, pp. 81-96.
Thomas, N. W. (1906), Kinship Organization and Group Marriage in
Australia, Cambridge University Press.
Torday, E. (1925), Causeries Congolaises, Vromant, Brussels.
Vererk-Pistorius, A. (1871), Studien over de inlandische
Houshouding in de Padangische Bovenlander, Holland.

A. I. Richards 289
19 R. N. Pehrson
Bilateral Kin Groupings1

R. N. Pehrson, ‘Bilateral kin groupings as a structural type: a preliminary

statement’, University of Manila Journal of East Asiatic Studies, vol. 3,
1964, pp. 199-202.

A good deal of anthropological theory concerning social organiza¬

tion is derived from investigations of societies based upon uni¬
lineal descent groups. The lineage looms large in our picture of
non-Western societies. It is frequently assumed that societies
based upon the bilateral principle of descent are ‘amorphous’,
‘unstructured’, ‘loosely organized’ or ‘infinitely complex’. I wish
to examine the local group organization and marriage residence
patterns of one bilaterally organized society in order to suggest
that there is structure to this type of society but that the structure
may be obscured by approaching it from a unilineal viewpoint.
My model of bilateral organization derives from research
among the North Lapp reindeer nomads who migrate near
Karesuando in Sweden, Enontekio in Finland and Kautokeino in
Norway. The fundamental bilateralism of Lapp society first
becomes apparent in the reindeer nomads’ inheritance rules and
kinship nomenclature.
Among the Lapps, property is owned individually. Wealth is
determined by the number of reindeer possessed and these are
inherited and transmitted equally through the male and female
lines. The reindeer are herded collectively by the migratory local
group on pasturelands allocated according to customary rights of
usage.
Neither is there any unilineal emphasis in the kinship termin¬
ology as shown by the complete bilaterality and symmetry of Lapp
consanguineal kinship terms and by the rigid differentiation into
l.This paper was originally presented to the annual meeting of the
Central States Anthropological Society at Urbana, Illinois, in May 1953.
I am grateful for suggestions on the paper made by Professor E. Adamson
Hoebel, University of Utah, and Mr McKim Marriott, University of
Chicago.

290 Kin Groups

generations. A third significant feature is the equivalence of sib¬
lings, terminologically expressed by classifying cousins with sib¬
lings (in Ego’s generation, the term of cousin is derived from the
sibling term). This equivalence of siblings also occurs in the affinal
terminology; Ego classes together his or her spouse’s sister and
female cousin and brother and male cousin and brother’s wife and
male cousin’s wife.
These terminological features are paralleled in local group
organization which, among the Lapps, takes the form of migratory
units or bands. Within the band, each generation has its own sphere
of activities, rights and obligations. Band members of the same
generational level are unified by the classification of cousins with
siblings. The principle of the equivalence of siblings also acts to
increase the number of band members available for economic
activities. In examining kinship behavioural patterns, I came to the
conclusion that sibling solidarity is the fundamental kinship bond
of Lapp society.
A cursory glance at Lapp band genealogies does not, at first,
reveal any clear organizational pattern. The reason for this, I
believe, is that a Lapp migratory unit is not a corporate body as
such is generally conceived in social anthropology. That is, it
lacks such corporate attributes of unilineal kin groupings as
perpetuity through time, collective ownership of property and
unified activity as a legal individual. Instead of thinking of the Lapp
band as a corporate body, we must think of it as a series of alliances
between sibling groups. Since these alliances have varying degrees
of permanence, a Lapp may be a member of several bands during
his life. Within the band, every person is related by blood or mar¬
riage to every other person either directly or through a third person.
However, the Lapp band is not an ‘ extended family ’ but, to repeat,
an alliance between sibling groups, one of which is dominant. The
dominant sibling group provides a nucleus to the genealogical
structure of the band since it is the basic point of reference of that
structure. The dominant sibling group also gives continuity to
leadership since the leader is one of the dominant siblings and his
successor is chosen from his sons or sons-in-law.
One of the problems involved in analysing local group organiza¬
tion is the problem of marriage residence patterns. (I might add
that this is not the only problem involved in local group organiza-

R. Pehrson 291
tion as some social anthropologists have assumed. It is also
necessary to account for the presence of such unmarried adults as
the hired herders found in many Lapp bands.) In analysing mar¬
riage residence, I found no clear-cut matrilocal or patrilocal rules.
I also found that the bands are not exclusively exogamic or endo-
gamic. The concepts of complete exogamy or endogamy in relation
to local groups may be useful in dealings with societies based on
unilineal descent. However, where the local group is neither
exclusively exogamic or endogamic (and this seems to be true of
many bilateral societies in addition to the north Lapps), the prob¬
lem becomes one of determining effective range of relationship
covered by incest taboos. This is so because the Lapp band is not a
corporate entity. By analysing Lapp sociological concepts, it be¬
comes apparent that the Lapp conceives of himself as a point in a
network of kinship relations and not as a member of any corporate
entity greater than the sibling group. In selecting a spouse, the
Lapp must determine his position on this network of kinship ties
in relation to the position of his prospective mate rather than
whether or not she belongs to his local group. Thus, if two people
in the same band are not directly related or if they are not too closely
related, then they may marry. Ideally, incest taboos extend to third
cousins but I found in analysing genealogical connections between
spouses that there is a certain amount of cousin marriage. Here
there was a discrepancy between the statements of informants and
a statistical analysis of the actual situation.
I noted the same discrepancy in statements as to change of
residence upon marriage. The Lapps invariably state that at mar¬
riage, the woman should join her husband’s band but an analysis
of all marriages shows that about equally often, the man joins his
wife’s band and remains there. Occasionally, the spouse who moves
brings his or her parents and parental siblings and families into the
new band. Thus, it became apparent that the Lapps have no simple
rules of residence, that the matrilocal and patrilocal characteriza¬
tions do not apply to a society where parents change residence upon
the marriage of their children and that each case of residence change
has to be investigated to determine the sociological factors at work.
Such an investigation reveals the following factors underlying
marriage residence patterns (although not necessarily in the
following order of importance):

292 Kin Groups

 Relative wealth. The spouse with less reindeer often moves to the
band of the richer spouse. Rich Lapps frequently consider it
desirable to marry olf their children to poorer Lapps since, by so
doing, they gain an addition to their labour force and do not lose
their children’s reindeer.
Relative status of the spouses’ parents or siblings. Status is not
determined exclusively by the individual’s wealth. It may also
derive from membership in the dominant sibling group of a
particular band.
Relative labour convenience. A person who has several siblings may
join his spouse’s band if the latter lacks sufficient manpower to
herd efficiently.
Relative age considerations. These are of two sorts. First, one must
consider the age of the partners relative to their respective siblings.
The eldest son tends to remain at home upon marriage if he is the
son of a band leader or wealthy man. In poor families where the
mother is widowed, the youngest son often remains at home upon
marriage so that he may herd his family’s reindeer while the elder
siblings are employed as servants in other bands. One must also
consider the age of husband and wife relative to each other. If there
is a great age difference, the younger spouse may join the band of
the elder spouse since the latter has had more time to accumulate
property.
Demographic factors are also important in determining resi¬
dence after marriage. If a herding leader has only daughters, then
most of the adult males in the band are liable to have come from
other bands (as Mr Ian R. Whitaker has shown).
Finally, ecological factors help to determine marriage residence.
For example, the son of a rich man may decide to join his wife’s
band if his father and his own siblings have too many reindeer in
relation to the available pastureland.
Now, throughout this consideration of factors determining
residence, you may have noted that people may move in relation to
their siblings instead of in relation to their parents. This may be
correlated with the tendency for Lapps to marry at a relatively late
age or when they have attained enough reindeer to set up their own
household. By the time a Lapp is ready to marry, band and family
leadership may reside in the person of a sibling or cousin rather
than in a member of the parental generation. In other words, a

R. Pehrson 293
Lapp’s relation to his siblings may be as important as his relation
to his parents in determining local group membership. Therefore,
the terms ‘matrilocalism’ and ‘patrilocalism’ do not correctly
characterize the whole situation. The terms ‘virilocal’ and ‘uxori-
local’ are more useful here. Virilocalism means that the married
couple lives at the locality of the husband’s kinsmen, uxorilocal-
ism that the married couple lives at the locality of the wife’s
kinsmen. The uxorilocal-virilocal characterization emphasize rela¬
tionships within one generation, a relationship shown in Figure 1.

Figure 1

This relationship I believe to be crucial in bilateral society. The

matrilocal and patrilocal characterizations emphasize the relation¬
ship over several generations, a relationship important in unilineal
societies shown in Figure 2.

A—n A —O

Figure 2

Thus, when dealing with a bilaterally organized society which

emphasizes sibling solidarity it seems apropos to use the terms
‘virilocal’ and ‘uxorilocal’ in place of the terms ‘matrilocal’ and
‘patrilocal’ with their implications of unilineality.
Leonhard Adam (1948, p. 12) suggests that virilocal and uxori¬
local are preferable to matrilocal and patrilocal which he considers
etymologically incorrect and logically misleading. I also found
the former terms more helpful in analysing marriage residence
patterns in a bilateral society. In using them to characterize all
marriages, it became clear that Lappish marriage residence pat¬
terns are, in fact, bilocal, that there appears to be equal chances of
uxorilocalism and virilocalism with the issue settled in each case
by the various factors I have discussed before (as was suggested to
me by Mr Ralph Bulmer).

294 Kin Groups

 The structural features of Lapp bilateral organization which
emerge from these considerations of kinship, marriage and local
group affiliation may be summarized as follows:

1. The society is horizontally separated into generations.

2. The sibling group is the fundamental structural unit of the
society. The sibling group is the basic point of reference of Lapp
social structure.
3. When the society must act as a unit, these sibling groups form
alliances by bilaterally tracing relationship over a network of
kinship ties. The temporary nature of these alliances gives flexi¬
bility to the structure.

I have not tried in this paper to trace all of the structural ramifi¬
cations of bilateral organization. Rather, I have attempted to
establish that bilateral organization has more structure than has
been allowed and that this structure must be studied in its own
terms.

References
Adam, L. (1948), ‘“Virilocal” and “uxorilocal”’, Man, vol. 48, art. 13,
pp. 12 ff.
Whitaker, I. R. (1955), Social Relations in a Nomadic Lappish
Community, Utgit a Nors Folkemuseum.

R. Pehrson 295
Part Ten Kin Terms

In this section I begin with an extended introduction to try to

clarify certain of the issues in the study of terminology, in
particular the technical vocabulary itself. This essay is followed
by an excerpt from Radcliffe-Brown (Reading 21) on the
relationship between terminology and behaviour, a subject
which continues to be of critical interest even with the
development of more formal methods of analysing kin terms as
a linguistic set. Componential analysis of this kind has been
developed by W. H. Goodenough and F. G. Lounsbury, to
whose papers the reader is referred for further information
about aims and methods. In this selection I have included
Hammel’s introduction to a symposium on ‘Formal semantic
analysis* (1965), which included papers by both these authors.
20 J. Goody

The Analysis of Kin Terms

A paper specially written' for this volume.

The subject of kinship terminologies has for long played a domin¬

ant part in the study of kinship. At the beginning of the eighteenth
century Father Lafitau (1724, p. 552) described the matrilineal
Iroquois and Hurons of North America, where ‘all the children
of a cabin regard their mother’s sisters as their mothers’; this ob¬
servation was the basis of L. H. Morgan’s ‘ classificatory ’ system
of terminology (1871, pp. 468, 470), which he distinguished from
the ‘descriptive’ variety (i.e. Western European) which gave diff¬
erent terms to lineal and collateral relatives (e.g. father and father’s
brother).
Morgan’s typology was criticized by A. L. Kroeber (1909) on
logical grounds. He suggested an analysis of kinship nomencla¬
tures by means of the following more specific criteria:

(a) the difference of generation;

(b) the difference between lineal and collateral kin;
(c) the difference of age within one generation;
(d) the sex of the relative;
(e) the sex of the speaker;
(f) the sex of the connecting relative;
(g) the difference between consanguinity and affinity;
(h) whether the connecting relative was alive or dead.

For comparative purposes these criteria were somewhat

cumbrous and attention was directed at the way only two sets of
relatives were grouped, those of a person’s own generation and
those of his parents. With regard to the senior generation, Robert
Lowie (1921, p. 57) and Paul Kirchhoff (1931) developed a
scheme which classified males into four categories as in Table 1.

J. Goody 299
Table 1

System Relatives Criterion

Father Father's Mothers

brother brother

Generational Father Father Father One ‘father’

term
Bifurcate Father Father Mother’s Two terms:
merging brother partial
merging
Bifurcate Father Father’s Mother’s Three terms:
collateral brother brother no merging
Lineal Father Uncle Uncle Two terms:
merging
only of
collaterals

Leslie Spier (1925) attempted a similar classification of relatives

of a man’s own generation, that is, ‘cousin terms’. The system was
developed by various writers, but especially by G. P. Murdock,
whose latest classification in the Ethnographic Atlas (1967) I
present below. In classifying cousins there are four main possi¬
bilities (a, b, c, d) which involve grouping the following five
categories:
1 2 3 4 5

EGO FS, MS FBS MZS FZS MBS

2 and 3 are ‘parallel cousins’; 4 and 5 are ‘cross-cousins’. The

possibilities taken care of in the code for the Atlas are:
(a) Same term for all, i.e. 1,2,3,4,5; this is the Hawaiian type.
(b) All differentiated by descriptive terms, i.e. 1/2/3/4/51; this is
Descriptive.
(c) A single term for all except 1 (parental siblings), i.e. 1/2,3,4,5;
this is the Eskimo type.
(d) Cross-cousins differentiated from parallel cousins, i.e. either
1,2,3/4,5, or 1/2,3/4,5; these are the Iroquois types.

1. / shows the division in the named categories.

300 Kin Terms

 The terms for parallel cousins (2 and 3) are often the same as
those for siblings or else constitute a separate pair. But those for
cross-cousins are frequently differentiated from one another. Two
of the common ways of doing this skew the terms with respect to
the generation structure, since they identify persons of an indi¬
vidual’s own generation with those of the adjacent generation.
These methods are:
(e) FZS = F and/or MBS = S; this is the Crow type;
(f) the opposite to this is M B S = M B and/or F Z S = S; this is the
Omaha type.
These are the main types of kinship terminology for cousins
distinguished by Murdock, though he also adds an entry for
Sudanese terms and one for mixed or variant patterns.
Although whole sets of kinship terms are sometimes described
by the same words (e.g. Crow, Omaha) as the features of the cousin
terminologies listed above, there is only a limited correspondence
between, say, the terminology for ego’s own generation and that
for the senior generation. However, some work is proceeding in
defining the variations found in sets of Eskimo (Dole, 1960), Crow
and Omaha terms (Lounsbury, 1964).
When a set of kin terms has been established for a particular
group, there are several methods of analysing them.

1. One can relate them to aspects of the past. W. H. R. Rivers and

other writers tried to deduce pre-existing systems of marriage and
descent from the presence of certain kin-terms; much of this work
was rightly described as pseudo-history. More important is the
analysis of cognate systems in order to understand their genetic
divergence or to reconstruct their ancestral forms, as for example,
in the kind of work carried out by Hoijer on Athapascan terms
(1956) or by Friedrich on proto-Indo-European (1966). Such
reconstructions carry dangers when used to deduce facts about the
earlier societies (see Goody, 1959). But equally, the study of kin
terms within a given region, according to the principles of historical
linguistics, can help to prevent the investigator, whether the
original collector or a subsequent discussant, from drawing du¬
bious conclusions about their usage; comparative or etymological
analysis can indicate the possibility of homonyms. With regard to
the lengthy discussion between Leach, Lounsbury and Powell about
the meaning of the Trobriand kinship term, tabu, and its assumed

J. Goody 301
relationship with the Oceanic word ‘taboo’, Ann Chowning has
recently written that if anthropologists, including the original
investigator, Malinowski, had possessed a general knowledge of
Oceanic languages, ‘it would probably never have occurred to
them that there might be anything “forbidden” about the Trob-
riand kinsmen called tabu, and they might have devoted more
time and ingenuity to dealing with the actual rather than the im¬
agined content of the kinship system’ (1970, p. 310).

2. One can relate them to aspects of present behaviour. The func¬

tionalists, Malinowski and Radcliffe-Brown, rejected the doctrine
of survivals implicit in some earlier approaches, because of both
practical and theoretical difficulties. If kin terms are related to
other social institutions, then we should look for present correla¬
tions, not evidence of past associations. Radcliffe-Brown (1930)
saw Australian kin terms as related to the existing system of
marriage; elsewhere he particularly emphasized the relationship
between kin terms and the structure of unilineal descent groups
(1941). Murdock and others have established a high correlation
between terminologies that distinguish siblings from cousins (e.g.
Eskimo) and the presence of an isolated elementary family and
also of neolocal residence (1949, p. 153; see also Marsh, 1967,
ch. 3). Such terminologies are found among the simple hunters as
well as among the most complex industrial civilizations. So that
this correlation supports the curvilinear relationship between
economy and extensive kinship groups (especially unilineal
descent groups) suggested by Radcliffe-Brown (1935), Forde
(1947) and Fortes (1953), and examined cross-culturally by
Nimkoff and Middleton (I960).2
Both these approaches have proved fruitful not only in several
comparative studies but also in the analysis of specific social
systems, for example, in F. Eggan’s study of the Hopi and Pueblo
kinship systems (1950). Indeed any study of kinship roles is bound
to deal with the terms people use to address or refer to kin, and the
way these terms relate to other aspects of social action.
3. There is the study of kin terms as a bounded set, delineating the

2. For an attempt to test a related hypothesis concerning the relationship

of lineal terminologies to systems of inheritance that keep property within
the elementary family, see Goody (1969; 1970).

302 Kin Terms

principles behind their employment. Essentially this is what
Kroeber set out to do (like McLennan, in the nineteenth century,
he claimed there was no close correlation with other aspects of
society, though this view he later abandoned) and it is the exam¬
ination of internal coherence that forms the basis of the more
systematic, linguistically-derived study of such sets, known as
componential analysis.

Componential analysis is at present much in vogue in the

United States, though some of its most distinguished practitioners
warn against expecting too much from its use. The pioneering work
in this field was done by W. H. Goodenough and F. G. Lounsbury.
In an article on the formal analysis of Crow and Omaha-type kin¬
ship terminologies, Lounsbury (1964) raises the general question
of the ‘meaning’ of kinship terms, discussed by many authors,
which is seen as critical to the definition of a set and is related to
the problem of homonyms, polysemy or metaphor (Wallace and
Atkins, 1960; Greenberg, 1949; Goodenough, 1956; Edmonson,
1957). In particular he asks whether kinship words have as their
primary referent categories of kin (i.e. members of a clan) of senior
generation or more immediate, genealogically, defined roles (e.g.
father). This problem touches upon another, that is, on the mean¬
ing of kinship (rather than of kinship terms), which again turns
partly on the problem of the ‘biological’ or ‘social’ reference of
terms and the status of notions of conception and procreation.
Lounsbury’s position leads him to oppose a lineage explanation of
Crow terms (e.g. Radcliffe-Brown, 1941, who points to the relation¬
ship with matrilineal descent groups) to one in terms of inheritance
which he assumes to occur between close kin. The latter position is
more in keeping with his need to define a bounded set for the pur¬
pose of analysing the components. The former approach is more
in keeping with Radcliffe-Brown’s emphasis on the necessity of
understanding the lineage context (and indeed the wider social
context) of the terms in order to understand their ‘meaning’.
The argument is somewhat academic, since there is no way of
testing primacy (unless we refer to the meaning a child learns first
of all, and this is not necessarily a useful guide). Moreover, there is
something suspicious about a controversy that leads those who
start from a genealogical (or familial) basis to find the evidence

J. Goody 303
supporting the ‘extensionist’ approach, while those who start
from a group (or ‘ structural ’) basis find the evidence supporting
the categorical view of kinship terms. As far as the alternative
‘explanations’ of Crow terms are concerned, it is difficult to see
how one could separate the factor of lineage membership from
that of inheritance.
The data which any investigator has to analyse consists of a
whole range of usages, part of which is specifically metaphoric
(that is, recognized by the actors as such), and part may be crypto-
phoric (e.g. dead metaphors). This range we have to understand as
best we can. Lounsbury is possibly correct in seeing close rather
than distant relationships as constituting the core set of meanings
from an overall analytic standpoint; indeed this view may be
entailed in the definition of the field of kinship itself. But the
conclusions he draws, in terms of the right to neglect other mean¬
ings in defining specific kinship sets seems highly questionable;
so too does his espousal of Malinowski’s phylogenetic reduction-
ism, which again ‘ fits ’ with his linguistic approach. Like Malinow¬
ski, he appears to confuse an analysis of the structure of a system
(e.g. a kinship set) with the developmental process by which it is
learned.
Each of these approaches, the historical, the structural-func¬
tional and the formal, has something to offer and no competent
analyst will overlook their various insights. But from the socio¬
logical standpoint, the most important task must be the investiga¬
tion of the relationship between kinship terms and other aspects of
social action, both by the intensive study of particular societies and
the extensive comparison of regional groups and world samples.
Such analysis must begin at the stage of data collection; it is too
often assumed that for every ‘society’ there is one kinship ter¬
minology ;3 or that for every relative there is one usage. It some¬
times seems that kinship terminologies have played so prominent
a part because people have regarded them as so simple to collect.
For instance, L. H. Morgan sent his questionnaires around the
world and produced his famous book on Systems of Consanguinity
on the basis of the data so acquired. Something can certainly be
done by this means; but it is important not to over emphasize the
3. On the class basis of some English kinship terms, see J. Goody (1962).

304 Kin Terms

 results of this and more sophistica ted work by assuming that in the
study of kinship terms one has the key to the ‘social structure’,
either in the present or in the past.

References
Chowning, A. (1970), ‘Taboo’, letter in Man, vol. 5, pp. 309-10.
Dole, G. E. (1960), ‘The classification of Yankee nomenclature
in the light of the evolution of kinship’, in G. E. Dole and R. L.
Cameiro (eds.). Essays in the Science of Culture, Crowell, pp. 162-77.
Edmonson, M. S. (1957), ‘Kinship terms and kinship concepts’,
Amer. Anthropol., vol. 59, pp. 393-433.
Eggan, F. (1950), Social Organization of the Western Pueblos,
Chicago University Press.
Forde, C. D. (1947), ‘The anthropological approach in social science’.
Advance Sci., vol. 4, pp. 213-24.
Fortes, M. (1953), ‘The structure of unilineal descent groups’, Amer.
Anthropol., vol. 55, pp. 17-41.
Friedrich, P. (1966), ‘Proto-Indo-European kinship’, Ethnol, vol. 15,
pp. 1-36.
Goodenough, W. H. (1956), ‘Componential analysis and the study
of meaning’. Language, vol. 32, pp. 195-216.
Goody, J. (1959), ‘Indo-European society’, Past and Present, vol. 15,
pp. 88-92.
Goody, J. (1962), ‘On nannas and nannies’, Man, vol. 62, art. 288.
Goody, J. (1969), ‘Inheritance, property and marriage in Africa and
Eurasia’, Sociol., vol. 3, pp. 55-76.
Goody, J. (1970), ‘Cousin terms’, Southwestern J. Anthropol., vol. 26,
pp. 125-42.
Greenberg, J. H. (1949), ‘The logical analysis of kinship’, Philos.
Sci., vol. 16, pp. 58-64.
Hoijer, H. (1956), ‘Athapascan kinship systems’, Amer. Anthropol.,
vol. 58, pp. 309-33.
Kirchhoff, P. (1931), ‘Die Verwandtschaftsorganisation der
Urwalstamme Sudamerikas’, Zeits. Ethnol., vol. 63, pp. 85-191.
Kroeber, A. L. (1909), ‘Classificatory systems of relationship’, J.
roy. anthropol. Inst., vol. 39, pp. 77-84.
Lafitau, J. F. (1724). Moeurs des Sauvages Ameriquains, vol. 1, Paris.
Lounsbury, F. G. (1964), ‘A formal account of the Crow- and Omaha-
type kinship terminologies’, in W. H. Goodenough (ed.) Explorations
in Cultural Anthropology, McGraw-Hill.
Lowie, R. H. (1921), Primitive Society, ?
Marsh, R. M (1967), Comparative Sociology, Harcourt, Brace & World
Morgan, L. H. (1871), Systems of Consanguinity and Affinity of the
Human Family, Smithsonian Institution Contributions to Knowledge,
No. 17.
Murdock, G. P. (1949), Social Structure, Macmillan Co.

