Вы находитесь на странице: 1из 34

The Structural Implications of Matrilateral Cross-Cousin Marriage

Author(s): E. R. Leach
Source: The Journal of the Royal Anthropological Institute of Great Britain and Ireland, Vol.
81, No. 1/2 (1951), pp. 23-55
Published by: Royal Anthropological Institute of Great Britain and Ireland
Stable URL: http://www.jstor.org/stable/2844015
Accessed: 15-03-2015 21:23 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Royal Anthropological Institute of Great Britain and Ireland is collaborating with JSTOR to digitize, preserve and extend
access to The Journal of the Royal Anthropological Institute of Great Britain and Ireland.

http://www.jstor.org

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

23

THE STRUCTURAL IMPLICATIONS OF MATRILATERAL CROSS-COUSIN MARRIAGE


CurlPrizeEssay, 1951
BY E. R. LEACH, M.A., PH.D.

CONTENTS
Part 1

BASIC ASSUMPTIONS AND SPECIAL CONCEPTS

i
ii
iii
iv
v

Part 2

THE LITERATURE ON KACHIN TYPE MARRIAGE, 1920-1951

i
ii
iii
iv

Part 3

Early Theories
TheMurnginControversy
LaterDiscussionof OtherVarietiesofKachinTypeMarriage
Le'vi-Strauss's
Theory

A NEW ANALYSIS OF THREE KACHIN TYPE MARRIAGE SYSTEMS

i
ii
iii
iv

Part 4

Local DescentGroups
DiagramLines: Local Linesand DescentLines
TypeMarriages
and Objectives
Assumptions
Diagrams

The GenuineKachinSystem
Batak
Lovedu
The ContentofPrestations

CONCLUSION

(845)

B 4

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

24

E. R. LEACH

INTRODUCTION

not mean to arguethatwomenhave no partto play


in the arrangement
of a marriageor that remotely
At firstglancethe themeof thisessay mightseem
situated
kinsfolk,
are
whollyignored;1 merelymean
excessively
narrowand pedantic;in fact,as I hope to
that
the
corporate
group
of personswho have the
show,it is veryappropriateto the termsof the Curl
most
decisive
in
say
bringing
about an arranged
Bequestcompetition.Firstlyit is a topicthatlies at
is
marriage
always
a
group
of
co-residentmales
the very heart of anthropologicalkinship theory.
representing,
as
a
rule,
three
genealogical
generations,
Secondlyit is a branchof kinshiptheorytq which
namely:
the
old
men
or
grandfathers,
the normal
a numberof significant,
and perhapsveryimportant,
adults
or
fathers,
and
the
young
adults
or
sons.
contributions
have been made duringthe past ten
In
practice,
membership
of
such
groups
is defined
years. Thirdlyit is a fieldto whichI myself,from
descent
as
well
as
by
In
residence.
this
essay
I am
my own experience,can make a new and original
concerned
only
with
systems
of
unilateral
(and
double
contribution.
unilateral)descentso thatI can formulatethe above
proposition
as follows:
The essayis arrangedin foursections.
Part 1-serves to establishcertainbasic definitions, In a unilaterallydefineddescentsystemwhere a
clan, or large scale lineage,ceases, for one
assumptiohns
and theoreticalobjectives.
reason
or another,to be a localised group,
Part2-reviews the literatureof the themeunder
then,
in
general,it ceases to be a corporate
discussion.
unitforthepurposesof arranging
a marriage.
Part3-provides materialfrommy own fieldwork,
The corporategroup which does arrangea
and examinesthe relevanceof this new
marriageis, in such circumstances,
always a
materialforthe analysisof twQotherwell
group of males who, besidesbeingmembers
documented societies which have not
of the same lineageor clan, sharea common
previouslybeen consideredfromquitethis
place of residence.
pointof view.
Part4-summarises the conclusionsthat may be In thisessayI shall referto corporategroupsof this
drawn from this review of theoryand kindas local descentgroups,or moresimply,wherever
ethnographic
fact,and specifiesa seriesof the contextis unambiguous,simplyas 'groups.'
Logicallyspeakinglocal descentgroupsthusdefined
propositions,whichnot only accord with
can
come about only in a limitednumberof ways.
thefactsas now known,butwhichare in a
The
followingwould appear to be the most likely
formwhich permitsof furtherempirical
possibilities:
testingin thefield.
(a) withpatrilinealdescentand patrilocalresidence.
PART 1. BASIC ASSUMPTIONSAND SPECIAL

CONCEPTS

(b) with matrilinealdescent and ' avunculocal"


residence(i.e., residencein the community
of
the mother's brother); succession to male
authoritybeing from mother'sbrotherto
sister'sson.2

Local descentgroups
Thereare two kindsof marriage. The firstresults
from the whims of two persons acting as private (c) with matrilineal descent and 'matrilocal'
organised
individuals;the second is a systematically
of
residence(i.e., residencein thecommunity
affairwhich formspart of a series of contractual
the wife) coupled with matrilateralcross
obligationsbetween two social groups. When I
cousin marriage (father's sister's son mention an institutionalisedor 'type ' form of
mother'sbrother'sdaughter);successionto
that
marriage,it is to thislatterkindof arrangement
I-refer.
In my view the social groups which ' arrange'
IThis termhas beenproposedby Murdock(1949,p. 17).
such a marriagebetweenthemselvesare, in almost
2
(1932,pp. 10,83).
rule;Malinowski
ThenormalTrobriand
all societies,of essentiallythe same kind. The core
villagewhenadolescent
brother's
A manmovesto hismother's
of such a group is composed of the adult males of and thenbringshis wifeto join himin thatvillage;cf. the
a kin groupall residentin one place. By thisI do CongoMayombesystemas analysedbyRichards(1951).

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

The Structural
Implications
ofMatrilateralCrossCousinMarriage
male authoritybeing fromfather-in-law
to
son-in-law.3

co

25

Cs

DiagramLines: local linesand descentlines


Figs. 1, 2a and 2b illustratediagrammatically
the
notionof local descentgroupsas resulting
fromeach
of the above situations. In thisessayI shall referto
suchdiagrammatic
local descentgroupsas local lines.
C3 - =
C03
X3
thelinesA.1-A.2-A.3, B. 1-B.2Thus in thesefigures
B.3; C.1-C.2-C.3 each representlocal lines. The FIG. 2b.-The line C.1-C.2-C.3 indicatesa matrilineal
local
kinshiprelationbetweeni
any two individualsin such descentgroupresulting
frommatrilocalresidenceand matrilateralcrosscousinmarriage
a diagram is always intendedto be classificatory
In
C2
succeedsC1 becausehe is husbandof x2
this
system
ratherthan actual; thus A.2 is classificatory
son of
cross
and potentialhusbandof x,. Onlywithmatrilateral
A.1: B.2 is classificatorv
sister'sson to B.1.
are C1 and C2 ofthesamedescentgroup
cousinmarriage
This notion of a local line is to be distinguished
fromthe parallelconceptof a descepnt
line (' line of a diagrammaticdevice for displayingthe categories
descent') whichhas frequently
beenusedbyRadcliffe- ofthekinshipsystemin relationto a centralindividual
Brown and his pupils. Descent lines have nothing calledEgo. The numberofbasic descentlinesin such
whateverto do withlocal grouping,theyare merely a diagramdependsmerelyupon how manydifferent
kinds of relativeare recognisedin the grandfather's
generation.It has nothingto do withthenumberof
At
local descentgroupsexistingin the society.4
as
betweenthe notionof local
Failure to distinguish
7 7
a local descentgroup)and descentline
line(indicating
42
0
2
(indicatinga set of kinshipcategories)has been the
sourceof muchconfusion.
betweenthese
difference
One particularly
important
A3
A
descent
line
of
is
this:
commonly
two
diagram
types
FIG. 1.-The line A.1-A.2-A.3 indicates a patrilineal local
e.g., grandfather,
comprisesat least fivegenerations,
frompatrilocalresidence
descentgrouplresulting
Ego, son,grandsonandeachoftheseindividuals
father,
is givenequal weight. A local lineon the otherhand
seldomcontainsmore than threegenerationsat any
S
b_
one time.5 As a child,Ego is a memberof a system
comprisingthe local descent groups of which his
*82
?i2
were membersby birth;
parentsand grandparents
as an adult,Ego is a memberof a systemcomprising
the local descentgroupsof himselfand his wife,and
the local descentgroupsinto whichhis childrenare
FIG. 2a.-The line B.1-B.2-B.3 indicates a matrilineallocal married. These two systemsof kinshipassociation
descentgroupresultingfromavunculocalresidence
overlapbutdo notnormallybothexistin theirtotality
one and the same time. Ego's fatherand grandat
3 The Garo pattem. Hodson (1921); Bose (1936). The
and their contemporariesare mostly dead
father
Bemba and Yao systemsappear to be partly,ifnot consistently,
are born. A diagram
before Ego's grandchildren
of thistype; cf. Richards(1951).
Of the other logical alternatives,consistentlypatrilineal- de?ignedto show descentlinesinsteadof local lines
matrilocal societies do not, I think, occur. Matrilinealfact.
patrilocal societies are reported but are probably ' ethno- tendsto obscurethisveryimportant
graphicerrors' in that these are reallyeithercases of double
descentor else not consistently
patrilocal. The Ila and Ashanti,
for example,have both at timesbeen describedas matrilineal(1951, p. 43).
4See Radcliffe-:irown
patrilocal,but in both societiesthereis an elementof double
5 A man may have classificatory
and classificagrandfathers
descent while the Ashanti have no ' normal' pattern of
residence; cf. Richards (1951) on the Ila, Fortes (1949, 1950) tory grandchildrenalive at the same moment,but it is not
on the Ashanti. cf. also de Josselinde Jong(1951), p. 190.
likelythattheywill bothbe membersof Ego's own local group

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

26

E. R. LEACH

Typemarriages
Readers of this essay will be familiarwith the
notion of typemarriageswhichhas been developed
to describethe variousformsof
by Radcliffe-Brown
institutionalised
marriage,regulationfound amotig
Australian tribes (Radcliffe-Brown,
1930, passim;
1951, pp. 41-42). Type marriagesform a very
convenientshorthandnotationand in this essay I
shallemploythefollowingseries:
1. Kariera type-' symmetrical
cross cousin marriage '. This systemapproves the simultaneous

a woman to group A, the serviceis never


in kind,thoughit mayof course
reciprocated
be reciprocatedin other whays-e.g., by
marriagepayments. In the ideal typea man
marriesthe mother'sbrother'sdaughter;he
is forbiddento marrythe father'ssister's
daughter.
My Kachin type includes the Australian
Karadjeri type and thus incluqes the much
discussedMurnginsystem.It shouldbe noted
howeverthat,in general,Kachin typesystems
lack thosefeaturesoftheKaradjeritypewhich
Australianmakethelattercharacteristically
e.g., the formaldivisionof the local descent
group into sectionscomposedof alternating
1951, pp. 43,
generations(Radcliffe-Brown,

or nearlysimultaneousexchangeof women
betweentwo local descentgroups. In the
ideal type a man marries the mother's
brother'sdaughterwho is sisterto his own
55).7
sister'shusband.
I amnotconcerned
withtheotherAustralian
and objectives
symmetrical
type marriagesbut it may be Assumptions
notedthattheArandatype,and theKumbainFor the purposesof this essay I assume that the
geri type,both approve marriagewith the threevarietiesofcrosscousinmarriagedefinedabove,
sisterof a man's own sister'shusband. They the Karieratype,theTrobriandtypeand the Kachin
differ
fromtheKarieratypeonlyin excluding type,can usefully,for purposes of comparison,be
frommarriagecertaincategoriesof women treatedas institutional
in the
isolates. I am interested
who would be admissible as 'mother's implicationsof such institutionalised
behaviourfor
sister'sdaughter' thesocietiesin whichsuchrulesoccur.
brother'sdaughter/father's
in the Kariera systemof kinship.6
The total literatureon the subjectof cross cousin
marriageis verylarge; the major part of it has been
2. Trobriand type--'asymmetricalcross cousin
recentlyreviewedby Levi-Strauss(1949). In this
marriage(patrilateral)'. This systempre- essayI am primarily
concernedonlywiththatpartof
cludesthe reciprocalmarriageof a man with thismaterialwhichdeals withKachin typemarriage;
the sisterof his own sister'shusband,but it I shall concernmyselfwith Kariera and Trobriand
to a systematic
amountsnevertheless
exchange typemarriagesonlyso faras is necessaryto provide
of womenbetweentwo local descentgroups. contrastsand comparisons.
The exchangeis completedonlyaftera time
lag ofone generation.In theideal typea man
7It is worthnotingthattheSouthIndian regulationdescribed
marriesthe father'ssister'sdaughter;he is by.AiyappanwhichforbidsmarriagewithF.sis.d. but approves
with both m.B.d. and own sister's daughteris not
forbiddento marrythe mother'sbrother's marriage
a case of Kachin typemarriageas it can resultin an exchange
daughter.
a.2 and b.3 are
of women betweengroups. See Fig. 'x'.
This kind of marriageregulationoccursin both simultaneouslym.B.d. and sis.d. to their respective
patrilinealas well as matrilinealsocieties. I husbands B.2 and A.2. (Cf. Aiyappan, 1934.)
shall use the description Trobriandtype
marriagein bothcases.

B,

Al

3. Kachin type-'

asymmetricalcross cousin mar-

riage (matrilateral)'. This systemprecludes


altogetherthe exchangeof women between
two local descentgroups. If group B gives

A23

A3i

J
3)-J

Ba
3

The same can be said of the somewhatanomalous Ambryn FIG. 'x '.-Patrilineal descent marriage with m.B.d. andior
system. (Deacon, 1927,pp. 328-9; Seligman,1927,p. 374.)
own sis.d.
6

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

27

Marriage
ofMatrilatqralCross-Cousin
The Structural
Implications

sense.
The particular
aspectofKachintypemarriagewhich the' function'of such rulesin a mathematical
interests
me is this. Wheresuch a systemof institu- For example:Givena rulesuchas thatwhichdefines
andgivenvariousothercommon
tionalisedmarriagerules exists in associationwith Kachintypemarriage,
local descentgroups,thena groupB whichprovides elementsbetweensocietyA and societyB, can we
thatsome otherunknown
wivesforgroupA is notcompensatedin kind. There infer,by logicalarguments,
'x' must also be commonto our two
characteristic
are thenthreepossibilities:
does not apply societies? And if we thinkwe can do this,how far
(1) That theprincipleof reciprocity
at all and thatgroupB obtainsno compensa- do empiricalfactsjustifysuch a claim?
tion;

Diagrams
is achievedby group A giving In my discussion of the literaturein the later
(2) thatreciprocity
groupB some formof economicor political sectionsof this essay it will becomeapparentthat
work serious misunderstandings
compensation-e.g.,marriagepayments,
have constantlyarisen
service,politicalfealty;
froma tendencyto confusestructural
dia:grams
with
reality.9In my own argumentI shall
(3) thatthreeor moregroupsA. B, C, makemutual ethnographic
referto diagramssuch as those of Figs. 3
arrangements
to ' marryin a circle'-C giving constantly
thatthe readershouldclearly
wives to B, who give wives to A, who give and 4. It is important
just how thesediagramsrelateto reality.
wivesto C again. In thiscase thewivesthat understand
In a systemofunilinealdescent,eitherpatrilinealor
C givesto B are,in a sense,compensationfor
matrilineal,Kachin type marriagehas the effectof
the wivesthatB givesto A.
groupingEgo's relativesinto at least threemutually
The implications
ofthesealternative
form
possibilities
exclusivecategories,namely:
thesubjectmatterof thisessay.
A. Groupscontaining'father'ssister'sdaughters'.
But let me be quite clear about what I mean by
B. Groupscontaining' sisters
implication.I am not concernedwiththe originsof
C. Grbups containing 'mother's brother's
institutional
rules. It seemsto me probablethatsuch
daughters'.
marriagerulesas we arediscussing
mayhaveoriginated
is permitted
to marryonlyintogroupsC.
Ego
(male)
in quitedifferent
societies. I am also
waysin different
is
to marry only into
Ego
(female)
permitted
in whatMalinowskimighthave
not greatlyinterested
A.
groups
called the overtfunctionof such behaviour. I have
no doubtthatin different
societiesone and the same
WithTrobriandtypemarriage,on the otherhand,
rule will servedifferent
immediateends; comparison a groupwhichcontainsa father'ssister'sdaughterin
in termsofsuchendscan therefore
onlylead to purely one generationwill contain a inother's brother's
in howeveris
negativeresults.' WhatI am interested
8

9 Cf.Radcliffe-Brown
ofLawrence
(1951,passim)forcriticism
and Murdockon thisscore.

This I thinkis clearlyshownby Hsu (1945).

AA

a, _____

*,
*

A
r -

CC

, -____,

,;"' :a2,d,A1=
--

43

AEG
A3

82
B1
baBa

ClC2
S1

e~~~~~~~~~~~~~~
C3

C3

.
a

FIG. 3.-Kachin typemarriagesystem(patrilineal).


Lines A, B, and C alone can be thoughtof as local lines. If the whole schemebe consideredincludinglines AA
and CC thenthe verticallines are descentlines(see text).

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

E. R. LEACH

28

Bt

A'
I~~~~~~~~~~~~~

A3
FIG.

