The Concept of Kinship: With Special Reference to Mr. Needham's "Descent
Systems and Ideal Language"
Ernest Gellner
Philosophy of Science, Vol. 27, No. 2. (Apr., 1960), pp. 187-204.
Stable URL:
http://links.jstor.org/sici?sici=0031-8248%28196004%2927%3A2%3C187%3ATCOKWS%3E2.0.CO%3B2-Q
Philosophy of Science is currently published by The University of Chicago Press.
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained
prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in
the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/ucpress.html.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.
http://www.jstor.org
Tue Mar 25 07:25:17 2008
THE CONCEPT OF KINSHIP*
With Special Reference To Ms. Needham's
"Descent Systems and Ideal Language"
ERNEST GELLNER
The London School of Economics and Political Science
The purpose of this paper is not merely to reply to Needham's arguments1
and to correct his errors, but also, in the course of this, to throw some light
on the anthropological concept of kinship. For simplicity of reference, it will
be useful to number Needham's errors.
(1) Needham (p. 97): "Biology is one matter and descent is quite another,
of a different order."
This is not so. This mistake of Needham's is particularly important, for
two reasons: it plays a crucial role in the internal economy of Needham's
argument, and it is a mistake which is not merely a personal idiosyncrasy of
Needham's. It is a dangerous travesty of a valid idea, and a mistake which is
perhaps shared by others: thus its discussion serves a wider purpose.
That it is a mistake can best be seen as follows. Suppose an anthropologist
observes, in a society he is investigating, a certain kind of recurring relationship
between pairs of individuals or of groups. (It may be a one-one relationship
or a one-many or a many-many one. It may be a relationship of authority,
or a symmetrical one of, say, mutual aid, or of avoidance, or whatnot.) Suppose
the autochtonous term for the relationship is blip.
The crucial question now is: Under what conditions will the anthropologist's
treatment of the blip-relationship fall under the rubric of kinship structure ?
It will be so subsumed if the anthropologist believes that the blip-relationship
overlaps, in a predominant number of cases, with some physical kinship
relationship. Otherwise, naturally, the blip-relationship will be subsumed
under some other rubric, such as of "authority" or "economy". What, other
than at least partial overlap with physical kinship, could conceivably lead a
relationship to be classified as a part of "kinship structure" ?
The remark which Needham makes immediately after the assertion of
his crucial error is important: "They (i.e. biology and descent) will usually
be concordant to some degree . . ." Needham believes this statement to be
(i) a proposition which reports, as it were, a de facto usual concordance, and
(ii) a minor concession which does not affect his main contention, namely
that "the defining character of descent systems is social".
In fact, however, the remark about the "usual concordance" of biology
and descent unwittingly contains the operational definition of "kinship
structureya,or "descent system", giving away the condition under which a
recurrent relationship can come to be classed under those terms. Far from
* Received November, 1959. 

' See Philosophy of Science, Vol. 27, no. 1, pp. 96-101. 

187
188 ERNEST GELLNER
it being the case that, as it were, it so happens that biology and kinship are
c r
usually . . . concordant to some degree", the fact that some social relationship
is "usually concordant to some degree" with physical kinship is, on the
contrary, the main condition for that relationship being classed as "kinship".
This is not primarily a discovery about societies, but rather about the anthropologist's
use of terms. Moreover, it is not a minor concession compatible with
Needham's argument, but a decisive refutation of it.
My point can be seen not merely from the consideration of what "kinship
structure" could possibly mean, but equally also from examining the actual
use of kinship-structure classifications of anthropologists. Needham goes on
to claim that Jzis point can be "seen in some institutions as . . . unilineal
descent reckoning . . . adoption . . . leviratic marriage . . . ghost marriage".
But Needham's own examples and indeed the very notions conveyed by each
of these terms in fact prove the very opposite from what Needham imagines
they illustrate.
Let us take Needham's examples, sticking also to his own accounts of the
meaning of the relevant notions:
"Leviratic marrage, in which a man marries a deceased brother's widow
and raises descendants in his brother's name". This is the simplest case of
all: the leviratic relationship and its offsprings are, as Needham's own definition
clearly shows, a function of kinship. The anthropologist's kinship term
ccleviratic"is only applicable when certain real kinship relationships obtain.