T-K-L J. Goody 305

Murdock, G. P. (1967), ‘The ethnographic atlas: a summary’,
Ethnol., vol. 6, pp. 109-236.
Nimkoff, M. F., and Middleton, R. (1960), ‘Types of family and
types of economy’, Amer. J. Sociol., vol. 46, pp. 215-25.
Radcliffe-Brown, A. R. (1930), ‘Social organization of Australian
tribes’, Oceania, vol. 1, pp. 34-63, 206-46, 322-41, 426-56.
Radcliffe-Brown, A. R. (1935), ‘Patrilineal and matrilineal
succession’, Iowa Law Rev., vol. 20, pp. 286-303.
Radcliffe-Brown, A. R. (1941), ‘The study of kinship systems’,
J. roy. anthrop. Inst., vol. 71, pp. 1-17.
Spier, L. (1925), Distribution of Kinship Systems in North America,
University of Washington Publications in Anthropology, vol. 1. pp.
69-88.
Wallace, A. F. C. and Atkins, J. (1960), ‘The meaning of kinship
terms’, Amer. Anthropol., vol. 62, pp. 58-80.

306 Kin Terms

 21 A. R. R add if fe-Brown
Kin Terms and Kin Behaviour

Excerpt from A. R. Radcliffe-Brown, ‘The social organization of

Australian tribes’, Oceania, vol. 1, 1930, pp. 426-56.

The social organization of Australian tribes affords material of

capital importance for the science of comparative sociology. We
find an organization of a single specialized type over the whole
continent, and the type has been elaborated into a large number of
different varieties. A comparative study of all the details of these
variations affords an opportunity for sociological analysis which
is perhaps not equalled in any other part of the world. This is one
of the chief reasons why it is of such importance to science to
obtain an adequate record of the Australian aborigines before they
and their culture disappear.
It is not possible in the space here available to undertake a
detailed sociological analysis of the Australian organization. But a
brief discussion seems desirable in order to remove misconceptions
that have arisen in theoretical discussions.1
The first question that requires to be dealt with is that of the
relation between social organization and the terminology of kin¬
ship. There are two views on this subject that I wish to controvert.
One is the view of Lewis Morgan, adopted from him by Howitt
and Sir James Frazer, which is to the effect that the kinship
terminology of Australian tribes is not correlated with the existing
social organization but is correlated with and has its origin in a
hypothetical condition in which individual marriage did not exist,
but groups of men were united in some sort of marriage bond with
groups of women. The second view is one which is held by Profes¬
sor Kroeber, that there is in general no very close correlation
1. Practically all the theoretical discussion of Australian social organiza¬
tion has been directed towards providing hypothetical reconstructions of
its history. Even Durkheim, though approaching the subject as a sociologist,
devotes his attention to questions of historical development. The more
modest but really more important task of trying to understand what the
organization really is and how it works has been neglected.

A. R. Radcliffe-Brown 307
between the kinship terminology of a people and their social
institutions (1909; 1917).
So far as Australian tribes are concerned it can be laid down as
definitely proved that the kinship terminology of a tribe is an
integral and essential part of the social organization. At every
moment of the life of a member of an Australian tribe his dealings
with other individuals are regulated by the relationship in which
he stands to them. His relatives, near and distant, are classified into
certain large groups, and this classification is carried out by means
of the terminology, and could apparently not be achieved in any
other way. Thus in any part of the continent when a stranger comes
to a camp the first thing to be done, before he can be admitted
within the camp, is to determine his relationship to every man and
woman in it, i.e. to determine what is the proper term of relation¬
ship for him to apply to each of them. As soon as he knows his
relation to a given individual he knows how to behave towards
him, what his duties are and what his rights.
The case against Professor Kroeber is, I think, proved conclu¬
sively by the fact that variations in the kinship terminology from
tribe to tribe are directly correlated with variations in the social
organization, including variations in the regulation of marriage.
As against Morgan and those who follow him it can be shown
that there is a very thorough functional correlation between the
kinship terminology of any tribe and the social organization of
that tribe as it exists at present. If this is so there is no reason
whatever to suppose that the kinship terminology is a survival
from some very different form of social organization in a purely
hypothetical past.2
2. The conclusive criticism of Morgan’s theories and others of the same
kind was stated forty years ago by Starcke (1889, p. 18) ‘Many learned
men are too much disposed to seek for the explanation of a given custom in
conditions of former times which have now perhaps disappeared. It is
certain that customs persist by the force of habit, even when the conditions
which first gave birth to them have long ceased to exist, yet it is scarcely
necessary to remark that this appeal to early times can only be effective
when it has been shown to be impossible to discover the cause of such
custom in the conditions under which they still continue. If this main
principle is not accepted, we shall be led astray by every idle delusion. If we
are able to trace the cause of a custom in existing circumstances, we must
abide by that cause, and nothing but a definite historical account of the
prior existence of the custom can induce us to seek for another explanation.’

308 Kin Terms

 I propose therefore to consider briefly some of the principles
that are active in the Australian classification of kin. The most
important of these principles is one which is present in all classifi-
catory systems of kinship terminology. Morgan applied that term
to all systems which apply the same term to lineal and collateral
relatives by regarding two brothers as equivalent, so that if a man
stands in a certain relationship to Ego his brother is regarded as
standing in the same relationship. This principle may be spoken of
as that of the equivalence of brothers. It applies, of course, equally
to two sisters. Now this principle is universally applied in all
Australian systems of terminology. Everywhere the brother of a
father is called ‘father’, and therefore his children are called
‘brother’ and ‘sister’, and similarly the sister of a mother is
called ‘mother’ and her children are also called ‘brother’ and
‘sister’.
This principle is not merely a matter of terminology. It is a most
important sociological principle which runs through the whole of
Australian life. It depends on the fact that there is a very strong,
intimate and permanent social bond between two brothers born
and brought up in the same family.3 This solidarity between
brothers, which is itself an expression or result of family solidarity,
is a very obvious thing to anyone who studies the aborigines at
first hand. It shows itself moreover in certain institutions. The
levirate is, I believe, universal in Australian tribes. By this custom,
when a man dies, his wife or wives and his dependent children pass
to his brother, in some tribes only to his younger brother. When
possible it is the man’s own brother who succeeds him, but if he
has no brother of his own his place is taken by someone who stands
in the classificatory relation of ‘ brother ’ to the deceased.
The function of this custom in terms of social integration is
fairly obvious. A marriage and the birth of children sets up certain
social relations, a certain structural arrangement. The wife and
children are dependent on the husband and father and their posi¬
tion in the society is fixed by that dependence. The man’s death
causes a disruption of the social structure and the society needs to
restore it with a minimum of alteration of the structure as a whole.
3. For an account of the relation between brothers see Warner (1930).
This article gives the best account of the actual working of an Australian
kinship system in the everyday life ct the tribe.

A. R. Radcliffe-Brown 309
This is done by replacing the dead individual by a person who is as
nearly as possible his social equivalent. The substitution of one
brother for another thus permits the social structure to be restored
with a minimum of change after the death of an individual.
Professor Sapir (1916) has suggested that there is a correlation
between the custom of the levirate and the general principle of
classificatory systems of terminology. In that I think he is right,
but I think he is in error in suggesting a direct causal relation
between the two whereby the custom of classifying the father’s
brother with the father is regarded as the effect of the levirate. In
general I believe that it is a false procedure to look for the cause of
one social institution in another particular institution. In the
present instance my own view is that both the levirate and the
classificatory principle in terminology are the results of the action
of a single sociological principle, namely that which I have called
the principle of the social equivalence of brothers. This principle is
at work, I believe, wherever we find the levirate and wherever we
find a classificatory system of terminology. Its action is more effec¬
tive in some societies than in others, and it is combined with the
action of other principles. Thus in some societies we may find a
classificatory system without the levirate, and in others we may
find the latter without the former.
The principle is obviously far more effective in the simpler
societies than in the more complex. In such a simple society as that
of an Australian tribe the intimate and close relationship between
brothers lasts right through life. Two brothers necessarily belong
to the same social groups, the same horde, the same clan, etc. The
only exception to this would be in age groups, when older and
younger brothers might belong to different groups. Two brothers,
therefore, occupy similar positions in the total social structure.
Their social personalities are almost precisely the same. This is
rarely the case in our own complex societies.
The principle of the equivalence of brothers as an active principle
determining social structure may be regarded as a special example
of a more general tendency the presence of which is readily dis¬
covered in the social structure of the simpler cultures. Wherever
the structure includes small groups of strong solidarity and having
important and varied functions, when an individual is brought into
some close social relation with one member of the group, there is a

310 Kin Terms

 tendency to bring him into close relation with all the other mem¬
bers of the group. An instance of this tendency is to be seen in the
special close relation that is set up in many societies between a man
and the group (family, clan, etc.) from which he obtains a wife.
In terms of persons, if there is a strong, intimate and permanent
bond between two persons A and B, then when a third person C
is brought into an important social relation with B there is a
tendency to bring him into close relation with A. The resulting
relation between C and A will depend, of course, on the kind of
relation that already exists between A and B.4 In terms of the
Australian social organization I am by the fact of birth and up¬
bringing brought into a specific relation with my father. Since
between him and his brother there is the special intimate relation
that we have seen I am brought into a very close relation with my
father’s brother in which he becomes for me another ‘father’.
This would seem to be the essential principle of the classificatory
system of terminology and of the Australian social organization.
A similar custom to the levirate is that known as the sororate.
The form that this takes in Australia is that when a man marries
the elder of two or more sisters he becomes entitled to marry the
younger ones also. In many Australian tribes the ideal arrangement
is considered to be that a man who marries the eldest of the sisters
should also marry the second and that he should then transfer his
right to the third and fourth to his younger brother. In this custom
of the sororate we have sisters treated as being socially equivalent,
just as with brothers in the levirate. The existence of this close bond
between sisters is shown also in the custom of some tribes, for
example the Yaralde, whereby a special, strong and intimate bond
is set up between two men who marry two sisters. In the Yaralde
tribe there is a special term of relationship for two men thus con¬
nected.
Without considering in any way how the Australian social
organization may have arisen in a distant past about which we shall
never obtain any direct knowledge, we may say that as it exists at

4. Thus it can be shown, I think, that it is this tendency which in the

instance of a man and his wife’s mother finally results in the custom,
universal in Australia, whereby the man must avoid all social contact with
the woman while still regarding her, in the phrase of a native, as his ‘best
friend in the world ’.

A. R. Radcliffe-Brown 311
present an analysis of it reveals this important active principle of
the solidarity of brothers, and we may say that on this principle
the existing system is built. By applying the principle the father’s
brother comes to be regarded for social purposes as similar to the
father, and the two are classified under a single kinship term,
without, however, any confusion between the real father and his
brother. Similarly the mother’s sister and the father’s brother’s
wife are classified as ‘mother’ and the behaviour towards them is
modelled on that towards the mother. Carrying forward to the
descending generation a man treats the children of his brother in a
similar way to that in which he treats his own children, and calls
them ‘son’ and ‘daughter’, just as they call him ‘father’. Passing
to more distant relationships the brother of the father’s father is
classified with the latter, both in terminology and for social pur¬
poses, and his son is therefore in turn classified with the father.
In this way the Australian native creates a stable social structure
by which all the details of social intercourse between one person
and another are regulated. Since relationships are traced without
any limit an individual stands in some definite relationship to every
person whom he meets in the course of his life.
Within a single class of relatives some are near and some are
distant and the degrees of nearness, though not usually expressed
in the terminology, are of course recognized for social purposes
and such recognition is an integral and essential part of the system.
Thus a man cannot marry, or show any familiarity towards the
daughter of any man he calls ‘father’. He could not fight with his
own father, nor, I think, with his father’s brothers or any of his
nearer ‘fathers’, but he may quite well on occasion fight against a
distant ‘ father ’, and indeed much more readily in some tribes than
against a distant ‘mother’s brother’.
A second important principle of the Australian system is the
distinction between the father and the mother, and therefore
between relatives through the father and relatives through the
mother. Father and mother are treated as two different kinds of
relative, though it is difficult to give any simple statement as to
what the difference consists in. Throughout Australia it seems that
the personal bond between a child, even a son, and the mother,
is regarded as stronger than that between child and father. By
virtue of the act of suckling, if for no other reason, the personal

312 Kin Terms

relation of child and mother is a peculiarly intimate one, especially
in the early years of life, and this creates a permanent bond of
solidarity which has great importance in Australian life and in
determining the social structure.
When we come to the brother of the mother and the sister of the
father the classificatory principle takes a new form. Since there is a
close bond between a child and its mother, and another bond
between the mother and her brother the child is brought into a
close personal bond with the mother’s brother. The latter is not
treated in any way as similar to the father or father’s brother, but
is treated as a sort of male ‘mother’. Similarly the father’s sister is
treated as a sort of female ‘father’. In all Australian tribes the
actual mother’s brother and the actual father’s sister of an indi¬
vidual have important places in his life, and the whole system can
only be understood when this is fully recognized. Thus the distinc¬
tion in terminology between mother’s brother and father and
between father’s sister and mother is correlated with social distinc¬
tions of the greatest importance. The tendency to treat the mother’s
brother as a sort of male ‘mother’ is the result of the action of the
same principle that results in the father’s brother being treated as a
‘father’.5
Another important principle of the Australian system is con¬
nected with the relations between persons of different generations.
The relationship of generation has its origin in the family in the
relation of parents to children. It becomes of importance in general
social life because social continuity requires that the body of
tradition possessed by the society shall be handed on by one
generation to the next, and this handing on of tradition entails a
relation of superiority and subordination as between one generation
and the next. The generation of parents must have authority over
the generation of children. We find this in one form or another in
every human society.
As between persons who are separated by an intervening genera¬
tion a new situation arises. If we call the generations 1,2 and 3, then
those of generation 1 exercise authority over those of 2 and those
of 2 over those of 3, but by a tendency which is apparent in many
of the simpler societies and is perhaps really universal, persons of 1
5. The tendency can be seen in many classificatory systems in different
parts of the world. See Radcliffe-Brown (1924).

A. R. Radcliffe-Brown 313
and 3 are brought together into a different kind of relationship
which, in spite of the difference of age, links them together on terms
of familiarity and almost of equality. It is possible to demonstrate
the reality of this tendency and its effectiveness in influencing the
social structure in many parts of the world. It is certainly effective
and important in Australian society. It shows itself in the termin¬
ology in two ways. In a few tribes the father’s father is called ‘elder
brother’ and the son’s son is called ‘younger brother’, but this
procedure is rare in Australia. More commonly a single reciprocal
term is used between grandparents and grandchildren. Thus a
father’s father and his son’s son address each other by the same
term. Now it seems that very frequently in classificatory systems of
terminology the use of a single self-reciprocal term between two
relatives indicates that the social relation between them is sym¬
metrical, whereas the use of two terms one reciprocal to the other
implies that the social relationship is asymmetrical. By a sym¬
metrical relationship is meant one in which, as between two rela¬
tives A and B, A behaves towards B in the same way as B towards
A, whereas in an asymmetrical relationship A behaves in one way
towards B and B behaves in a different but correlated way towards
A. Thus the relation of father and son is a typical asymmetrical
relationship in Australia and apparently everywhere. In Australia
also the relationship between two brothers is always in some
respects asymmetrical, and therefore in the terminology there is
usually no word for brother but one term for elder brother and
another for younger brother. Now in the case of grandparents and
grandchildren, or at any rate in that of father’s father and son’s
son, it does seem that the use of a single self-reciprocal term be¬
tween them is associated with a tendency to group them together
on terms of familiarity, and if not equality at any rate of social
equivalence. This is borne out by the way in which, in certain
kinship terminologies, a given individual applies the same term of
relationship to two men who are father’s father and son’s son to
one another. One of the significant features of the section system is
that it brings together into the same position in the social structure
the father’s father and his son’s son.
The principle that is here indicated enables us to understand a
very strange feature of the terminology of some tribes. The father’s
father’s father is called by the same term as ‘son’, and the son’s

314 Kin Terms

son’s son is called by the same term as ‘father’. Since I include
under a single relationship my father’s father and my son’s son,
the sons of all relatives of that kind should fall together and can be
called ‘ father ’, while the fathers of all of them should equally be
classified together and may therefore be called ‘son’.
Another most important principle in the Australian system is
that of reciprocity in marriage. This is merely a special instance of
a much wider principle of reciprocity. What underlies it is the fact
that when a marriage takes place there is a change of social
structure, certain existing social ties being broken or changed and
other new ties created. The group from which the bride is taken,
whether we regard the family only, or the horde, suffers a loss or
damage. For this they must be compensated or indemnified. It is
this aspect of marriage that affords the explanation of a great many
of the ritual and other customs connected with marriage in all
parts of the world. In Australia it results in a custom whereby
marriage is normally an exchange in which each side loses a
woman and gains one. In the majority of tribes this takes the form
of sister exchange. A man receives a wife from a certain family and
horde and his own sister goes in exchange to his wife’s brother.
Amongst the tribes of Gippsland. who have no moieties, the
exchange of sisters is regarded (according to Howitt) as the only
legitimate form of marriage. In the Yaralde and other tribes where
ther local patrilineal clan is a very important group the exchange is
not between families but between clans. Where there is a system of
moieties one of the functions of this is that every marriage,
whether by exchange of sisters or not, is an exchange between one
moiety and the other. So also, in the section system all marriages
are parts of a continuous series of exchanges between the two
sections or sub-sections of a pair.
Most Australian systems of terminology are dependent on this
reciprocity in marriage. Where there is sister exchange the father’s
sister and the mother’s brother’s wife are classified together under
a single term, and similarly the wife’s brother is classified with the
sister’s husband. In the exceptional tribes in which sister exchange
is not permitted these relatives are distinguished.

References
Kroeber, A. L. (1909), ‘Classificatory systems of relationships’,
J. roy. anthropol. Inst., vol. 39, pp. 77-84.

A. R. Radcliffe-Brown 315
Kroeber, A. L. (1917), Californian Kinship Systems, University of
California Publications.
Radcliffe-Brown, A. R. (1924), ‘The mother’s brother in South
Africa’, South African J. Sci., vol. 21, pp. 542-55.
Sapir, E. (1916), ‘Terms of relationship and the levirate’, Amer.
Anthropol. vol. 18, pp. 327-37.
Starcke, C. N. (1889), The Primitive Family, Routledge & Kegan Paul.
Warner, W. L. (1930), ‘Morphology and functions of the Australian
Murngin type of kinship’, Amer. Anthropol., vol. 32, pp. 207-56.

316 Kin Terms

22 E. A. Hammel
Formal Semantic Analysis

Excerpt from ‘Introduction’, in E. A. Hammel (ed.), ‘Formal semantic

analysis’, American Anthropologist, Special Issue, vol. 67, no. 5, pt 2,
1965, pp. 1-8. Revised for this edition by the author.

Formal analyses in anthropology are distinguished from others in

two senses: the methodological and the theoretical. In the former
sense they are characterized by particular attention to rigor and
internal form, and in the latter by explicit recognition of one or
more superordinate levels of analytic determinants in a given
domain. This style of analysis has multiple and tangled historical
roots, particularly because of its close connection with kinship
studies on the one hand and with descriptive linguistics on the
other. In one sense, it is but the most recent outgrowth of a long
interest, in anthropology and philosophy, in the relationships
between language and other aspects of culture, particularly per¬
ception and cognition. Wilhelm von Humboldt suggested such
a connection as long ago as 1836, and Boas (1911) also suggested
that the proper entree into folk psychology was an appreciation of
the internal form of language.
The contributions of Sapir, Whorf, Lee and others need not be
summarized here. With respect to kinship studies, it is important
to note that Kroeber (1909) took a Boasian view of the nature of
kinship terminologies, insisting that internal form was indicative
of a particular ‘psychological’ orientation toward the phenomena
so described, although he later clarified his definition to include
cognition in the broadest sense (1952).
A second current of explanation (in kinship studies) was mani¬
fested by Morgan (1871), Rivers (1914), Seligman (1917), Gifford
(1916) and others in their ascription of facets of kinship termin¬
ologies to particular forms of marriage and in their use of ter-,
minologies to reconstruct sociological history. Similar weight was
placed on sociological determinants by Radcliffe-Brown (1941),
Tax (1937), Lowie (1930; 1932) and others with greater attention

E. A Hammel 317
to total systems of prestations and statuses, and (by some) to socio¬
logical factors in reconstruction. Murdock’s discussions (e.g. 1949)
reject the particular sociological determinants employed by some
of his predecessors, but return to the use of others in reconstruc¬
tion. As different as all of these approaches seem, they are united
in that they attempt explanation of terminological systems by
reference to factors which do not seem, in our Western European
or traditional anthropological ideology of kinship, internal to the
terminological systems themselves. In contrast, most ‘componen-
tial analyses ’ of kinship terminologies have been concerned with
internal form and have specified as analytic determinants certain
features of genealogical reckoning which the analysts felt in some
way to be naturally ‘inherent’ in kinship. Whether genealogy is
more or less relevant to kinship is moot, and we will consider the
question further below.
Although the notion of ‘components’ and ‘factors’ was wide¬
spread in science, the evolution of a self-conscious methodology of
formal semantic analysis seems to have taken place largely within
descriptive linguistics, and almost all of the anthropologists who
concern themselves with such analyses are linguists, or were
trained by linguists, or were markedly influenced by linguistic
theory. (Biological systematics also played a role, particularly in
ethnoscience.) Bloomfield’s insistence (1933) on the use of meaning
only as a constant, in order to determine the variety of linguistic
forms and the nature of phonological and morphological redund¬
ancy, obstructed the analysis of meaning until relatively recently.
Three kinds of developments seem to have stimulated removal of
the obstruction. First, out of the pioneering work of Jakobson in
distinctive feature analysis (1928), of Hockett’s concept of the
portmanteau morph (1947; 1955), of Harris’ analysis of a Hebrew
paradigm (1948) and of Hjelmslev’s hints at stratification (1943)
there grew the employment of a hierarchy of superordinate and
interacting analytic levels, populated by hypothetical constructs,
and among which was a sememic stratum. Second, the notion of
what we might loosely call ‘syntactic control’ was expanded from
the simple utterance-matching procedures used in specifying
minimal pairs in phonology, to include taxonomic specification of
relationships of contrast and inclusion (Goodenough, 1956;
Conklin, 1962;Frake, 1961) and eventually to include a wide range

318 Kin Terms

 of social and cultural criteria (Gumperz, 1958; Hymes, 1962;
Tyler, 1966). Neither of these concepts - that of higher levels in
linguistic structure and interacting strata, now considerably
elaborated in one way by Lamb (1964a; 1964b) and in another by
Chomsky (1957) or the widened specification of syntactic control -
is necessarily at variance with the basic analytic attitudes expressed
by Boas, Bloomfield and other grandfathers of descriptive linguis¬
tics. The third factor in the evolution of formal semantic analysis
was simply a reversal, but not a discard, of Bloomfield’s dictum
about meaning. Rather than holding a referential meaning con¬
stant and determining which linguistic forms may occur with it,
we now also hold a linguistic form constant and determine which
elements of referential meaning may occur with it.
The interweaving of linguistic developments with anthro¬
pological ones is complex, but we might say that the two were
closely associated while linguistics and ethnology were themselves
hardly separate, that is, for example, when Boas spoke of language
and folk psychology (1911) and Kroeber of the nature of classi-
ficatory systems of relationship (1909). Anthropological emphasis
until the early 1950s remained with ‘external’ analyses, while
further methodological progress was made in linguistics itself.
By that time, the influence of the linguistic development made itself
felt in Goodenough’s monograph on Truk (1951), his paper on
componential analysis (1956) and Lounsbury’s on the Pawnee
system (1956). The two papers of 1956 were published as papers in
linguistics.
From that point on, events are too numerous and complex to
trace here. Nevertheless, it is worth pointing out that there are still
two very strong trends evident within formal semantic analysis:
one concerns itself with the rigorous analysis of seemingly internal
form, and the other with the utility of such analyses in providing
insight into other areas of behaviour. These two trends are them¬
selves interrelated and cannot be separated from the three major
issues raised by Gardin (1965): (1) delimitation of the corpus of
phenomena to be investigated, (2) selection of the language of
description and analysis, and (3) the relevance of the analysis.
The interrelationships of these three points can be seen more
clearly if we admit that most formal analyses are grammars of
correspondence between two languages, a target language (e.g. a

E. A. Hammel 319
kinship terminology) and a reference language (e.g. kin type
designation) (cf. Hammel, 1964).1 The first task, delimitation of the
corpus in the target language, is usually achieved by syntactic
control in the use of a superclass lexeme or phrase - for example, in
the control question, ‘Is your — a relative?’ The difficulties
attendant on this approach are pointed out by Schneider (1965),
Bright and Bright (1965), and Burling (1964). When data from the
literature are being analysed, one must often assume that some
kind of control question was in fact used, so that, for example, all
kin types reported would have fallen under some term for ‘kins¬
man Control questions can be good for some purposes and bad
for others. Delimitation of the corpus can also be attempted by
referential control, for example, by insisting that ‘kin terms’ in the
target language may only apply to persons genealogically linked,
or to persons who live nearby, or to persons with whom Ego has
social relationships of particular types, and so on. The essence of
this procedure is that external factors are used in the delimitation.
The essence of both procedures is that some rule of inclusion and
exclusion must be used to define the corpus.
An analysis may proceed further in an essentially syntactic
fashion in the target language, that is, solely on the grounds of
internal form without reference (other than perhaps in delimitation)
to any external descriptive grid. Burling’s second analysis of
Burmese (1965) follows this course, using internal lexical rela¬
tionships. This internal approach, the search for intrinsic internal
pattern without reference to an external descriptive grid or to
sociological or psychological ‘function’, is characteristic of much
research in ethnoscience and also of those analyses of kinship
terminology that lean heavily on the process of enculturation (e.g.
Burling, 1970) or the theory of information processing (Sanday,
1968). Most analyses, however, shift at this point to a reference

1. The term ‘language’ is used here in its most general sense. In using
‘target language’ to refer to a kinship terminology I mean not only the set of
unit lexemes referring to kinsmen in a natural language but also the rules
which govern their combinations as polylexemic utterances and those which
describe their relationships of contrast and inclusion. In using ‘reference
language’ to refer to a system of kin type designations, I mean not only
the primitive ‘lexical’ elements F, M, B, Z, S, D, but also the rules which
govern their combination into logically possible and nonredundant com¬
pounds, such as B S, M B D, etc.