B3

(matrilineal).
system
4.-Trobriandtypemarriage

Lines A', B', C' can be thoughtof as local lines,

daughterin the next.10Thereis thusno categoryof will be necessary. In the actual Kachin systemthis
local descentgroupswhich are ' non-marriageable' is the case (Leach, 1945),and also in the Australian
forEgo (male) althoughoutsidehis own clan.
Yir-Yorontsystem
(Sharp,1934). The muchdiscussed
Thisdistinction
betweenKachintypeand Trobriand Murnginsystemrequiresin all no less than 7 lines
.typeis made clear in Figs. 3 and 4. In Fig. 3 the (Warner, 1930-31), so that the diagram contains
threelines,A, B, C, can be takento representthree furtherlines AAA and CCC to the left and right
accord- respectively.
patrilineallocal descentgroupsintermarrying
But thiswide extensionof thedescentline diagram
ing to Kachin type marriage. In Fig. 4 the three
lines A', B', C1, can be taken to representthree does not necessarilyimplythat an equal numberof
matrilineal
local descentgroupsintermarrying
accord- local descentgroupsare associated with Ego's own
ing to Trobriandtypemarriage. In thefirstcase the group. If Fig. 3 denotes local lines (local descent
relationshipof groupB to groupA and of groupB groups)insteadof descentlinesthenthereis no reason
to-groupC is quite distinct. B receiveswivesfrom whyAA should not coincideor at any rate overlap
C and giveswivesto A. In thesecondcase thegeneral with B. This is the crux of,Murdock's and Leviof the Murnginsystem
typeof relationship
betweenB' and A' is thesame as Strauss's misunderstanding
that which exists betweenB' and C1, but merely whichhas beencriticised
byRadcliffe-Brown
(1951)."
If we are concernedwithdescentlinesit is always
shiftedone generation. This argumentfirstappears
truethatthe most satisfactory
clearlystatedin Fortune(1933).
diagrammodel which
In Kachin type systemsthe division of Ego's will representthe whole of a Kachin typemarriage
relativesinto threemutuallyexclusivecategoriesis a systemwill consistof some unevennumberof lines,
minimum;theremay be further
categoriesof a like Ego's own line beingcentrallyplaced. In contrast,
kind. In theoryfor exampleit mightseem, as in any systemof marriageregulationswhichapproves
Fig. 3, thattheremustalwaysbe a further
groupAA the marriageof a man with his sister'shusband's
relatedto A in the same way as A is relatedto B, sister(e.g., Kariera,Aranda,Ambryntypes)can be
and thattheremustalwaysbe a groupCC relatedto mosteasilyrepresented
bya diagrammodelcontaining
C in the same way as C is relatedto B.
an even numberof lines,Ego's own position.being
If we are merelyseekingto displaythe categories immaterial.Thisfactalso has led to muchconfusion.
of the kinshipsystemby a diagramof descentlines, In Fig. 3 the centralpart, lines A, B, C, can be
it is veryprobablethattheseextralinesAA and CC takenas a diagramof local lines;butthefullscheme,
includinglinesAA and CC, can onlybe a diagramof
10E.g., in Fig. 4 cl2 is father's
sister'sdaughter
to B12 but
c'3 is mother's
brother's
daughter
to B'3

'1

See also Part 2 (ii) of thisessay.

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

Marriage
ofMatrilateralCross-Cousin
The Structural
Implications
descentlines-it merelyshows the categoriesinto
whichEgo's relativesnecessarily
fall. It says
- nothing
at all about the totalityof Ego's society. That total
societymay contain any numberof local descent
groups; Fig. 3 only specifiesthreeof them,namely,
A, B, and C; the remaindermightpotentiallyfitin
anywhere.For instance, suppose there is a local
descentgroup X to whichB is eitherunrelated,or
only remotelyrelated; thenif Ego (male) marriesa
womanof X, X will thereafter
be ratedas category
C-' wifegiving'; but if Ego (female)marriesa man
of X, X will thereafter
be rated as categoryA'wife taking.'

PART 2.

29

THE LITERATUREON KACHIN TYPE MARRIAGE

1920-1951

Earlytheories
BeforeI cometo myownanalysisoftheseproblems
I must reviewthe literaturein which this type of
institutionalised
behaviourhas pyeviouslybeen discussed; forI recognisethatthe theoriesof 1950 are
only elaborationsof the theoriesof 1920 and 1,930.
If my commentsseem almost uniformly
adverse,it
is partlyforthefollowingreason. The writers
whom
I am about to mentionhave all in one wayor another
propoundedtheoriesabout 'cross cousinmarriage',
but theyhave done so withsundrydifferent
ends in
The majorityof writers
who have discussedKachin view. Some have been interestedin the originsof
typemarriagesystemshave failedto understandthis. human society, some in the algebra of kinship
Insteadtheyhave been led to assumethata diagram terminology,
some in assertingdogmaticallysome
such as Fig. 3 can serveto representnot merelythe principleof ethnographic
cause and effect,
and some
wholekinshipsystem,but thewholeof Ego's society. again simplyin denyingsuch assertions. Only one
Once this assumptionis made certain erroneous of them,it seems to me, namely,ProfessorLeviinferences
appear to followimmediately.In thefirst Strauss,has developed his theoryin the spiritof
place iflinesAA and CC denotelocal descentgroups, logical deductionand demonstration
which I have
thenAA has no husbandsand CC has no wives; it outlinedabove as my own objective. That beingso
mustthenfollowthatCC take wivesfromAA. The I am oftenin sympathy
withthe viewsof Professor
systemthenbecomescircular. Furthermore
the five Levi-Straussbecause I thinkI understandwhathe is
lines AA, A, B, C, CC, now cease to denotemerely trying
to do; likewiseI am oftenout ofsympathy
with
categoriesof Ego's relatives,they become actual other writersmerelybecause I do not understand
segmentsof the total society,and can be thoughtof whattheyare trying
to do.
as ' marriage classes', or perhaps as five strictly In Part 1 we have seennot onlythatthereare three
exogamousphratrieswhichby some mysticalprocess distinct
typesofcrosscousinmarriagebutthateach of
alwaysmanageto remainofexactlyequal size and sex thesetypeshas quite different
structural
implications.
composition.
Althoughthe significant
literatureon ' cross cousin
Radcliffe-Brown
(1951,passim)has rightly
criticised marriage'extendsback at least to the earlyyearsof
(Rivers,1907)theclearappreciation
ofthe
Murdock's analysisof the Murnginsystemon this thiscentury
is
importance
of
these
distinctions
a
-late
relatively
score; I shall likewisecriticisethe workof Hodson,
Westermarck
and
development.
others
Frazer,
had
Mrs. Seligman, Granet, Levi-Straussand others.
before
1920
vol.
indeed,
98
(Frazer,
1918,
2,
pp.
ff.),
But I would like to make it clearthattheproblemat
noted
occurrence
of
the
widespread
cross
asymmetrical
issue is not simplyone of understanding
the ethnographicfacts. In a numberof societieswkhich
possess cousin marriageof both varieties,but thesephenoKachin typemarriagesystemsthe nativeinformantsmena were not then generallyregardedas separate
12
fromreciprocal
crosscousinmarriage.
themselveshabituallyexplainthe intricaciesof their isolatesdistinct
As
late
1929
in
as
an
Britannica
article
Encyclopaedia
kinshipsystemby sayingthatthe societyconsistsof
3 or 5 or 7 clans which' marryin a circle', and it entitled" Cousin Marriage" Miss Wedgwood(1929)
requiresthe collectionof genealogiesto prove that (clearlyunderthe aegis of Malinowski)confusesall
thisdescriptionis a fiction. Furthermore,
cases can threetypesofcrosscousinmarriage,and merelytakes
be foundwherethreeor morelocal descentgroupsdo noteof theoccurrenceofthe Kachintype" in a tribe
in fact,as well as in theory,'marryin a circle' on a in Assam." Malinowski'sdogmatismwas perhaps
continuousand more or less exclusivebasis. The
as wellas thecontradictions
correspondences
between 12 Rivers(1921) howeverstressesthat'we need evidence
theexactdistribution
ofthethreevarieties,
ideal model and empiricalfact thereforecall for notonlyconcerning
butalso . . . . we need to knowwithwhatotherpracticeseach
commentand analysis.
formofmarriage
is associated."

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

30

E. R. LEACH

partlyresponsibleforthislack of discrimination.In the one hand and the ' wife givinglineage' on the
simplecause and effect other.
the heydayof functionalism
wereall thatweresoughtforin theway
relationships
The paperspublishedinthedecade1920-1930which
of explanationfor structuralphenomena. If an seem to have had the greatestinfluenceupon subarrangementcould be shown to sequenttheoryare thoseby Hodson (1921-25),Mrs.
institutionalised
"satisfya need," it was not considerednecessaryto Seligman(1928), and Warner(1930).
look further.In Trobriandmarriage,as practisedin
It was Hodson who seems to have been the first
the Trobriands,a man's heiris his sister'sson. As to recognisethatwhileKarieratypemarriagecan be
two exogamousgroups,
can be seen fromFig. 4, the sister'sson of a sister's. operatedby, at a minimum,
son is Ego's own grandson(son's son) and he will Kachin type marriageinvolvesat least three such
to have advocated a
daughter); groups.15 He seems therefore
(daughter's
marry
Ego's owngranddaughter
thusthistypeof marriageservesto conserveproperty classificationof Assam tribesinto those with dual
organisation,16
titlesin the patrilinealline in defianceof matrilineal organisationand thosewithtripartite
to the type of cross cousin marriage
inheritanceand descent.13 This led Malinowski corresponding
(1932, p. 86) to assertthat,in the Trobriands,such adopted. Hodson found in the Kachins an ideal
marriage-"is undoubtedlya compromisebetween example of the latter kind of organisation. He
two ill adjustedprinciplesof motherrightand father assumes this societyto consist of five exogamous
right; and this is its main raison d'etre." Miss ' divisions'marryingin a circle.All theethnographers
and assertedof have made it clear that the Kachins do not in fact
Wedgwood(1929) wentmuchfurther
rules,but
cross cousin marriagein general that " its most adhereat all closelyto theirowntheoretical
is on thetransmission
ofproperty."14Hodson apparentlyassumedthisto be due merelyto
effect
important
customs.
At the date of Miss Wedgwood's articleKachin a recentdecayof traditional
typemarriagehad alreadybeentreatedas an indepen- Mrs. Seligmanin 1928 had not, it would seem,
Hodson's papers. Her approachto the
dent isolate by one or two writers. The earliestof encountered
theseappearsto have been Gifford
(1916) in his study topicof Kachintypemarriagewas somewhatindirect.
of the Miwok. Giffordseems to have been mainly In 1927 therehad been publisheda posthumous
interestedin the correlationbetweenmarriagerule paper by Deacon (1927) whichindicatedthat there
and kinshipterminology.In the fashionof his day, existedon Ambrynan anomalousformof marriage
afterthemodel of Rivers,he perceivedthe systemas withthe sisterof the sister'shusbandwhichresulted
one of secondarymarriages. His own summaryof in the divisionof societyinto six 'marriageclasses'
is:
or sectionsvery much on the Australianpattern.
theargument
"The rightof a man to marryhis wife'sbrother's Studyof thissystemled Mrs. Seligmanto re-examine
published
daughterwas relegatedto his son, who thus theschemeofPentecostkinshipterminology
fit the
might
she
thought
which
(1914),
Rivers
by
marriedhis father'swife'sbrother'sdaughter,
the
she
reached
conclusion,
Instead
pattern.
Ambryn
in otherwordshis owncrosscousin(mother's
brother'sdaughter)" (Gifford,1922,p. 256). on purelyinductivegrounds,thatthemarriagesystem
on Pentecostmustbe oftheKachintype(matrilineal).
Gifforddoes not appearto have recognisedthatsuch
Her explanationof this rule was both novel and
an artangementwould imply a special continuing
curious.
relationshipbetweenthe 'wife taking lineage' on
Mrs. Seligman proposed to distinguishthree
systemsof descent,unilateral,bilateraland asymdescentwas deemedto be a
metrical. Asymmetrical
13 In termsof Fig. 4 B13whois thesecondgeneration
heir
of the othertwo.
features
combining
system
to B%-'sland titlesis son'ssontoB]3'and will marrya-3 who

is daughter'sdaughterto the same man. This of course


seemsto be implicitin severalpapersby
to therealratherthanclassificatory
father's 15 The argument
assumesmarriage
clearly
from1921onwards.It is formulated
Hodsonpublished
sister'sdaughter.
14 This type of functional
explanationof cross cousin in Hodson(1925,pp. 173-4).
' arethoseusedby
' dual' and ' tripartite
16 The expressions
was firstmadeby Hill-Tout(1907,p. 145) and was
marriage
from1921onwards
arguedin greaterdetailby Richards(1914). The argument Bose (1934). Huttonin variouscontexts
validinsomecases.In theTrobriands writesof Assamsocietieswith' dual' and ' triple' divisions
mayofcoursebe perfectly
formofcross
' at all butdoes notassociatethesewithanyparticular
are ' arranged
ofmarriages
forexampleonlya minority
seems
and theseare usuallyin householdssuchas thoseof chiefs cousinmarriage;cJfHutton(1921a). Hodsonhimself
usuallyto writeof ' dual' and ' multiple'organisation.
property.
inheritable
whichpossesssubstantial

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

Cross-Cousin
The Structural
Implications
ofAMatrilateral
Marriage

31

"Descent maybe said to be asymmetrical


whenone
boy has only his father'sblood and the girl
has only her mother'sblood. . . . But the
form'worksin a submerged
mannerwhilethe
dominantformonly is responsiblefor clan
motherhas the blood of her brother'sson
becausethelatter,beinga son, has theblood
organisation
(or any otherformof grouping).
In thisformof descentthe dominantformis
of his father.... In otherwordsa girland
so far as
her motherare conventionalised,
recognisedby both sexes, but . . . the submatingis concerned,into siblings,but they
mergedformis recognisedby one sex only.
are not practicalmembersof the paternal
Thus, with dominantlymatrilinealdescent,
sib."17
men and women both recognisematrilineal
descent,but men also recognisepatrilineal Kulp's wording is far from clear, but what he
descent while women do not" (Seligman, apparently
intendsto say is thata womanis identified
withher motherand a man withhis father,so that
1928,p. 536).
A fewpageslaterthissituationis further
explainedby sex relationsbetweena man and his father'ssister's
sayingthatin a matrilineal
incest.
society" a womanmarries daughterare the equivalentof brother-sister
haveclaimed
intothegroupof herfather,a man intothegroupto Mrs. Seligmanmight,I think,justifiably
whichneitherbelongs." Thesetwostatements
descent.
do not thisas a concreteexampleof asymmetrical
even if an argumentof thiskindmay
appear to be quite consistentbut what seems to be
Nevertheless
intendedis that,in termsof Fig. 5, Ego (male) has sometimeshelp to explain the mental associations
membershipin B and C 'groups' simultaneously;wherebymarriagewiththe father'ssister'sdaughter
whilehissisterEgo (female)is a memberofB ' group' may come to be thoughtof as incestuous,it tellsus,
only.
implicationsof
nothingabout any positivestructural
the Kachin typemarriagerule. That beingso, I do
A
BC
notfeelmyself
thatMrs. Seligman'spapercontributed
verymuch to the immediatetopic of our present
discussion,even thoughit may have been important
as a contribution
to thetheoriesofincestand exogamy.
EG;O
eg
Howeverthatmay be, theconceptof asymmetrical
descent,and the associated notion that the intermArrfeae of'grup
C-th exc aueo
h
marrying
'groups' in a Kachin typesystemmay be
'gous in quetio isFIG.
obcr.Serahrhstt
S
regardedas ' marriageclasses,'has undoubtedlyhad
In Mrs. Seligman'sown diagram(I1928,p. 542) the very considerableinfluence,both upon subsequent
as circular-malesof ' group' A theoryand upon subsequentfieldwork-notalways
systemis represented
marryfemalesof ' group' C- theexactnatureof the withveryfortunate
results.18
' groups' in questionis obscuire. She ratherhesitatinglysuggeststhaton Pentecostthese'groups ' may The Murngincontroversy
be regardedas 'marriageclasses' on the Australian
The firstmodernstudyof a Kachin typemarriage
pattern,and seemsto arguesimilarly
forKachin type systemin actual operation was Warner's report
marriagesystemsin Assam. At the same time she (1930-31)on theMurngin'9,
publishedin 1930. Rival
as Hodson had done,thatwhiletheremust interpretations
recognises,
of the data were publishedby Elkin
be at leastthree' groups' in sucha system,
theremay (1933) and Webb (1933) in 1933,and controversy
has
wellbe morethanthree(1928,pp. 550, 553).
raged ever since. The latest contributionto this
Mrs. Seligmanseemsto have evolvedherprinciple debate a remarkablebroadside Radcliffe-Brown
is
by
of 'asymmetricaldescent' as an inference
fromthe (1951),which,whileeffectively
theearlier
demolishing
studyof kinshipterminologies.She did not claim
thatany such systemhad been reportedby anthrolessambiguous
as itincludes
pological fieldworkers.In point of fact however 17 Kulp'soriginaltextis slightly
references
to
a
diagram
here
omitted.
of thesorthad alreadybeenreportedfrom 18 See Part2 (iii) below.
something
China. Kulp (1925, p. 168) had explaineda South
19 Murnginis a termappliedbyWarner
to thepopulation
of
China Kachin typemarriagesystemas follows:
eastern
Arnhem
Land. ThoughWarner
speaksoftheMurngin
'tribe' thisis a misnomer.As Thomsonpointsout " tribal
"The latterniarriage(i.e., withtheF.sis.d.)is taboo organisation
is conspicuousby its absencefromtheintricate
because of the traditionalattitudethat the socialorganisation
of thearea" (Thomson,1949,p. 11).