The relationship, and its offspring, can only be identified by the anthropologist
as crleviratic" because the anthropologist knows that the fiction by which
offsprings are raised "in the dead man's name" is indeed a fiction. If the
anthropologist did not know this, if he too accepted the social fiction of the
dead man's name, he would not be able to notice and identify the phenomenon
which, as Needham's account lucidly shows (contrary to his intention) is
defined in terms of a systematic, regulated disparity between physical and
social kinship. The identity of the deputy for the dead man is also fixed in
kinship terms.
One of Needham's confusions (error 2) is the idea that a 'Yunction" is
always "identity". Identity is indeed a functional relation, and perhaps the
simplest one, but of course there are innumerable others.
Take "adoption". Again, as with leviratic marriage, the very use of the
notion, the possibility of classifying offspring as adoptive, depends on the
observer's knowledge of the disparity between the social and the physical
relationship, and it is this disparity which gives the term its meaning. It is
true that in this case the social kinship relation is only negatively a function
of the physical one: only a disparity is required, and the applicability of the
term "adoption", unlike "leviratic marriage", does not also require specific
There is one complication here: it might be objected that the brotherhood relation between
the deceased man and his substitute-genitor might itself not be physical but a consequence
of an earlier social "fiction". In individual cases, certainly, but if it happened regularly, this
would itself affect the classificatioil of the kinship structure in which it happens.
189THE CONCEPT OF KINSHIP
physical kin relationship (e.g. the dead man's brother being the physical
father). And note: it is not just contingently true that a physico-social disparity
occurs in the case of adoption, but it is, roughly, what "adoption" means.
With regard to "ghost marriage7' similar observations apply.
Finally, there is the example of "unilineal descent reckoning". (I have spent
some considerable time studying obe society of this kind, and can hardly
be supposed to be as ignorant of the phenomenon or its implications
as Needham supposes . . .) This, again, is a clear case. In such a situation, a
person's membership of a lineage is determined by the lineagemembership
of one of his parents and theirs in turn similarly, and so on. Again, there
could hardly be a clearer case of social kinship being a function of physical
kinship. The "function", the rule specifying the connection, is that a person's
lineage is that of his male (or female) parent only, and physical and social
parentage overlaps.
The situation can also be made clear as follows: anthropologists frequently
say that, for instance, kinship is of great importance in simpler societies,
or that in some societies of this kind a man's position in the social structure
is determined by his birth. Suppose for a moment that Needham were right,
and that this meant (merely) social kinship and (merely) social birth, and
that the connection of these with physical kinship or birth were merely
contingent and sociologically irrelevant. The meaning of the statements cited
would then degenerate into saying something almost wholly vacuous, namely
that simpler societies have some kind of structure of relationships, and that
a man's social position is determined by something . ..These almost vacuous
statements however would hardly differentiate one kind of society from
another. The only content possibly left to these assertions might then be,
that the structure of relationships or the social position of an individual in
simpler societies are conceived by their members in terms borrowed from
physical kinship. But the original statements meant far more than this, and
in any case, this degenerate statement, to the effect that social relationships
are described by participants in terms borrowed from physical kinship, would
be highly suspect: for the ethnographer has to decide how he translates the
autochtonous social terms, and the main reason he can normally have for
translating them as kin terms is that their application does in fact overlap
with physical kinship. If it did not, or if the overlap were contingent and
irrelevant, the merely terminological overlap which, by implication, we should
be left with on Xeedham's analysis, would be at the mercy of the vagaries
of an inherently arbitrary tran~lation.~
For instance: in a certain society, the same term is used by a woman to describe her husband's
father, and (by everyone) to describe the chief of the clan. Of course there is a tenuous
connection between the two notions, but one would not, on the strength of this terminological
overlap, place an account of clan chieftancy (which in this case is not a function of the kinship
position of its holder) in the chapter dealing with kinship structure. If the overlap or otherwise
with physical were really irrelevant, one would then have to deal with chieftancy under kinship
in this context, for one could say that the term is borrowed from kinship.
190 ERNEST GELLNER
Needham's next error (3) is the assumption that if one thing is a function
of another, it is a function of that other thing and of nothing else. Functional
dependence may be complex. A thing may "be a function of", i.e. regularly
vary with, a number of other "things" or variables.