320 Kin Terms

 language and continue within it. The reference language most
commonly used for kinship studies is one of kin types (e.g. F, M, B
and the relative products FB, MBS, etc.); the mapping of the
target language onto the reference language is achieved by a listing
of the ‘lexemes’ of the reference language which are included
within the lexemes of the target language. It is worth noting that
the reference language must be more primitive than the target lan¬
guage.2 Some analyses such as those of Romney and D’Andrade
(1964), Romney (1965), Hammel (1965a; 1965b) shift to a still
more primitive reference language based on the logical genea¬
logical definition of kin types.3
Up to this point, because the elements of the reference language
are always equal to or more numerous than the corresponding
elements of the target language, there is little in the analysis that
can be called ‘componential’. As the subsets in the reference lan¬
guage, which are externally delimited by their rules of inclusion
under lexemes of the target language, come to be internally defined
through their own distinctive features or through internal rules of
inclusion and contrast, the componential character of the analysis
emerges. It is at this point that the analysis shifts from its correla¬
tional focus, although the change may be only implicit. For
example, in Lounsbury’s analysis of Crow-Omaha systems (1964)
or HammePs of Comanche (1965a) there are defined certain rules
of transformation which effect the reduction of the members of a
subset of kin types to one such member and thereby define the
nature of internal redundancies in the reference language. The
basic member of each set (root, kernel) to which all other members
are reduced (or from which they may be generated) can in fact be
regarded as having a superordinate locus, the realizations of which
constitute the members of the subset. If the class of all such super-

2. The use of a reference language which is more primitive than the target
language simply allows one to examine a set of reference-language lexemes
(simple or compound) which are grouped under a single target-language
lexeme for the presence of distinctive features.
3. I have not included references to a number of studies, frequently much
more abstract or mathematical, that discuss the theory ol lormal analysis
or the comparative applications of formal analysis per se, without analysing
particular terminologies as well. See for example Kay (1965; 1967; 1968),
Tyler (1966a), as well as the comparative implications of Lounsbury
(1964), Hammel (1965a; 1965b), Romney (1965).

E. A. H ammel 321
ordinate units or ‘archi-kin-types ’ is itself internally defined by the
distinctive features of its separate member archi-kin-types, state¬
ment of and classification by those distinctive features constitute a
componential analysis, and the distinctive features themselves can
be regarded as components on a still higher level. Some transforma¬
tional analyses are intended only as a statement of rules of re¬
dundancy; their purpose can be to make eventual specification of
distinctive features easier and more rigorous. In many componen-
tional analyses, isolation of the components is accomplished by
working out the rules of redundancy in the analyst’s head instead
of on paper, and ‘proving’ them by giving listings of correspon¬
dences between, say, kin types and features of genealogical reckon¬
ing like lineality, generation level, and so on.
Clearly, these procedures depend on the nature of the delimita¬
tion of the corpus in the target language and on the selection of the
reference language(s). Up to this point, the analyses can be judged
by criteria of accuracy, consistency and parsimony. Disputes about
the worth of formal analyses, however, seldom rely on these
criteria but relate to Gardin’s third point (see p. 319) of ‘what can
the analysis teach us?’ In my view, these questions are but a con¬
tinuation of the earlier ones and pose a problem of the correspond¬
ence between one reference language and another. If our concern
is the relevance of formal analyses (or indeed of any analyses),
relevance must be taken to mean correspondence with a set of
propositions or important questions in still another language of
description. That language may concern itself with the nature
of cognition independently assessed, with pragmatics, with resid¬
ence, inheritance or succession rules, with the native ideology as
independently expressed, or with a host of other matters. These
are important questions if formal analyses are to be incorporated
into the larger theoretical structure within which we think we are
operating as social scientists. They determine, in fact, the rules
according to which the lexical corpus of the target language is to be
selected and those according to which the initial reference lan¬
guages are to be specified. Each particular question allows us to
determine whether, for it, the analytic domain is in fact a semantic
domain or whether an analysis is a semantic analysis at all (cf.
Gardin, 1965; Schneider, 1965) for an analysis can have meaning
only with respect to some such question.

322 Kin Terms

 Up to this point in the history of the method there has been
necessary and inevitable concern with technical detail, and the
reference languages have been those which were easy to determine
and profitable to use in an analytic sense. Thus, for example,
analyses of kinship terminology have depended largely on genea¬
logical criteria for several reasons: because of our own sanguine
ideology of kinship, the detail of terminologies has most frequently
been specified in genealogical terms, with only scattered informa¬
tion on other matters. Analyses often have tended to base them¬
selves on genealogy for the same ideological reasons, as well as
because of the nature of the data reported. Leach (1958), Beattie
(1964), Schneider (1964) and others have taken issue with this
genealogical and nuclear-family-based extensionist view of kin¬
ship, a view which underlies several of the analyses cited above and
which is specifically espoused by Lounsbury (1964; 1965).
But we must observe that formal semantic analyses are not
immediately concerned with the nature of ‘kinship’ as that word
may be understood by social anthropologists, but rather with
kinship terminologies as they may be found in ethnographic
sources and data and delimited by their users. Further, ‘kinship’
is obviously many things, not only to many analysts but also to
many as participants in their own social organization. True, we
cannot assume that genealogy must underlie ‘kinship’, but by the
same token we cannot reject all genealogical referents a priori or
because incautious workers have made translation errors. A set
of objects, such as the people grouped under a given kin term, can
be defined simultaneously by rules based on very different
characteristics; if these rules are in some way correlated, one kind
of rule can be used as a diagnostic of the others and some super¬
ordinate rule may state their correspondence. So far, genealogically
based analyses of kinship terminologies have provided the clearest
and most elegant solutions. That regularity cannot be dismissed
out of hand but must be justified, not only on the basis of formal
closure but also on that of inclusion in a wider theoretical frame¬
work. Whether it exists because genealogically based criteria do
underlie kinship, or because sociological, psychological, or other
kinds of data were inconsistent or insufficient, or because there is
some second-order correspondence between sociological and
genealogical variables is at this juncture moot. Not enough

E. A. Hammel 323
comparisons have been made. There are of course some welcome
exceptions, such as the paired analyses by Goodenough (1965)
and Schneider (1965), Lounsbury’s discussion of Leach’s work
(1965), Graburn (1964), Pospisil (1964), and Tyler (1966b). But
the task of sociological analysis is clearly more difficult than that
of the ‘genealogical’ analyst, largely because of the lack of ade¬
quate reference languages.
In this sense and for these purposes of semantic analysis of a
terminological corpus, the dispute over whether kinship ‘is’ or ‘is
not ’ genealogical is no dispute at all but a statement of alternative
strategies. If one is interested in the neatness, the closure, the
sufficiency of an analysis in a formal sense, then genealogically
based analyses clearly give the best results at this time. If one’s
concern lies in other areas, the reference language of analysis is
determined by the theoretical question posed. The problems of
formal analysis center now in improvement of the regularity of
analyses, in construction of full parallel analyses using different
reference languages, in expansion of the general method to other
analytic domains, and most importantly in asking why the pheno¬
mena should exhibit particular regularities or indeed any regular¬
ity at all. The answer to this last question demands sensitivity
to general theoretical issues, but it also demands one thing that
formal analyses have been able to provide better than any others,
regardless of the particular reference language involved - namely
that the regularities are in fact demonstrable. The primary virtue
of a formal analysis, whether on the grounds here outlined or any
other, unlike those analyses that make vague appeals to theories
of relevance, is that it may be subjected to the precise and public
criticism that is the hallmark of scientific communication.

References
Beattie, J. (1964), ‘Kinship and social anthropology’, Man, vol. 64,
art. 130.
Bloomfield, L. (1933), Language, Holt, Rinehart & Winston.
Boas, F. (1911), ‘Introduction’, Handbook of American Indian Languages,
Bull. 40, The Smithsonian Institution: Bureau of American Ethnology,
pt 1, pp. 1-83.
Bright, W., and Bright, J. (1965), ‘Semantic structures in Northern
California and the Sapir-Whorf hypothesis’, in E. A. Hammel (ed.),
‘Formal semantic analyses’, Amer. Anthrop., Special Issue, vol. 67,
no. 5, pt. 2, 3249-57.

324 Kin Terms

Burling, R. (1964), ‘Cognition and componential analysis: God’s
truth or hocus-pocus?’, Amer. Anthrop., vol. 66, pp. 20-8.
Burling, R. (1965), ‘Burmese kinship terminology’, in E. A. Hammel
(ed.), ‘Formal semantic analysis’, Amer. Anthrop., Special Issue, vol.
67, no. 5, pt 2, pp. 106-17.
Burling, R. (1970), ‘American kinship terms once more’,
Southwestern J. Anthrop., vol. 26, pp. 15-24.
Chomsky, N. (1957), Syntactic Structures, Mouton.
Conklin, H. C. (1962), ‘Lexicographical treatment of folk taxonomies’,
in F. W. Householder and S. Saporta (eds.), Problems in Lexicography,
Indiana University Research Center in Anthropology, Folklore, and
Linguistics, publication no. 21.
Frake, C. O. (1961), ‘The diagnosis of disease among the Subanun of
Mindanao’, Amer. Anthrop., vol. 63, pp. 113-32.
Gardin, J. C. (1965), ‘On a possible interpretation of componential
analysis in archaeology’, in E. A. Hammel (ed.), ‘Formal Semantic
analysis’, Amer. Anthrop., Special Issue, vol. 67, no. 5, pt. 2, pp. 9-21.
Gifford, E. W. (1916), Miwok moieties. University of California
Publications in American Archaeology and Ethnology, vol. 12, no. 4.
Goodenough, W. H. (1951), Property, kin and community of Truk,
Yale University Publications in Anthropology, no. 46.
Goodenough, W. H. (1956), ‘Componential analysis and the study of
meaning’. Language, vol. 32, pp. 195-216.
Goodenough, W. H. (1965), ‘Yankee kinship terminology: a problem
in componential analysis ’, in E. A. Hammel (ed.), ‘ Formal semantic
analysis’, Amer. Anthrop., Special Issue, vol. 67, no. 5, pt. 2, pp. 259-86.
Graburn, N. H. H. (1964), Taqagmiut Eskimo Kinship Terminology,
Northern Coordination and Research Centre, Department of Northern
Affairs and Natural Resources, Ontario.
Gumperz, J. J. (1958), ‘Dialect differences and social stratification in a
north Indian village’, Amer. Anthrop., vol. 60, pp. 668-82.
Hammel, E. A. (ed.) (1964), ‘Further comments on componential
analysis’, Amer. Anthrop., vol. 66, pp. 1167-71.
Hammel, E. A. (1965a), ‘Formal semantic analysis’, Amer. Anthrop.,
Special Issue, vol. 67, no. 5, pt. 2, pp. 1-7.
Hammel, E. A. (1965b), ‘A transformational analysis of Comanche
kinship terminology’, in E. A. Hammel (ed.), ‘Formal semantic analysis’,
Amer. Anthrop., Special Issue, vol. 67, no. 5, pt. 2, pp. 65—105.
Hammel, E. A. (1965c), ‘An algorithm for Crow-Omaha solutions’, in
E. A. Hammel (ed.), ‘Formal semantic analysis’, Amer. Anthrop.,
Special Issue, vol. 67, no. 5, pt. 2, pp. 118-26.
Harris, Z. S. (1948), ‘Componential analysis of a Hebrew paradigm’.
Language, vol. 24, pp. 87-91.
Hjelmslev, L. (1961), Prolegomena to a Theory of Language,
University of Wisconsin Press. Translated by F. J. Whitfield.
Hockett, C. F. (1947), ‘Problems of morphemic analysis , Language,
vol. 23, pp. 321-42.

E. A. Hammel 325
Hockett, C. F. (1955), ‘A manual of phonology’, Internat. J. Amer.
Linguistics, memoir 11.
Von Humboldt, W. (1836), Uber die Kawisprache, pt. 1, Berlin.
Hymes, D. H. (1962), ‘The ethnography of speaking’, in T. Gladwin
and W. Sturtevant (eds.), Anthropology and Human Behavior, The
Anthropological Society of Washington.
Jakobson, R. (1928), ‘Quelles sont les methodes les mieux appropri£es
a un expose complet et practique de la phonologie d’une langue
quelconque?’, Actes du ler Congris International de Linguistes,
Paris.
Kroeber, A. L. (1909), ‘Classificatory systems of relationship’, /.
roy. anthrop. Inst., vol. 39, pp. 77-84.
Kroeber, A. L. (1952), ‘Introduction to “Classificatory systems of
relationship”’. The Nature of Culture, University of Chicago Press.
Lamb, S. M. (1964a), ‘The sememic approach to structural semantics’,
Amer. Anthrop., vol. 66, no. 3, pp. 57-78.
Lamb, S. M. (1964b), ‘On alternation, transformation, realization and
stratification’, Monogr. Lang. Ling., Georgetown University Institute
of Languages and Linguistics, no. 17, pp. 105-22.
Leach, E. R. (1958), ‘Concerning Trobriand clans and the kinship
category tabu ’, in J. Goody (ed.), The Development Cycle of Domestic
Groups, Cambridge Papers soc. Anthrop., no. 1.
Lounsbury, F. (1956), ‘A semantic analysis of the Pawnee kinship
usage’, Language, vol. 32, pp. 158-94.
Lounsbury, F. (1964), ‘A formal account of the Crow- and Omaha-
type kinship terminologies’, in W. H. Goodenough (ed.), Explorations
in Cultural Anthropology, McGraw-Hill.
Lounsbury, F. (1965), ‘Another view of Trobriand kinship categories’,
in E. A. Hammel (ed.), ‘Formal semantic analysis’, Amer. Anthrop.,
Special Issue, vol. 67, no. 5, pt. 2, pp. 142-82.
Lowie, R. H. (1930), ‘The Omaha and Crow kinship terminologies’
in Grossman and A. B. Antze (eds.), Verhandlungen des 24.
Internationalen Amerikanisten Kongresses, Friederichsen, De Gruyter,
pp. 103-7.
Lowie, R. H. (1932), ‘Kinship’, Encyclopedia of the Social Sciences,
Macmillan, vol. 8, pp. 568-72.
Morgan, L. H. (1871), Systems of Consanguinity and Affinity of the
Human Family. Smithsonian Institution Contributions to Knowledge,
no. 17.
Murdock, G. P. (1949), Social Structure, Macmillan Co.
Pospisil, I. (1964), ‘Law and societal structure among the Nunamiut
Eskimo’, in W. H. Goodenough (ed.), Explorations in Cultural
Anthropology, McGraw-Hill.
Radcliffe-Brown, A. R. (1941), ‘The study of kinship systems’,
J. roy. anthrop. Inst., vol. 71, pp. 1-18.
Rivers, W. H. R. (1914), Kinship and Social Organization, London
School of Economics.

326 Kin Terms

Romney, A. K. (1965), ‘Kalmuk Mongol and the classification of lineal
kinship terminologies’, in E. A. Hammel (ed.), ‘Formal semantic
analysis’, Amer. Anthrop., Special Issue, vol. 67, no. 5, pt. 2, pp. 127-41.
Romney, A. K., and D’Andrade, R. G. D. (1964), ‘Cognitive aspects of
English kin terms’, Amer. Anthrop., vol. 66, no. 3, pt. 2, pp. 146-70.
Sand ay, P. R. (1968), ‘The “psychological reality” of American English
kinship terms: an information-processing approach’, Amer. Anthrop.,
vol. 70, pp. 508-32.
Schneider, D. M. (1964), ‘The nature of kinship’, Man, vol. 64,
art. 217.
Schneider, D. M. (1965), ‘American kin terms and terms for
kinsmen: a critique of Goodenough’s componential analysis of
Yankee kinship terminology’, in E. H. Hammel (ed.),
‘Formal semantic analysis’, Amer. Anthrop., Special Issue, vol. 67,
no. 5, pt. 2.
Seligman, B. Z. (1917), ‘The relationship system of the Nandi, Masai
and Thonga’, Man, vol. 17, art. 46.
Tax, S. (1937), ‘Some problems of social organization’, in F. Eggan
(ed.). Social Anthropology of North American Tribes, University of
Chicago Press.
Tyler, S. A. (1966a), ‘Parallel/cross: an evaluation of definitions’.
Southwestern J. Anthrop., vol. 22, pp. 416-432.
Tyler, S. A. (1966b), ‘Context and variation in Koya kinship
terminology’, Amer. Anthrop., vol. 68, pp. 693-707.

E. A. Hammel 327
Part Eleven Fictive Kinship

The study of kinship is not limited to biological relations, nor

yet to those relationships that different societies define
genealogically. In most societies there are a number of other
relationships that are seen as approximating to kinship, but
which stand somewhat apart from the main set. These
relationships are seen either as a kind of sibling relation (e.g.
blood-brotherhood) or as a kind of parent-child relation
(e.g. adoption, fostering, ritual co-parenthood, step-parenthood,
etc.). Such relationships are not only of widespread significance
for particular societies, but are receiving increasing attention for
theoretical reasons, since their study is seen to define ‘the
nature of kinship * itself. Esther Goody lays out a paradigm
for the study of fictive kinship, in so far as parent-child
relationships are concerned. Mintz and Wolf offer a historical
and comparative analysis of ritual co-parenthood (‘god-parents ’),
an institution that was widespread in Mediterranean Europe
and has become of great significance in Latin America.

329
 23 E. N. Goody

Forms of Pro-Parenthood: The Sharing and Substitution

of Parental Roles

A paper specially written for this volume.

Parenthood is institutionalized in every known society. Parent-

child relationships are thus among the most ubiquitous of all
roles; every society has these to build on, and every society does so,
albeit in widely differing ways. Broadly speaking, the parent-child
relationship faces two ways. Looking inward, it constitutes one
axis of the ‘nuclear’ family, i.e. a couple and their children. While
a very few societies are so organized that the nuclear family is not
at any time a residential unit, none exists without at least one
parent-child dyad (usually mother-child) as a localized focus of
kinship and domestic institutions. Extreme dependency and abso¬
lute power characterize child and parent respectively at the begin¬
ning. As the child matures the balance becomes less unequal, and
may even invert if the parents reach old age. Intense, often ambi¬
valent, feelings are also characteristic of the internal (and internal¬
ized) aspects of parent-child relations.
But every family is set in a nexus of other institutions: economic,
political, religious and those arising from ties between neighbours.
In its outward-facing aspect, parenthood serves to prepare the
child for participation in these institutions, and more than this,
the parents tend to mediate entry into the sphere of economic,
religious and political activity. Initially it is as his father’s son, or
her mother’s daughter, that the child acts and is perceived by
others.
Given that parental roles are universally institutionalized and
are central both to the internal, domestic system and in mediating
between this and other institutions, it is hardly surprising that
many societies use the parent-child dyad as a model for other
relationships. I want here to discuss some of the main ways in
which the peculiar characteristics of parenthood have been adapted

E. N. Goody 331
to forge new links between people and to reinforce existing ones. I
shall argue that there are several kinds of cement which together
bind parent and child in this uniquely durable relationship. Of the
various pro-parental institutions considered, some make use of one
kind and some of another, and a few of more than one. But in
each case, the choice of social cement is directly related to the
functions which the institution serves.
The parent-child relationship is multi-bonded because of the
many functions which it fulfils. Where the child grows to maturity
with his biological parents, they fulfil all of the following role
elements: genitor/genetrix; source of status identity (pater/mater);
nurse; tutor in moral and technical skills; and sponsor in the
assumption of adult status. The rights, obligations and experi¬
ences associated with each role element generate a characteristic
bond between parent and child. The bond between a parent and his
or her biological offspring is recognized in most, but not all,
societies. This alone is not available for transfer or delegation, and
hence only the biological parents can fill the full set of parent-child
role elements.
In most societies the provision of status identity is based on birth
to ‘legally married’ parents, and passes through one or other
parent to the child. There are a few societies (e.g. the Lozi and
the Gonja) where assumption of a full set of adult statuses is not
linked to the marriage of the parents, so long as certain prohibitions
are respected.1 But however a given society defines the conditions

1. E. Kathleen Gough uses the concept of ‘birth status rights’ to mean

‘... all the social relationships, property-rights, etc. which a child acquires
at birth by virtue of his legitimacy, whether through the father or through
the mother’ (1959, p. 32). Among the Gonja (E. Goody, 1962; 1970) and
Lozi (Gluckman 1950), these rights are based on physiological paternity
(i.e. no distinction is made between social and physical paternity) and they
do not depend on marriage. They are, nonetheless, defined in terms of who
the parents, father and mother, were. It is the designation of adult status
rights in terms of the parents ’ identity that I refer to as ‘ birth status identity ’.
In some societies full birth status identity only accrues to children born in
marriage. In other societies this is not the case.
Of course in matrilineal societies, like the Nayar, the major status reci¬
procities will link a child and his matrikin, and not involve the father.
Typically, bestowal of birth status identity follows from the formal identi¬
fication of the mother’s genetricial powers with her lineage in an ‘initiation’
or ‘puberty’ rite and does not depend on the marriage of the parents.