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

E. R. LEACH

32

ofMurdock(LawrenceandMurdock,1949), theyare also distinctfromone another. It is, then,


arguments
that readersof Warner'spaper
veryunderstandable
stillleaves somemattersunexplained.
I do not proposeto recapitulatethe whole of this shouldhave supposedthateach of the sevenlinesin
argument,but will merelyexaminefurthercertain his diagramdenoteda separategroupor setof groups
aspectsof the matterthat have so far tendedto be of actualpeople.
That granted,it was inevitablethat Lawrence,
somewhatneglected.
The crux of the debate is this:-Warner displays Murdock and Levi-Straussshould all inferthat the
the Murnginkinshipsystemby means of a diagram whole system is really circular. Lawrence and
of 7 descent lines on the model of my Fig. 6. Murdock(1949), in pointoffact,notonlyassertthat
and hisimmediatepupils,e.g.,Sharp 'each ofthesevenlinesofWarner'sdiagramrepresents
Radcliffe-Brown
(1934), Thomson(1949), take it forgrantedthatthe a separatedescent group, they claim that thereis
diagramimpliesno morethanthis. Otherreadersof also an eighthsuch descent group which Warner
Warner'spaper,in particularLawrence,Murdockand failed to notice and that the whole eight groups
Levi-Strauss,have howeverassumedthatthe 7 lines 'marry in a circle' in the manner proposed by
represent,not merely descent lines, but actual Hodson forthe Kachins. Not onlythat,but each of
theseimaginarydescentgroupsis further
subdivided
segmentsof thetotalsociety.
ridiculedMur- into 4 sections to produce in all 32 'classes.'
Radcliffe-Brown
(1951) has recently
but I thinkit is a Apparentlythe theoryis that Ego can only marry
dock for this misunderstanding,
moot point how far Warnerhimselfwas clear upon into one of theseclasses,whichseemsto implythat
the matterat issue. Althoughin the firsthalfof his he mustmarrythemother'sbrother'sdaughterwho is
referto his also father'sfather'ssister's daughter'sdaughter's
paper Warnerdoes, fairlyconsistently,
diagramas if it werea systemof descentlines only, husband'ssister'shusband'ssister'shusband'ssister's
husband's sister-a proposition which Radcliffein thesecondhalfhe also assertsthat
regardsas ludicrous!
" all perpendicular
are strongand Brown(1951, p. 53) rightly
relationships
if
were
case
thatthe Murnginwere
Actually,
it
the
unbreakablesincetheyare patrilineallinesof
divided
16
easily
recognisabledistinct
into
8,
or
32
father,sons, daughters,brothersand sisters
a
fantastic
of this kind
even
system
local
groups,
belongingto the same totemic clans and
would deserveserious examination. But since this
families"
interlocking
kind of orderingof local groupsdoes not, so far as
and
"each of the sevenlinesof descentis builtout we know,exist,the Murdockversionof the Murngin
considered.20
need not be further
of the restrictedfamily,whichpreservesits social structure'
continuity
bythepatrilineallaws thatregulate 20 Considerableadditionaldata is necessary
beforea fully
analysisof the Murnginsituationcan be made,
descentamong fatherand sons" (Warner, satisfactory
butthisextrainformation
is nowpromised
(cf Elkin& Berndt,
1931,p. 172).
But,as we have seenin Part 1 ofthisessay,ifthelines
representlocal descent groups then group A and
groupC are not onlydistinctfromEgo's own group
AAA

AA

1951). It seemspossiblethatthekeyto thesituation


maybe
in thisarea, of numerous'linguistic
foundin theexistence,
of whichdoes not coincidewiththe
groups'thedistribution
distribution
of populationby locality(cf. Elkin & Berndt,
1950: Firth,1951).

B
MARlMO0

7L\

=5

momo

1=)\=)
cirnd

A=

cc

c
A

GA

mori
mokul

MARt
MAR
ELKE

cc
momo NTIEL

elkM
=/
arndi

EGO

FIG. 6.-Diagram of same typeas Fig. 3 givingskeletonof Murngindescentline diagram.

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

GAW

0mart

ER

Omokul

The Structural
Implications
of-Matrilateral
Cross-Cousin
Marriage

cc
MARIEI

MARI42

ELKER

ccc

Bc
I;