T o sum up this part of the argument: "kinship structure" or "descent
system7', and the more specific terms used in connection with them, are
anthropologists' terms (not terms used by the people studied). Their meaning,
as implicit in the actual practice of anthropologists, is that set of social relationships
which largely, though not completely, overlaps with some of the physical
kinship relationships of the people studied. They can be said to be those
social relationships which are a function of physical kinship relationships,
though the function (i.e. the rule specifying just which kinship relations
correspond to which social ones) need not be a simple one, and indeed often
is complex.
It goes without saying that social kinship relations are not the function of
each and every physical kin relationship obtaining in a society. There are
systematic omissions, and the manner in which the significant physical kin
relationships are selected is of course just that which it is the job of the
anthropologist to find out.
Moreover, the functional relationship between physical kinship and social
kinship roles is such that it needs only hold in the predominant number of
cases, whilst allowing of individual exceptions: for instance, an undetected
infidelity on the part of a wife may lead to a disparity between the physical
and the social relationship of fatherhood, without this having any social
consequences and hence without being of any great interest to the social
anthropologist.
It might be supposed-erroneously-that difficulties arise for this account
of what "kinship structure" means, from two kinds of sources: (a) the fact
that, as stated, social kin relationships needs only overlap with the corresponding
physical ones in a predominant number of cases, but not invariably, and
(O) from the fact that discussions of kinship structure also include accounts of
some relationships which, as relationships, do not, at any rate directly, overlap
with any physical kin relationship at all. In other words: not only may
individual cases of a relationship be exceptions to the concordance, but a
whole class in type of relationships may be exceptional in this respect. Neither
of these difficulties invalidate the case, but they will have to be examined
in detail.
(a) The issue may best be discussed with the help of an example. In our
society, the (social) father of a child is generally also his physical genitor,
and it may be said that the social role is a function of the position of the two
individuals on the physical kinship map. But it is also true that occasionally
there may be a disparity between social paternity and the genetic one, a
disparity of which the persons concerned may or may not be aware. Does this
possibility show that social paternity is not, after all, a function of physical
kinship ?
191THE CONCEPT OF KINSHIP
Not at all. Suppose for a moment that our society changed: suppose
promiscuity became so widespread that, instead of disparity between social
and physical paternity being fairly rare, it became so general that it would
be unusual for a person to know the identity of his physical father. At the
same time, suppose the social institution of "fatherhood" to remain unchanged.
If it were true, as Needham claims, that a descent system is a thing "of a
different order" from physical kinship, and has no connection, or only a
contingent one, with it, we should then be forced to say that the kinship
system of the second society is the same as that of our present one. For, ex
hypothesi, nothing has changed in the social roles and relations.
But, on the contrary, the "kinship structure", or "descent system" such
as our present one, in which social paternity generally is ascribed to the
physical genitor, albeit with some exceptions, both conscious and unconscious,
is quite different, and would by any social anthropologist rightly be counted
as different, from a system in which physical paternity were unidentified or
disregarded, and social paternity then ascribed in virtue of some fact (necessarily)
other than physical paternity. It would count as different despite the
the fact that the social role of the "father" vis-a-vis his "offspring" and society
might in all other respects be similar to that of a father in our society.
This gedankenexperiment shows that, although occasional divergences
between social and physical kinship do not affect the principle that a social
kinship role is a function of a physical kin relationship, a regular divergence
results in a difference in what counts as kinship structure. The kind of
convergence there is, or lack of it, is part of what anthropologists mean by
the kinship structure of a society, and not something contingently true of
it and sociologically irrelevant.
(b) There are other terms which one may encounter in discussions of
kinship structure, such as, say "clan" or "godfather", where the independence
of the meaning of such a term from physical kinship is of a different kind.
Let us take each of these terms in turn. The point about a "clan" is this:
its members may believe themselves to be descended from a single common
ancestor, without necessarily knowing just how, whilst this belief is probably
false and in any case irrelevant. We have here a case which seems to favour
Needham's argument: a "kinship" concept which has a social significance,
defining a co-operating group, etc., by means of a kin notion, whilst the
truth about the physical links does not correspond to the notion and is
irrelevant to it. In brief, we cannot here give an account of the logic
of this anthropological concept similar to that given in connection with
Ir
leviratic marriage" etc.