332 Ritual and Fictive Kinship

under which it may be transmitted, full birth status identity carries
with it rights and obligations (a share in, or management of
property, worship of shrines, often burial of the previous status-
holder) which may be termed status reciprocities.
Analytically separate from these are the reciprocities of rearing.
These arise from the two phases of rearing: care and nurturance in
infancy and early childhood, and education, both moral and
technical, which characterizes later childhood and adolescence.
Rearing reciprocities are based on the ambivalent affect which is
generated by the dependency and discipline of the phases of
nurturance and education. Characteristically, they depend for
their strength on moral obligations, unlike the status reciprocities,
which tend to be strongly sanctioned, and can fairly be classed as
jural in nature.2
Sponsorship consists in the provision of a youth with the posi¬
tion and resources necessary for assumption of adult status. Just
what is required varies widely from one society to another. Most
often it is a spouse, and frequently this in turn entails the provision
of bridewealth or dowry and often also the resources from which
to support the new family. In rural Ireland a house and farmland
must be available before the ‘boy’ can marry; a Yak 6 youth must
have access to land for his yam farm; a Fulani husband cannot
bring his wife and child to live with him until enough cattle are
available to provide them with milk.
The protege in return is usually expected to acknowledge him¬
self subordinate to his sponsor, perhaps expressing this by living
with him, or by contributing economic assistance. Or it may suffice
that he serve his sponsor by granting him public respect, or as a
political follower. As this suggests, the sponsor-protege relation¬
ship not only provides for the transition between minority and
adulthood, but extends the filial relationship into the adult world,
into the external system. For the protege enters the wider social
system by making use of a share of his sponsor’s resources and
position in it.

2. Any definition of jural rights which is independent of the institutions of

a concrete social system must be framed in extremely general terms. By jural
rights and obligations are meant those which are accepted as the basis for
claims in respect of ownership, inheritance and succession and membership
in restricted groups, and to services based on membership in such groups.

E. N. Goody 333
 It is very often the case that the status identity conveyed
through the pater carries with it rights to position and resources
needed to achieve full adulthood. Here the role of sponsor and
pater are merged. But there are a number of societies where this is
not so, usually because birth status is narrowly defined. In Euro-
American society, for instance, the training received in adolescence
determines to a large extent the position to be occupied in adult¬
hood, and the intervention of a well-to-do sponsor who provides
for a good education can completely transform the opportunities
available to a working-class youth. This same phenomenon is seen
in present-day Ashanti where either the father or the mother’s
brother pays for secondary schooling and university on the basis
of which a man is enabled to enter the educated elite. Significantly,
in Ashanti, the protege incurs an obligation to support other mem¬
bers of the family later on. Or to take a somewhat different case,
a Yako youth who lives with his mother and her kin has the right
to farmland which enables him to marry, despite the fact that land
is usually reserved for agnates. But he may use this land only so
long as he lives with and assists his matrikin (Forde, 1963). Per¬
haps the Tallensi case is the clearest of all. Here too a man can live
with and farm on land belonging to his maternal kin. But so firmly
is status identity tied to social paternity that he may never ‘ own ’
this land nor serve his mother’s lineage ancestors. He has these
rights only in patrilineal land and ancestors. Matrikin can sponsor
him socially and economically, but they are unable to make him
fully one of themselves (Fortes, 1949).
By far the most common situation that we find is for all five
of these parental roles to be played by the same persons. When this
happens the bond of begetting between genitor and biological
offspring is combined with the status reciprocities linking pater and
filius, with the reciprocities of rearing which unite nurse, and later
tutor, and child, and finally, with the ongoing obligations of sub¬
ordination and assistance required by the sponsor of his protege.
This is the ‘ordinary’ parent-child relationship, and it is very
strongly cemented indeed.
It is, I suggest, the very over-determination of the ordinary
parent-child relationship which accounts for the widespread
occurrence of pro-parental institutions. For one or more elements
of the total set can be split off and used to create a fictive parent-

334 Ritual and Fictive Kinship

child tie without dissolving the remaining links between true
parent and child. And indeed, in many of these institutions it is
the continued existence of the links between true parent and child
which give meaning to the reassignment of other parental roles.
Which of the possible roles is transferred and which retained will
determine the type of links between true parent and child, and
between pro-parent and child. Together with other factors, such
as their respective positions in the wider social system, the distri¬
bution of parental roles will also determine the relationship between
true parent and pro-parent themselves. In order to illustrate how
this works, three pro-parental institutions will be very briefly dis¬
cussed: these are fosterage, ritual sponsorship and adrogation.
Each will be shown to depend in a different way on the splitting of
parental role elements between two sets of parents. Finally, we
will consider the very different situation of adoption, where the
substitution of one full set of parental roles for another, rather than
the sharing of roles, is the object.

Fosterage
The verb ‘to foster’ is derived from the Old English root fod, for
food. To foster is ‘to nourish, feed, or support; to cherish; to
bring up with parental care’ {Shorter Oxford Dictionary). In the
present context fosterage can be defined as the institutionalized
delegation of the nurturance and/or educational elements of the
parental role. Thus we would expect the ties which link foster
parent and foster child to be essentially affective and moral in
character, based as they are on the reciprocities of rearing.
For some purposes it is also convenient to distinguish between
fostering which is initiated of necessity, typically because the natal
family of orientation has dispersed due to death or divorce, and
fostering entered into voluntarily. Such a distinction has nothing
to do with which parental role elements are delegated, but is only
concerned with the basis on which the original decision to foster
was made.3

3. A further distinction which I have developed elsewhere (E. Goody,

1968) turns on the difference in nurturance requirements of infants and
young children on the one hand and older children and adolescents on the
other. In the first case rearing means providing food and care and early
socialization in control of impulses and bodily functions. This can con-

E. N. Goody 335
 Equally important in defining fosterage is the designation of
what it is not. As used here, fosterage does not effect the status
identity of the child, nor the jural rights and obligations this
entails. Fosterage concerns the process of rearing, not the jural
definition of status or relationships.

The fostering of children by kin in Gonja

The Gonja of northern Ghana do not have the segmentary unilineal
descent groups which characterize many African societies, and
rights in children are held, and exercised, by kin of both parents.
It is these joint rights in children which are made explicit in the
institution of fosterage. Roughly one half of the adults in five
samples had been reared during childhood and adolescence by kin
other than their true parents.* * * 4 A would-be foster parent may claim
a sibling’s child or a grandchild at its birth, or later during infancy
or early childhood. The child remains with the biological parents
until the age of five or six when it joins the foster parent in his or
her home, sometimes in the same village, but often in a distant
one, there to remain until ready to marry, except for visits home
from time to time.
The Gonja do not think of physiological parenthood as separ¬
able from social parenthood. They do not distinguish between
pater and genitor; both are determined by conception. Nor can
either element be transferred by legal fiction. The act of begetting
permanently fixes the status identity of the child, and with it his
eventual eligibility for political and many ritual offices. Nor is the
nurturance of the infant and young child delegated unless a crisis
renders this unavoidable. It is the education of the older child and

veniently be termed nurturant fosterage. The older child also requires food
and shelter, but the primary task of foster parents at this later period is
training in adult role skills and the values of the society. Here the terms
educational or apprenticeship fostering are appropriate, depending on the
emphasis given to the institution in a given situation. The institution of the
wet nurse in medieval Europe is a type-instance of nurturant fosterage, while
apprenticeship as commonly practised in the same period provides an
illustration of educational fosterage which merges in many cases with
sponsorship as defined on p. 333.
4. The proportion in a sixth sample from a different part of the country
was much lower. See E. Goody (1970), for an account of Gonja fostering.

336 Ritual and Fictive Kinship

 adolescent that is transferred to kin, and the Gonja are quite
explicit about this. ‘It is so that the child will have a good character
that we find another to rear (belo) him.’ But it is also held to be
important that every child knows both his mother’s and his
father’s kin, and their respective home towns. This too is accom¬
plished by the dispersal of the sibling group between true parents
and their relatives in fosterage. The fostered child retains loyalty to
and affection for his own parents and siblings, but adds to this a
deep involvement with the kin of the foster home. In this way the
sibling group, the members of which remain in close contact
throughout their lives, includes persons who are literally at home
with the different branches of the family. In the terms outlined
above, respect for the genitor, status reciprocities and early rearing
reciprocities are firmly anchored on natal kin, while the rearing
reciprocities of the educational phase, and in some cases the
obligations of protege towards his sponsor, are located in the
foster parent/foster child relationship.
It is characteristic of Gonja fosterage that foster parent and
foster child often belong to different social estates, youths of the
ruling group being, for instance, reared in commoner households,
and vice versa. There is thus no clear pattern of using foster child¬
ren to build up a following, perhaps in part because sons are ex¬
pected always to return eventually to their own fathers’ households.
Nor was it possible to show that fostering provided a direct avenue
of advancement in the acquisition of either traditional or modern
office (E. Goody, 1970, p. 65 ff.). On the whole it seems best to
accept the Gonja view of fostering as arising out of the joint claims
kin have on one another’s children, as serving several specific
functions (companionship, domestic assistance, help with farm
chores, and particularly character training) and as linking together
socially and physically dispersed kin through their common bonds
with children. That is, fostering here operates within a framework
of fixed status identity and the jural reciprocities on which this is
based. It is the affective and moral reciprocities of rearing which
are manipulated, here shared between two sets of parents.

Ritual sponsorship
The historical development of ritual co-parenthood (<compadrazgo)
has been well documented by Mintz and Wolf (1950), who also

T-K-M
E. N. Goody 337
consider the various functions of this institution in Europe and
Latin America. They see co-parenthood as the creation (by ritual
and legal fictions) of obligations similar to those imposed by
parenthood. Essentially these involve the conversion of non-kin
relationships into quasi-kin ties complete with constraints on
marriage and sexual relations, the prohibition of quarrels and the
obligation to provide economic assistance. Where ritual co¬
parents are selected from among existing kin this serves to impose
specific obligations where previously diffuse ones obtained. Mintz
and Wolf point out that co-parenthood can be based on lateral ties
within a single social stratum in such a way that cohesion within
the group is reinforced. But it can also be used to create a lien on a
member of an adjacent stratum. The latter form allows the
development of links between otherwise independent elements -
farmer and shopkeeper, field-hand and foreman. Here an element
of status imbalance is involved, deference being traded for social
and economic advantage.
A second characteristic of ritual co-parenthood is that the effec¬
tive ties are between parent and co-parent, with their common child
a relatively passive member of the triad. This illustrates well the
manipulative aspect of the institution. Ostensibly its purpose is to
ensure the ritual well-being of a child. Yet repeatedly it has been
socially elaborated in such a way that the child is the least import¬
ant person involved. What element of the parental role complex is
being used here ? Clearly there is no alteration of the identity status
of the child, and the central status reciprocities are not affected.
Similarly, unless the parents cease for some reason to be able to
care for the child, neither the nurse nor the tutor element is dele¬
gated. The child remains with the true parents who raise it, and
with whom the affective and moral obligations of the rearing
reciprocities are generated. There remains the role of sponsor. The
special characteristic of this role is its mediation between the child
and the external system. Originally meant to mediate between the
child and the supernatural, there appears to be a recurrent ten¬
dency for the link between ritual co-parent, the child, and its own
parents to be extended into other spheres. Economic cooperation
is common between co-parents, and so also is the stating of claims
to respect, importance and assistance. Here the prominence of the
co-parents and the passivity of the child become more understand-

338 Ritual and Fictive Kinship

able, for children cannot operate effectively in the external system,
although others can, and do so, on their behalf.
Like fosterage, co-parenthood seems designed for working in
the gaps between the formal elements of the system. Neither
institution affects jurally defined status identity, yet both make use
of elements of the parental role to create reciprocities which form
the basis of links between individuals and groups.

Ritual sponsorship in the Balkans: komstvo

A recent analysis of ritual sponsorship in the Balkans (komstvo) by
Hammel (1968) provides an illustration of how the links based on
sponsorship can be institutionalized to form continuing links
between families. As Hammel describes it, komstvo has significance
in determining the relative status of local descent groups, in that a
sponsor is sought from an equal or higher status group, never one
of lower position. Further, the ‘giving’ of komstvo, that is, the
offering of the sponsorship role, conveys prestige, and this is acted
out at each occasion on which the sponsor is present, for he is
greatly deferred to and given precedence in ceremonies (especially
in southern and eastern Serbia). Hammel is further able to show that
komstvo links tend to be repeated in successive generations and are
seen as affecting relations between kin groups in the same sort of
way as ties of marriage and of descent. However, in komstvo,
as with the compadrazgo complex, sponsorship, and indeed only
ritual sponsorship, is the single parental role element to be trans¬
ferred to the pro-parent; only by means of this link are joint in¬
terests created and maintained between the two sets of parents.
And even more clearly than with compadrazgo, komstvo relations
are extra-domestic, affecting as they do the relative statuses of
local descent groups.

Adrogation
The final parental role element to be discussed is the provision of
status identity. In order to avoid confusion, I suggest that the
Roman term adrogation be used where the main purpose of creat¬
ing Active parenthood is the substitution of a new status identity
for that which was acquired at birth. An extreme, and a very clear,
example of this sort of re-allocation of the status-bestowing aspect
of parenthood occurs in the adoptions by will which were accepted

E. N. Goody 339
practice among the Roman nobility. It was in this way that
Augustus became the ‘son’ of Julius Caesar. This may have been
technically considered as ‘ inheritance on condition of taking the
testator’s name’ (Crook, 1967, p. 112). But either way it illustrates
how status identity can be altered without affecting the other
parental role elements.

Roman ‘ adoption’ and alliances based on adrogation

‘Adoption’ in Republican and Imperial Rome was basically a
legal and ritual transaction by which a person passed from one
potestas (legal authority) to another. ‘It had nothing to do with the
welfare of children’ says J. A. Crook, but was ‘the characteristic
remedy for a family in danger of dying out ’ and \ .. those adopted
were often adults’ (1967, p. 111). Similarly, it has been argued that
‘mainly the institution (adoption) was one whereby the great
families provided themselves with heirs to their property and
worship, successors to office or a political following’ (J. Goody,
1969, p. 60).
The evidence for adoption in ancient Rome concerns the
nobility almost exclusively. They had not only vast fortunes to
transmit, but also the splendid reputations of their ancestors,
whose wax effigies were celebrated at each new funeral and to
whose descendants the consulship was effectively restricted. Under
such circumstances it is hardly surprising that there were fathers
willing to lose their sons in adoption. That it was a loss is indicated
by the fact that the son ceased to be a legal member of his natal
family, and should the other members all die, the adopted son
could not return to save it from extinction.
Yet there were advantages. In a study of party politics at the
end of the Republic, L. R. Taylor writes
To unite families, adoptions were as important as intermarriages . ..
By such adoptions the two (noble) houses were brought closely to¬
gether, and the man adopted was thought to belong to both families.
Thus the younger Scipio Africanus, son of Aemilius Paullus . . . speaks
of both his fathers, and is proud of the family tradition of both his
houses (Taylor, 1966, p. 34).
There is an apparent contradiction in these accounts, for on the
one hand both Taylor and Crook speak of the fact that the
adopted son must renounce any legal or religious link with his

340 Ritual and Fictive Kinship

natal family, and yet both also refer to the use of adoption to
create an alliance between two noble houses. The key to the prob¬
lem lies in the fact that, as both writers again note, adoption often
concerned adult men rather than children. In terms of the distinc¬
tions we have been using, the reciprocities of rearing, and of course
the bond of begetting, are already firmly forged between the son
and his natal kin. There remains the transfer of status identity so
explicitly arranged by the Romans, and in some cases the sponsor¬
ship of the young adult.5 The bonds of parenthood are again split
between two families, in this case in line with the distinction we
have made between jural and moral reciprocities. Significantly, the
adopting father did not seek an infant or young child with whom
he might expect to fill an almost complete set of paternal roles.
Instead he often appears to have preferred a son for whom, since
he was already adult, the nurturant and educational aspects of
parenthood were closed. Such a son combines in his person loyal¬
ties to two great families; his adoption secures the basis for a lasting
alliance as well as an heir to name, traditions and estate.
The characteristic feature of Roman adoption, then, is the con¬
cern to substitute one legal status for another. Adoption of a
jurally independent person, one whose father and father’s father
were dead, was referred to by the distinctive term, adrogation. For
adrogation meant the disappearance of the separate identity of the
adrogee and his descendants, and each instance apparently re¬
quired a special law (Taylor, 1966, pp. 88, 91). There was not a
direct correspondence between adrogation and the adult status of
the new son in ancient Rome, since children might be orphaned,
and adults remained in the potestas of their father until his death.
However the term does correctly emphasize the substitution of
status-identity independently of the reassignment of other parental
roles. In addition, in Roman usage, it commonly was adults who
were ‘adopted’, and their persisting ties of affection and duty
towards their natal kin were utilized to forge alliances central to

5. The peculium was a sum of money placed by his pater at the disposal
of the adult who was still under potestas and thus unable legally to own
property on his own account. It fits exactly the functions we have indicated
for sponsorship, allowing the young man to establish his own household,
and thus gain a measure of independence even though legally a jural
minor.

E. N. Goody 341
political life, while their recruitment to the adopting family insured
the legal continuity of its reputation and estate.

Adoption
Adrogation is seldom institutionalized, as the special conditions
which render it advantageous are relatively rare. It is far more often
the case that where an heir to office, property or shrines is needed
those concerned hope also to find a child whom they may rear and
train, and who will love and respect them as well as observe the
legal formalities of status reciprocities. While the bond of begetting
is not transferable, we have seen that the remaining parental role
functions may be filled by others besides the biological parents.
I suggest it is where the complete set of available parental role
elements is transferred from natal to pro-parents, that the term
adoption is best employed. This has the advantage of consistency
with popular usage. But more than that, it allows us to distinguish
institutions which depend for their effectiveness on the sharing of
parental role elements, from the case where ‘total parenthood’ is
renounced by one couple and assumed by another, i.e. adoption.
Of course ‘total parenthood’ is not transferable. For the fact
of physiological parenthood is not open to social manipulation.
But the desire for a total transfer can be inferred in respect to
adoption in western society from the efforts made to keep the
identity of the real parents secret. Writing about contemporary
America, V. Carroll remarks on the concern with secrecy lest the
natural parents try to reclaim their child from its adoptive parents
(1970, p. 4). But he goes on to speak of the lengthy jural processes
which, once completed, render such an attempt illegal. The adop¬
tive parents appear to fear that the bond of begetting may prove
stronger than both the ties of affection and moral obligations
arising out of child rearing and the legal claims of status recipro¬
cities. Thus there is in adoption, as it is practised in our society, no
basis for the sharing of parental role elements, just because adop¬
tive parents are seeking to fill all these roles themselves. They want
a whole child, not part of one.
Eskimo adoption provides another instance of institutionalized
preference for total transfer of parental roles. For despite the
difficulties of rearing an infant apart from its mother, adoptions
were ideally arranged during the pregnancy and completed shortly

342 Ritual and Fictive Kinship

after the child’s birth. When, as sometimes happened, an older
child was orphaned, even close relatives were reluctant to adopt it
(Rousseau, 1970, pp. 5, 40). As with Western parents, the Eskimo
make every effort to secure rights to nurse the adopted child in
infancy, thus preventing the development of affective ties between
natural parents and their child which might later weaken the
attachment to the adoptive parents.

The care of orphans and destitute children

The definition of who is obligated to care for whom is a function
of both legal and moral norms, and these in turn are based on
rights in property and in persons. But inevitably, the definition of
what should be done to care for deprived children is related to the
existing modes of delegating parental roles. Thus, in Gonja, where
fostering by kin is voluntarily arranged, fostering is also used as a
way of caring for children if their natal family of orientation dis¬
perses. Kin who have a right to claim a child in voluntary foster¬
age, have a duty to accept the role of foster parent in a crisis.6 But
in both cases it is the roles of nurse and tutor which are assumed,
with the identity of the sponsor more likely to be filled by the foster
parent if the true parents are dead. Since these roles are separated
from that of pater/genitor in voluntary fostering, there is no reason
to redefine the child’s jural identity when fostering is necessitated
by family crisis. In Gonja, rearing children has no effect on who
they are.
A similar situation typically obtains where one finds institutional¬
ized co-parenthood. The privilege of being a ritual sponsor carries
with it the obligation to provide assistance in crisis if this becomes
necessary. Ordinarily the co-parent is concerned only with partial
sponsorship, and not with either jural status or rearing. Should
his protege be orphaned the sponsor may take over his care and full
sponsorship obligations if kin are not available to fill these roles.
But the main function of the sponsor in crisis is to be an interested
party, to see that suitable arrangements are made, and to ratify
6. Because moral and legal rights and obligations seldom coincide
exactly, the individuals who act voluntarily and those who come forward
in a crisis may not be identical. The degree of fit is a matter for investigation
in any given society. In Gonja there is a high correspondence, and the in¬
stitutional charter is the same for both voluntary and crisis fosterage.

E. N. Goody 343
these as being appropriate. Again there need be no question of
redefining the jural identity of the orphan, although for other
reasons (e.g. that the role of pater is vacant, and the sponsor would
like to fill it) this may be done.
Adrogation, where jural status only is altered, does not serve to
provide for the care and training of an orphaned child. Adrogation
of an orphaned adult does, however, mean that there will be no
persisting ties to parents to act as the basis for claims on the adro-
gated son.
We have defined adoption as the transfer of all available parental
roles. As such it is clearly an effective way of providing for the
various needs of destitute children. But there is a special sense in
which adoption is appropriate in such cases. For by definition,
where the parents are dead or have actively rejected the child, they
present less of a threat to the adoptive parents’ monopolization of
parental roles. Adoption brings together those who need parents
and those who need children. But more than this, adoption of
orphans and illegitimate children brings together those who lack
someone to fill key parental roles, and those who wish to pre-empt
these roles completely. Potential conflict between natural and
adoptive parent is thus eliminated or minimized. Such a rigid
separation of natural and adoptive parents is the antithesis of the
pattern found in the institutions of kinship fosterage, ritual
sponsorship and adrogation.
Because in our own society all parental roles are concentrated
within the nuclear family we tend to see this as both right and
necessary. Yet these roles are potentially available for sharing
among kin or even with unrelated neighbours or friends. To do so
is one way of spreading the task of caring for deprived children.
But it is also an effective way of forging links between adults -
parent and pro-parent - and between generations - child/parent
and child/pro-parent. Where sharing of parental roles is institu¬
tionalized, as in the giving of foster children and ritual sponsorship,
such links are systematically created. To understand these as
simply ways of coping with crisis situations, or of arranging to
cope with them should they occur, is to fail to recognize the way in
which many societies make use of the unique strength of the bonds
between parent and child.

344 Ritual and Fictive Kinship

References
Arensberg, C. N., and Kimball, S. T. (1940), Family and Community
in Ireland, Harvard University Press.
Carroll, V. (ed.) (1970), Adoption in Eastern Oceania, Hawaii,
Bishop Museum.
Crook, J. A. (1967), Law and Life in Rome, Thames & Hudson.
Forde, D. (1963), ‘Unilineal fact or fiction: an analysis of the
composition of kin groups among the Yako’, in I. Schapera (ed.).
Studies in Kinship and Marriage, Royal Anthropological Institute.
Fortes, M. (1949), The Web of Kinship among the Tallensi, Oxford
University Press.
Gluckman, M. (1950), ‘Kinship and marriage among the Lozi of
northern Rhodesia and the Zulu of Natal’, in A. R. Radcliife-Brown
and D. Forde (eds.), African Systems of Kinship and Marriage,
Oxford University Press.
Goody, E. (1962), ‘Conjugal separation and divorce among the Gonja of
northern Ghana’, in M. Fortes (ed.). Marriage in Tribal Societies,
Cambridge University Press.
Goody, E. (1968), ‘Paradigm for the analysis of pro-parental
relationships’, unpublished report submitted to the Social Science
Research Council.
Goody, E. (1970), ‘Kinship fostering in Gonja: deprivation or
advantage?’, in P. Mayer (ed.), Socialization: The Approach of Social
Anthropology, Tavistock.
Goody, J. R. (1969), ‘Adoption in cross-cultural perspective’. Comp.
Stud. Soc. Hist., vol. 11, pp. 55-78.
Gough, E. K. (1959), ‘The Nayars and the definition of marriage’,
/. roy. anthropol. Inst., vol. 89, pp. 23-34.
Hammel, E. A. (1968), Alternative Social Structures and Ritual Relations
in the Balkans, Prentice-Hall.
Mintz, S. W., and Wolf, E. R. (1950), ‘An analysis of ritual co¬
parenthood {compadrazgo)', Southwestern J. Anthropol., vol. 6, pp.
341-65.
Rousseau, J. (1970), U Adoption chez les Esquimeaux Tununermuit,
Centre d’fitudes Nordiques, Universite Laval.
Stenning, D. J. (1958), ‘Household viability among the pastoral
Fulani’, in J. Goody (ed.). The Developmental Cycle in Domestic
Groups, Cambridge University Press.
Taylor, L. R. (1966), Party Politics in the Age of Caesar, University of
California Press.