momo

arn i

EGO'S

33

MARIKMO omol

=2

oarndi

~ ~

~~~AI

=1AI

mari

mar'

ELKER

mokul G WELmPUI GAWEL

EGO
MOIETY

OPPOSITE

MOIETY

FIG. 7.-Top lefthandcornerofFig. 6 rearranged.B; CC; C; and CCC areherefourlocal descent


groups. B and CC in one

C and CCC in theother.


moiety,

But while I concede that Radcliffe-Brown


has againsta similarpair of linkedlocal groups,that of
successfullydemolishedthe notion that the seven the mother'sbrother(GAWEL) and that of the
descentlinesof Warner'sdiagramrepresent
segments mother'smother'smother'sbrother'sson (NATIof the total Murnginsociety,he has so farfailedto ELKER).21make clearjust whatis the degreeof correspondence The local descentgroup CC whichis paired with
betweenthese descentlines and the local descent Ego's own descentgroup 13need not necessarilybe
groupswhichconstitute
actualMurnginsocietyat any a part of Ego's own patrilinealhordebut it may be,
and thisis the crucialfact,clearlystatedby Warner,
particularpointin time.
of the
This point can I thinkbe cleared up by a little which invalidatesMurdock's interpretation
reasoningfromfirstprinciples.
facts. In Fig. 7 MARI (line CC) and MARIKMO
My Fig. 6 is a skeletonof Warner'sfull diagram (line B) cannotbe one and the same personbut they
of 7 descentlines. I have filledin onlythe kinship can be clan brothers.22
termswhichdenoterelativeswho are genealogically Crucialto Warner'soriginalanalysisis a discussion
linkingthethreeindividuals
relatedto Ego whilehe is stilla youth.The persons ofthestatusrelationships
denoted by the lower and left hand portions of Ego (line B), GAVVEL(line C) and MARI (line CC).
formulated
by
Warner'sdiagramonlybecomesociallyimportantto
The generalprinciplesof reciprocity
Ego afterhe has marriedand acquiredoffspring.
thissametop righthand cornerof
Fig. 7 represents
betweenthe different
Fig. 6 redrawnto show the moietyarrangement. 21 To make clear the relationship
alteredWarner'sspelling:
In thislatterfigurethelinesB, C, CC and CCC need kinshiptermsI have
Warner
Leach
no longerbe thoughtof simplyas descentlines; they
Natielker
Natchiwalker
Momo-elker
Momelker
heredenotelocal descentgroupsof real people alive
Mari-elker
Marelker
at the same time and related to Ego duringhis ThisI thinkis legitimate
and spelling
in viewofthecomments
childhoodand youth. This diagramshowsthe logic of Radcliffe-Brown
(1951,p. 49) and Thomson(1949,p. 77).
werediminutives
notesthatthesethreetermsof the kinshipterminology
much more clearlythan Warnerhimself
any of the constructions
providedby Warner. Ego's of22Nati,Momo,and Marirespectively.
Warner(1931,p. 180) " theybelongto thesamemoiety
ownlocal groupB is alliedwiththatofEgo's mother's and frequently
to thesameclan." Cf Warner(1937,pp. 17,
mother'sbrother(MARI), group CC, and balanced 28-29).
(845)

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

34

E. R. LEACH

is represented
as one ofequilibrium.
Mauss in L'Essai surle Don would lead us to expect theoverallsystem
Thereremainsomegaps in Warner'sargumentand
thatKachintypemarriagewillnormallybe associated
with some scheme of giftexchangebetween'wife it would stillappearthathis explanationrequiresthat
side of Ego's system
giving' and ' wifereceiving' local descentgroups such the groupson thewifereceiving
supplying
himwithvaluablesforwhich,
that,on balance,the 'wife giving' groupwillreceive are constantly
theyhaveno sourceof supply. This slight
fortheloss ofthewomanand her ultimately,
somecompensation
offspring.In everyempiricalcase of Kachin type paradox is howeverlargely resolved by a recent
fromThomson(1949) who indicatesthat
marriagefor whichthe ethnographicdata is at all contribution
adequate,thisdoes appearto be thecase; though,as the exchangesbetweenmother'sbrotherand sister's
we shall see, theformof compensationis not always son are but one elementin a widesystemof economic
quitewhatmightbe expected.
exchanges,and that the ultimatesource of those
on the
This immediately
raises a problemof theoretical valuableswhichare not easilymanufactured
foreigntradewithIndonesiansand
importance.If,in termsof Fig. 3 or Fig. 6, groupA spot is continuing
is, on balance, always givinggiftsto group B and Europeans23.
groupB to groupC, does notthissuggesta permanent This discussionof theeconomicand statusrelations
status difference
in which C is seniorto B and B implicitin a Kachin typemarriagesystemis, to my,
iftheoverallsystemis not mind,the mostimportantthemeto be derivedfrom
seniorto A ? Furthermore
' circular,' what systemof reciprocities
permitsthe Warner's study,but it is a theme which has in
goods which C accumulatesto returnback to A generalbeen ignoredin the later literatureon the
Murngin. Instead,thewholeweightofthedebatehas
again?
Warner examinedthis problem in the Murngin centredon how far the eightsub-sectionsfoundin
situation and his conclusions are on the whole Murnginsocietycan be regardedas marriageclasses
convincing. On his analysis,personsin Ego's own and " sur la maniere dont se fermele cycle des
group (Fig. 7, line B) are, on balance, constantly mariages" (Levi-Strauss,1949, p. 245). The basic
has already
givingvaluablesto personsin GAWEL's grouplineC, errorinthese' Murdockstyle' arguments
and GAWEL does occupy a position of status been explained.
FinallyI wouldremarkthat,to mywayofthinkirig,
superiority
towardsEgo. Butin thesamewaypersons
in line C are givingvaluablesto personsin line CC, Radcliffe-Brown
(1951,p. 55) is ratheroveranxiousto
and MART occupiesa positionof statussuperiority see the Murnginsystemas merelya variantof the
towardsGAWEL. ButbetweenEgo and MARI there generalAustralianpattern. Here I am in agreement
forEgo is a kindofritualsuccessor withLevi-Strausswho regardsKachin typemarriage
is greatsolidarity,
to MARI, inheritinghis names (and possibly his as a legitimateisolate. The fact that the Murngin
systemfallsinto the Kachin typecategorymakes it
valuables).
Warner's argumentthereforeseems to be that comparablein importantrespectswithKachin type
whetheror not groupB and CC are actuallysegments systemsoccurringoutsidetheAustralianfield.
of the same clan they are close allies and stand
togetherin balancedoppositionto the two groupsC Later discussionof other varietiesof Kachin type
and CCC whichare similarlyallied and withwhich, marriage
Writersfromotherareas wereequallyslow to take
takenas pairs,thereis an exchangeofwomen-though
not an exclusiveone. This is thepositionwhileEgo up Warner'shintthatKachin typemarriageis likely
of themother'sbrother's
is still a youth. Later in life he becomes firsta to implya statussuperiority
' mother's brother' and then later a 'mother's local descentgroupoverthatof Ego's own.
The topic of Kachin type marriage has been
brother'sbrother' and other systemsof four local
is
to
on a numberof occasions since 1930 by
discussed
sectors
into
of
lines come
say
being-that
and otherIndiananthropologists
linesA, AA, and AAA in Fig. 6. In each case there K. P. Chattopadhaya
will be this same balance of a pair of closelyallied (Chattopadhaya,1931; Das, 1935, 1945; Bose, 1934,
local descentgroupswhich,takenas a pair,exchange
wives with anothersimilarpair on a non-exclusive 23 Readers of Thomson'smonographmay note that he
" whereasWarnertranslates
translates
gurrutoas " relatives
basis.
I think,
Thisservesto confirm,
system."
gurratuas " kinship
the
status differencethatWarnerhimselfwas confusedas to how farhis abstract
In this mannieralthough
' is admitted, kinshipsystem
between' wifegivers' and ' wifereceivers
denotedlivingpersons.

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

35

Marriage
Implications
ofMatrilateralCross-Cousin
The Structural

1935, 1937a, b; Roy, 1936). The argumentsput in both the archaicand earlyclassicalperiods. The
forwardby these writersall derive directlyfrom data used forthispurposeare kinshipterminologies,
Rivers,Hutton,Hodson and Mrs. Seligman. Two forms of ritual adopted from ancestor worship,
themesconstantlyrecur; the claim to demonstrate quotationsfromtheChineseclassicsand so on. The
that Kachin type marriageis an outcome of the conclusionsare that the archaic systemwas of the
conquest by a patrilinealminorityof a matrilineal Kariera type but based on matrilocalinstead of
societyhaving dual organisation,and an insistence patrilocalmarriage-i.e, a matrilocal4 sectionsystem
upon the empiricalexistenceof 'marriage classes' -while at a later,earlyclassical,periodthisorganisaof the type postulated by Mrs. Seligman. Both tion was somehow convertedinto a systemof 8
argumentsare expoundedwithingenuityand much patrilineally
defined'categories' (in effect'marriage
learnedalgebia, but appear otherwiseto lack merit. classes') whichmarriedin a circle accordingto the
early
On thecontrary,
theassumptionthattheclansamong Kachintyperule. In otherwords,theimaginary
theOld Kuki ' tribes' of Manipurare in someway a classicalChinesesystemof Granetwas a 'Murdockspeciesof' marriageclass ' has now so prejudicedthe Murngin' system. Granet's reconstructionis an
of thisregionthatforcom- impossibility
ethnographic
description
forjust the same reason as Murdock's
of the Murnginsystemis an impossiparativepurposesthematerialis almostuseless. For interpretation
examplesystemsof 'marriagein a circle' have been bility.
As Granetdisdainsto citetheworkofotherscholars
claimedfortheChiru,theChawte,thePurumand the
Tarau. In not a single case does the empirical and insists upon describingfairlysimple anthroevidenceprovidedby the ethnographers
tendto sup- pological situationsby means of a highlycomplex
notationof his own invention,it is
port thisproposition(Das, 1934, 1945; Bose, 1934, diagrammatical
to trace preciselythe source of his ideas.
19355.1937a, b; Roy, 1936). The only positive difficult
evidencethatdoes emergeis that in any one village It seemsreasonablyclear howeverthathe musthave
patrilineagesstand in more or less stable 'wife been familiarwiththeworkof both Radcliffe-Brown
giving'-' wife receiving' relationship, and that in and Warner. Levi-Strauss'slavish praise of his
thusseemsunwarranted.
any one villagethereare statusdifferences
between originality
patrilineages.These howeverare not the inferences The chiefmeritof Granet'sworkis thathe brings
whichtheauthorsthemselves
drawfromtheirdata.
the discussionback to the point whereWarnerhad
Similarargumentsto the effectthat Kachin type left it. He stressesthe theme that an arranged
marriageis necessarilycorrelatedwith a systemof marriageis not a one sided transaction;it is part of
threefoldpatrilinealmarriageclasses marryingin a an exchange. In Kariera typesystemsit is a direct
it is an
circle,and that it representsan evolutionfroman exchangeof women; in Kachin type.systems
thegiftsin
earlier matrilineal dual organisation have been exchangeof womenforgifts(prestations),
propoundedbyOlderogge(1946) and Mrs. Ruhemann turnbeingexchangedwithanothergroupforfurther
(1948), but I do not findtheirarguments
convincing. women. Thus logicallywe shouldbe led to consider
Van Wouden (1935) has formulateda theoretical Kachintypemarriagenot simplyas a phenomenonof
schemeof 16 marriageclasses marryingin a circle kinshipin isolation,but as a phenomenoninvolving
which he claims as the basis of modern East- the inter-relation
betweenkinshipinstitutionsand
But Granetdoes not pursue
Indonesiansocialsystems. de Josselinde Jong(1951) economicinstitutions.,
has recentlyattempteda rathersimilaranalysisfor thisaspectof thematter.
Like Hodson, Granetfoundin the Kachin system,
Minangkabau. These schemes are purely hypoan ideal modelfor
theticaland do not correspondat all closelywithany as describedby theethnographers,
crosscousinmarriage.
facts.24
recordedethnographic
a systembased on matrilateral
Anotherversionofthesamethemeis themonograph Even so itis notablethathe does nothesitateto adjust
recordwhereverthe reportedfacts
publishedby Granet(1939). This sets out to recon- thisethnographic
structthekinshiporganisationof theancientChinese failto fittherequirements
ofhishypothetical
historical
reconstructions.The Kachin systemas reportedby
Granetis much more remotefromrealitythan the
24 Another
Dutchscholarwhohas written
extensively
on the
Kachin
systemas understoodby Hodson.
of Indonesiais Fischer,buthis
Kachintypemarriage
systems
havenot,I think,advancedthetheoretical
writings
position. ThoughGranet'smonographprobablytellsus little
it had theusefuleffect
about ancientChinesehistory,
4f.Fischer(1935and 1936).
c2

(845)

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

36

E. R. LEACH

a familyof lower standingto a familyof


higherstanding. For it is the customthat
to come
parentsprefertheirdaughters-in-law
from lower familieswhile they wish their
daughtersto marryinto higherfamiliesthan
themselves.m.B.S. marriagemeansfora girl
to enterin a lowerfamilythantheone of her
origin....

to enquire into the


of stimulatinganthropologists
Kachintypemarriageamongmodern
factsconcerning
Chinese25

Though nowhere elevated to the status of an


absoluterule,theredoes appear to be a verygeneral
in certainareasofChinato approvemarriage
tendency
withthe mother'sbrother'sdaughterwhile banning
marriagewiththefather'ssister'sdaughter.Fei (1939)
and Hsu (1945) have both looked for 'functional'
explanationsof a simplecausal type.
Fei's explanationis thatmarriagewiththe F.sis.d.
is barred because, in such a system,Ego (male)'s
grandmother
(F.m.), who is the tyrantof Ego's own
household would then be of the same lineage as
from
Ego's bride(F.sis.d.). Ego's motherhas suffered
and would tend to
the tyrannyof the grandmother,
persecutetheyoungbrideinrevenge.m.B.d.marriage
and daughterhas the reverseeffect;mother-in-law
in-laware of thesame lineage. The ban on marriage
with the F.sis.d. is thus said to promoteharmony
between mother-in-lawand daughter-in-law(see
Figs. 8 and 8a).
As0

2. MarriageoftheF.sis.d.typewillshrinkthecircle
of kinshipand reducethenumberof relatives
who maybe of help....
3. In all Chinese provincesthe custom is for a
marrieddaughterto returnto her parents'
fromtimetotimeforperiodsofsojourn..
family
She enjoysa definiteprivilegedplace in this
house as contrastedwith her place in her
husband's house where she is the least
privilegedof all members. Now F.sis.d. and
m.B,S. marriagewill place a mother'sand
daughter's position in jeopardy; in one
householdthe motheris privilegedand not

6C2
02a

C3

B,

LC2

B2

LC3

B3i

C=
OC

'-

C2

C3iC3

marriage,
cl, c2 and
FIG. 8.-With'Trobriandtype (patrilineal)marriagecl is FIG. 8A.-WithKachintype(patrilineal)
fromsame local descentgroupas c3. a2 whohas C3 are all of samelineageand should(in theory)be friendly
originally
to one another.
on c3.
herself
at handsofcl willrevenge
suffered
persecution

Hsu has rightlyargued that explanationsof this


kindinvolvea largenumberof unstatedand unverito show
assumptions.In an effort
fiablepsychological
that no one functionalexplanationcan have any
explanations
generalvalidityhe cites five different
which were given to him at various times by
" sophisticatedyoung scholars, and old fashioned
school teachers,elderlyladies withfixedviews and
youngerladies of modem attitudeand extraction,
middle aged labourers and their peasant wives."
" of Kachin type
These several alleged " functions
marriageare of considerableinterest. I quote:
from
" 1. Marriagefora womanmeanstransference
25

reviewofGranet'sbook.
Hsu (1940)in a critical

thedaughter;whilein anotherthepositionis
reversed. It makespsychologicaladjustment
betweenthetwo difficult....
4. F.sis.d. and m.B.S. typeof marriageis actually
a returnof bone and flesh....
Childrenand parentsas well as brothers
and sistersare bone and fleshto each other.
The father'ssisterby givingherowndaughter
in marriageto her brother'sson will have
whatvirtuallyis her own permanent
effected
returnto her parents'home. . . it is an
ominousforecastof divorce....
5. The Kiangsutypeof argumentthatF.sis.S. and
m.B.d.marriagemakesforharmoniousfamily
relationshipand F.S.d.-m.B.S. marriage

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

Marriage
ofMatrilateralCross-Cousin
The Structural
Implications

37

destroyssuch harmony-whichFei -endorsed topicofthisessay,butit also embracesa muchvaster


"26
field.
ofthebook is thatit attempts
My principalcriticism
Whatseemsto me veryinteresting
about thislistis
that the firstthreeexplanationsare all essentially far too much. Instead of being contentto tryto
structural
explanations,in thattheyare expressedin establish correlationsbetween particular kinship
termsof persistingrelationsbetweenlocal descent structuresand a limited variety of institutional
groups; whilethelast two explanations,
whichoddly dimensions,the authorseemsto aim at establishing,
thegenerallaws of development
enoughare theonlyoneswhichseemto have received or at leastindicating,
societies,ancientand modern,
all
Asiatic
governing
support from the anthropologists,are essentially
This enormous prosophisticated.
and
primitive
psychologicalexplanations,in that they relate to
grammeis only covered by adopting a decidedly
temporaryrelationships
betweenindividuals.
It is onlythestructural
typeofexplanationthathas cavalier attitudetowards the facts of historyand
any real relevanceto our presentdiscussion. Of the ethnography.In the courseof a long,thoughrapid,
of all Australiaand
latter,explanations1 and 3 are structurally
the same journeythroughtheethnography
scatters in
Levi-Strauss
Asia
mainland
of
most
-they both derive from the assumptionthat the
' wifereceivinggroup' willbe of higherstatus than profusion analytical suggestionsof the greatest
the' wifegivinggroup.' A rephrasing
ofexplanations brilliance. But too oftentheseideas are misapplied,
detail,or
2 and 3 mightread " everylocal descentgroupmust eitherbecause of weaknessof ethnographic
a
to get
hurry
of
in
much
too
is
the
author
because
,have ' wife receivingpartners' and 'wife giving
I
am
concerned
excitingthings.
and
more
on
to
bigger
partners' and these two categoriesmust not be
here only withthat part of Levi-Strauss'sargument
confused."
which
deals withKachintypemarriage. Levi-Strauss
Fromthisthefollowingsimpleinference
seemsperon thistheme
seemsto considerhis inferences
himself
missible:-ChinesewhopractiseKachintypemarriage
logicaldeductionsfromtheworkofHodson,
takeitforgrantedthatin termsofFig. 3 seniority
runs as strictly
fromleftto right. GroupAA rankshigherthanA, Warnerand Granet,aided by his own re-examination
A higherthanB, B higherthanC, C higherthanCC. of the originalKachin sources. But are theselogical
It willbe notedthatthisis the reversedirectionof deductionsin factvalid?
Levi-Strauss'sreading of the Murnginsituation
seniorityfrom that previously inferredfor the
to resembleclosely that of Murdock, and
appears
Murngin. With the Murnginthe mother'sbrother
(group C) rankshigherthan Ego (group B) and on mustbe dismissedforthereasonsalreadygiven. He
notes Warner'sviews about the superiorstatus of
balance receivesgiftsfromhim.
It is then highlysignificant
that in the Chinese Murnginmother'sbrotherto sister'sson and the
between
bythesolidarity
systemthebalance of paymentsgoes fromwifegivers balancingoftherelationship
sister's
son,
and
daughter's
brother
mother's
mother's
to wifereceiversand thattheexpressfunctionof this
interas
a
psychological
the
argument
but
dismisses
dowryis to raise the statusof the bridein her new
pretationof social facts.28 I concludefromthisthat
home (Fei, 1939,pp. 43-44).27
either I or Levi-Strausshas misunderstoodwhat
was tryingto say. If myprecisof Warner's
Warner
Le'vi-Strauss's
Theory
a
argument(p. 32 f.) is a valid one, thenit is strictly
In 1949 Levi-Strausspublishedhis formidableand
structuraland not a psychologicalargumentand
ambitious work Les Structureselementairesde la
indeed it is an argumentof just the same kind as
Parentc. This is at once a contributionto incest
in other
himself
putsforward
thosewhichLevi-Strauss
to all
theory,a studyof the relevanceof reciprocity
partsof his book. The confusionI thinkhas arisen
institutionalised
formsof marriage,an analysisof the
and in part
in partfromWarner'sunclearstatements,
structural
implications
oftheseveralvarietiesof cross
fromLevi-Strauss'sown tendencyto confusedescent
cousin marriage,and a general theory of social
lineswithlocal lines.
evolution. The scope of the workthusincludesthe
For his Kachin arguments,Levi-Strauss,having
of Hodson and Granet,goes
notedthe contributions
In this fieldI can claim
to
sources.
back
original
26Hsu (1945). It willbe seenthat4 is essentially
thesame
as thatgivenbyKulp (1925); see supra,p. 31.
argument
27 Ideally
thedowryis supposedtobe doublethebride-price.

28

Le'vi-Strauss(1949), q. 236.

(845)

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

c3

38

E. R. LEACH

confinemy attentionto his


special knowledge,havingmyselfcarriedout field- followsI will therefore
work in the Kachin area at various timesbetween theoretical
propositions.
1939and 1943. Havingthisspecialknowledge,I am
Shornofitsethnographic
background
Levi-Strauss's
bound to statethatLevi-Strausshas oftenseriously thesisseemsto runsomething
like this:
misunderstoodhis sources and that in important (a) In the traditionof Tylor the exogamyof the
is due to quite
particularsthis misunderstanding
individualfamilyis seen as an,expressionof
inexcusablecarelessness. Nevertheless,
despitethese
the positivesocial necessityto ' marryout',
errorsof ethnographic
factit appears to me thathe
of negativeincest
ratherthan as a reflection
has put forwardseveralwhollyoriginaltheoretical
prohibitions.
whichare not onlyempirically
suggestions
valid,but
(b) A marriageconsideredby itselfis a relation
are of the utmostimportancefor a proper underbetween individuals. But sociologicallya
standingof theKachin situation.
marriagedoes notexistin isolationbutas part
Firstletmemakegood myaccusationofinexcusable
of a series of marriagespast and future.
carelessness. In his veryextensivereferences
to the
Fromthispointofviewa marriageis but one
Kachins Levi-Strauss relies, in the main, upon
incidentin a seriesof reciprocaltransactions
standard sources (Wehrli, 1904; Gilhodes, 1922;
betweengroups.
Hanson, 1913; Carrapiett,1929), but he also refers
two generalmodesin which
(c) We can distinguish
at a numberof crucialpoints to a work by Head
reciprocalrelationsbetweengroups are exentitledHandbookof theHaka ChinCustoms(1917).
pressed. Firstlyin the transferof goods;
From the contextin whichhe uses thissourcethere
of women.
secondlyin thetransfer
can be no doubtat all thathe has assumedthatHead's
statements
abouttheHaka Chinsare applicableto the
(d) Hence in logic we,need to considerthreetypes
Kachins29. Therecan be no excuseforthisblunder.
of exchangerelationship:
Not only are the Chins geographically
remotefrom
(i) Relationshipbetween social groups is
the Kachins,theydo not so far as we know even
or nearly
expressedbythesimultaneous
practiseKachin typemarriage30.
simultaneousexchangeof goods-the
As Levi-Strauss(1949, p. 322) himselfargues,the
TrobriandKula mightbe takenas an
crux of his whole Kachin analysisturnsupon two
example.
apparentparadoxes:
(ii) Relationshipbetween social groups is
"Nous sommesdonc-en presencede deux opposiornearly
expressedbythesimultaneous
tions: l'une entrela simplicitedes reglesde
simultaneous
exchangeof women--the
l'union preferentielle
et la complexitedu
AustralianKariera and Aranda marsysteme des prestations; l'autre entre la
be takenas examples
might
riagesystems
etla richesse
pauvretedes termesde references
(echange restreintin Levi-Strauss's
des termesd'appellation."
terminology).
But thecomplexsystemof prestations
whichhe cites
(iii) Relationshipbetween social groups is
is ChinnotKachin,and thesupposedpovertyofterms
expressedby the exchangeof women
of reference
and richnessoftermsof addressis simply
ofasymforgoods-formalisedsystems
an errorof theliterature3
.
metricalcrosscousinmarriage(echange
In mattersof Kachin ethnographyLevi-Strauss
generalisein Levi-Strauss'sterminomust then be deemed whollyunreliable. In what
logy)provideexamplesof thispattern.
(e) Seen fromthispointofviewthetwovarietiesof
29 Cf. Levi-Strauss(1949, pp. 297, 322 if., 377, etc.).
asymmetricalcross cousin marriage
30 The Haka Chins are neighboursto the Lakher on one
have quite different
implications:
side, who do practiseKachin typemarriage,and to the Zahau
the effect
has
thatas
Kachin
marriage
type
Chins on the other,who do not. Concerningthe Haka thembetweentwo groups A and B, A will conselves thereis no evidence.
"I Cf. Leach (1945). There are 18 termsof referencenot 14,
tinuously give goods to B while B will
as Levi-Strauss supposed.
The " termes d'appellation"
continuously
givewomento A.
referred
to are in factpropernames. Far frombeingnumerous
that
Trobriandtypemarriagehas theeffect
thereare only 9 foreach sex, a factwhichresultsin a proliferabut
in
B
A
women
to
one
generation,
gives
tion of " nicknames."

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

39

Marriage
ofMatrilateralCross-Cousin
Implications
The Structuiral

thiswill lead to an accumulationof women


B gives women back to A in the following
in one part of the circuitratherthan in
generation(Levi-Strauss,1949,ch. 27).
another,with a consequentdevelopmentof
(f) It appears to be Levi-Strauss's(1949, p. 554)
differentials:
bride-price
is,relatively
viewthatTrobriandtypemarriage
"4
donc a la conclusion que
arrive
on
since it
speaking,sociologicallyunimportant
l'echange
generaliseconduit,de faron
does not result in long term structural
a 1'hypergamie,
in6luctable,
presque
continuities. Two marliages in successive
entreconjoints
au
mariage
ci'est-'a-dire
generationsmerely constitutea reciprocal
status
differents."
de
exchange between two biological families.
he attempts
suggestion
Havingmadethistheoretical
ofthesecondmarriagethe
Withthefulfilment
whichhe
material
in
the
Kachin
it
terms
of
to
validate
whole transactionis complete.
their
despite
notes
that
He
discussed.
has
previously
In Kachin typemarriageon theotherhand
reported
theunilaterally
defineddescentgroupsA and formofmarriagetheKachinsarenevertheless
society. He notesalso that
B standin a permanent
persisting
relationship as beinga class stratified
of ' wife givinggroup' and ' wifereceiving thereare proverbswhichapproveand otherswhich
group.' The whole structureof societyis disapproveof polygyny. He seems to argue that
thatensuetheacquisition
builtupon the assumptionof thiscontinuity.despitethesocial difficulties
must be highlyvalued in Kachin
women
of
several
(g) We have seen that most of Levi-Strauss's
The readeris clearlyintendedto inferthat
predecessorsin this fieldhave assumedthat society.
thewifereceiving
group
thesocialstructure
throughout
Kachin type marriagenecessaiilyimpliesa
willrankhigherthanthewifegivinggroup,thenobles
circularsystemof marriageclasses of the
ofwomen.32 Hence
at thetop beingtheaccumulators
generaltype A marriesB, B marriesC, C
if we take the lines of Fig. 3 to represent5 intermarriesA. To a considerableand often
marrying'groups ' AA, A, B, C, CC, these groups
confusingextent Levi-Straussaccepts this
can be thoughtof as ' exogamous castes' which
pQsition;yetthereare momentswhenhe sees
marriage,AA being the most
hypergamous
practise
beyond it. For suppose the circle is not
senior.
closed,whatthen? Withthe wealthobjects
Laterin thebook (1949,pp. 518 ff.)he arguesthat
involved in marriage transactionsalways
fivemajor clans (grandsgroupesfondamentaux)
the
movingin the same direction,does not this
whichKachin societyis supposedto be divided
into
imply an ultimate differencein economic
as theseexogamouscastes. And he
be
identified
can
statusbetweenthe wealthgivingand wealth
Kachin typemarriage
goes on to arguethata simila#
receivinggroups?
rule may have been a factorin creatingthe Indian
(h) Followingthis line of reasoning,Levi-Strauss caste systemas we now knowit.
reaches,on purelytheoreticalgrounds,the
With such extremespeculationsI am no more
conclusion:
followinginteresting
IdeallyKachin typemarriageoughtalways
32
to operatein a circle. Providedit does so,
(1949) does not appearto statespecifically
Levi-Strauss
above
wealth objects will simplycirculatein one whetherhe supposes the 'wife receivers'to rank
the 'wife givers' or vice versa; but since he drawsclose
directionwhile women will circulatein the 'analogiesbetweenthe Kachin systemand that of the
the former
other, the status of the componentlocal Chineseand also withIndian caste hypergamy,
seemsto be assumed(p. 587). He notes with
alternative
descentgroupswillremainequal.
to operatein
surprisethatParryreportsLahkerhypergamy
In practice such a systemmay well be thereverse
fromthis.
(p. 336),butmakesno inference
direction
workableprovidedthereare, say, onlythree In actualfact,amongtheKachins,as amongtheLahkers,
groups in the circle. But in theory the 'wifegivers'rankhigherthan'wifereceivers.'
To supporthis view that Kachinsat timesset a high
in either value
systemmightbe extendedindefinitely
(p. 326)citestheproverb:
Levi-Strauss
on polygyny
direction(as in Fig. 3 or 6). The morelocal
du numshi; taratnummali
womanfour.
chiefwomanten;commoner
it will
groupsthereare themoreimptacticable
may
havetenwives,a com'
chief
a
as
interprets
This
he
be to keep all transactions
withinthecircle.
monerfour.' In facttherealmeaningof theproverbis 'the
In practice, argues Levi-Strauss (1949, priceofa chief'swifeis tencattle;thepriceof a commoner's
forwomen; wifefourcattle."
p. 325), therewillbe competition
c4

(845)

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

40

E. R. LEACH

In Kachin ideology if a man of patrilineageA


thanwiththoseofGranet,whichareindeed
concerned
of muchthe same kind,but it is worthdiscussingin marriesa woman of patrilineageB, it is properand
some detailhow far Levi-Strauss'sinitialhypothesis expectedthat at a later date therewill be a further
marriagebetweenanotherman of lineage A and
does in facttallywiththe Kachin situation,
a marriage
anotherwomanoflineageB. Furthermore
betweena manoflineageB and a womanoflineageA
law.
PART3. A NEW ANALYSISOF THREEKACHINTYPE wouldbe a breachof customary
The relationship
betweenthetwolineages.A and B
MARRIAGE
SYSTEMS
is thus specificand has structuralcontinuity.With
ThegenuineKachinsystem
respectto persons of lineageA lineageB is mayu;
Kachin33 society as described in the standard withrespectto persons of lineage B, lineage A is
ethnographic
accountsand by Levi-Straussis made dama. At every marriagethe dama purchasefrom
up of 5 exogamouspatriclanswhichmarryin a circle mayuthepotentialoffspring
of thebride.
in themanneralreadydiscussed. In a paperpublished A male Ego is related to the fellows of his
in 1950 (Leach, 1945) I showed that this circular community
in threeprincipalways. Personsof his
marriagesystemdoes not represent
empiricalfactbut own local descentgroupand of otherlineagesof his
is simplya kind of verbalmodel whichthe Kachins own clan are 'brothers' (hpu-nau);personsof local
themselves
use to explainthegeneralpatternof their descentgroupsinto whichhe and his male siblings
system. The empiricalsituationis as follows:
are expectedto marryare mayu; persons of local
Kachins clearlythinkof theirsocietyas a widely descent groups into which his female siblingsare
ramifying
dispersedclan system.The majorclansare expectedto marryaredama. Mostcloseacquaintances
intodispersedlineages,and thesesegmented fall into one of thesethreecategories,thosewho do
segmented
into furtherdispersedsub-lineagesand so on. On not do so by strictcognateor affinalrelationship
are
firstenquiryone getstheimpression
thatclanexogamy treatedmuchas iftheywereremoterelativesof Ego's
is intendedto prevailthroughout,
but in practicethis own clan. Thus themayuof themayuare classedas
is simply'a mannerof speaking.' The exogamous 'grandparents'; thedama of themayuare classedas
unit is really the lineage at the lowest level of 'brothers'; the dama of the dama are classed as
segmentation.In essentialsthissmallestlineageis a ' grandchildren.'In termsof our diagramFig. 3 the
local descent group associated with a particular orderof seniority
thusrunsfromleftto right,group
politicaldomain(mung),thoughsome membersof it AA are' grandchildren,'
A aredama,B are 'brothers,'
mayresideelsewhere.
C are mayu,CC are ' grandparents.'34This suggests
Kachins habituallyspeak as if the 'wife giving'- that if thereis a difference
of statusbetweenwife
' (mayu-dama)
'wife receiving
thatresults giving(mayu) and wife taking(dama) groups,it is
relationship
frommarriagewas the concernof major lineagesor theformerand not thelatterwho are of higherrank.
even clans. But in practicemarriageis the affairof This is in factthecase.
local descentgroupsonly. If therehappento be two
The Kachin ideal is that a series of mayu-dama
groupsin the same domain which,fromthe lineage lineages should marryin a circle; the explanatory
point of view, are segmentsof the same clan, it is mythspecifiesfivemajor lineages (clans), but any
morethanlikelythattheywill intermarry
and be in numbergreaterthantwowillserve. Circularsystems
mayu-damarelationship. In the account which embracingthreeand occasionallyfourlocal descent
followsmy use of the terms' lineage' and ' local groups are not uncommon,but, as Levi-Strauss
descent group' is intended to conformto this perceived,the systembecomesincreasingly
unstable
distinctionbetween Kachin ideology and Kachin as thenumberof unitsin a singlenetworkofrelationpractice.
ship is increased. Levi-Strausssuggestedthat the
will arisefromcompetition
instability
to accumulate
women for polygynousmarriages; the emnpirical
33 Theterm
Kachinappliesto a populationofabout300,000
arisesfromcompetition
for
scattered
overa vastarea (some 55,000squaremiles)in the situationis thatinstability
Assam-Burma-Yunnanfrontier
areas. The samenetwork bridewealth.

of kinshiprelationsoperatesoverthewholeof this territory


butthepopulation
is nottightly
unified
politically
orculturally. 34 It shouldperhapsbe stressedthatEgo is permitted
to
There are howeverno clearlydistinguishable
tribal sub- marrythese classificatory'grandparents and 'grandofthepopulation.
divisions
children.'

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

The Structural
Implications
of MatrilateralCross-Cousin
Marriage

41

has a fairly animal -killed in sacrificeor in the hunt and by


The empiricalKachin local community
thatthesamepersonsare underobligation
structural
formthoughits dimensionsin recognition
standardised
area,populationand component to providefreelabour forthe chiefon certainstated
termsofgeographical
segmentsare veryvariable. The politicalunit is a occasions,as at theclearingof thechief'sfieldor the
singlecontiguousarea ruledover by a chief.35This buildingof the chief'shouse. The villageheadman's
area I shall call the domain (mung)of a particular ownershipon theotherhand is recognisedin thefact
lineage of chieflyrank. The officeof chief(duwa) thathe disposesof thecultivation
rightsin thevillage
wvill
alwaysbe held by a memberof'thislineage,and lands amongthehouseholdersof his village. Neither
the title will normallypass fromyoungestson to typeof ownershipcan properlybe said to includethe
youngestson in the'male line. The natureof the rightof alienation,thoughlands can be transferred
rightspertainingto this officeof chiefwill be con- from one chiefto anotheror fromone village to
anotherin certainspecial circumstances.
sidered presently.
The population of the domain will normally The politicalrelationship
betweenchiefand village
lineages,some headman--andforthatmatterbetweenvillageheadcomprisemembersof many different
of highrank and some of low rank,but withinthat man and villager-has considerableresemblancesto
domain no otherlocal descentgroup can rank as thatofEnglishfeudaltenure. If thechiefbe regarded
highas thatofthechief. All thelocal descentgroups as Lord of the Manor, thenthe statusof the village
withina domain tend to be relatedto one another headmanresemblessomethingbetweenthe statusof
holdinghis land in fee tail, and the
in a quite definitemanner,explainedbelow. This a freeholder,
of the status of a tenantin villenageholdinghis land in
patternis littleaffectedby the ramifications
customaryfreeholdor copyhold. If we accept this
lineagesystemoutsidethe local area.
the area of a domain is commonly analogy,then we mightexpect to find some clear
Territorially
of class status as between chiefs' local
intoseveralsectionswhichI haveelsewhere difference
segmented
called village clusters(mare); the village clustersin descentgroups,villageheadmen'slocal descentgroups,
local descentgroups. Thisdifference
turnare segmentedinto villages(kahtawng). Struc- and commoners'
are homologous. The relation- of class status exists, though it is importantto
turallythesesegments
aredefinedmainly
shipbetweena villageand its parentvillageclusteris emphasisethattheclassdistinctions
prestigesymbols,36
in almost all respectsidenticalwiththat betweena in termsofrightsto non-utilitarian
in the economic standards of
villageclusterand itsparentdomain. Thus,although and the difference
and commonersis normallyveryslight.
some politicalunits compriseonly a singlevillage, aristocrats
If we neglectthe complicatedsubjectof slavery,37
a village
whileothersincludeseveralvillagesforming
recognisedclassesin Kachin
cluster,and others again embrace several village thereare threeordinarily
in each ofthese society,whichwe can call chieflyclass, aristocratic
theprinciplesof organisation
clusters,
class and commonerclass. The qualityof class is
politicalsystemsis the same.
an attributeof a lineageratherthan of
To illustratethese principleslet us consider a theoretically
domainwhichconsistsof a singlevillage a person. Thuslineagesare describedas beingdu baw
hypothetical
clustercomprisingfour villages. Each of the four amyu-' chiefly
lineages,'ma gam amyu-' eldestson
headman(salangwa) whose lineages,' and darat amyu-' commoner lineages.'
villageshas an hereditary
in thesamewayas thatof thechief. Sincethecommoners
outnumber
thearistocrats
officeis inherited
greatly
greatlyoutnumber
thechiefsthere
One of the fourheadmenis also chiefof the whole and thearistocrats
domain. The lineageof the chief'owns' the whole is necessarilya procedurewherebythe upper classes
of the domain. The local descentgroupof
territory
of his village 36 E.g., therightto makea particular
a villageheadman' owns' the territory
kindofsacrifice
or to
but thesetwo typesof ownershipare of.a differentputup a particular
kindofhousepost.
37 Kachin 'slavery' was officially
is recognisedbythefact
order. The chief'sownership
abolishedby British
action. Formerly
therewereseveralcategories
thatall personsin thedomain,who are not ofhis own administrative
of' slaves' (mayam),
but themajority
werevoluntary
serfslineage,must presenthim with a hind leg of any or