E. N. Goody 345
24 S. W. Mintz and E. R. Wolf
Ritual Co-Parenthood (compadrazgo)

Excerpts from S. W. Mintz and E. R. Wolf, ‘An analysis of ritual

co-parenthood (compadrazgo)', Southwestern! Journal of Anthropology,
vol. 6, no. 4, 1950, pp. 341-65.

As anthropologists have been drawn into the study of Latin

American cultures, they have gathered increasing amounts of
material on the characteristic cultural mechanisms of compadrazgo.
This term designates the particular complex of relationships set up
between individuals primarily, though not always, through partici¬
pation in the ritual of Catholic baptism.
This rite involves, among its various aspects, three individuals or
groups of individuals. These are: first, an initiate, usually a child;
secondly, the parents of the initiate; third, the ceremonial sponsor
or sponsors of the initiate. It thus involves three sets of relation¬
ships. The first links parents and child, and is set up within the
confines of the immediate biological family. The second links the
child and his ceremonial sponsor, a person outside the limits of his
immediate biological family. This relation is familiar to most
Americans as the relation between godfather or godmother and
godchild. The third set of relationships links the parents of the
child to the child’s ceremonial sponsors. In Spanish, these call
each other compadres (Latin compater-commater, Spanish com-
padre-comadre, Italian compare-commare, French compere-com-
mere, German Gevalter-Gevatterin, Russian kum-kuma, etc.),
literally co-parents of the same child. The old English form of
this term, godsib, is so unfamiliar to most English-speaking people
today that they even ignore its hidden survival in the noun
‘ gossip ’ and in the verb ‘ to gossip ’. In English, as in the Ecuadorian
compadrear, the meaning of the term has narrowed to encompass
just one, if perhaps a notable characteristic of compadre relations.
Most other aspects of this relationship have, however, fallen by the
wayside. In contrast, in medieval Europe, the'compadre mechanism
was of considerable cultural importance, and in present-day Latin

346 Ritual and Fictive Kinship

America, its cultural role is attested by its frequent extensions
beyond the boundaries of baptismal sponsorship. [...]

Functional analysis
The Catholic ceremonial complexes, when carried to the New
World, were to develop under conditions very different from those
of fifteenth-century Europe. Alienation of Indian lands through
such devices as the repartimiento and the encomienda proceeded
concurrently with the wholesale conversion of millions of native
peoples to Catholicism. The functioning of such mechanisms as
compadrazgo in Latin American communities is strongly colored
by four hundred years of historical development within this new
setting. Yet there is little material on the cultural significance and
usages of the compadre mechanism during the colonial period.
Certainly considerable research needs to be carried out on the
processes of acculturation following early contact. Analysis of the
social functioning of compadrazgo in its American beginnings is
but a minor aspect.
Historical sources attest that baptism of natives had proceeded
from the time of first contact. Fray Toribio de Benavente writes
that in the fifty-five year period between 1521 and 1576 more than
four million souls were brought to the baptismal font (Gonzalez,
1943, p. 193). The evidence is good that emphasis was not on prior
instruction in the catechism, but rather on formal acceptance of
the faith. Father Gante and an assistant, proselytizing in Mexico,
claims to have baptized up to fourteen thousand Indians in a single
day. In all, Gante and his companion stated that they baptized
more than two hundred thousand souls in a single Mexican
province (Bancroft, 1883, p. 174).
Baptism was a sacrament designed to remove the stigma of
original sin. The acquisition of godparents purported to guarantee
to the initiate religious guidance during the years following his
baptism. Actually, Spaniards who were members of exploring
parties frequently served as sponsors for Indian converts, and thus
fulfilled but a formal ritual necessity (Espinosa, 1942, pp. 70 ff.).
We can assume that most of the social implications of the compadre
mechanism developed but slowly at first, if for no other reason than
this.
Yet the baptismal ceremony established an individual in the

S. Mintz and E. R. Wolf 347

Catholic universe, and perhaps by virtue of its symbolic simplicity,
it was readily accepted by many native populations. Redfield and
Villa R (1934), Parsons (1936), Foster (1948) and Paul (1942, pp.
79 ff.), among others, have sought to differentiate between abori¬
ginal and Catholic elements in the modern Latin American ritual.
Parsons, Redfield and Paul have felt further that certain deriva¬
tions of the modern godparental ritual have come from the
adaptation of this ceremonial form to pre-Columbian ceremonies
and social patterns. The Maya of Yucatan possessed a native
baptism so like the Catholic ritual that, according to one authority,

some of our Spaniards have taken occasion to persuade themselves and

believe that in times past some of the apostles or successors to them
passed over the West Indies and that ultimately those Indians were
preached to (Lopez Medel, quoted in Landa, 1941, p. 227).

The Aztecs also had a kind of baptism, and in addition, god¬

parents of sorts were chosen in an indigenous Aztec ear-piercing
ceremonial, according to Sahagun (1932, pp. 34-5). Paul feels that
there may even have been an aboriginal basis for the compadre as¬
pect of the complex in the existence of various kinds of formal
friendship among native peoples (1942, pp. 85-7).
But it is impossible to generalize about the ease with which
aboriginal ceremonial procedure could be accommodated to the
new sacrament, as endorsed by the Church. The most important
modern social result of the baptismal ceremony in practice - the
creation of a security network of ritual kin folk through cere¬
monial sponsorship - seems rather to be due to the institution’s
inherent flexibility and utility, than to any pre-existing pattern
with which the new complex might be integrated. Present-day
folklore concerning the fate of an unbaptized child (Redfield and
Villa R, 1934, p. 169; Parsons, 1945, p. 44; Paul, 1942) suggests that a
strong emphasis on the moral necessity for baptism was made from
the start. In modern practice, however, whether the people in a
given culture will feel that baptism requires the official approval
and participation of representatives of the Church varies consider¬
ably. The evidence is that once the secular utility of this sacred
institution was established, the native populations could count on
the fulfilment of these reciprocal obligations which godparentage
and compadrazgo entailed, the Church might not even be con-

348 Ritual and Fictive Kinship

suited. Makeshift ceremonies, consummated without orthodox
clerical approval, became so widespread as to be illegalized by
ecclesiastical ruling in 1947, except in cases where the child’s death
seemed imminent before official baptism.
The mechanism of godparenthood took shape originally as a
means for guaranteeing religious education and guidance to the
Catholic child. This aim was achieved through the ritual kinship
established between the newborn infant, its parents, and its god¬
parents, at the baptismal ceremony. The relationship frequently
was reinforced, or extended with new sponsors, at other life crisis
ceremonies, including confirmation and marriage.
From the original Catholic life crisis ceremonial sponsorship,
godparenthood has been elaborated in various Latin American
communities into the ceremonial sponsorship of houses, crosses,
altars, or carnivals (Gillin, 1945, p. 105), circumcision (Beals,
1946, p. 102), the future crop (Parsons, 1936, p. 228, n. 96),
commercial dealings (Zingg, 1938, pp. 717-18), and so on. Gillin
lists fourteen forms of compadrazgo for a single community (1945).
In certain cases, it cannot be said with any certainty whether the
new adaptation was developed locally, or constitutes a carryover
of some kind from some older European elaboration.
In general, ritual ties between contemporaries seem to have
become more important than those between godparents and god¬
children. This point is elaborated by Gillin in his discussion of the
Peruvian community, Moche. He writes:
The essence of the system in Moche is an ‘artificial bond’, resembling a
kinship relationship, which is established between persons by means of
a ceremony. The ceremony usually involves a sponsorship of a person
or material object by one or more of the persons involved, and the
ceremony itself may be rather informal. However in Moche it seems to
be placing the wrong emphasis to label the whole system ... ‘cere¬
monial sponsorship.’ ... The emphasis in Moche is upon the relations
between sponsors of an individual or thing, and between them and
other persons - in other words, relations between adults rather than
between adults and children or things (1945, p. 104).
While the custom derives primarily from a conception of spiritual
parenthood, modern Latin American emphasis seems to be rather
on ritual co-parenthood; the compadre-compadre relationship out¬
weighs the godparent-godchild relationship.

S. Mintz and E. R. Wolf 349

 The ritual complex has been demonstrated to be of so flexible
and adaptable a nature that a wide group of individuals can be
bound together ceremonially. Paul makes the points that the
mechanism of compadrazgo may be used either to enlarge numeri¬
cally and spatially the number of ritually related kin on the one
hand, or to reinforce already existing blood or ritual ties on the
other. These contrasting motives he calls ‘extension’ and ‘intensi¬
fication’ (Paul, 1942, p. 57). The authors of the present article feel
that whether the compadre mechanism will be used prevailingly to
extend or to intensify a given set of relationships will be determined
in a specific functional-historical context.
In modern Latin American communities, there is clear pattern¬
ing of choice. Compadres may be chosen exclusively from within
one’s own family, or perhaps blood kin will be preferred to out¬
siders. In other communities, on the other hand, one pair of god¬
parents may serve for all of one’s children, or compadres chosen
from outside one’s own family may be rigidly preferred. The
present writers are convinced that the rare usages of compadrazgo
in inheritance indicate the lack of utility of this mechanism in
dynamically affecting prevailing patterns of ownership. It is a
mechanism that can be used to strengthen existing patterns, but
not to change them. In the two cases in which compadrazgo plays
any role in determining land inheritance, land is held by the village
community, and all that is inherited is temporary right of use
(Wisdom, 1940; Villa R, 1945). Marital impediment under Canon
Law, a factor of continuing importance in much of the New
World (Herskovits, 1937, p. 98), and the selection of compadres
within the kin group or outside it, are also factors bound together
functionally and historically. This problem lies beyond the scope
of the present article.
Compadrazgo, once accepted by a social grouping, can be
moulded into the community way of life by many means. It is a
two-way social system which sets up reciprocal relations of variable
complexity and solemnity. By imposing automatically, and with a
varying degree of sanctity, statuses and obligations of a fixed
nature, on the people who participate, it makes the immediate
social environment more stable, the participants more inter¬
dependent and more secure. In fact, it might be said that the
baptismal rite (or corresponding event) may be the original basis

350 Ritual and Fictive Kinship

for the mechanism, but no longer its sole motivating force. Some
brief examples will demonstrate the institution’s flexibility.
In Chimaltenango (Wagley, 1949, p. 19) two compadres

will lend each other maize or money (‘as much as six dollars’).... Two
comadres should visit each other often and they may borrow small
things readily from one another. When one is sick, or when one has
just had another child, her comadre should come bringing tortillas for
the family, and she should work in her comadre's house Tike a sister.’

In Peguche (Parsons, 1945, p. 45), ‘white compadres are an

asset for anyone who has business in Otovalo or Quito.’
In Tzintzuntzan (Foster, 1948, p. 264),

On the economic level, the compadrazgo system forms a kind of social

insurance. Few are the families which can meet all emergencies without
outside help. Often this means manual help at the time of a fiesta, or
the responsibility of a carguero. Sometimes it means lending money,
which near blood relatives do not like to do, because of the tendency
never to repay a debt. But compadres feel obliged to lend, and no one
would have respect for a man who refused to repay a compadre.

In San Pedro de Laguna (Paul, 1942, p. 92),

The practical purpose motivating the selection of Ladinas as comadres
is the belief that they can cure infant illnesses and have access to the
necessary medicines. The Indians store no medicine. But the Ladinas -
by virtue of their cultural tradition and their greater income - custom¬
arily have on hand a number of drugstore preparations. The godparent
bond imposes on the Ladina the responsibility of coming to the medical
aid of her Indian godchild. The first year or two is correctly considered
to be the most critical period of the infant’s life. Hence the natives
sacrifice long-run considerations in favor of providing a measure of
medical protection during the infancy of the child. .. .

Evidence from studies of two communities in Puerto Rico

suggests that the compadre relation may be invoked to forestall
sexual aggressions (Wolf, 1950; Manners, 1950). Cases are
mentioned where a man concerned about the attentions of a
family friend to his wife, sought to avoid trouble by making his
friend his compadre. Thus a new and more sacred relationship was
established.
Among the Huichol (Zingg, 1938, p. 57), the compadre relation¬
ship

S. Mintz and E R. Wolf 351

unquestionably strengthens Huichol social organization outside the
family, which is not strong. Though compadres are not under economic
bonds to each other, the injunction to be kind and friendly prevents
drunken fights and brawls, which are the greatest source of weakness in
Huichol society.
One form of compadrazgo is
specifically organized to avoid aggression between two compadres: ‘el
compadrazgo de voluntad’. People say that where there are two bullies
in the same barrio, they will conclude a ‘non-aggression pact’ and make
themselves compadres de voluntad, which means that they can no
longer fight each other (Wolf, 1950).
The persistence of compadrazgo in very secularized contexts, and
its existence in such cases even without the sponsorship of a person,
object or event, is evidence of its frequently high social and secular
plasticity.
The formal basis for selecting godparents for one’s children -
religious guidance, and if necessary, the adoption of orphaned
children - is sometimes carried out. Gamio mentions this tradi¬
tional usage in the Valley of Teotihuacan (1922, vol. 2, p. 243),
Redfield and Villa R for Chan Korn (1934, p. 250), Villa R for Tusik
(1945, p. 90), Gonzales for the Mixe and Zoque (1943, pp. 204-5),
and Wisdom for the Chorti (1940, pp. 293-4). Among the Chorti,
the godfather often acts in every way as the actual father in the event
of the latter’s death. He gives his ward advice, gets him out of diffi¬
culties, sometimes trains him in a man’s work, and may act as his
parent when he marries. The same is done by the godmother for her
female godchild. If both parents die, and the godchild is young,
the godparent may receive the portion of the property which the child
inherited, and put it to his own use, in return for which he must bring
up the child as one of his own family. As soon as the young man or
woman becomes eighteen years of age, his inheritance is made up to
him by his godfather. Where there is more than one minor child, each
godfather receives his ward’s share out of the total property, each
child going to live in the home of its own godfather, leaving the adult
children at home.
This usage is of particular interest because the compadre
mechanism can be seen here as a link in the process of inheritance.
Yet final property rights in this society are vested in the village, and
not in the individual. A single case of the same kind of usage is
mentioned by Villa R for the Maya Indian community of Tusik

352 Ritual and Fictive Kinship

(1945, p. 90). Yet compadrazgo cannot override the emphasis on
group land tenure in either of these societies. The mechanism is
flexible and adaptable specifically because it usually carries with it
no legal obligations - particularly regarding inheritance. Paul
makes this point clearly when he writes that,

Unlike the involuntary ties of kinship those of ritual sponsorship are

formed on the basis of choice. This enables godparenthood to serve
as the social link connecting divergent income groups, disparate social
strata, and separated localities. Affinity too may cut across class and
locality through the practices of hypergamy and intermarriage. But the
frequency with which such irregular forms of marriage occur through¬
out the world is sharply limited by strong social pressures operating to
keep the unions within the class or community. This is understandable
in view of the fact that marriage is the means by which the in-group
perpetuates itself. Because no such considerations of social recruit¬
ment impede the formation of godparent bonds between persons of
different social strata, godparenthood more readily serves as a
mechanism for intergroup integration (1942, pp. 72-3).

It may be fruitful to examine cases of compadrazgo as examples

of mechanisms cross-cutting socio-cultural or class affiliations, or
as taking place within the socio-cultural confines of a single class.
The authors believe such pattemings will prove to be determined,
not haphazard in character, nor determined solely along con¬
tinuums of homogeneity-to-heterogeneity, or greater-to-lesser
isolation. Rather they will depend on the amount of socio-cultural
and economic mobility, real and apparent, available to an indi¬
vidual in a given situation. There is of course no clear-cut device
for the measurement of such real or apparent mobility. Yet the
utility of compadrazgo might profitably be examined in this light.
The aim would be to assess whether the individual is seeking to
strengthen his position in a homogeneous socio-cultural com¬
munity with high stability and low mobility, or to strengthen cer¬
tain cross-cutting ties by alignment with persons of a higher
socio-cultural stratum, via reciprocal-exploitative relationships
manipulated through compadrazgo. Some examples may illu¬
minate the problem.
The Maya Indian people of Tusik (Villa R, 1945), a community
in east central Quintana Roo, Yucatan, are homogeneous in a tribal
sense, rather than having a mono-class structure. Says Villa R:

S. Mintz and E. R. Wolf 353

There are no classes here in the sense that different groups of people
have different relations to the production and distribution of economic
goods; in the sense that some people own land on which other people
work, or that some people are engaged in producing goods while
others are engaged in distributing them or in servicing the rest of the
population. As we have already pointed out, everyone in the subtribe
has the same relation to the land as everyone else; the land is commonly
held by the subtribe, and a man’s rights to a piece of land rests only on
the right that he has put agricultural labor into the land and is en¬
titled to the products of his labor. Every man makes milpa - even the
sacred professionals earn their living as farmers - and since the secular
division of labor is practically nonexistent, there are no merchants or
artisans.

The economic life of the group centers about maize, and the
people consume all that they produce. Labor for other men is
rare, and when done, payment in kind prevails. The only cash
commodity is chicle. Says Villa R:

Apparently all the people of the subtribe enjoy the same economic
circumstances. Nothing one observes in their ordinary, daily behavior
suggests the existence of differences in accumulated wealth. . . . The
acquisition of wealth is related directly to the personal ambition of the
individual, for there are no differences in opportunity and no important
diflerences in privilege. The principal source of wealth is the extraction
of chicle, which is within the reach of all. . . . This equality of op¬
portunity is a recent matter, for some years ago when the chiefs had
greater authority, the lands of the bush were distributed by them and the
best portion preserved for their own use. In some cases men were thus
able to enrich themselves through special advantage.

Regarding compadrazgo, the grandparents of the child to be

born, prefeiably the paternal ones, are chosen. If they are not
alive, chiefs oi maestros cantores, as persons of prestige and good
character, are selected. It is noteworthy that no mention is made
of any choice ol travelling merchants as compadres, although

the travelling merchants are the natives’ main source of contact with
the outer world. It is they who bring into the region . . . the most
important news fiom the city. .. . The arrival of the merchant is the
occasion for the people to gather together and excitedly discuss the
events he relates to them, and in this atmosphere the merchant’s own
friendly ties with the natives are strengthened.

354 Ritual and Fictive Kinship

Chicle is sold, and commodities bought through these travellers,
but apparently ritual kinship is not used to bind them with the
community.
In marked contrast to the isolated, subsistence crop, tribal
culture of Tusik, we may examine two communities which exhibit
cultural homogeneity under completely different conditions. They
are fully integrated economically, and to a great degree welded
culturally into national cultures. The first of these is Poyal (Mintz,
1950).
Barrio Poyal is a rural community on the south coast of Puerto
Rico, in an area of large-scale sugar cane production, with
corporate ownership of land and mills. The lands are devoted
exclusively to the production of the single cash crop. While the
barrio working population forms what is practically a mono-class
isolate, compadres could be selected among the foremen, adminis¬
trators, public officials, store owners, and so forth. Instead, there
is an overwhelming tendency to pick neighbors and fellow workers
as compadres. A man who seeks a wealthy compadre in Poyal is
held in some contempt by his fellows; a wealthy compadre would
not visit him nor invite him to his house. People remember when
the old hacienda owners were chosen as godparents to the workers’
children, but this practice is totally outmoded now. A local land¬
owning group no longer exists in Poyal.
Compadre relationships generally are treated reverentially;
compadres are addressed with the polite Usted, even if they are
family members, and the compadre relationship is utilized daily in
getting help, borrowing money, dividing up available work
opportunities, and so forth. However, as more and more Poyal
workers migrate to the United States, the utility of many compadre
ties is weakened.
Another example of the same category is Pascua (Spicer, 1940),
a community of essentially landless, wage-earning Yaqui Indian
immigrants who, with their descendants, form a village on the
outskirts of Tucson, Arizona. The economic basis of Pascua life
bears certain striking resemblances to Barrio Poyal: the almost
total lack of subsistence activities, the emphasis on seasonal
variation, the emphasis on wage-earning as opposed to payment in
kind, and so on. Says Spicer:
Existence is wholly dependent on the establishment of relationships

S. Mintz and E. R. Wolf 355

with individuals outside the village. If for any reason the economic
relations of a Pascuan with outside persons are broken off for an
extended period, it becomes necessary to depend upon other Pascuans
who have maintained such relations.
While the economic linkages are exclusively with external sources
of income and employment, the compadre structure is described
as an all-pervasive network of relationships which takes into its web
every person in the village. Certain parts of the network, here and
there about the village, are composed of strong and well-knit fibers.
Here the relationships between compadres are functioning constantly
and effectively. Elsewhere there are weaker threads representing
relationships which have never been strengthened by daily recognition
of reciprocal obligations. These threads nevertheless exist and may
from time to time be the channels of temporarily re-established com¬
padre relationships.
Spicer notes that:
sometimes in Pascua sponsors are sought outside the village in Libre
or Marana, or even among the Mexican population of Tucson.
But everything suggests that the ritual kinship system here func¬
tions predominantly within the wage-earning, landless mono-class
grouping of the Yaqui themselves. Spicer’s description of com-
padrazgo is probably the most complete in the literature today, and
the Pascua system appears to be primarily between contemporaries
in emphasis, and as in Barrio Poyal horizontal in character.
These three cases, Tusik, Barrio Poyal, and Pascua, illustrate the
selective character of compadrazgo and some of its functionings,
within small ‘homogeneous’ groupings. The mechanism plainly
has considerable importance and utility and is treated reverentially
in all three places. Yet while Tusik is isolated and lacks a class-
character, Barrio Poyal and Pascua are both involved in wage¬
earning, cash crop, world market productive arrangements where
the homogeneity is one of class membership only, and isolation is
not characteristic.
In Tusik, compadrazgo is correlated with great internal stability,
low economic mobility, ownership of land by the village, and the
lack of a cash economy and class stratification. In Barrio Poyal
and Pascua compadrazgo correlates with homogeneous class
membership, landlessness, wage-earning, and an apparent growing
identity of class interest.

356 Ritual and Fictive KinshiD

 An interesting contrast is provided by Gillin’s study of Moche.
This is a Peruvian coastal community which, according to the
Foreword,

is in the last stages of losing its identity as an Indian group and of

being absorbed into Peruvian national life. ... Surrounded by large,
modem haciendas, Moche is ‘Indian’ only in that its population is
largely Indian in a racial sense, that it has retained much of its own
lands, that it exists in a certain social isolation from surrounding
peoples, retaining a community life organized on a modified kinship
basis, mainly of Spanish derivation. ... Its lands, however, are now
owned individually, and they are being alienated through sale and
litigation. It is on a cash rather than subsistence basis economically.
... Many Mocheros even work outside the community for wages,
and some are in professions. . . . Formal aspects of native social
organization have disappeared, and contacts with the outside world
are increasing (Gillin, 1945).

In Moche, the compadre system would expectably be subject to

the same stresses as those suffered by any other local social institu¬
tion. Yet

the whole idea of this type of relationship has been carried to extremes
in Moche. There are more types of padrinazgo [i.e. godfatherhood] in
this community than in any other concerning which I have seen
reports. This fact may be linked with the absence of spontaneous
community organization and solidarity.

Gillin finds evidence for fourteen different kinds of compadre

relations. As to the choice of compadres, Gillin says:

Godparents may be blood relatives, but usually the attempt is made to

secure persons who are not relatives of either of the parents. Not only
Mocheros, but in these days, trusted forasteros [i.e. outsiders] are
chosen. From the point of view of the parents it is desirable to choose
godparents who are financially responsible, if not rich, and also
persons who have ‘influence’ and prestigeful social connections. The
real function of godparents is to broaden, and, if possible, increase the
social and economic resources of the child and his parents and by
the same token to lower the anxieties of the parents on this score.