evenadoptedsons-of theirmastersratherthanchattel
Similar systemsof so-called 'slavery' have often been
reportedfromthisregion(cf. Parry,1932,p. 223; Hutton,
35 Whatfollows
pp. 145if.,385if.). In theKachinmarriage
to theKachingumsa type 1921(b),
appliesprimarily
hierarchy
of politicalorganisation.In an alternative
formedtwo additionalclassesjunior
typeof system the 'slaves ' nominally
different.
to thecommoners.
is somewhat
knownas gumlaothestructure

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

42

E. R. LEACH

FIG. 9.-Kachin politicalstructure.


who' own' threeseparatepoliticaldomains. A is dama
LinesA, B, C arethelocal descentgroupsofthreerulingchiefs,
toB; BisdamatoC; CisdamatoA.
headman
DomainA comprises
4 villagesrepresented
bylinesA, b, c, d. The chiefofthedomain,A1,is simultaneously
ofhisownvillage.
Thethreesubordinate
headmenmarry
in a circle. d is damato b; b is damato c; c is damato d. In additionat leastone
oftheseheadmen,
namelyd, is damato thechiefA.
In thesamewaythevillageof headmand comprises
fourlocal descentgroups. Threeofthese,ac, 3,y,are ofcommoner
class. oc, , y marryin a circleand at leastone ofthem(a) is
class; thefourth
d is theheadman'sownand is of aristocratic
damato theheadman'sgroupd.

class. Threechiefsin mayu/dama


circle
Chiefly
forma loosepoliticalfederation.

- -

Aristocraticclass. Three village headmen


withthechief,as fourth
together
villageheadman,
ofdomain.
controlpoliticalaffairs

Commoner
class. Seniormembers
ofcommoner
localdescent
groups,
together
withvillageheadman,
as representing
his local descentgroup,control
ofVillage.
affairs

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

TheStructural
Implications
ofMatrilateralCross-Cousin
Marriage

43

shedtheirsurplusintotheclassbelow. Thisprocedure enhancedby specifying


bridepricesof huge (and quite
dependsupon thewellunderstoodprincipleoflineage imaginary)dimensions.
fission. Whena lineagehas acquireda 'depth' of
Thereare thusalwayssomewomenofa chief'slocal
fouror fivegenerationsit tendsto split,but of the descentgroupwho marryawayfromthedomaininto
two residuallineagesonlyone retainsthe class status otherchiefly
lineages;othershoweverwillmarrywith
of the parentlineage the othertendsto ' go down men of the aristocratlineages of the chief's own
hill' (gumyuyu ai). In theory,the seniorlineageis domain. Typicallythelocal descentgroupofa village
alwaysthe youngestson line-i.e., the youngestson headmanis ofthearistocrat
classandwillbe dama(wife
of the youngestson, etc.-hence lineageswhichhave taking)in respectto the chief'slocal descentgroup.
lineagesand therebyassumed
splitaway fromchiefly
level.
Similaralternatives
operateat thearistocratic
a subordinatestatus are eldest son lineages38-i.e.,
The typicalaristocraticlineage is the lineage of a
aristocrats. Similarlythe lineageswhichsplit away villageheadman. Some aristocratic
maleswillmarry
fromaristocratic
lineagestendto 'go downhill' and womenof the chief'slocal descentgroup; some will
becomecommonerlineages.
local descentgroups
marrywomenofotheraristocratic
Matrilateral
crosscousinmarriageplaysan integral in the vicinity-especiallythose of other village
but headmenin the domain. Some aristocraticfemales
thisclass structure
partnot onlyin maintaining
betweenchiefs, marryaristocratic
in defining
the 'feudal' relationships
males,othersmarrycommonersof
headmen,and commoners.
theirown village. At each levela limitedthoughnot
The two most generalprinciplesthat governthe exclusivecircularsystemof the hkau wanghkuttends
Kachin marriagesystemare that a man will do to establishitself.-The principlecan be illustrated
by
an imaginary
everything
possible to avoid marryinginto a class resortto a diagram(Fig. 9) representing
beneathhim,and thata man will seek to make the political federationcomprisingthree chiefsone of
maximumprofit-eitherin terms of bridepriceor whom has underhim threeheadmen,one of whom
political advantage-out of the marriage of his has underhimthreecommonerlocal descentgroups.
is highly The structure
viewthatpolygyny
itmaybe notedis not,as Levi-Strauss
daughters.Levi-Strauss's
in theIndian caste
esteemedforits own sake is erroneous. The factors supposed,analogousto hypergamy
reversed. Womenmaymarry
whichinfluencechiefs(and sometimesvillage head- system,
buthypergamy
the intotheirown class or theclass below,but neverinto
men) to acquire more than one wifeare firstly
ofhavinga male heirto carryon thelocal theclass above.39
importance
descentgroup,and secondlythe politicaladvantage
that comes frommaintainingrelationswithseveral
different
mayu(wifegiving)groupsat the same time.
39 Chiefsand headmen
maytakewomenfroma lowerclass
A chief,in orderto maintainstatus,mustmarry- as secondarywivesto raisean heirbuttheoffspring
of such
as his firstwife-a woman from another chiefly womendo nothave thefullclass statusof themale parent;
rigidity
as opposedto thetheoretical
thepracticalflexibility
lineage,that is to say a woman fromsome other of the Kachinsystemdependsupon thispoint(see note 38
chieflydomain. Such marriagescan formthe basis above).
thatthisexplanation
thatall readerswillappreciate
of large-scalepoliticalalliances. It is not uncommon I presume
of matrilateral
crosscousinmarriagefor
of the significance
to findthreeneighbouring
chiefly
groupsA, B, and C, thehierarchy
intheKachingumsatypeofpolitical
ofauthority'model'
linkedby the rule that chiefA marriesprincessB, organisation
verysimplified
is basedon a necessarily
chiefB marriesprincessC, chiefC marriesprincessA. of any empiricalreality.But the ' model' is one which,in
themselves.Gumsa
effect,the Kachins have constructed
Such a systemis called by the Kachins ' cousincircle Kachinsdo appearto organisetheirliveson thetheorythat
path' (hkau wanghku). The threechiefsare all of lineagesare hierarchically
orderedin the mannerI have
equal status. Women go round the circle in one described. They insist most vigorouslythat commoner
cannevergainstatusand becomearistocratic
lineages,
direction; bridewealthin the other. Incidentally, lineages
notthecase. Objectively
thisis clearly
eventhough
empirically
sincethemarriagetransactionsare in a sense,some- thesituationseemsto be thatin gumsa Kachinsocietyall the
status
havea ranking
what nominal,the prestigeof all concernedcan be lineagespresentin anylocal community
-i.e., of anytwolineages,one rankshigherthantheother;
are generally
thisis trueevenif boththelineagesconcerned
lineages.Butthisrankorderis notstable.
In Kachin theorythe status of a lineage is defined ratedas commoner
seekingto gainstatus
so thata lineagecan onlylose status Everyheadof a householdis constantly
absolutely
bygenealogy
ofrecognised
andforhislineage. Thereis a variety
and nevergain it. The practicalsituationis much more forhimself
flexible
ofthisthememustbepostponed techniquesof achievingmeritin this way-a man may for
thanthisbutdiscussion
examplegive a ' wealthfeast' (sut manau), but the most
to a laterpublication.
38

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

44

E. R. LEACH

is resumed. This has someThe class stratification


and theformalised
marriage mayu-damarelationship
links betweenclasses ties in with the land tenure times meant that the defeatedgroup,settlein the
of thevictorsas theirdama tenants.40
organisationin the followingmanner. Ideally the territory
patternof Kachin residenceis patrilocal; Kachins
Whena chiefgoesto warhe has theinstitutionalised
oftentalk as if a son, on growingup, automatically rightto call hisdamato hisaid. The promisedreward
settleddown in the village of his parents. With of victorymay be villageterritory.Once again the
commoners
whohaveno prestigeto maintain,
practice finaloutcomeis thatthesuperiorchiefand histenants
relationship.
conformsfairlywellwiththeory,but personsof rank are in mayu-dama
Finallyin areas wherepopulationis sparsea chief
are at least as likelyto settlematrilocallyas patrilocally. One majorfactorhereis thatthehighranking who has much territorybut few followersmay
status of the youngest brother does not favour explicitlyask the followersof anotherchiefto come
amicable relationsbetween -adultmale siblingsin and join him and will seal the compactwiththe gift
of a daughter.41
chiefly
families.
But if a chief'sson settlesin thealien villageof his
Generalising,it is fair to say that where the
father-in-law,
he places himselfin an inferior
position 'tenants' or followers of a village headman or
and admitshis inferiority.Probablyhe will get his domainchiefare not regardedas clan brothersof the
wifefora muchreducedbrideprice,
but thisin itself 'landlord,' they are in the status of 'son-in-law'
is a disgrace. In effect
a man who settlesmatrilQcally(dama) to the ' landlord.' The procedure for
becomesthefollowerand tenantof his father-in-law.acquiringland rightsof anykindis in almostall cases
a womanfromthelineageof
to marrying
Yet in settlingmatrilocally
in thisway a man founds tantamount
a new patrilineallocal descentgroup. If the mayu- thelord. In Kachin terms,therightsthatanytenant
dama relationship
thusinitiatedcontinuesforseveral has in hisland are expressedin thefactthathe is dama
generationswith orthodoxpatrilocalresidence,the to his immediatesuperiorin the 'feudal' hierarchy.
Thisanalysisresolveswhatappearedto Levi-Strauss
ultimatepositionwill be thatthe descendantsof the
father-in-law
and the descendantsof the son-in-law as an outstandingparadox. As Levi-Straussunderwill be livingside by side in the same community
in stood thepositiontheKachin systemis ideallyone of
marriageclasses and thusessentially
landlord-tenant
relationship. This probablyis the 5 intermarrying
egalitarian. Since a man's prospectivebrideappears
mostusual historyof presentday Kachin villages.
inadvancebytherulesofthemayu-dama
There are also otherways in whicha mayu-dama to be specified
statusbetweentwo local descentgroupscan come to system,it is surely paradoxical that brideprice
reflect
a landlord-tenant
relationship
in a feudalsense. paymentshould be large and complex?42 LeviBlood feudsforexamplenormallystartoverwomen. Strauss's(1949, p. 327) explanationseemsto be that
Typicallya feud is betweenmayu and dama. The -these large brideprice payments are a kind of
the " conflitentre
appropriateending of a feud is a marriage-the pathologicalsymptomrepresenting
de 1'echangegeneralise,
et ses
les conditionsegalitaires
consequencesaristocratiques." He perceivesthat a
effective
and permanent
procedure
forsocialclimbing
is that marriageclass systemof theformalcomplexity
which
of arranging
an advantageous
marriageforone's son.
Kachin
the
is
to
unworkable.
system
he
attributes
By wayof further
clarification
I mustalso stressagainthe
point emphasisedin my earlierpaper (Leach, 1945). The He infersthatthemarriageclasseswillconvertthemKachinmarriage
preferred
is not betweenmother'sbrother's selvesintoprivileged
class!esbut he considersthatthis
daughterand father'ssister'sson but betweenclassificatoryis " en contradictionavec 16 systeme,et doit donc
and classificatory
mother'sbrother's
daughter
father's
sister's
sa ruine" (p. 325).
son (Kachinnamand Kachingu). It is onlyamongthechiefs, entrainer
whosecircleofkinsfolk
is necessarily
somewhat
restricted,
that
Yet thesystemas I have now describedit is neither
an orthodox
withnamimpliesmarriage
marriage
witha 'real' contradictory
nor self-destructive.
It is truethat,on
mother's
brother's
daughter
or anynearrelative.A cominoner
bridewealth
moves
in the same
always
balance,
male normallyhas a wide rangeof nam to choose from,

forexampleanylineagesisterof thewifeof anyof


including
his father'slineage brothers. In many cases the actual
between
butin Kachin
relationship
gu andnamis veryremQte,
40
Cf. the Lakher examplecited by Parry(1932, p. 219).
eyesanygu-nam
marriage
is strictly
orthodox.An important 41
Cf. Kawlu Ma Mawng (1942, pp. 55, 58).
corollaryof thisis thatthe class statusimplications
of the
ruleoperatewithmuchmoremarkedeffect 42 Although Levi-Strauss confused Kachin practice with
Kachinmarriage
classes(whotendon thewholeto marry Chin, he mightstill I think argue that the scale of Kachin
amongthearistocratic
classeswhodo not. bridepricepaymentsis paradoxicallylarge.
nearrelatives)
thanamongthecommoner

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

Marriage
Implications
ofMatrilateralCross-Cousin
The Structural

45

directionfromcommonerstowardsaristocrats,
from asymmetry
of thekinshipsystemthewhole organisaaristocrats
towardschiefs. And it is thustruethatif tionis in politicaland economicbalance.
bridewealthwere composed whollyof irreplaceable It will I hope be agreed that the analysisI have
commoditiesof the Trobriand vaygu'a type, the givenis decidedlymoresatisfactory
thanthatprovided
systemcould be self-destructive,
since in time the by Levi-Strauss.It is worthpointingout whereinthis
total sum of 'bridewealthcurrency' would come superiority
lies.
intothehandsofthearistocrats.But in factthemain
The original theorisingof Hodson and Mrs.
itemin a brideprice,
or in any of the othercomplex Seligmanwas defectivenot merelybecause of the
legal obligationsto whichLevi-Strauss(1949, p. 326) inadequacyof theirempiricaldata but because they
refers(des prestationset des e'changes,
des " dettes"', consideredkinshipsimplyas a systemin itself. If a
des cre'ancesetdesobligations),
is a giftof cattle;and kinshipschemebe consideredwithoutreference
to its
cattle,among the Kachin, are a consumablecom- political,demographicor economicimplications
it is
modity. On balancethechiefdoes tendto accumulate inevitably
thoughtof.as a logicallyclosed system. If
wealthin theformof cattle. But prestigedoes not it is not closed,it cannotwork, Hence theeagerness
come fromthe owningof cattle; it derivesfromthe withwhichthese earlywritersaccepted any ethnoslaughterof animalsin religiousfeasts(manau). If a graphicevidencewhichseemedto suggestan arrangechiefbecomesrichas a consequenceof marriagesor mentof mechanically
articulatedmarriageclasses.
otherlegaltransactions
he merelyholdsmanauat more
Levi-Straussfollowingthe lead given by Warner
frequentintervalsand on a larger scale, and his and Granetmakesa greatadvanceupon these 'pure
whopartakeofthefeast,benefit
followers,
accordingly. kinship' theoriesbecause he takes into accountthe
Herethenis theelementwhichis necessary
to complete reciprocity
aspectsof kinship. He is not contentto
thecycleofexchangetransactions,
theabsenceofwhich see Kachintypemarriagesimplyas a variantfromthe
struckLevi-Straussas paradoxical.
' classic' systemsof Australia;he considersalso the
of L'Essai sur le Don. Nevertheless
implications
he
Let me recapitulate
myanalysis:
of prestationsas symbolsof
stressesthe significance
(1) From a politicalaspect,chiefis to headmanas relationship
ratherthanas economicgoods.
feudal Lord of the Manor is to cus,tomary Levi-Strauss(1949, p. 606) rightlyarguesthatthe
freeholder.
structuralimplicationsof a marriagecan only be
(2) From a kinshipaspect,chiefis to headmanas understoodif we thinkQfit as one itemin a whole
betweenkin groups. So far,so.
mayu to dama, that is as father-in-law
to seriesof transactions
son-in-law.
good. Butin noneoftheexampleswhichhe provides
(3) From a territorial
aspect,the kinshipstatusof in his book does he carrythisprinciplefar enough.
of kinshipobligationare not merely
theheadman'slineagein respectto thatofthe The reciprocities
of
symbols
alliance,
theyare also economictransacchiefis heldto validatethetenureof land.
tions, political transactions,chartersto rightsof
(4) From an economicaspect the effectof matri- domicileand land use. No usefulpictureof 'how
lateral cross cousin marriage is that, on a kinshipsystemworks' can be providedunlessthese
balance, the headman's lineage constantly severalaspectsor implications
ofthekinshiporganisapays wealthto the chief'slineagein theform tionare consideredsimultaneously.But Levi-Strauss
of bridewealth.The paymentcan also, from supposesthatthereareonly" deuxformules
d'echange
the analyticalpointof view,be regardedas a reel" (1949, p. 582). He is concernedsimplywith
rentpaid to theseniorlandlordby thetenant. whetherthe alliance is directlyreciprocal.group A
The mostimportant
partofthispaymentis in exchangingwomenwithgroupB (&hange restreint),
the formof consumergoods-namely cattle. or multiple,severalgroupsexchangingwomenin a
The chiefconvertsthisperishablewealthinto network(6changegeneralise). Fundamentallyhe is
imperishable
prestigethroughthemediumof not reallyinterested
in the natureand significance
of
spectacularfeasting.The ultimateconsumers the counter-prestations
that serveas equivalentsfor
of the goods are in this way the original womenin the systemshe is discussing. Because of
producers, namely, the commoners who thislimitationof his viewhe is led to attribute
to the
attendthe feast.
whichit does not in fact
Kachin systeman instability
Structurally
speakingtherefore
despitethe seeming possess, and fromthis point he wandersfar afield

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

46

E. R. LEACH

Kachin
Batak44
history
intowild speculationsabout the evolutionary
of halfthekinshipsystemsof Europe an:dAsia (1949,
Kinship
pp. 585-90).
A patriclan
system
segmented ditto.(44a).
Nevertheless,
althoughI consideritquiteillegitimateintolineagesandsublineages.
to treattheKachin systemas ifit werea fundamental Thelineageat thelocalgroup ditto.(b).
typein a long termhistoricalsequence,I findthata levelis exogamous;thepatricomparisonleads to clanis not.
morecautioustypeof structural
results.
veryilluminating
The systemis idealised as ditto,at anyrateoftheKaro
Some of the societieswhich Levi-Strausshimself consistingof 5 major clans Batak(c).
by rule of
comparesto the Kachin mightwell repayre-analysis intermarrying
on the lines I have now given,but I will confine matrilateralcross cousin
myselfon this occasion to showingthe relevanceof marriage.
myKachinformulation
to twosocietieswhichare not It is in fact the localised ditto.
consideredby Levi-Straussat all, namely,the Batak lineageswhichpractisethis
kind of marriage,and are
of Sumatraand theLovedu of SouthAfrica.
thereby
pairedintowifegiving
The firstof thesesocietiesis a kind of structural groups
(mayu) Batak-hulahula.
duplicateof that of the Kachin; the second, in a and wife receivinggroups
manner,seemsto have
salutaryand thought-provoking
(dama) Batak-anak boru(d).
muchthesame kindof structure
in reverse.
Authority

Batak43

ditto. For 'village ' read

" . . . the executiveradja of a

The ethnography
of theBatak is veryextensiveand 'village cluster'; for 'dis- villageis chosenfromamong
of a certaineligible
dates back to the 18thcentury. Of severalpublished trict' read ' domain'; for members
summariesof this work the most comprehensive
is radja read duwa; for 'sib ' family... The familyfrom
' patriclan
for whichtheradjais chosenmust
';
showsthat, read
Loeb (1935). Analysisof the literature
' family
' read 'localised be a part of a sib whichis
if one ignores entirelycultural and demographic patrilineage,' or 'local spokenof as the'rulingsib'
factors,thepurelystructural
parallelsbetweenKachin descent group.'
...
Among the Karo Batak
close.
the fivemain sibs are to be
and Batak organisationare remarkably
foundineveryvillage,although
The descriptionof Kachin societygiven in the
in everydistrict
a certainone
was
the descriptionof
previoussectionof thisessay
is -in the majorityand is
a model; I said nothingabout cultureor languageor