In a later section, however, Gillin states that he does not feel that
socially defined classes as such exist in Moche (1945, pp. 107,113).
It is extremely noteworthy that the mechanism of compadrazgo

S. Mintz and E. R. Wolf 357

has maintained itself here in the face of what appears to be pro¬
gressively accelerating social change. We wonder whether the
elaborations of the mechanism’s forms may be part of the com¬
munity’s unconscious effort to answer new problems. It must
increasingly face the insecurity of growing incorporation into the
national structure and increasing local wage-based, cash crop
competition. This may call forth an increased emphasis on tech¬
niques for maintaining and strengthening face-to-face relation¬
ships. Eggan’s study of Cheyenne kinship terminology (1937)
suggests that the kinship structure is sensitive to rapid social change
if the changing terminology reflects genuine structural modifica¬
tions. Ritual kinship structures may react to the weakening of
certain traditional obligations by spreading out to include new
categories of contemporaries, and therefore potential competitors.
Other examples suggest that vertical phrasings of the compadre
system may take place in situations where change has been slowed
at some point, and relationships between two defined socio¬
cultural strata, or classes, are solidified. San Jose is a highland
coffee and minor crop-producing community of Puerto Rico
(Wolf, 1950). The frequency distribution of land shows a con¬
siderable scatter, with 55 per cent of the landowners holding
10 per cent of the land at one extreme, and 5 per cent holding 45
per cent of the land at the other. Thus, while Tusik people hold
their land communally, Barrio Poyal and Pascua people are land¬
less, and Moche people are largely small landowners with no farm
over four acres, San Jose people are in large part landowners with
great variability in the size of holdings. While a large part of the
agricultural population is landless, agricultural laborers in San
Jose may be paid partly in kind, and frequently will be given in
addition a small plot of land for subsistence farming. Production
for wages is largely of the main cash crop, coffee.
In the rural zones of this community, a prevailing number of the
compadre relations tie agricultural workers to their landholding
employers, or small landholders to larger ones. Thus a large
landowner may become compadre to twenty smaller landowners
Jiving around his farm. In isolated areas, where the ‘community’
is defined entirely in familial terms, most compadre relationships
take place within the family. Yet it must be recognized that
members of the same family, and brothers of the same filial

358 Ritual and Fictive Kinship

 generation, may be variously landowners, sharecroppers (jmedia-
news), and laborers.
Compadrazgo in San Jose may help in the stabilization of pro¬
ductive relations between large and small landholders, or between
landholders and their sharecropping employees and laborers.
Interesting in this connection is the fact that the economic basis in
San Jose is much less exclusively cash than in Tusik, Barrio Poyal,
or Pascua. The land tenure pattern in San Jose does not appear to
be changing rapidly. Compadrazgo relations are phrased vertically,
so as to cross-cut class stratification, quite probably serving in this
connection to solidify the relationships of people to the land.
There is evidence of landowners getting free labor out of their
laborer brothers who have been made compadres. Contrariwise,
laborers bound by compadrazgo to their employers are accustomed
to rely on this bond to secure them certain small privileges, such as
the use of equipment, counsel and help, small loans, and so on.
The authors know of no fully documented study of compadrazgo
in the context of an ‘old-style’ plantation or hacienda. Siegel’s
material on the Guatemalan plantation community of San Juan
Acatan indicates that the Indians there often invite Ladinos with
whom they come in contact to sponsor the baptisms of their
children. But Siegel adds that the relationship in this community
is ‘virtually meaningless’ (quoted in Paul, 1942, p. 72). The
authors of the present article would in general predict that
plantation laborers, either bound or very dependent on the planta¬
tion, with daily face-to-face contacts with the owner or hacendado
would seek to establish a reciprocal coparental relationship with
the owner. Historical material from old informants in Barrio
Poyal offer evidence of this tradition, now markedly altered in the
pure wage, absentee ownership context.
The mechanism may be contrasted, then, in several distinct
contexts. In the first context are Tusik, Barrio Poyal and Pascua.
These communities are alike in their ‘homogeneity’, and the hori¬
zontal structuring of the compadre system; yet they are markedly
different in other respects. Tusik is tribal and essentially isolated
from the world market, while Barrio Poyal and Pascua are in¬
corporated into capitalistic world economies, and are fully formed
working class strata.
In the second context is San Jose, with its varied land ownership

S. Mintz and E. R. Wolf 359

pattern, its mixed (cash and subsistence) crop production and its
several classes. Through the vertical phrasing of its compadre
system, San Jose demonstrates a relatively stable reciprocity,
economic and social, between the landed, large and small, and the
sharecroppers and laborers.
In the third context is Moche. Land is held predominantly in
small plots; the crops, as in San Jose, are both cash and subsistence,
and while Gillin doubts the existence of classes, certainly the
compadre system is described as a vertical structuring one. Here,
too, the elaboration of face-to-face ceremonialism may help to
slow the accelerated trend towards land concentration, a cash
economy, and incorporation into the world market.

Conclusion
Thus in cases where the community is a self-contained class, or
tribally homogeneous, compadrazgo is prevailingly horizontal
(intra-class) in character. In cases where the community contains
several interacting classes, compadrazgo will structure such rela¬
tionships vertically (inter-class). Last, in a situation of rapid social
change compadre mechanisms may multiply to meet the acceler¬
ated rate of change.

References
Bancroft, H. H. (1883), History of Mexico, The History Society of San
Francisco.
Beals, R. L. (1946), Cherdn: A Sierra Tarascan Village, Smithsonian
Institution publication no. 2.
Eggan, F. (1937), ‘The Cheyenne and Arapaho kinship system’, in
F. Eggan (ed.), Social Anthropology of North American Tribes,
Chicago University Press.
Espinosa, J. M. (1942), Crusaders of the Rio Grande, Chicago University
Press.
Foster, G. N. (1948), Empire's Children: The People of
Tzintzuntzan, Smithsonian Institution publication no. 6.
Gamio, M. (1922), La Poblacidn del Valle de Teotihuacan, Secretaria de
Agricultura y Fomento, Mexico.
Gillin, J. (1945), Moche: A Peruvian Coastal Community, Smithsonian
Institution publication no. 3.
Herskovits, M. J. (1937), Life in a Haitian Valley, Knopf.
de Land a, D. (1941), Landa’s Relation de las Cosas de Yucatan, in
A M. Tozzer (ed.). Papers: Peabody Museum of American Archaeology
and Ethnology, Harvard University, vol. 18.
Manners, R. A. (1950), A Tobacco and Minor Crop Community in
Puerto Rico, unpublished ms.

360 Ritual and Fictive Kinship

Mintz, S. W. (1950), A Sugar-Cane Community in Puerto Rico,
unpublished ms.
Parsons, E. C. (1936), Mitla: Town of the Souls, Chicago University
Press.
Parsons, E. C. (1945), Peguche: A Study of Andean Indians, Chicago
University Press.
Paul, B. D. (1942), ‘Ritual Kinship: With Special Reference to
Godparenthood in Middle America’, doctoral thesis, University of
Chicago.
Red field, R., and Villa R, A. (1934), Chan Kom: A Maya Village,
Carnegie Institution of Washington, publication no. 448.
Rojas, G. F. (1943), ‘La institution del compadrazgo entre los Indigenas
de Mexico’, Rev. Mexicana Sociol., vol. 5, no. 1.
de Saha gun, B. (1932), A History of Ancient Mexico, Nashville.
Spicer, E. (1940), Pascua: A Yaqui Village in Arizona, Chicago
University Press.
Villa R, A. (1945), The Maya of East Central Quintana Roo,
Carnegie Institution of Washington, publication no. 559.
Wagley, C. (1949), The Social and Religious Life of a Guatemalan
Village, Memoir, American Anthropological Association, no. 71.
Wisdom, C. (1940), A Chorti Village in Guatemala, Chicago University
Press.
Wolf, E. R. (1950), A Coffee-Growing Community in Puerto Rico,
unpublished ms.
Zingg, R. (1938), Primitive Artists: the Huichols, University of Denver
Contributions to Ethnology.

S. Mintz and E. R. Wolf 361

Part Twelve Changing Patterns of Kinship

The kinship behaviour of simple hunters is clearly different

from that of groups practising irrigation agriculture. In the first
place, kinship is often used to channel the inheritance of
property, so that where the means of production change, kin
relationships are likely to do so as well. Equally, the industrial
process brings with it an individualization of the family income
and stress on spatial and social mobility, so that it is associated
with changes both of a demographic and of a structural kind.
It is particularly important to investigate these changes if we
are to understand kinship and family patterns in the developing
world; in many parts the kin groups of an earlier period are
being shattered by political, economic, religious and educational
changes, affecting a whole network of interpersonal behaviour
and patterns of dependence and co-operation. The classic work
on these changes is Goode’s World Revolution and Family
Patterns (1963), from which our final Reading comes.
/
25 W. J. Goode

World Changes in Family Patterns

Excerpts from W. J. Goode, World Revolution and Family Patterns,

Free Press, 1963, pp. 1-15.

For the first time in world history a common set of influences - the
social forces of industrialization and urbanization - is affecting
every known society. Even traditional family systems in such
widely separate and diverse societies as Papua, Manus, China and
Yugoslavia are reported to be changing as a result of these forces,
although at different rates of speed. The alteration seems to be in
the direction of some type of conjugal family pattern - that is,
toward fewer kinship ties with distant relatives and a greater
emphasis on the ‘nuclear’ family unit of couple and children. In
the Western Culture Complex - the New World, Australia, New
Zealand and Europe west of the Urals - where a conjugal family
system has already been in operation to some extent, the direction
of change seems to be toward its wider spread.
If it is true that the rough outlines of a conjugal system are
beginning to emerge in such disparate cultures as China and the
Arab world, we are witnessing a remarkable phenomenon: the
development of similar family behavior and values among much
of the world’s population. This inquiry will try to ascertain
whether and why, this process is indeed occurring.
Such an investigation raises a host of complex theoretical prob¬
lems. But important theoretical problems direct any fruitful study
of the facts, just as facts in turn serve to sharpen theory. Therefore,
an outline of the broad, relevant, theoretical issues should preface
our analysis of family changes in six major world cultures.

1. Even if the family systems in diverse areas of the world are

moving toward similar patterns, they begin from very diflerent
points, so that the trend in one family trait may differ from one

W. J. Goode 365
society to another - for example, the divorce or illegitimacy rate
might be dropping in one society but rising in another.
2. The elements within a family system may each be altering at
different rates of speed. While some were greatly strained under
the traditional system, others were buttressed by many institutional
supports; new influences, therefore, encounter more resistance at
some points in the family system than at others.
3. Just how industrialization or urbanization affects the family
system, or how the family system facilitates or hinders these
processes, is not clear.
4. It is doubtful that the amount of change in family patterns is a
simple function of industrialization; more likely, ideological and
value changes, partially independent of industrialization, also have
some effect on family action.
5. Some beliefs about how the traditional family system worked
may be wrong. Even to measure change over the past half-century
requires a knowledge of where these family systems started from.
We may learn, for example, that although some family patterns are
thought to be new, they have in fact always been part of the tradi¬
tional system - for example, the reported breakup of the Indian
joint family. Reports that the joint family system is breaking up
have been prevalent for sixty years. It now seems likely that al¬
though most Indian families are ‘joint’ at some phase in their life
cycle, they also go through a stage of breaking up when the
married sons move away to found their own households. Thus,
the ‘breakup’ is really one stage in the process of formation and
growth of individual families, and has characterized the Indian
family system for decades.
6. Correlatively, it is important to distinguish ideal family patterns
from re*?/family behavior and values, and it is especially necessary
to differentiate the social habits of upper-class families from those
of the majority of the society.
7. Finally, since the aim of research is precision and understanding,
it is necessary to examine carefully whatever numbers and counts
can be obtained in order to be sure that they are in fact descriptions
of reality and not accidents of poor recording procedures. Es¬
pecially when inquiring into the past, we must question the validity
of the data.

366 Changing Patterns of Kinship

 Those involved in the process of social change often exaggerate
its radical character. If public opinion polls had been taken in every
important historical period, they would probably have revealed
that the participants thought all the old traditions were dissolving,
although from our perspective the alterations of each period seem
minor. Western people, for example, tend to believe that morals
have deteriorated greatly over the past hundred years. However,
the illegitimacy rate has actually dropped slightly; and some
evidence suggests that the percentage of women who are pregnant
when they marry is about the same as it was in the past - although
perhaps not so high as it was two centuries ago.
Unfortunately, the data are not yet available for most of the
problems with which we are concerned here - a fact to be noted
when considering the opinions of social analysts, who tend to
assume that changes in economic or industrial systems are the
primary cause of alterations in family systems. This relationship
has been taken for granted, but has never been scientifically
demonstrated.
Social science developed only one comprehensive theory of
family change, one based on nineteenth-century evolutionary
ideas. (The science has since evolved beyond that theory without
developing a substitute.) Intended to explain both the observable
differences among the world’s family patterns and their history,
the theory asserted that in the course of man’s development the
family had ‘ progressed ’ from the primitive sexual promiscuity of a
semi-animal horde, through group marriage, matriarchy and
patriarchy in some polygynous form, to culminate in the highest
spiritual expression of the family, Victorian monogamy.1 But
growing philosophic doubts that the nineteenth-century family
and religious systems were indeed the highest levels reached by man
began to bring into question the validity of this sweeping theory. In
addition, as knowledge about human societies over the world
accumulated, the core of the evolutionary theory - that each
advance in technology was accompanied by an advance in family
patterns and religion - dissolved. No such correlation actually
emerged.
E. B. Tylor, who was otherwise so shrewd in his understanding of
1. For a concise statement of nineteenth-century family theory see Kdnig
(1955).

W. J. Goode 367
human societies, formulated the main technique being used in the
nineteenth century, the interpretation of certain social patterns,
such as ‘ marriage by capture ’ or the custom of doffing the hat as a
greeting, as ‘social survivals’ or social fossils (Tylor, 1888). He
claimed that just as we can ascertain what Eohippus was like from
his fossil remains, so can we learn what Stone Age man was like
from the survival of some of his social patterns - for example, the
relic of marriage by capture appears in some primitive societies in
the form of a mock fight between the kin of bride and groom before
the bride is carried away. Thus, by looking at Australian aborigines
or Polynesians, we can learn about Stone Age man; and the Haw¬
aiian marriage patterns ‘ proved ’ for a while the former existence of
group marriage.
Curiously, the determined anti-evolutionists never attacked this
great reconstruction systematically. But the growing use of social
research techniques at the end of the nineteenth century added to
suspicions about the validity of such reconstructions of man’s past.
About the turn of the century the evolutionist approach was being
labeled irrelevant, foolish, and amateur, and it became itself a
‘survival’ in the writings of dilettanti and Communist apologists
through the succeeding half-century. It was seen that ‘primitive’
societies like those of the Australian aborigines might have very
complex religious and kinship systems, and that modern stone¬
using societies do not all have the same social systems. Moreover,
since all men presumably come from the same evolutionary line
and are equally distant historically from Cro-Magnon men, all
societies are equally old.
Social scientists came to see, then, that contemporary primitive
societies were not necessarily ‘primeval’, nor necessarily closer to
the social behavior of Neolithic man than to civilized societies. The
family patterns of both Neolithic and Paleolithic man are forever
lost to us. We cannot reconstruct them either from the behavior of
the gorilla or chimpanzee (about which almost no valid informa¬
tion was available, in any event, at the turn of the century), or from
the technologically least advanced societies still in existence.
Technological systems may be ranked as more or less advanced,
and, with somewhat greater theoretical difficulty, so may economic
systems; but no specific family forms seem to be correlated with the
specific ‘stages’ of the economic and technological evolution. For

368 Changing Patterns of Kinship

 example, the Eskimo types of kinship and that of the United States
are very similar2. Both religious and family systems seem to vary
independently of economic and technological systems3. This state¬
ment does not imply that family systems have not gone through any
evolutionary sequence - that is, some determinate series of changes
- or even that there is no set of determinate relations among family
and economic or technological variables. At the present time, how¬
ever, no such set of determinate sequences or relations has been
demonstrated- although they are often assumed to be well known.
Marx did not aim at a comprehensive theory of family social
change; but in explaining the origins of industrial England in the
mid-nineteenth century he took the trouble to analyse the effects
of the machine upon the working class family in a society that
accepted rational calculation as the basis for economic decisions.
As Marx pointed out, the essence of the machine is not that it is
driven by non-human power, but that it can perform the work of
the human hand, and can thus be multiplied or speeded up. Con¬
sequently, women, children or unskilled men could be used in the
labor force, causing the wages of skilled men to drop to the level
of those paid to women or children. Furthermore, since all could
now work, the wage level could be reduced to the point where a
composite family income was needed to produce the wages
formerly earned by one male household head. Consequently,
either everyone worked or all starved. A manufacturer could pay
higher wages only at the risk of being driven out of business by his
less scrupulous, but more rational, competitors.
Women were forced into factories and thus separated from their
children and households. The results were inevitable: infant and
child mortality rates rose, the working day lengthened, children
were kept out of school, young girls did not learn the necessary
domestic arts and fathers sold their children’s labor on harsh
terms.4 Thus, a century ago, did Marx document industrialization’s
2. See Murdock (1949), p. 226-7. Not all Eskimo groups fall into the same
kinship category: see Hughes (1959).
3. Attempting to locate underlying variables by scalogram analysis.
Winch and Freeman found that two variables would not ‘fit’ into a scale of
societal complexity (1957). Recently, Meyer F. Nimkoff has begun a cross-
cultural attempt to fit different forms of the family with different economic
and technological traits.
4. Marx’s analysis of the process may be found in Capital, chapter 15.

T-K-N W. J. Goode 369

destructive effect on the lower-class family, using the numerous
research reports of his time. In fact, however, the processes were
much more complex, as Smelser has indicated. In the earlier phases
of the factory system, men were able to supervise their own children
and thus maintain their authority as household heads. (Smelser,
1959).
But though Marx was able to show a few effects of the factory
system on the family, his analysis was limited by his political com¬
mitment. One could not have predicted from Marxian theory all
the changes in the family that have occurred since his time in both
industrial and non-industrial cultures. Certainly, he failed to prove
that particular forms of the family emerge when certain economic
or technological stages are reached in the history of a given society.
William F. Ogburn also gave major attention to the process of
social change, emphasizing technology’s impact on the family.
Although he is prominently identified with one version of Marxian
theory - that the prime mover is technology (‘material culture’)
and that after a period of time (‘culture lag’) the non-material
elements of culture adjust to it - his analyses are actually eclectic,
recognizing a wide range of innovations as sources of family
change: ideologies, birth control, urbanism, steam engines, air¬
planes, etc.5 Ogburn not only pointed out the probable effects of a
given invention or social innovation on the family but also traced
specific family changes - the increasing divorce rate, for example -
to its various probable causes. According to his analysis, with
economic production as well as education, religious training, and
protection being taken away from the home, the family in the
United States was ‘losing its functions’. This opinion has been
widely accepted by family analysts, although its precise meaning is
unclear and, for the most part, simply incorrect. Ogburn’s contri¬
bution was not theoretical, and in any event, our aim here is not to
criticize large-scale theories of family change, which could only be
premature. Ogburn was particularly effective in aggregating masses
of empirical data, or in using his mastery of research techniques to
test easy assumptions about some detail of behavior. For example,
he shows that although the average United States household
dropped from about 5-5 persons in 1850 to 3-5 in 1950, the large
5. Ogburn’s earlier formulation may be found in Social Change (1922).
See also his more mature analysis (1955, chs. 11 and 12; 1946, ch. 15).

370 Changing Patterns of Kinship

 household made up of many persons was not typical even a
century ago (Ogburn, 1955).
Similarly, he was able to ferret out data to show that the ‘labor-
saving devices ’ in the United States home seem to have saved no
labor at all, since housewives who owned them continued to work
as many hours per week as those who did not (Ogburn, 1955). In
his mature work, he no longer held that the machine was a primary
factor in altering society, but rather that society was made up of
intricately interwoven elements so that almost any part could be
linked causally with any other.
Such technically careful summations of fact, however, constitute
neither a theory of family change nor even a charting of the effects
of industrialization on the family. Ogburn’s later expositions are
guided not so much by a theory as by a general orientation - that
to understand family change it was most fruitful to examine the
nexus between a specific technological innovation and its several
possible effects on family relations (as evaluated by commonsense
expectations). This examination proceeded under the assumption
that the acceptance of an innovation was to be taken for granted
on grounds of rationality. That the responses of different cultures
to such innovations would vary did not constitute an inquiry of
research for Ogburn. He saw the upward march of technology as
both inevitable and theoretically unproblematic.
This has been a brief summary assault upon the attempts to show
a simple correlation between economic or technological develop¬
ment and particular stages or forms of family systems. As against
this necessary skepticism, one crude observation derived from
modem events seems accurate enough: wherever the economic
system expands through industrialization, family patterns
change. Extended kinship ties weaken, lineage patterns dissolve,
and a trend towards some form of the conjugal system gener¬
ally begins to appear - that is, the nuclear family becomes a more
independent kinship unit. Modem commentators have reported
this process from many parts of the world, some interpreting it as
one aspect of the ‘Americanization ’ of Europe, or even of the
world. [...]

W. J. Goode 371
The conjugal family as an ideal type
As now used by family analysts, the term ‘conjugal family is
technically an ideal type; it also represents an ideal.6 The concept
was not developed from a summary or from the empirical study of
actual United States urban family behavior; it is a theoretical
construction, derived from intuition and observation, in which
several crucial variables have been combined to form a hypothetical
structural harmony. Such a conceptual structure may be used as a
measure and model in examining real time trends or contemporary
patterns. In the ensuing discussion, we shall try to separate the
fundamental from the more derivative variables in this construc¬
tion.
As a concept, the conjugal family is also an ideal in that when
analysts refer to its spread they mean that an increasing number of
people view some of its characteristics as proper and legitimate, no
matter how reality may run counter to the ideal. Thus, although
parents in the United States agree that they should not play an
important role in their children’s choice of spouse, they actually
do. Relatives should not interfere in each other’s family affairs,
but in a large (if unknown) percentage of cases they do. Since,
however, this ideal aspect of the conjugal family is also part of the
total reality, significant for changes in family patterns, we shall
comment on it later as an ideology.
The most important characteristic of the ideal typical construc¬
tion of the conjugal family is the relative exclusion of a wide range
of affinal and blood relatives from its everyday affairs: there is no
great extension of the kin network.7 Many other traits may be
derived theoretically from this one variable. Thus, the couple
cannot count on a large number of kinfolk for help, just as these
kin cannot call upon the couple for services. Neither couple nor
kinfolk have many rights with respect to the other, and so the
reciprocal obligations are few. In turn, by an obvious sociological
principle, the couple has few moral controls over their extended
kin, and these have few controls over the couple.
The locality of the couple’s household will no longer be greatly
6. It is an ideal type in the Weberian sense; see Weber (1949, pp. 88 et
seq.) also a brief summary in Goode (1947, pp. 473-5).
7. For a list of additional variables for comparing family systems see
Goode (1959, pp. 178-91).

372 Changing Patterns of Kinship

 determined by their kin since kinship ties are weak. The couple will
have a ‘neolocal ’ residence, i.e. they will establish a new household
when they marry. This in turn reinforces their relative independ¬
ence, because it lowers the frequency of social interaction with
their kin.
The choice of mate is freer than in other systems, because the
bases upon which marriage is built are different: the kin have no
strong rights or financial interest in the matter. Adjustment is
primarily between husband and wife, not between the incoming
spouse and his or her in-law group. The courtship system is there¬
fore ideally based, and, at the final decision stage, empirically as
well, on the mutual attraction between the two youngsters.
All courtship systems are market or exchange systems. They
differ from one another with respect to who does the buying and
selling, which characteristics are more or less valuable in that
market, and how open or explicit the bargaining is. In a conjugal
family system mutual attraction in both courtship and marriage
acquires a higher value. Nevertheless, the elders do not entirely lose
control. Youngsters are likely to marry only those with whom they
fall in love, and they fall in love only with the people they meet.
Thus, the focus of parental controls is on who is allowed to meet
whom at parties, in the school and neighbourhood, and so on.8
When such a system begins to emerge in a society, the age at
marriage is likely to change because the goals of marriage change,
but whether it will rise or fall cannot be predicted from the charac¬
teristics mentioned so far. In a conjugal system, the youngsters
must now be old enough to take care of themselves, i.e. they must
be as old as the economic system forces them to be in order to be
independent at marriage. (Alternative solutions also arise: some
middle-class youngsters may marry upon the promise of support
from their parents, while they complete their education.9) Thus,
if the economic system changes its base, e.g. from agriculture to
industry, the age at marriage may change. The couple decides the
number of children they will have on the basis of their own needs,
not those of a large kin group, and contraception, abortion or

8. For an analysis of love-control in various societies see Goode (1959a,

pp. 38-47).
9. Additional help patterns are described in Sussman (1953, pp. 18-
22).

W. J. Goode 373
infanticide maybe used to control this number. Whether fertility will
be high or low cannot, however, be deduced from these conjugal
traits. Under some economic systems - for example, frontier agri¬
culture - the couple may actually need a large number of children.
This system is bilineal or, to use Max Gluckman’s term, multi¬
lineal: the two kin lines are of nearly equal importance, because
neither has great weight. Neolocality and the relative freedom
from control by an extended kin network prevent the maintenance
or formation of a powerful lineage system, which is necessary if one
line is to be dominant over the other.10
Since the larger kin group can no longer be counted on for
emotional sustenance, and since the marriage is based on mutual
attraction, the small marital unit is the main place where the emo¬
tional input-output balance of the individual husband and wife is
maintained, where their psychic wounds can be salved or healed.
At least there is no other place where they can go. Thus, the emo¬
tions within this unit are likely to be intense, and the relationship
between husband and wife may well be intrinsically unstable,
depending as it does on affection. Consequently, the divorce rate is
likely to be high. Remarriage is likely because there is no larger kin
unit to absorb the children and no unit to prevent the spouses from
re-entering the free marriage market.
Finally, the couple and children do recognize some extended
kin, but the husband recognizes a somewhat different set of
kindred than does his wife, since they began in different families.
And the children view as important a somewhat different set of
kindred than do their parents: the parents look back a generation
greater in depth than do the children, and perhaps a greater dis¬
tance outward because they have had an adult lifetime in which to
come to know more kin. That is, each individual takes into account
a somewhat different set of kindred, though most of them are the
same for all within the same nuclear unit.
The foregoing sketch is an ideal typical construction and thus
must be compared with the reality of both behavior and ideal in
those societies which are thought to have conjugal family patterns.
To my knowledge, no such test has been made. Very likely, the
ideals of a large proportion of United States families fit this con-
10. For other kinship characteristics of this ‘Eskimo’ type see Murdock
(1949, pp. 226-8).

374 Changing Patterns of Kinship

struction very well. Some parts of the construction also fit the
behavior of a considerable, but unknown, fraction of United
States families - e.g. the emphasis on emotionality within the
family, the free choice of spouse, and neolocality, bilineality and
instability of the individual family. On the other hand, data from
both England and the United States indicate that even in lower-
class urban families, where the extension of kin ties might be
thought to be shorter (following the ideal type), many kin ties are
active. No one has measured the intensity and extensivenes of kin
ties in a range of societies in order to ascertain how Western
family patterns compare in these respects. It is quite possible that
those countries thought to be closest to the conjugal pattern do in
fact have a less extended kin network.
Nevertheless, the ideal type conflicts sharply with reality and
theory in one important respect. Theoretical considerations sug¬
gest that, without the application of political pressure, the family
cannot be as limited in its kin network as the ideal typical con¬
struction suggests. Both common observation and theory coincide
to suggest that: (1) grandparent-grandchild ties are still relatively
intense and (2) emotional ties among siblings are also strong.
Consequently (3), parents-in-law interact relatively frequently
with their children-in-law (see Sussman, 1953, pp. 22-8), and (4)
married people have frequent contacts with their brothers- or
sisters-in-law. It follows, then, that (5) children maintain contacts,
at least during their earlier years, with their uncles and aunts, as
well as with their first cousins. Without question, of all types of
‘visiting’ and ‘social occasions’, the most common, even in the
urban United States, is ‘ visiting with relatives ’.
If no active ties are maintained with the categories of kin
mentioned above, the family feels that some explanation is called
for, and pleads some excuse (‘ They live too far away,’ or ‘ We’ve
never got along well’).
In addition, perhaps most families have some tie with one or
more relatives still further away in the kin network. Those noted
above seem to be linked to the nuclear family in an inescapable
way; it is difficult to ignore or reject any of them without simul¬
taneously rejecting a fellow member of one's own nuclear family.
The child can not ignore his uncle without hurting one of his own
parents, and reciprocally. A girl may not neglect her sister-in-law

W. J. Goode 375
without impairing her relationship with her brother. Of course,
brother and sister may combine against their own spouses, and
social interaction may continue even under an impaired relation¬
ship. Cousins are dragged along by their parents, who are siblings
and siblings-in-law to one another. The extension of the family
network to this point, then, seems determined by the emotional
ties within the nuclear family unit itself. To reduce the unit to the
nuclear family would require coercive restriction of these ties
between siblings or between parents, as the Chinese commune has
attempted to do.

The ‘fit’ between the conjugal family and the

modern industrial system
The argument as to whether political and economic variables, or
the reverse, generally determine family patterns seems theoretic¬
ally empty. Rather, we must establish any determinate relations
(whichever direction the causal effect) among any particular
family variables and the variables of other institutional orders -
not a simple task. Even the relation between the conjugal family
and industrialization is not yet entirely clear. The common
hypothesis - that the conjugal form of the family emerges when a
culture is invaded by industrialization and urbanization - is an
undeveloped observation which neglects three issues: (1) the
theoretical harmony or ‘fit’ between this ideal typical form of the
family and industrialization; (2) the empirical harmony or fit
between industrialization and any actual system; and (3) the
effects upon the family of the modern (or recently past) organ¬
izational and industrial system, i.e. how the factors in the system
influence the family.
At present, only the first of these can be treated adequately.
The second has been dealt with primarily by researchers who have
analysed a peasant or primitive culture with reference to the
problem of labor supply, and who suggest that family systems
other than the conjugal one do not adequately answer the de¬
mands of an expanding industrial system. Malinowski asserted,
for example, that although young Trobriander men could earn
more by working on plantations than by growing yams, they
preferred to grow yams because this activity was defined as re¬
quired for their family roles. Similarly, a head tax was necessary

376 Changing Patterns of Kinship

to force young men to leave their families to work in the South
African mines.11 Men’s objections to women leaving the home for
outside jobs have limited the labor supply in various parts of the
world, especially in Islamic areas. On the other hand, within
conjugal or quasi conjugal systems such as those in the West, the
strains between family patterns and industrial requirements have
only rarely been charted empirically (see, however, Smelser, 1959,
chs.9,10,11).
This last task would require far more ingenious research de¬
signs than have been so far utilized. It requires that the exact
points of impact between family and industrial organization be
located and the degree of impact measured. Succeeding chapters
will devote some attention to this problem. Specific decisions or
choices need to be analysed, in which both family and industrial
variables are involved.
Nevertheless, if we are to achieve a better understanding of
world changes in family systems, it may help if we can correct the
theoretical analyses of the first problem, the fit between the ideal
typical form of the conjugal family and industrialization. It seems
possible to do this through some reference to common observa¬
tions about both United States and European family patterns.
Let us consider first the demands of industrialization, which is
the crucial element in the complex types of change now occurring
in even remote parts of the world. Although bureaucratization
may occur without industrialization (witness China), and so may
urbanization (for example, Dahomey, Tokugawa, Japan),
neither occurs without some rise in a society’s technological level,
and certainly the modern system of industry never occurs without
both urbanization and bureaucratization.
The prime social characteristic of modern industrial enterprise
is that the individual is ideally given a job on the basis of his
ability to fulfill its demands, and that this achievement is evalu¬
ated universalistically; the same standards apply to all who hold
the same job. His link with the job is functionally specific; in other
words, the enterprise cannot require behavior of him which is
not relevant to getting the job done.
Being achievement-based, an industrial society is necessarily
11. See also Moore’s (1951, chs. 9, 10, 11) report on the relation between
traditional family patterns and industrial demands in Mexico.

W. J. Goode 377
open-class, requiring both geographical and social mobility. Men
must be permitted to rise or fall depending on their performance.
Moreover, in the industrial system, jobs based on ownership and
exploitation of land (and thus on inheritance) become numerically
less significant, again permitting considerable geographical
mobility so that individuals are free to move about in the labor
market. The neolocality of the conjugal system correspondingly
frees the individual from ties to the specific geographical location
where his parental family lives.
The conjugal family’s relationship to class mobility is rather
complex. Current formulations, based on ancient wisdom, assert
that by limiting the extensiveness of the kin network, the indi¬
vidual is less hampered by his family in rising upward in the job
structure. Presumably, this means that he owes less to his kin
and so can allocate his resources of money and time solely to
further his career; perhaps he may also more freely change his
style of life, his mode of dress and speech, in order to adjust to a
new class position without criticism from his kin. On the other
hand, an industrial system pays less attention to what the
individual does off the job, so that family and job are structurally
somewhat more separated than in other systems. Consequently,
one might reason that differential social or occupational mobility
(as among siblings or cousins) would not affect kin ties. Yet the
emotional ties within the conjugal system are intense, compared
to other systems, so that even though there are fewer relatives, the
weight of kin relationships to be carried upward by the mobile
individual might be equivalent to that in a system with more, but
less intense, ties.12
An alternative view must also be considered. Under some cir¬
cumstances the kin network actually contributes greatly to the
individual’s mobility, and ‘social capillarity’ as a process (that is,
that individual rises highest who is burdened with least kin)
moves fewer people upward than does a well-integrated kin
network.13 A brief theoretical sketch of this alternative view also
12. Parsons (1949, ch. 5) has analysed some parts of this relationship
between role needs of job and family, and has particularly used the two
pattern variables of ascription - achievement and universalism - particu¬
larism to differentiate among four types of societies.
13. For a good analysis of mobility rates in a kin-based society see Marsh
(1961).

378 Changing Patterns of Kinship

throws light on the supposed ‘adjustment’ between the needs of
the small conjugal family and those of a modern industrial system.
First, in the modern industrial system, the middle and upper
strata are by definition more ‘ successful ’ in the obvious sense that
they own it, dominate it, occupy its highest positions and direct
its future. One must concede that they are ‘well adjusted’ to the
modern industrial society. Paradoxically, their kin pattern is in
fact less close to the ideal typical form of the conjugal family than
is the family behavior of the lower strata. The upper strata
recognize the widest extension of kin, maintain most control over
the courtship and marriage choices of their young, and are most
likely to give and receive help from one another.
Consequently, the lower strata’s freedom from kin is like their
‘freedom’ to sell their labor in an open market. They are less
encumbered by the weight of kin when they are able to move
upward, but they also get less help from their kin. Like English
peasants, who from the sixteenth to eighteenth centuries were
gradually ‘freed’ from the land by the enclosure movement, or
nineteenth-century workers, who were ‘freed’ from their tools
by the development of machinery, the lower strata have fewer
family ties, less family stability and enjoy less family-based
economic and material security. The lower-class family pattern is
indeed most ‘ integrated ’ with the industrial system but mainly in the
sense that the individual is forced to enter its labor market with
far less family support - his family does not prevent industry from
using him for its own goals. He may move where the system needs
him, hopefully where his best opportunity lies, but he must also
fit the demands of the system, since no extended kin network will
interest itself greatly in his fate. The job demands of the industrial
system move the individual about, making it difficult for him to
keep his kin ties active; and because his kin are also in the lower
strata he has little to lose by relinquishing those ties. In short,
lower-strata families are most likely to be ‘conjugal’ and to serve
the needs of the industrial system; this system may not, however,
serve the needs of that family pattern. This means that when
industrialization begins, it is the lower-class family that loses
least by participating in it and that lower-class family patterns
are the first to change in the society.
We might speculatively infer further that now, a century after

W. J. Goode 379
the first great impact of industrialization on the lower-class
family in the Western urban world, family patterns of Western
middle and upper classes may be changing more rapidly than
those of the lower. (Whether rural changes may not be occurring
equally rapidly cannot be deduced from these inferences.) How¬
ever, although this inference may be empirically correct, the
available data demand a more cautious inference. Whether or not
the middle and upper strata are now changing more rapidly in the
Western world, they do have more resources with which to resist
certain of the industrial system’s undermining pressures (e.g.
capital with which to support their youngsters through a long
professional training) and a considerable interest in resisting them
because their existing kin network is more active and useful. We
would suppose, then, that in an industrializing process both the
peasants and primitives are forced to adjust their family patterns
to the demands of industrial enterprise more swiftly, and see less
to lose in the adjustment. By contrast, the middle and upper strata
are better able to utilize the new opportunities of industrialization
by relinquishing their kin ties more slowly, so that these changes
will occur only in a later phase of industrialization, such as the
United States is now undergoing.
Continuing now with our analysis of the ‘fit’ of the conjugal
family to industrial needs; the more limited conjugal kin network
opens mobility channels somewhat by limiting the ‘closure’ of
class strata. In general, rigid class boundaries can be maintained
partly by the integration of kin bonds against the ‘outsider’
through family controls. When the network of each family is
smaller, the families of an upper stratum are less integrated, the
web of kin less tightly woven, and entrance into the stratum
easier. Since the industrial system requires relatively free mobility,
this characteristic of the conjugal pattern fits the needs of that
system. This general principle also holds for classical China,
where an empirically different system prevailed. A successful
family would normally expand over generations, but thereby have
insufficient resources to maintain so many at a high social rank.
That is, the reciprocal exchanges necessary for tightness and
closure of the kin system could be kept up only by a few individual
families in the total network. If all the families in the network
shared alike as kinsmen (which did not happen), the entire net-

380 Changing Patterns of Kinship

 work would lose social rank. If the few well-to-do families helped
their kin only minimally and maintained ties with other upper
stratum families the integration of the stratum was kept intact
and the stratification system was not threatened.
The modem technological system is psychologically burden¬
some on the individual because it demands an unremitting disci¬
pline. To the extent that evaluation is based on achievement and
universalism, the individual gets little emotional security from his
work. Some individuals, of course, obtain considerable pleasure
from it, and every study of job satisfaction shows that in positions
offering higher prestige and salaries a higher proportion of people
are satisfied with their work and would choose that job if they
had to do it again. Lower-level jobs give little pleasure to most
people. However, in higher-level professional, managerial and
creative positions the standards of performance are not only high
but are often without clearly stated limits. The individual is under
considerable pressure to perform better than he is able.
The conjugal family again integrates with such a system by its
emphasis on emotionality, especially in the relationship of hus¬
band and wife. It has the task of restoring the input-output emo¬
tional balance of individualism in such a job structure. This is so
even for lower-strata jobs where the demands for performance are
kept within limits by an understood quota but where, by contrast
with upper-strata jobs, there is less intrinsic job satisfaction. Of
course, the family cannot fully succeed in this task, but at least
the technological system has no moral responsibility for it and
can generally ignore the problem in its work demands.
Bilateral in pattern, this family system does not maintain a
lineage. It does not concentrate family land or wealth in the hands
of one son through whom the property would descend, or even in
the hands of one sex. Dispersal of inheritance keeps the class system
fluid. Daughters as well as sons will share as heirs, and a common
legal change in the West is towards equal inheritance by all chil¬
dren (as is already the situation generally in the United States).
Relatively equal advantages are given to all the sons, and although
even United States families do not invest so heavily in daughters
as in sons (more boys than girls complete college), the differences
in training the two sexes are much less than in other family systems.
Consequently, a greater proportion of all children are given the

W. J. Goode 381
opportunity to develop their talents to fit the manifold oppor¬
tunities of a complex technological and bureaucratic structure.
The conjugal system also specifies the status obligations of each
member in much less detail than does an extended family system,
in which entrepreneurial, leadership or production tasks are
assigned by family position. Consequently, wider individual varia¬
tions in family role performance are permitted, to enable members
to fit the range of possible demands by the industrial system as
well as by other members of the family.
Since the young adult is ideally expected to make his own
choice of spouse and the young couple is expected to be eco¬
nomically independent, the conjugal system by extending the
adolescent phase of development, permits a long period of tute¬
lage. For example, it is expected that the individual should be
grown up before marrying. Note, however, that it is not the family
itself that gives this extended tutelage, but public, impersonal
agencies, such as schools, military units, and corporations, which
ideally ignore family origin and measure the individual by his
achievement and talent. This pattern permits the individual to
obtain a longer period of training, to make a freer choice of his
career, and to avoid the economic encumbrance of marriage until
he has fitted himself into the industrial system. Thus, the needs
of the industrial system are once more served by the conjugal
family pattern.

References
Goode, W. J. (1947), ‘Note on the ideal type’, Amer. Sociol. Rev.,
vol. 12, pp. 473-5.
Goode, W. J. (1959), ‘The sociology of the family’, in R. K. Merton,
L. Broom and L. S. Cottrell (eds.), Sociology Today, Basic Books.
Goode, W. J. (1959a), ‘The theoretical importance of love’, Amer.
Sociol. Rev., vol. 24, pp. 18-28.
Hughes, C. C. (1959), ‘An Eskimo deviant from the “Eskimo” type of
social organisation’, Amer. Anthrop., vol. 60, pp. 1140-47.
Konig, R. (1955), ‘Familie und Familiensoziologie’, in W. Bernsdorf
and F. Bulow (eds.), Worterbuch der Soziologie, Stuttgart.
Marsh, R. (1961), The Mandarins, Free Press.
Marx, K. (1936), Capital, Modern Library.
Moore, W. E. (1951), Industrialization and Labor, Cornell University
Press
Murdock, G. P. (1949), Social Structure, Macmillan Co.
Ogburn, W. F. (1922), Social Change, Viking, rev. edn, 1952.

382 Changing Patterns of Kinship

Ogburn, W. F. (1946), The Social Effects of Aviation, Houghton
Mifflin.
Ogburn, W. F. (1955), Technology and the Changing Family, Houghton
Mifflin.
Parsons, T. (1949), The Social System, Free Press.
Smelser, N. (1959), Social Change in the Industrial Revolution,
University of Chicago Press.
Sussman, M. B. (1953), ‘The help pattern in the middle class family’,
Amer. sociol. Rev., vol. 24, pp. 38-47.
Tylor, E. B. (1888), ‘On a method of investingating the development of
institutions: applied to laws of marriage and descent J. anthrop.
Inst., vol. 18, pp. 245-72.
Weber, M. (1949), ‘“Objectivity” in social science and social policy’,
in The Methodology of the Social Sciences, Free Press. Translated by
E. A. Shils and H. A. Finch.
Winch, R. F., and Freeman, L. C. (1957), ‘Societal complexity: an
empirical test of a typology of sciences, Amer. J. Sociol., vol. 62,
pp. 461-6.

W. J. Goode 383
Further Reading

Introduction
The best introductory book on kinship is:
Fox, R., Kinship and Marriage, Penguin, 1967.
Radcliffe-Brown, A. R., ‘Introduction’ in African Systems of Kinship
and Marriage (ed. A. R. Radcliffe-Brown and C. D. Forde), Oxford
University Press, 1950, remains a most stimulating survey. So too, in
a different vein, does:
Levi-Strauss, C., Les Structures elementaires de la parente, 1949,
translated into English as The Elementary Structures of Kinship,
1969, Eyre & Spottiswoode.
Another general work is:
Fortes, M., Kinship and the Social Order: the Legacy of Lewis
Henry Morgan, Aldine Press, 1969.

The Family
Bott, E., Family and Social Network, Tavistock, 1971, 2nd edn.
Levy, M. J., and Fallers, L. A., ‘The family; some comparative
considerations’, American Anthropologist, vol. 61, 1959, pp. 647-51.
Malinowski, B., ‘Kinship’, Man, vol. 30, 1930, pp. 19-29.
Parsons, T., and Bales, R. F., Family, Socialization and Interaction
Process, Free Press, 1955.
Smith, R. T., The Negro Family in British Guiana: Family Structure
and Social Status in the Villages, Routledge & Kegan Paul, 1956.
Smith, R. T., ‘Culture and social structure in the Caribbean: some
recent work on family and kinship studies ’, Comparative Studies in
Society and History, vol. 6, 1963, pp. 24-46.
Spiro, M. E., ‘Is the family universal?’, American Anthropologist,
vol. 56, 1954, pp. 839-46.
Spiro, M. E., Children of the Kibbutz, Harvard University Press, 1958.

Incest and Sex

Aberle, D. et ah, ‘The incest taboo and the mating patterns of
animals’, American Anthropologist, vol. 65, 1963, pp. 253-65.
Davies K., ‘Intermarriage in caste societies’, American
Anthropologist, vol. 43, 1941, pp. 376-95.
Fox, J. R. ‘Sibling incest’, British Journal of Sociology, vol. 13,
1962, pp. 128-50.

384 Further Reading

Kortmulder, K., ‘An ethnological theory of the incest taboo and
exogamy’, Current Anthropology, vol. 9, 1968, pp. 437-49.
Malinowski, B., Sex and Repression in Savage Society, Routledge
& Kegan Paul, 1927.
Merton, R., ‘Intermarriage and social structure; fact and theory’.
Psychiatry, vol. 4, 1941, pp. 361-74.
Middleton, R., ‘Brother-sister and father-daughter marriage in
Ancient Egypt’, American Sociological Review, vol. 27, 1962, pp.
603-11.
Parsons, T., ‘The incest taboo in relation to social structure and the
socialization of the child’, British Journal of Sociology, vol. 5, 1954,
pp. 101-17.
Weinberg, S. K., Incest Behaviour, Citadel Press, 1955.

The Developmental Cycle

Collver, A., ‘The family cycle in India and the United States’,
American Sociological Review, vol. 28, 1963, pp. 86-96.
Fortes, M., ‘Introduction’ in The Developmental Cycle in Domestic
Groups (ed. J. Goody), Cambridge University Press, 1958.
Fortes, M., Time and Social Structure, Athlone Press, 1970.

Joking and Avoidance

Goody, J. ‘The mother’s brother and the sister’s son in West
Africa’, Journal of the Royal Anthropological Institute, vol. 89,
1959, pp. 61-88. Reprinted in Comparative Studies in Kinship,
Routledge & Kegan Paul, 1969.
Marshall, L., ‘The kin terminology system of the IKung Bushmen’,
Africa, vol. 27, 1957, pp. 1-25.
Marshall, L., ‘Sharing, talking and giving: relief of social tensions
among IKung Bushmen’, Africa, vol. 31, 1961, pp. 231-49.
Radcliffe-Brown, A. R., ‘The mother’s brother in South Africa’,
South African Journal of Science, vol. 21, 1924, pp. 542-55.
Reprinted in Structure and Function in Primitive Society, Cohen &
West, 1950.
Radcliffe-Brown, A. R., ‘On joking relationships’, Africa, vol. 13,
1940, pp. 195-210. Reprinted in Structure and Function in
Primitive Society, Cohen & West, 1950.

Marriage Transactions
Bohannan, L., ‘Dahomean marriage: a revaluation’, Africa, vol.
29, 1949, pp. 273-87.
Evans-Pritchard, E. E., Kinship and Marriage among the Nuer ,
Clarendon Press, 1951.

Further Reading 385

Gray, R. R., ‘Sonjo bride-price and the question of African “wife-
purchase”’, American Anthropologist, vol. 62, 1960, pp. 34-57.
Yalman, N., Under the Bo Tree: Studies in Caste, Kinship and
Marriage in the Interior of Ceylon, University of California Press,
1967.

Plural Marriage
Clignet, R., Many Wives, Many Powers: Authority and Power in
Polygynous Families, Northwestern University Press, 1970.
Gough, E. K., ‘The Nayars and the definition of marriage’, Journal
of the Royal Anthropological Institute, vol. 89, 1959, pp. 23-34.
H R H Prince Peter of Greece and Denmark, A Study of Polyandry,
Mouton, 1963.
Smith, M. G. /Secondary marriage in Northern Nigeria’, Africa,
vol. 23, 1953, pp. 298-323.