I was concernedonlywith
populationor geography,
of thesystemsof kinship, 44 Bataksources:
interrelations
thestructural
social class and
land tenure,economicdistribution,
(a) Loeb (1935, p. 46).
(b) Loeb (1935, p. 47).
politicalorganisation.If we analyseBatak societyin
(c) Loeb (1935, p. 47).
terms of these same dimensionsthe patternthat
(d) Fischer(1950); Tideman (1922); Loeb (1935, p. 53).
emergesis almost exactlythe same. There is no
(e) Loeb (1935, pp. 39-40).
(f) Haar (1948,pp. 206,208).
featureof my Kachin model whichis not duplicated
(g) Cole (1945, p. 273).
or closelyparalleledin theBatak situation.
(h) Loeb (1935, pp. 17-38).
The purpose of my Batak analysisis simplyto
(i) Cole (1945-p. 273); Joustra(1911, p. 11).
(j) Loeb (1935, p. 29).
as described,
thatKachinorganisation,
is
demonstrate
(k) This seems a legitimateinferencefromLoeb (1935,
in no sensea freaksystem.My purposewilltherefore
1
p. 43).
best be served by listing the Kachin structural
(1) Warneck(1901, p. 532); Loeb (1935, pp. 58, 42).
(m) Loeb (1935, p. 58).
to whichI havealreadyreferred
characteristics
against
(n) Loeb (1935, p. 59).
Batak features.
the corresponding
(o) Loeb (1935, p. 59).
43Batakis a collectivename for a populationof over
in thearea of Lake Toba in NorthCentral
resident
1,000,000
dividestheBatakintovarious
Sumatra.Thoughtheliterature
'tribes'-Toba, Karo,Timor,etc.,the same kinshipnetwork
seemsto prevailthroughout.As withtheKachins,however,
' embracesa considerable
the' Bataksystem
rangeofcultural
diversity.

(p) Loeb (1935, p. 55).


(q)- Loeb (1935, p. 40).
(r) An inferencefrom Fischer (1950) and Warneck
(1901, p. 542); Loeb (1935, p. 61).
(s) Loeb (1935, p. 53).
(t) Loeb (1935, p. 39).
(u) Loeb (1935, p. 42).

(v) Warneck(1901).

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

47

Marriage
ofMatrilateralCross-Cousin
Implications
The Structural
Batak
Kachin
generally
saidto be theoldest This -orthodox type of
one in theregion,as well as marriage
is patrilocal.
of
therulingone. Irrespective
actual power or following Analternative,
andillthought
of of,modeof getting
everymale representative
a wifeis
this-sibcallshimself
radja...
by labour service. A man
In some villages thereare worksin thehouseholdofhis
moreradjasthansubjects(e). futurefather-in-law
for an
agreednumberof years in
Land Tenure
or at least
lieu of brideprice
Normal successionrule is Normal successionrule is partofit.
with primo- primogeniture
with ultimoultimogeniture
of his service
On completion
in someareas(.f).
in someareas.
geniture
geniture
he may,in theory,take his
The land of the political ditto. The" owning" lineage wifebackhome. In practice
domain is all ' owned' by is called namora-mora
(" origi- he commonlystays on as
the lineageof the chiefby nal ") (g).
tenantof his father-in-law.
rightof conquestor original
settlement.
Kachin

is one
The settlement
pattern
ofhomologous
segments.
At the largestextensiona

ditto.
Kachin

Batak

domain (mung) comprises

mung

urung

made up of villages (kahtawng).

kahtawng

kesain (h).

severalvillageclusters(mare) mare

Constituentvillages (kaht- ditto. (i).


awng) are ' owned' by
lineageswhich,locally,areof
inferior
statusto the chief's
lineage.Thosewhicharenot
ofthesameclanas thechief's
in ' sonlineagearenormally
to that
in-law' relationship
lineage.
Rentand Brideprice
Onlytokenrentis paidto the ditto. (j).
chiefintheformofmeatfrom
animalskilled,workin the
to
chief'sfield,contributions
thechief'shousebuilding.
The paymentof brideprice ditto. (k).
from dama to mayu can

howeverbe construedas a
formofrent.

huta

Batak
ditto. (m).

ditto. (n).

In a specialformof marriage
calledambilanak,a manpays
no bridepricebut liveswith
and raises
his father-in-law
heirs to the father-in-law's
he laterpays
line. Ifhowever
brideprice-even after the
death of the father-in-law(or someofthem)
thechildren
become his own and he
rightsin
acquirespermanent
the whole or part of his
land. (o).
father-in-law's

clear,but
borntoa womanfor Positionnotentirely
Children
ditto. (p).
whom bridepricehas not seemingly
beenpaid are n-ji (bastards).
Theybelongin effectto the
woman'sgroupunlesslegitimisedintoherlover'slineor
husband'sline
intoherfuture
as theresultofpayments.
Socialclass
statesthatthere
The generally recognised Theliterature
classes of Kachin society areonlythreeclasses-nobles,
slaves. The disapplyto lineagesratherthan commoners,
madebetween'ruling
individuals.The classesare: tinction
' and ' ruling sib '
(du baw),aristocratic family
Chiefly
thatthereis infactan
(darat) suggests
(ma gam),commoner
class
slaves(mayam). additionalintermediate
and formerly
correspondingto Kachin
aristocrats.(q).

intotheir
Kachinmalesmarry
own class or into the class
is statedto bepaid above. Kachinfemales
Brideprice
Apartfromobjectsofsymbolic
marry
value only,the mainpartof inmoney.Butbya system
of into theirown class or into
is paidin cattle. pawning or debt-bondagetheclassbelow.
a brideprice
Cattle are eventuallycon- lowerclasscanalwaysborrow
of
sumed in prestigefeasting. fromupper-class. As these Thusifthereis a difference
On balance bridewealth debts are seldom fullyre- classbetween
mayuanddama
cattle tend to move from deemedbut are cancelledby it is the mayu who rank
lower class towardsupper death,the systemis a dis- higherthanthedama.
class,butas upperclass give guisedformof rentrefund.
moreand biggerfeastscon- (1).
is equalised.
sumption

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

ditto. (r).

ditto. hulahularankhigher
thananak boru. If circumstancesresultin a hulahula
groupbeingpoorerthanEgo's
marriages
owngroup,further
are avoidedand therelationshipcomesto an end. (s).

48

E. R. LEACH

Kachin
Batak
Practically
thedifferences
be- ditto. (t).
tweensocialclassesare indiofprestige
catedbyownership
symbols
andrights
overland,
ratherthan any significant
in economicstandifference
dards. Chiefsdo however
sometimes get rich by
theirspecialopporexploiting
tunities
as traders.

we have encounteredamong the Kachins and the


Batak, the implicationsof these featuressometimes
reflecttheKachin patternin reverse. In particularit
between
appearsthatwhenthereis a statusdifference
the' wifegiving' and 'wife receiving' local descent
groups,it is thelatterand not theformerwhichrank
thehigher. This ofcourseis also theChinesepattern,
but whereaswiththe Chinese,thisorderof seniority
witha paymentofdowry,theLovedusystem
correlates
involvespaymentof brideprice. Beforediscussing
Formerlyonly the chiefs
the Lovedu
let me summarisebriefly
thisdistinction,
ownedslaves.
situation.
I will again confinemy summarydescriptionto
chiefsgainedrepu- ditto. (u).Although
tationbyhavingmanyslaves;
kind and will ignoreculture,
factorsof a structural
therewere many economic
geographyand demography.If thissummaryseems
advantagesin beinga slave.
thatI am here
undulybare, it mustbe remembered
Bondslaverywas a meansof
onlyconcernedto discoveithedegreeto whichKachin
economic
credit
and
obtaining
type marriage,here associated with a segmentary
fromthe
politicalprotection
chief.45
of bridesystemof patrilineagesand an institution
'
'
type
political
feudal
with
a
correlated
is
also
price,
Polygynyis rare and a ditto. (v).
and a hierarchy
of social classes.
structure,
perquisiteof the chiefsand
aristocrats.It is a devicefor
The Kriges'account(Krigeand Krige,1943,p. 164)
ensuringcontinuityof the
of the Lovedu is not presentedwith this type of
lineageand themaintenance
analysisin mind. Althoughtheyrecognisethat" the
thanan
ofpoliticaltiesrather
systemis not a thingapart,standingaloof
political
endin itself.
frommarriageand the social structure"theyin fact
It is clearthenthatin fourmajorinstitutional
how theiranalysisof themarriage
fields, failto demonstrate
namely,those of kinshipbehaviour,land settlement,systemcan be correlatedwiththeiranalysisof the
of consumergoods, politicalsystem.
social class, and the distribution
the 'model' or structural
The Lovedu are a SouthernBantutriberesidentin
patternof Batak societyis
similarto thatoftheKachin. In both northeast Transvaalnumberingsome 33,000 in an
quitestrikingly
societiesKachin type marriage,as manifestedin a area of 150 squaremiles. The Lovedu properare an
sectionof thispopulation,the remainder
specifickinship relation between affinallylinked aristocratic
lineages,is associatedwiththe same typeof 'feudal' being segmentedinto various sub-tribesof alien
of the
land tenure. In both cases the intermarriage
of origin,but all alikerecognisetheparamountcy
social class resultsin a Lovedu Queen (Krige and Krige, 1943, pp. 13-14).
localisedlineagesof different
generaltendencyto transfereconomic wealthfrom Lovedu and aliens alike (with certain exceptions)
the lower class to the upper. In both cases social systematically
practiseKachin typemarriage.
existwherebythiswealthis redistributed The whole societyis organisedinto a systemof
mechanisms
and thelowerclass is not permanently
impoverished. patrilineallocal descentgroups (Krige and Krige,
As a model systemthis total patternformsan 1943,p. 86). Bridepriceis paid in cattle; but these
integratedschemeand I cannot regard thesecor- cattle are not consumer goods (Krige, 1939, p. 395).46
respondencesbetweentwo such widely separated Exceptin freaksituationstheyare onlyused to secure
societiesas purelyfortuitous.
a wife for a brotherof the bride. Consequently
" ownershipof cattleis not the chiefor even a very
Lovedu
methodof reckoningstatus. A man will
importa-nt
The maininterestof mysecondexample,as I have complain,not because he has no cattle,but because
alreadyindicated,lies in the factthat,whileLovedu he cannotbrewbeerto maintainhisprestige" (Krige
social structure
duplicatesmanyof thefeatureswhich
0

Cf. The Chin tefa systemdescribedby Stevenson(1943).

for
46Goats on the otherhand are freelyslaughtered
as meat.
consumption

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

49

ofMatrilateralCrossCousinMarriage
TheStructural
Implications

thanthemanhimself
(1943,p. 101).
and Krige,1943,p. 42). On the otherhand,whereas thesocialhierarchy
among the Batak polygynyis somethingof an They also note that a giftof womenis the socially
amongtheLovedu it is not onlystrongly recognisedmode of offeringtributeto a political
abnormality,
thatthe givingof
approved of but even, it would seem, a statistical superior(1943, p. 95) and further
47
beeris typically
a gestureof honourand an approved
norm.
The objectiveof maritalpolicyseemsto be to build formof tribute(1943,pp. 18, 63, 287-88).
My suggestionis that we have here a kind of
up a villageof manyhutswhichhas a reputationfor
in reverse. No doubt,as withthe
givinglavish and frequentbeer-drinks.Polygynyis Kachin structure
Kachins,themajorityof marriagesare betweenlocal
a meansto thisend.
The Kriges make no direct referenceto earlier descentgroupsof equal status; but I postulatethat,
status,then
theoreticaldiscussionsof Kachin typemarriagebut whereverthe kin groupsare of different
they neverthelessinterpretthe systemas circular. thereis a verystrongtendencyforthe wifereceiving
Their ideal model comprises6 local descentgroups groupto rank the higher. Thus a marriedwoman
in a circlewithbridesgoing one way and belongsto a higherrankinglocal groupthanhercattlemarrying
cattlethe other(Krige and Krige,1943,pp. 66, 145). linkedbrotherand receivestokensof honourfromhis
of household accordingly. This reversed order of
Theyrecognisethatsucha modelis a simplification
thatactuallyoccursand theyrecognisethat seniorityis correlatedwith the fact that it is the
anything
theirregularities
of actualpracticemustcause " social acquisitionof extrawivesratherthanthe acquisition
stressesand-strains." But apparentlytheysuppose of extracattlethatis a majorvaluein Lovedu society.
Ifthishypothesis
be acceptedmuchthatis somewhat
will cancel out (Krige, 1939,
thattheseirregularities
pp. 411 ff.). Certainlytheyseemto have no inkling bizarreand freakishin the Kriges' accountbecomes
meaningful,while at the same time it
of Levi-Strauss'sconclusionthat,whensuch systems structurally
willtendto develop. becomesapparentthatthereare important
gapsin the
failto be circular,
classdifferences
whichtheKrigesprovide.
societyand severalof information
Yet Lovedu is a class stratified
The totalpoliticaldomainof the Lovedu Queen is
the cases which the Kriges (1939, pp. 413, 416)
each
withstrainedmarriagerelations subdividedintoa numberofpoliticalsubdistricts
mentionin connection
concernthe maritalaffairsof village headmen. In withits own districthead. Districtheadshipis, in
each case, it seems, the headman's wife and her theory,hereditarythough succession is liable to
' fromthe centre. The Queen herself
brothersstartby beingof inferiorstatusto thehead- ' manipulation
is an anomalous social person,for,thoughphysioman himself.
The specificwifegiving-wife
receivingrelationship logically female, she is sociologicallymale. Her
local descent groups tributarydependants-i.e., the districtheads and
between two intermarrying
boru) is 'marginal' foreignchiefswho wish to avail them(Kachin: mayu-dama;Batak hulahuila-anak
Between
two selvesof the Queen's rain magic-pay her tributein
for the Lovedu vamakhulu-vaduhulu.
such groupsgiftexchangesare continuous. In sum theformofwives(Krigeand Krige,1943,pp. 173-74).
' the Queen acceptsthese women
of beer- From 'foreigners
the vaduhulureceivewomenand offerings
heads of
often on a very large scale-and in returnthey as giftswithoutreciprocation;to thedistrict
contributecattle and various kinds of assistance the main Lovedu area she repaysthe giftwithcattle
includinggoat meat (Krige and Krige, 1943,pp. 27, brideprice. As a consequencethe Queen is in a sort
'
relationshipto the ' foreigner
63,77). The Krigesseemto arguethattheseexchanges of master-servant
(wifereceiving)
relationship
are exact equivalentsand tendto stressthatthe two headmen,butin vaduhulu
heads.
groupsare of equal standing(1943, p. 149). Yet in to thedistrict
othercontextstheystressthat a man's cattle-linked Aftera periodat Courtthe ' wives' of the Queen
sister(i.e.,thesisterbymeansofwhoma manreceives (vatanoni)are reallocated as true wives to other
bridewealth
cattleto acquirea wife)rankshigherin districtheads and to Lovedu Court officialsof high
rank. The new husband pays no bridepriceeither
Kriges found that 35 per cent. of men have more to the Queen or to the originalparentsof his bride,
4'The
than one wifeand thatthe ra-tioof marriedwomento married but he has an obligation,in due course,to give a
men was 156:100. They argue that the custom of polygyny daughter
ofhisbrideto theQueenas a further
vatanoni
createsshortagesof suitablecross cousinsforthe men and that
wife.
On
the
other
hand,
although
-he
has
paid no
thisis the reason whythe marriagerule is asymmetrical.The
brideprice
himself,
thenewhusbandis now considered
reasoningappears to be invalid (Krige, 1939, pp. 411-12).
(845)

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

50

E. R. LEACH

in vaduhulurelationshiptowards the originallocal is vaduhuluto groupA. Moreoverthe Queen herself


descentgroup of the bride (Krige and Krige, 1943, is in the status of 'cattle-linkedsister' to X (her
isthusQ-X-A---B-C,
kinsman).Onelineofranking
p. 98).
At firstsightthereis a striking
paradox about this Q being the highest. But in additionthe Queen is
account. Why should the Queen pay out cattlefor individuallythe superior of X as 'cattle-linked
herwives,butthengivethewivesawayagain without sister,'of A2 and B2 as vaduhulu,and of C as
getting
cattlein return? The purelystructural
answer tributeclaimingoverlord.
In due coursea3 and b3 are tenderedto the Queen
seemsto be thatin thissecondcase thegiftis repaid
notin cattlebat in thepersonofthewoman'sdaughter. as tributewivesand theprocessis repeated.
But if we ask whythisshouldbe so, the Krigeshave
This diagrammaticscheme seems to correspond
'no answer. I suggestthatthe logic of the situation closelywiththeactualpoliticalhierarchy
as described
fromtheLovedupointofview,is thatan inferior
does by the Kriges,thoughin thereal patternthereis one
complication.
not properlygive cattleto a superior. The proper further
gestureto personsofhigherrankis to givewomenand
Amongthe dependantsof the Lovedu Queen there
beer; vaduhulurankabove vamakhulu.
are certainShangana-TongagroupswhomtheLovedu
If this be so then the complicatedmatterof the consideras low caste and with whom marriageis
forbidden. The Queen accepts tribute
Queen'swivesmakespoliticalsense. Thiscan be seen theoretically
the situationby a diagramsimilarto wivesfromthesegroupsalso. But thesewomenare
if we represent
those already employed,in the earlierpart of this reallocatedto theLovedu nobilitynot as wivesbut as
daughters. Then, finally,they are marriedoff to
essay.
In Fig. 10 the lines A, B, C, representthe local Lovedu commoners,puttingthe said .commonersin
to thenobles. The Shanganarelationship
descent groups of three headmenA1, B1, C1. C1 vamakhulu

QUEEN

Ai

A2

2
a3

C,

a,

Al

b82

C2Z

FIG. 10.-Principles of Lovedu political structure


(forexplanationsee text).

is a 'foreigner' headman; A1 and B1 are Lovedu Tonga are thus,by a fiction,keptout of the kinship
altogether.
districtheads in the sense describedabove. X is a structure
I think,makesit clearthattheintimate
This
analysis,
Queen.
the
of
Lovedu nobleman,a kinsman
In the firstphase of the vatanoniprocedure,A1, and intricaterelationshipbetweenthe hierarchyof
B1, C1 give daughtersa2, b2, c2 to the Queen as rankand thepracticeofKachintypemarriage-which
tribute(hu lotva). The Queen accepts c2 simplyas in the Kriges'accountemergesas a kind of paradox
tribute;fora2 and b2 she pays cattle as brideprice. since theyhold that wife giversand wife receivers
These threewomenthenspenda periodat Court as shouldbe ofequal status-is in factbasic to thewhole
and is fullyintegrated
the Queen's ' wives.' Later theyare reallocatedas schemeof politicalintegration
real wivesat theQueen's discretion.We assumethat withthetotal systemof valuesin Lovedu society.
Yet we are left with importantelementsin the
c2 is givento group B, b2 to group A, a2 to the
unexplained. In thisschemethe
noblemanX. By thisactionthe Queen establishesa economicstructure
fotgroupB are now vaduhulu Queen-is for ever paying out bridepricecattle but
ofprecedence,
hierarchy
to groupB and X is apparentlydevoid of any sourcefromwhichthese
to groupC, groupA arevaduhulu

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

The Structural
Implications
ofMatrilateralCrossCousinMarriage

51

cattlemightbe acquired. On this point the Kriges must assume that, overall,both parties-the junior
group and the seniorgroupalike-are satisfiedwith
provideus withno information48.
theirbargain,and therefore
thattheexchangeaccount
The content
ofprestations
'balances.'
But we cannot predict from first
What then is the significanceof this Lovedu principleshow the balance will be achievedbecause
materialforour generalargument?
we cannot know how the different
categoriesof
It requires,I think,thatwe re-examine
just whatis 'prestation' will be evaluated in any particular
meant,by the notion propounded by Granet and society. A Chinesemay be so anxiousto securethe
Levi-Straussthatin Kachin typemarriagewe have a patronageof an influentialson-in-lawthat he will
regularexchangeof womenfor'prestations.' What pay handsomelyfor the privilege:a Kachin in the
? Levi-Strauss,
aretheseprestations