Marriage and Alliance

Ackerman, C., ‘Structure and statistics: the Purum case*,
American Anthropologist, vol. 65, 1964, pp. 53-6.
Coult, A. D., ‘An analysis of Needham’s critique of the Homans-
Schneider theory’, Southwestern Journal of Anthropology, vol. 18,
1962, pp. 317-35.
Homans, G. C., and Schneider, D. M., Marriage, Authority and Final
Causes, Free Press, 1955.
Leach, E. R., Political Systems of Highland Burma, Bell, 1954.
Needham, R., Structure and Sentiment, University of Chicago
Press. 1962.
Schneider, D. M., ‘Some muddles in the models’, in The Relevance
of Models for Social Anthropology, Tavistock, 1965.

Divorce and Marriage Stability

Ackerman, C., ‘Affiliations: structural determinants of differential
divorce rates’, American Journal of Sociology, vol. 69, 1963, pp. 13-20.
Ardener, E., Divorce and Fertility: an African Study, Oxford
University Press, 1962.
Fallers, L. A. ‘Some determinants of marriage instability in Busoga:
a reformulation of Gluckman’s thesis’, Africa, vol. 27, 1957, pp.
106-21.
Goode, W. J.. ‘Marital satisfaction and instability: a cross-cultural
class analysis of divorce rates’, International Social Science Journal,
vol. 14, 1962, pp. 507-26.
Goody, E. N., ‘Conjugal separation and divorce among the Gonja
of Northern Ghana’, in Marriage in Tribal Societies (ed. M. Fortes),
Cambridge University Press, 1962.

386 Further Reading

 Goody, J., and Goody, E., ‘The circulation of women and children
in Northern Ghana’, Man, n.s., vol. 2, 1969, pp. 226-48. Reprinted in
J. Goody, Comparative Studies in Kinship, Routledge & Kegan
Paul, 1969.
Leach, E. R., ‘Aspects of bridewealth and marriage stability among
the Kachin and Lakher’, Man, vol. 57, 1961, pp. 50-55. Reprinted
in Rethinking Anthropology, Athlone Press, 1961.

Kin Groups
For a summary of recent work, see my article on ‘Kinship: Descent
,
Groups’, in D. L. Sills (ed.), Encyclopedia of the Social Sciences
Macmillan Co. 1968.
For a discussion of basic concepts see: Schneider, D. M. ‘Some
muddles in the models: or, how the system really works’, in
M. Banton (ed.), The Relevance of Models for Social Anthropology,
Association of Social Anthropologists, monograph no. 1, Tavistock,
1965.

For the discussion of particular types of kin groups, see:

Davenport, W., ‘Nonunilinear descent and descent groups’, American
Anthropologist, vol. 61, 1959, pp. 557-72.
Firth, R., ‘Bilateral descent groups: an operational viewpoint’, in
Studies in Kinship and Marriage (ed. I. Schapera), pp. 22-37, Royal
Anthropological Institute, 1963.
Freeman, J. D., ‘On the concept of the kindred’, Journal of the
Royal Anthropological Institute, vol. 91, 1961, pp. 192-220.
Goodenough, W. H., ‘A problem in Malayo-Polynesian social
organization’, American Anthropologist, vol. 57, 1955, pp. 71-83.
Goody, J., ‘The classification of double descent systems’, Current
Anthropology, vol. 2, 1961, pp. 3-25. Reprinted in J. Goody,
Comparative Studies in Kinship, Routledge & Kegan Paul, 1969.
Scheffier, H. W., ‘Ancestor worship in anthropology: or,
Observations on descent and descent groups’, Current
Anthropology, vol. 7, 1966, pp. 541-51.
Schneider, D. M., and Gough, K., Matrilineal Kinship, University
of California Press, 1961.
Smith, M. G., ‘On segmentary lineage systems’, Journal of the
Royal Anthropological Institute, vol. 86, 1956, pp. 39-80.

For a comparison of systems of unilineal descent see:

Barnes, J. A., ‘African models in the New Guinea highlands’,
Man, vol. 62, 1962, pp. 5-9.

Further Reading 387

Fried. M. H.. ‘The classification of corporate unilineal descent
groups*. Journal of the Royal Anthropological Institute, vol. 87, 1957,
pp. 1—50.
Lewis. 1. M.. ‘Problems in the comparative study of unilineal
descent*, in M. Banton ted.). The Relevance of Models for Social
Anthropology, pp. 8'-112, Association of Social Anthropologists,
monograph no. 1. Tavistock, 1965.

For studies of societies with descent groups see:

America
Egg .in. F.. Social Organisation of the Western Pueblos, University
of Chicago Press. 1950.
Eggan. F. (ed.). Social Anthropology of North American Tribes,
University of Chicago Press, 1962.
Fox R., The Keresan BrLige, Athlone Press, 1967.

New Guinea
Glasse. R. M., and Meggitt, M. J. (eds.). Pigs, Pearlshells, and
Women: Marriage in the New Guinea Highlands, Prentice-Hall, 1969.
Straihem. A.. ‘Descent and alliance in the New Guinea highlands:
some problems of comparison’. Proceedings of the Royal Anthropo¬
logical Institute. 1968, pp. 37-52.

Australia
Warner. \Y. L.. A Black Civilization, Harper & Row, 1937. Revised
edn. 1958.
Meggitt, Nl. J., Desert People, .Angus & Robertson, 1962.

Africa
Evans-Pritchard, E. E.. The Nuer, Clarendon Press, 1940.
Evans-Pritchard. E. E.. Kinship and Marriage among the Nuer,
Clarendon Press. 1951.
Fortes. M., The Dynamics of Clanship among the Tallensi, Oxford
University Press. 1945.
Fortes. M., The Web of Kinship among the Tallensi, Oxford University
Press, 1949.

China
Freedman. M., Lineage Organisation in South-eastern China,
Athlone Press. 195S.
Freedman, M., Chinese Lineage and Society: Fukien and Kwangtung,
Athlone Press, 1966.

3SS Further Reading

Oceania
Firth, R., We, the Tikopia. Allen & Unwin, 1936.

For non-unilineal systems see:

Arensberg, C. M. and Kimball, S. T., Family and Community in
Ireland, Harvard University Press, 1940.
Barton, R. F., The Kalingas, University of Chicago Press, 1949.
Campbell, J. K., Honour, Family and Patronage, Clarendon Press,
1964.
Pehrson, R. N., ‘The bilateral network of social relations in
Konkama Lapp district’, International Journal of American
Linguistics, vol. 23, 1957, part 2.
Schneider, D. M., American Kinship: a Cultural Account,
Prentice-Hall, 1968.
Young, M. and Wilmott, P., Family and Kinship in East London,
Routledge & Kegan Paul, 1957, Penguin, 1962.

Kin Terms
Goodenough, W. H., ‘ Componential analysis and the study of
meaning’, Language, vol. 32, 1956, pp. 195-216.
Hammel, E. A. (ed.), ‘Formal semantic analysis’, American
Anthropologist, Special Issue, vol. 67, 1965.
Lounsbury, F. G., ‘The structural analysis of kinship semantics’,
Proceedings of the Ninth International Congress of Linguists,
Mouton, 1962.
Lounsbury, F. G., ‘A formal account of the Crow?- and Omaha-type
kinship terminologies’, in W. H. Goodenough (ed.), Explorations
in Cultural Anthropology: Essays in Honour of George Peter Murdock,
McGraw-Hill, 1964, pp. 351-93.
Lowie, R. F., and Eggan, F. R., ‘Kinship terminology’.
Encyclopaedia Britannica, vol. 13, 1964, pp. 407-409.
Murdock, G. P., Social Structure, Macmillan, 1949.

Ritual and Fictive Kinship

Carroll, V. (ed.), Adoption in Eastern Oceania, Bishop Museum,
Hawaii, 1970.
Goody, E. N., ‘Kinship fostering in Gonja: deprivation or
advantage?’, in P. Mayer (ed.), Socialisation: the Approach of Social
Anthropology’, Tavistock, 1970.
Goody, J., ‘Adoption in cross-cultural perspective’. Comparative
Studies in Society and History, vol. 11, 1969, pp. 55-78.

Further Reading 389

Hammel, E. A., Alternative Social Structures and Ritual Relations
in the Balkans, Prentice-Hall, 1968.
Pitt-Rivers, J., ‘Pseudo-kinship’, International Encyclopedia of the
Social Sciences, vol. 8, pp. 408^413, Macmillan, 1968.

Changing Patterns of Kinship

Goode, W. J., World Revolution and Family Patterns, Free Press,
1963.
Laslett, P., The World We Have Lost, Methuen, 1965.

390 Further Reading

 Acknowledgements

Permission to reproduce the Readings in this volume is

acknowledged to the following sources:
1 Thomas T. Crowell Co. Inc. and Richard N. Adams
2 George Allen & Unwin Ltd and the Macaulay Co.
3 Routledge & Kegan Paul Ltd and Humanities Press Inc.
4 Eyre & Spottiswoode (Publishers) Ltd and Beacon Press
5 Routledge & Kegan Paul Ltd and Stanford University Press
6 Cambridge University Press and Meyer Fortes
7 Routledge & Kegan Paul Ltd and Free Press
8 Oxford University Press for the International African Institute
9 Holt, Rinehart & Winston Inc. and Ernestine Friedl
10 Edward Arnold (Publishers) Ltd and J. Hajnal
11 Royal Anthropological Institute
12 Northwestern University Press
13 Royal Anthropological Institute
14 Southwestern Journal of Anthropology and D. Maybury-Lewis
15 Oxford University Press for the International African
Institute and Max Gluckman
16 Oxford University Press for the International African Institute
17 Athlone Press, American Anthropological Association and
Meyer Fortes
18 Oxford University Press for the International African Institute
19 University of Manila Press
21 Royal Anthropological Institute
22 American Anthropological Association and Eugene Hammel
23 Esther Goody
24 Southwestern Journal of Anthropology and Eric R. Wolf
25 Free Press and William Goode

Acknowledgements 391
Author Index

Ackerman, C., 219,235, 236, 237, D’Andrade, R. G. D., 321

238, 239, 241, 244 Davenport, W., 262
Adam, L., 278, 294 Davis, K., 174
Adams, R. N., 17 Dole, G. E., 19
Aiyappan, A., 159, 184 Dorjahn, V., 164, 171, 173
Dumont, L., 181, 200, 202, 204,
Bachofen, J. J., 11 215, 219, 222
Baker, T., 177 Durkheim, E., 57, 181, 233, 235,
Banton, M., 176 249, 307
Barnes, J. A., 234, 235, 244, 255,
259 Eggan, F., 101, 102, 302, 358
Beattie, J., 323 Engels, F., 10
Best, E., 51 Evans-Pritchard, E. E., 77, 114,
Bird, M., 176, 177 228, 230, 231, 233
Blake, J., 174
Bloomfield, L., 318, 319 Fallers, L. F., 235, 239, 240,
Boas, F., 317, 319 241, 242
Bohannan, L., 231, 251, 273 Firth, R., 262, 278, 280
Bott, E., 233, 237 Fischer, H., 151, 152
Braithwaite, R. B., 222 Forde, D., 156, 265, 268, 282,
Brass, W., 165, 173 283, 302
Briffault, R., 184 Fortes, M., 26, 66, 70, 71, 78,
Burling, R., 320 79, 83, 105, 115, 155, 231,256,
261, 302
Carneiro, R. L., 19 Fortune, R. F., 60, 76, 77, 79
Caroll, V., 342 Foster, G. N., 348
Chomsky, N., 319 Frazer, J. G., 181, 307
Chowning, A., 302 Freeman, J., 262
Clement, P., 176 Freud, S., 45, 76
Clerc, A., 232 Friedl, E., 9, 117, 149
Clignet, R., 10, 149
Cohen, R., 173 Galton, F., 11
Coult, A. D., 203, 204, 205, Gamble, D., 176
209, 211 Gardin, J. C., 319, 322
Crook, J. A., 340 Gautier, E., 141
Csoscan, J., 140, 141 Gifford, E. W., 317
Gillen, F. J., 217

Author Index 393

Gillin, J., 349, 357 242, 244, 301, 323, 324
Ginsberg, M., 11 Le Play, P. G. F., 144
Gluckman, M., 225, 233, 239, LeVine, R., 166
243, 250, 251, 254, 259, 374 Levi-Strauss, C., 45, 76, 79, 181,
Gonzales, R. F., 352 201, 207, 218, 219, 222
Goode, W. J., 9, 10, 363, 372, Levy, M. J., Jr, 20
373 Little, K., 176
Goodenough, W., 36, 88, 262, Linton, R., 55, 279
297, 303, 319, 324 Livingstone, F. B., 209, 211, 214
Goody, E., 234, 235, 329, 336 Lonsbury, F. G., 297, 301, 303,
Goody, J., 45, 302, 304 304, 319, 321, 323, 324
Gough, E. K., 153, 332 Lowie, R., 29, 299, 317
Grabura, N. H. H., 324 Lubbock, J., 50

Hajnal, J., 9, 10, 117 McLennan, J. F., 11, 50, 303

Hallam, H. E., 146 Maine, H., 11
Hammel, E. A., 209, 297, 321, Malinowski, B., 10, 12, 17, 45,
339 56, 57, 60, 65, 66, 69, 70, 76,
Harris, Z. S., 318 77, 261, 266, 302, 304, 376
Henry, L., 141 Malthus, T. R., 140
Hjelmslev, L., 318 Marriott, M., 290
Hobhouse, L. T., 11 Marsh, R., 378
Hockett, C. F., 318 Marx, K., 10, 369, 370
Hoebel, E. A., 290 Mauss, M., 101, 181
Hoijer, H., 301 Maybury-Lewis, D., 181, 205
Homans, G. C., 145, 203, 205, Mead, M., 55, 56
207 Mercier, P., 176
Howitt, A. W., 307 Middleton, J., 243
Middleton, R., 302
Jacobson, A., 244, 245 Mintz, S. W., 329, 337
Jakobson, R., 318 Mitchell, J. C., 225, 244, 272
Junod, H., 102, 232 Monteil, V., 175
Moore, F., 163
Kirchoff, P., 284, 299 Moore, W. E., 377
Kroeber, A. L., 267, 270, 299, Morgan, L. H., 10, 299, 304,
303, 307, 308, 317, 319 307, 308, 309, 317
Muhsam, H. V., 174
LaBarre, W., 19 Murdock, G. P., 11, 19, 20, 24,
Labauret, H., 105, 113 25, 26, 27, 28,31,34, 65, 66,
Lamb, S. M., 319 76, 163, 203, 204, 300, 301,
Lane, R. B., 209, 211, 212, 213 302, 318, 369, 374
Leach, E. R., 149, 202, 203, 204, Murphy, R. F., 222
205, 219, 222, 235, 240, 241,

394 Author Index

Nadel, S. F., 268 Smelser, N., 370
Needham, R., 199, 200, 201, Smith, A,, 121
202, 204, 206, 207, 209, 210, Smith, R. T., 23, 24, 25, 26, 27,
211, 215, 218, 219, 220, 221, 31, 32, 33, 35
222 de Sonsberghe, L., 212, 213
Nimkoff, M. F., 302, 369 Spencer, B., 217
Spicer, E., 355, 356
Ogburn, W. F., 370, 371 Spier, L., 300
Spiro, M. E., 20, 30
Parsons, E. C., 348 Srinivas, M. N., 195
Parsons, T., 24, 25, 26, 27, 28, Sussman, M. B., 373
31, 34, 76, 378
Paul, B. D., 348, 350, 353 Tardits, C;, 176
Paulme, D., 102, 105, 113 Tax, S., 317
Pedler, F. J., 104, 105 Taylor, L. R., 340
Pehrson, R., 262 Thomas, N. W., 278
Peter, H.R.H., Prince of Greece Thore, L., 176
and Denmark, 151, 152, 153, Torday, E., 284
155, 157 Tyler, S. A., 321, 324
Pospisil, I., 324 Tylor, E. B., 9, 11, 45, 60, 76,
Powell, H., 301 78, 99, 181, 367

Radcliffe-Brown, A. R., 12, 30, Von Humboldt, W., 317

31, 65, 66, 76, 99, 117, 152, Villa R, A., 348, 352, 353, 354
228, 263, 271, 297, 302, 303, Vinogradoff, P., 120
313, 317
Rathray, R. S., 66, 68, 69 Warner, W. L., 309
Redfield, R., 348, 352 Weber, M., 263, 372
Richards, A. I., 231, 252, 253, Westermarck, E. A., 17
254, 255, 259, 261, 269 Wheeler, G. C., 11
Rivers, W. H. R., 200, 279, 301, Whiting, J., 174
317 Whorf, B. L., 317
Romney, A. K., 321 Williams, F. E., 60
Wilson, G., 255, 256, 257, 258,
Salisbury, R. F., 211, 213, 214, 259
216 Winch, R. F., 170, 369
Sapir, E., 310, 317 Winter, E. H., 243
Sawer, J., 159, 160, 166 Wisdom, C., 352
Schneider, D. M., 203, 205, 207, Wolf, E. R., 329, 337
220, 221, 222, 233, 320, 323,
324 Zelditch, M., Jr, 166
Seligman, B. Z., 54, 76, 79, 317

Author Index 395

 Subject index

Adoption, 91, 342^t patrilateral, 200, 207-223

Roman, 340-41 bilateral, 211, 214, 214-23
Adrogation, 340-42
Adultery, 64-81, 128 Demographic factors, 10
payment, 67 Descent, 9, 11, 96-8, 181,
Alliance, 105, 106, 112-15, 263-74, 281
129-32, 181, 183-98 Descent groups, 205-23,
see also Marriage 263-76, 290-95
Ancestors, 132 matrilineal, 66-81, 227, 245,
Avoidance relationships, 99, 252
102-15 patrilineal, 66-81, 227, 267
Developmental cycle, 83, 85-98
Bilateral systems, 23, 290-95 Divorce, 149, 225, 23in,
Blood-brotherhood, 54-62, 112, 248-60
113 Domestic group, 85-98
Bridewealth, 117, 119-32, 169, Double unilineal descent, 156
250, 286 Dowry, 117, 119-31, 13A-9

Child-rearing, 89-98, 331-44 Endogamy, 11, 48-62, 157

Church, 121-32 Evolutionary theories, 367
Clans, 216, 277 Exchange, 113-15, 129-32
; Class, 379-82 Exogamy, 11, 42, 45, 48-62, 65,
Class hypogamy, 240 69, 77-81, 90, 134-9, 250,
Complementary filiation, 267-74 281-8
Componential analysis, 9, 297, Extended family, 280, 283, 291
303, 318
Conjugal family, 371-82 see also Family, 17, 19-36, 42-4, 64-81,
Family, Nuclear family 83, 86-98, 276-88
Corporate group, 229, 252, 270, ‘Father-right’, 241, 244
271 Fictive kinship, 329
Cousin terms, 300 Filiation, 181, 266, 272
Cross cousin, 213, 300 Formal analysis, 317
classificatory, 203 Fosterage, 335-44
Cross-cousin marriage, 11, 53, Fostering, 33 In¬
61, 104-15, 131-2, 157, 181, voluntary, 343
183-98, 200-223, 284
matrilateral, 184, 200, 201-23 Genetrical rights, 251-60

Subject Index 397

Grand family, 280 Lineage, 31, 91-3, 264, 265,
273, 277-88
Incest, 11, 45, 50, 51, 56-62,
64-81, 250, 274 Marriage, 9, 47-62, 88-9, 102-4,
Inheritance, 90, 134-9, 151-61, 119-32, 151-61, 181,227, 368
191-8, 229, 271-4 alliance, 47-62, 77, 183-98
Initiation, 44, 57, 94, 95, 119 father’s sister’s daughter,
200-13
Joking relations, 10, 99, 101-15 gifts, 191-8
Jural authority, 96, 251 see also Dowry
mother’s brother’s daughter,
Kibbutz, 20, 30 200-13
Kin groups, 9, 123-32, 261-2 patterns, 140-47
bilateral, 254, 290-95 payments, 10, 120-32, 177,
Kin network, 135-9, 336, 372, 380 227-32, 240, 244, 258, 286
Kin terms, 110-12, 297, 307-15, plural, 76, 149, 163-78
320 see Kinship terminology preferential, 203, 213, 221-3
Kinship, 10, 11, 17, 38, 47-62, prescriptive systems,
52, 65, 78, 83, 183-91, 249-55 181, 199-223
256, 323, 363 prohibitions, 55-62, 65-6
categories, 212, 220 regulations, 183-5
classificatory, 90, 107, 216 residence patterns, 134-9,
matrilineal, 255, 276-81 169-78, 227, 278-81, 290-95,
patrilineal, 255 368, 373
system, 30, 42-4, 53, 106-10, stability, 225, 228-32, 233-46,
227-46, 254 248-60, 286
Kinship terminology, 185-91, transactions, 117, 119-32
290, 297, 317, 358 transfer of rights at, 227
Arunta type, 217, 218 Maternal dyad, 20-36
Australian type, 309 Matrilineal belt, 276
Crow type, 215, 301, 303, 321
Eskimo type, 309 Nuclear family, 17, 19-36,
Hawaiian type, 300 331-44
Iroquois type, 300
Kariera type, 217, 218 Oedipus complex, 17, 43
Omaha type, 215, 216, 301, Orphans, 343-4
303, 321
Komstvo, 339 Parent-child relationship,
38-41, 329, 331-44
Land tenure, 134-9 Polyandry, 149, 151-61
Legitimacy, 38-41, 158 Polygyny, 10, 117, 149, 151-61,
Levirate, 11, 165, 234, 236, 239, 163-78, 284
243, 245, 310,311 sororal, 165, 166, 169

398 Subject Index

Polykoity, 151 20, 29, 30, 38-41, 112-15,
adelphic, 152, 157 119-32, 227-46, 307-15
Potest as, 128, 340-42 Sororate, 11, 243, 311
Prestations, 90, 124, 192-8 Sponsorship, 333, 346
see also Marriage gifts Status, 51, 167, 271, 272, 332,
333, 334, 339
Reciprocities of rearing, 333 Stem-family, 144
Reciprocity, 60, 315 Structural approach, 26-8, 28
Residence patterns, 87-98 Succession, 90, 252, 276-88
Rites de passage, 93, 94
Ritual co-parenthood Taboo, 45, 302
(compadrazgo), 329, 346-60 incest, 65-6
Ritual sponsorship, 337-9 sexual, 174
see also Komstvo
Unilineal descent groups,
Segmentation, 91, 264, 267 66-81, 90-98, 228, 261,
Sex ratio, 257 263-74
Sexual offences, 66-81 see also Double unilineal
Sexual prohibitions, 45, 72-81 descent
Sibling groups, 31, 96, 152, 267, Uxorial rights, 251-60
268, 291, 295
Slavery, 91, 284, 285 Virginity payments, 258
Social change, 85-98, 175-8,
284-60 Wedding, 120
Social structure (organization), Wido inheritance, 165, 240

Subject Index 399

Penguin modern sociology offers essent Jack Goody is a Fellow
Readings over the whoie range of thinking of St John's College ar
and research, both classical and Director of the Centre
contemporary,, in sociology and social for African Studies,
anthropology. The distinguished advisory Cambridge University.
board is international in composition and USA $3.25
this is reflected in the individual volumes,
Cover design by
which are edited by leading authorities and
Peter Stone
contain, in many cases, material receiving
its first translation into English.
Kinship,and family patterns vary as widely Cover desi'
as the societies of which they form a base. Peter Ston
The complex kin networks of many
pre-industrial societies, in particular, play a
vital role in safeguarding social and
cultural stability. Many of the stresses of
developing societies are consequently best
understood through an awareness of their
traditional kinship systems, since these are
breaking up under the pressures of
urbanization and industrialization. (The
family patterns of western societies, where
these processes are more advanced, are
surveyed in a companion to this volume,
Sociology of the family.)
Jack Goody's collection of Readings
illuminates many aspects of a very
diversified field of siudv. and includes texts

[06/08/201