thoughhe usesthe same situationpays to acquire an influential
fatherword,does not,so faras I can discover,examinethe in-law. Both are systemsof exchange,certainly,
but
natureof this categoryat all carefully. Yet it has theycan onlybe understoodif the categoriesof just
become clear in the course of this essay, that the whatis exchangedare carefully
assessed. In anysuch
' in a Kachin typemarriagesystemmay analysisit is veryimportantto distinguishbetween
' prestations
not onlytake on a varietyof forms,theymay have consumableand non-consumable
materials;it is also
structural
functions.
severalquite different
very importantto appreciatethat quite intangible
Considerforexamplethe case of two local descent elementssuchas ' rights' and ' prestige' formpart
groupslinkedin 'wife giving' and 'wife receiving' of thetotalinventory
of' things' exchanged.
and let us assume that one of theseis
relationship,
of higherstatusthan the other. Then what are the
PART 4. CONCLUSION
' prestations
' thatpass fromone groupto the other
What then can be inferredfromthe theoretical
in thefourcontexts:Kachin,Batak,Lovedu,Chinese? discussionsof thefirsthalfof thisessayand thethree
is givenin Table I.
The answerbriefly
briefstructural
analysesgivenin the second?
1. A reviewof the literature
has shownthatif we
TABLE I
ignorevariousquitehypothetical
historical
reconstrucof' prestation'
Statusofreceivinggroup
Variety
Kachin Batak Lovedu Chinese tionsthe mostpopular ' explanation' ofKachintype
..
..
junior junior senior senior
a. women
marriageis to see it as the hallmarkof an imaginary
..
senior senior
b. labour of men
typeof stiucturalsystemin which'marriageclasses'
ofwomen ..
senior senior
in a circle. Radcliffe-Brown
marry
has shown that
c. consumer
goodsis
invalid
for
this
the
I have shown
analysis
Murngin;
..
.
senior
beer
(t)
it to be equallyinvalidfortheKachins.
.
..
senior
junior
meat
(*)
..
senior
livestock .
2. Two major errorsare involvedin the 'marriage
(t)
..
.
senior
money
class'
argument. The firstis to suppose that the
d. non-consumer
goodswhichare pairedas 'wife giving' and 'wife
groups
..
junior(*)
ritualcattle
(t) receiving' by the marriagerule are major segments
..
jewellery ..
(t)
ritualobjects
*
(t)
of the society. In fact,in all cases closelyexamined,
e. territorial
rights .. junior junior junior
theyare local descentgroupsdomiciledin thesame or
politicalprotection.. junior junior junior junior
communities.The seconderror
f. 'prestige,' 'face,' senior senior senior senior closelyneighbouring
is to suppose that the marriagesystemin itself
'honour'
a closed system. In fact,as we have seen,
constitutes
* Livestockamong theLovedu must be consideredas Kachin typemarriageis onlyunderstandable
if it is
non-consumable
goods.
thoughtofas one ofmanypossibletypesofcontinuing
t In thesecases similarobjectsare presentedboth from relationship
betweenpairedlocal descentgroups.
seniorto juniorand vice versa accordingto circumstances.
3. Despitethe stimulating
qualityof Levi-Strauss's
but it illustratesthe argument,
This list is not comprehensive
his mainpropositionis back to front. He
one arguesthatthe fundamental
argument. In any such systemof reciprocities
characteristic
of Kachin
is
it
that
is
type
marriage
must
egalitarian-women
48 The onlypeoplewhomtheKriges(1943,p. 10) mention
for
be
on
a
sort
of
exchanged
fixed
goods,
price
as givingcattleto theQueenarepersonsofequal statusto the
basis. He perceivesthatinfactdifferences
equilibrium
Zulu and Pedikings.
Queenherself-e.g.,theformer
D 2

(845)

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

52

E. R. LEACH

marriageexchangethe equivalentof a woman is


alwaysanotherwoman. Levi-Straussarguesthatthe
equivalentof a woman may also be in goods and
labour and symbolicobjects; but he perceivesmuch
less clearlythatintangiblefactorssuch as rightsand
reputationcan play a part in this exchangewithout
damageto thesystem. The factthattheChinesewho
practiseKachintypemarriagepay on balancea dowry
and not a brideprice,while their neighboursthe
and not a dowryis a crucial
Kachins
pay a brideprice
4. Whentwo local descentgroupsA and B are in
of
which
Levi-Strauss's
fact
theorytakesno account.
relationthe 'things' which can be exchangedto
be
said
on
the
What
side? I suggest
can
positive
expressthisrelationshipcan be roughlycategorised
the
minimal
which
following
propositions
are capable
as follows:
of empiricaltestingin the field.
I. Tangibles
1. With Kachin type marriagethe relationship
(a) 'women' and 'men '49
betweenwife givingand wife receivinggroups is
(b) labourof menor women
hencedifferentiation
of statusone way
asymmetrical;
(c) consumergoods and money
or theotheris morelikelythannot. Such differentia(d) capitalgoods
tioncan be avoidedifa smallnumberofneighbouring
(e) ritualobjects-ofno intrinsic
value
local descentgroupsmarryin a circle,or if thereis
II. Intangibles
a systemof balancingrightsand obligations--aswith
(f) 'rights' of a territorialand political theMurngin;but anysuchsystemof balanceswillbe
nature
unstable.
(g) relative' status' or ' prestige'
2. If the status between wife givers and wife
The-lastitemcannotbe definedexceptin termsof the receiversis unequal one cannot predictfromfirst
culturalsituation;itis simply' thatkindofreputation principleswhichof thetwo groupswillbe the senior.
whichgains a man the admirationof his fellows,'it It seemsprobablehoweverthatin any one cultural
may be derivedfrommurderin one society,
-philan- situationthe positionwill be consistent.One would
thropyin a second,saintlinessin a third.
not expect that all wife giversare ranked high in
In every relationshipbetween individuals and one village,and all wifereceiversrankedhighin the
betweengroups,itemsin theabove listare exchanged. next.
It is in the nature of most such ' exchanges' that,
3. The statusrelationsbetweenwifegiversand wife
as regardsthetangibleitemsa, b, c, d, e thereis always
mustconformto thestatusrelationsimplicit
receivers
an imbalanceon one side or theother. The exchange
in other(non-kinship)institutions:e.g., wherewife
accountis balanced by the intangibleitemsf and g.
to wifereceivers,
one can
be admittedthentheroleofwomen giversare sociallysuperior
If thisargument
thatthepoliticaland territorial
of wife
predict
rights
in the total exchangesystemcannot be discovered
etc.willbe superiorto thoseofwifereceivers,
givers
fromfirstprinciples.
vice versa. In otherwords,whereKachin type
and
It is the errorof Hodson, Murdockand the other
it is partof thepoliticalstructure.
'marriage class ' enthusiaststo suppose that in a marriageoccurs,
4. It seems probable that a costlybridepricein
goodsand labourimpliesthatwife
terni of consumer
- '9 In mostof thisessayI havereferred
to theexchangeof
than wifereceivers. Conversely
rank
givers
higher
muchas iftheywerechattels.Thisis of
womenin marriage
in
consumer
a
dow
ry
expressed
goodsimpliesthatwife
Inmost
the
outcome
societies
coursea grossoversimplification.
is concerned
withvalidating
thestatus giversranklowerthanwifereceivers. A substantial
of marriage
exchanges
societies
thehusband's
ofthewoman'soffspring.
In,patrilineal
of objectsof symbolicand ritual
consisting
groupcan usuallybe said to 'buy' the woman'soffspringbrideprice
fromthewife'sgroup. Yet equallythereare caseswherethe value onlyprobablygoes withhighrankand equality
and ' buy' thesexualservices of status. The absence of eitherwork payment,
wife'sgroupretaintheoffspring
is
ofthehusbandfromthehusband'sgroup. Thisforinstance
andalso ofthe bridepriceor dowry suggestsa breakup of the
matrilocal
trueofthematrilineal
Manangkabau
Kachin typemarriageinstitution.
ambilanakformofmarriage
alreadymentioned.
ofpriceand differences
ofstatuswillresult,butregards
thisas the breakdownof the system,and henceas a
mechanismin the generalprocessof social evolution.
My own argumentis almostthe reverseof this, As
far as the marriagesystemis concerned,the status
relationsbetweengroupA and groupB mustbe taken
as givenfactorsin the situation;a marriageis only
one of many possible ways of 'expressing' those
relations.

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

53

The Structural
Implications
of MatrilateralCrossCousinMarriage

of all the societies Finally there emerges from this-discussion an


5. It is a notablecharacteristic
considered-Murngin, Kachin, Batak, Lovedu- importantprincipleof method. If anthropologists
of a tribalorganisation are to arriveat anyvalidprinciplesof social organisathatthe usual characteristics
are lacking. The network of kinship relations tion,the generalmethodmustbe comparative. But
in its
embracesa greatnumberoflocal groupsand tiesthem the originalcomparativemethod,exemplified
formbytheworkofFrazer,rested
into a kind of loosely knit political system; but mostoverwhelming
throughout
thepopulationso linkedthereis no strong on thecomparisonofculturaltraits.Undertheimpact
whichinsistedupon the analysisof
sense of social solidarityand in some cases thereare of functionalism,
oflanguageand culturebetween whole cultural systems,this type of comparative
evenwidedifferences
methodfellintodisreputesimplybecauseit appeared
different
partsof theone system.
that to be an impossibility-the
It seemsprobablethatthisis a characteristic
body of data thatwould
may normallybe expected to be associated with be involvedin an adequatecomparisonwas altogether
sincetheasymmetricaltoo vast.
Kachintypemarriagesystems,
and his
relationshipbetweenwife giversand wife receivers Since 1930, however, Radcliffe-Brown
of the system followers
havehad somesuccessin applyinga different
tendsalwaysto push the ramifications
in kind of comparison,namely,that of whole political
to widerand widerlimits. The whole structure,
fact, does bear close resemblanceto the feudal systems.In a comparativemethodof thislatterkind
organisationof medievalEurope,whichin the same culturalfeaturesare to a largeextentignored,and the
way drew a number of culturallydivergentcom- 'things' which are compared are reallysimplified
mtnities into a single political system, though modelsof the societiesunderdiscussion,as observed
theintegration
of thatsystemwas, at best, from a particularpoint of view. In practice'the
admittedly
weak.
extremely
particularpoint of view' has been that of kinship,
I certainly
do nothold,as Levi-Strauss
seemsto do, and despitethe verygreatvalue of works such as
is to be foundin Social Organisationof Australian Tribes, Social
thatthe originof feudalstructures
thebreakdownof Kachin typemarriagesystems;but Anthropologyof North American Tribes, African
it does appear to be the case that Kachin type Political Systems and other kindred studies the
marriagesystemscorrelateverywell with political generalisations
thatemergeare liable to be distorted
of a somewhatfeudaltype.
structures
on thisaccount.
correlations
withKachin
This seriesofhypothetical
My own argument,in whichto a greatextentI
is thatthecomparisonof models
typemarriageappear to fitall the cases examinedin followLevi-Strauss,
thisessayand also, so faras theethnography
permits ratherthan of 'whole cultures' is a necessaryand
one to judge, those otherKachin typesocietiessuch valid method-indeed I would go much furtherin
as the Gilyak,the Lakher and the Old Kuki which such abstractionthanhas usuallybeen the case with
uponbyLevi-Strauss;moreover the followersof Radcliffe-Brown.But at the same
havebeencommented
theydo so withoutstrainingthe factsto fita world timeI would insistthatthe comparisonmustalways
embracingtheory of social evolution-which can take into account the whole range of institutional
normally
hardlybe said of some of Levi-Strauss'sarguments. dimensionswith whichthe anthropologist
In themoregeneralfieldI suggestthattheconcept has to deal and muststartfroma concretereality-a
of local descentgroupdevelopedat the beginningof local groupof people-ratherthan froman abstract
thisessaymayhave important
analyticalimplications reality-suchas theconceptoflineageor thenotionof
in many cases- where the pure descent concept kinshipsystem.
'lineage' lacks precision. Moreover'thearguments The contentof thisessayshouldmakeit clearwhy
which have here been propounded regardingthe I hold this view. It also provides an excellent
relevanceof statusconceptsto the analysisof bride- illustration
ofhowverymisleadingcomparisonsbased
price~and dowry have relevance also outside the on the analysisof the kinshipdimensionalone are
immediatecontextof Kachin typemarriage.
liableto be.

(&45)

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

54

E. R. LEACH

AIYAPPAN,A.

..

..

1934

BOSE, J. K.

..

..

1934a
1934b
1936
1937a
1937b

CARRAPIETT,
W. J.S.
..
K. P. ..
CHATTOPADHAYA,
COLE,F. C.
DAS, T. C... ..

..

..

1929
1931

1945
1935

..

DEACON,A. B. ..
..
..
ELKIN, A. P.
ELKIN,A. P., and BERNDT,C. and R.
FEI, H. T... .
FIRTH, R. ..
FISCHER,H. TH.

..

..
..

..
..

1945

1927
1933

1950

1951
1939
1951
1935
1936

FORTES,M.

..

..

1950
1949
1950

FORTUNE,R. F. ..
FRAZER, SIR S. G.
GIFFORD,E. W. ..

..

GILHODES, C.
GRANET,M.

..

..

..
..

HAAR, B. TER
HANSON, 0.
HEAD, W. R.

..

..

..
..

..
..

..
..

..
..

HUTTON,J. H.

..

..

Hsu, F. L. K.

..

..

HILL-TOUT,C.
HODSON,T. C.

DE JONG,P. E.
JOSSELIN

JOUSTRA,M.

..

..
..

DE
..

KAWLU MA NAWNG

..

KRIGE,J. D.

..

..

KRIGE,E. J.,and KRIGE,S. D.


.
....
KULP, D. H

1933

1918
1916
1922
1922
1939

References

"Cross Cousinand Uncle-Niece


in SouthIndia." Congr.int.
Marriages

Sci. anthrop.ethn. London.


"Social Organisationof theAimol Kukis." J. Dep. Letters,25. Calcutta
" Dual Organisationin Assam." J. Dep. Letters,25. Calcutta.
" The Nokrom Systemof the Garos of Assam." Man, 36.
" Origin of Tri-clanand MarriageClasses in Assam." Man, 37.
" Marriage Classes among the Chirusof Assam." Man, 37.

TheKachinTribesofBurma. Rangoon.
"ContactofPeoplesAffecting
Rules." Pres.Add.anthrop.
Sec.
Marriage

IndianSci. Congr.

ThePeoplesofMalaysia. New York.


" Kinship and Social Organisationof the Purum Kukis of Manipur."
J. Dep. Letters,28. Calcutta.
The Purums. Calcutta.
" The Regulationof Marriagein Ambryn."J. Roy.anthrop.
Inst.,57.
" Marriage and Descent in Eastern ArnhemLand." Oceania, 3.

ArtinArnhem
Land. Melbourne
and London.

" Local Organisationin ArnhemLand." Man, 51.


Peasant Life in China. London.
Review of Artin ArnhemLand (Elkin and Berndt,1950) in Man, 51.
"De Aanverwantschapbij eenige volken van de Nederlands-Indische
Archipel." MenschenMaaische7ppij,
11.
Het asymmetrisch
cross-cousinhuwelijk'in NederlandsIndie."
Tijds.,
Indisch.Taal-, Land-, Volkenk.,76.
" The Concept of Incest in Sumatra." Amer.Anthrop.,
52.
" Time and Social Structure:an AshantiCase Study" in Social Structure
(Ed. Fortes, M.)
"Kinship and Marriage among the Ashanti" in AfricanSystemsof
Kinshipand Marriage. (Eds. Radcliffe-Brown,
A. R. and Forde, D.)
Oxford.

"A Note on Some Formsof KinshipStructure."Oceania,4.

Folklorein theOld Testament. London.


"Miwok Moieties." Amer.Archaeol.ethn.,12.
"California Kinship Terminologies." Amer. Archol. ethn,18.
The Kachins ^theirReligionand Mythology. Calcutta.
" Categories matrimonialeset relations de proximite dans la Chine
ancienne." Ann.sociol. Ser. B Fasc.1-3.
1948 Adat Law in Indonesia. NewYork.
1913
The Kachins: theirCustomsand Traditions. Rangoon.
1917
HandbookcoftheHaka Chin Customs. Rangoon.
1907
BritishNorthAmerica. London.
1921
" The Garo and Khasi MarriageSystemsContrasted." Man in India, 1.
Ranchi.
1922
The PrimitiveCultureof India. London.
1925
"Marriage of Cousins in India." Man in India, 5.
1921a Censusof India, 3, Assam. AppendixB.
1921b The Sema Nagas. London.
1940
"Concerning the Question of Matrimonial Categories and Kinship
Relationshipin China." Tien Hsia Monthly,11. Shanghai:
1945
" Observationson Cross Cousin Marriagein China." Amer.Anthrop.,
47
1951
andNegriSembilan. Leiden.
Minangkabau
1911 Batakspiegel. Leiden.
1942
The Historyof theKachinsin theHukawng Valley(ed. Leyden,J. L.).
Bombay.
1939
" The Significanceof Cattle Exchanges in Lovedu Social Structure."
Africa,12.

1943
1925

TheRealmoftheRainQueen.. Oxford.

CountryLife in South China: The Sociologyof Familism. Vol. 1.


York.

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

New

The Structural
Implications
ofMatrilateralCrossCousinMarriage
LAWRENCE,
W. E. and MURDOCK,G. P.

1949

..

..

..

..

1945

..

..

..

1949

..

..

..

..
..

MALINOWSKI, B. ..
..
MAUSS, M...
MURDOCK, G. P...

..

..

..

..

..

..

1932
1923
1949

LEACH, E. R.
LEVI-STRAUSS,C.
LOEB, E. M.

1935

D. A.
OLDEROGGE,

..

..

..

..

..

..

PARRY, N. E.

..

..

..
..

1932

..
..

1914

..

RADCLIFFE-BROWN,
A. R,
RICHARDS, F. J.
RICHARDS, A. I.

..

..

..

..

..

..

..

RIYERS, W. H. R. ..

1946

1930
1951
1951

1907

1914
1921
1936

Roy, R. C.
RuHEMANN, B.

..

..

..

..

..

..

..

..

1948

SELIGMAN,B. Z.

..

..

..

..

1927

1928

SHARP, LAURISTON

..

..

..

1934

STEVENSON,H. N. C.

..

..

..

THOMSON, D. F.

..

..

..

(1943)
n.d.
1949

TIDEMAN, J.
WARNECK, F.

..

..

..

..

..

WARNER,W. L.

..

..

______

WEBB, T. T.
WEDGWOOD, C.
WEHRLI, H. J.

........

References
"Murngin Social Organisation." Amer.Anthrop.,51.
"Jinghpaw Kinship Termfinology." J. Roy. anthrop.Inst.,75.
Les structures
e'lmentairesde la parente. Paris.
" Sumatra:
its History and People."
Wienew Beitr. Kulturges.
Linguistik,3.
The Sexual Life of Savages. (3rd ed.) London.
"Essai sur le Don."
l'Ann. sociol., N.S., 1.
Social Structure. New York.
" The Ring Bond between Clans or the Three-Clan Union (Gens
Triplex)." BriefCommunications,
Inst. Ethno., 1. Moscow.
The Lakhers. London.
" The Social Organisationof AustralianTribes." Oceania, 1.
" MurnginSocial Organisation." Amer.Anthrop.,
53.
"Cross Cousin Marriage in South India."
Man, 14.
"Some Types of Family Structureamongst the Central Bantu " in
AfricanSystemsof Kinshipand Marriage. (Eds. Radcliffe-Brown,
A. R.,
and Forde, D.) Oxford.
"The Marriage of Cousins in India."
J. Roy. asiat. Soc.
The Historyof MelanesianSociety. Cambridge.
" Kinship and Marriagein India." Man in India, 1.
" Notes on the Chawte Kuki Clan."
Man in India, 16.
" The RelationshipTerms of Some Hill Tribes of Burma and Assam."
Southwestern
J. Anthrop.,4.
" Bilateral Descent and the Formation of Marriage Classes." J. Roy
anthrop.Inst.,57.
"Asymmetry in Descent, with Special Reference to Pentecost." J. Roy.
anthrop.Inst.,58.
" The Social Organisation of the Yir-Yoront Tribe, Cape York Peninsula."

Oceania, 4.
The Economicsof the CentralChin Tribes. Bombay.

EconomicStructureand the CeremonialExchange Cycle in ArnhemLand.


Melbourne.

..

..
..

1922
1901

..

..

1930-31 " Morphology and function of the Australian Murngin type of Kinship."

1937
..

..

..

..

..

..

..
..

..

..

..

..

..

..

WOUDEN, F. A. E. VAN

55

1933
1929

1904
1935

Simeloengoen. Het land der Timoer-Bataks. Leiden.


" Das Eherecht bei den Toba-Batak." Bijd. Taal-, Land- Volkenke.
NederlandschIndie,53.

Amer.Anthrop.,
32, 33.
A Black Civilisation. New York.

" Tribal Organisation in Eastern Arnhem Land."


" Cousin Marriage." Encyclopaedia Britannica.

Oceania, 3.
(14th ed.)

" Beitrag zur Ethnologie der Chingpaw (Kachin) von Ober-Burma."

Int.Archiv.
Ethn.,16,Supp.

Sociale Structuurtypenin de Grote Oost.

(845)

Leiden.

D4

This content downloaded from 78.96.197.15 on Sun, 15 Mar 2015 21:23:04 UTC
All use subject to JSTOR Terms and Conditions

Вам также может понравиться