The correct account of the logic of such terms is this: indeed, a term such
as "clan" does not correspond to a physical kinship reality. But: "clan'' is
a concept essentially related to other concepts-they might be "sub-clan"
or "lineage" or "extended family"-which, in turn, do denote groups for
whom some social reality (co-operation, cohabitation, for example) does have
a reasonable and systematic congruence with some kinship affinity. A "clan"
192 ERNEST GELLNER
will be a co-operating set (justifying or expressing or reinforcing its co-operation
in terms of a kin myth) of sub-groups, which in turn are kinship groups
in the sense analysed above.
The reason why "clan" occurs within discussions of kinshipstructure is
not because clansmen subscribe to the myth of a common ancestry, (although
they do), but because the relationship of belonging-to-the-same-clan is one
which, when spelt out, reads something like this: belonging to a group, of
groups,, where group, is in fact socially defined (whatever its possible kinship
myth), whilst social groups, have cohesion-supporting kinship convictions
whose claims are parallel, on the whole, with physical reality. Hence, the
relationship of being-of-the-same-clan satisfies izdirectly the criterion we
set up initially for determining whether an unspecified social relationship,
bl+, does or does not fall under the rubric of kinshipstructure.
The case of a term such as "godfather" or "blood-brother" is different
from cases such as "clan". Here, it is generally plain both to participants and
to the observing anthropologist that there is no physical kinship. What there
is, generally, is the occurrence of some kind of ritual which establishes a
relationship either similar4 to, or systematically parallel5 with, relations
dependent on physical kinship. Indeed, anthropologists are liable to refer to
this kind of kinship as "ritual kinship". The point about calling this kind of
relationship ritual kinship is not the fact that its establishment is accompanied
by ritual, for that would not differentiate it from ordinary kinship relationships
which are similarly initiated, (though they are functions of the physical
position, they also depend on the confirming ritual); the force of the expression
"ritual kinship" is "only ritual kinship9'-or "kinship-but not really".
There is of course an element of truth in the view Keedham stresses
(though this element, far from being in conflict with what I originally said,
is precisely what my scheme was meant to bring out): namely, the fact that
social kinship systems are not identical with the reality of physical kinship,
but, on the contrary, systematically add to it, omit from it, and distort it.
The kind of situation which brings this out may be instanced by a tribe
in which genealogies become untrue at, or after, the fourth generation: men
will name correctly their grandfather and perhaps his father, but ancestors
beyond that are simply "arranged" so as to express, symbolise, subgroups
existing in the tribe now. There will, in the simplest case, be one alleged
common ancestor for the whole tribe, who will be said to have had n sons,
corresponding to the n sub-groups of the tribe. Each of these n minor ancestors
may in turn find himself attributed a number of sons corresponding to the
number of the sub-sub-groups of his sub-group, and so on, until these
fictitious or quasi-fictitious genealogies are as it were tied on to the more
accurate three or four generations deep genealogies of living men.
This is a fairly typical situation, and of course many complex variations
A blood-brother may have to behave like a real one.
5 A godfather does not do the same things as a father, but to every father there may correspond
one or a fixed number of god-fathers, performing duties complementary to the father's.
THE CONCEPT OF KINSHIP 193
of it occur. It is this kind of thing, amongst others, which makes anthropologists
insist on not equating kinship beliefs with kinship reality. But: in order
to see how the two diverge, and to say how kinship notions are "manipulated"
in the interests of the present group and of expressing its present alignments,
one has to know what the physical kinship reality is which is being distorted
and manipulated and how it is manipulated. Where there is a relationship,
say blip again, which does not in any reasonably regular way distort some
physical kinship relationship, (i.e. is not a function of it), an account of its
working will not fall into the account of a "descent system". In other words,
relationships which do fit Needham's claim that they are of a wholly different
order from physical kinship do not however fall under the rubric of social
kinship (or "descent system") either.
T o sum up the position: "kinship structure" or "descent systems" are,
by definition, systems of social relationships such as are functions of (are
regularly related to) physical kinship, bearing in mind that the function is
not identity; the rule relating the physical kinship and the social relation being
generally complex, involving additions, omissions and distortions; and all
this notwithstanding the fact that individual instances of the relationships
may occasionally diverge from the rule (e.g. consequence of undetected
infidelity), and also that individual concepts the system of social
concepts (e.g. "godfather") may fail to be related directly by any rule to
physical kinship (for they remain embedded in a system of concepts most
of which are so related).
It must be stressed that this definition is not a prescription or recommendation,
but an explication of what is implicit in the actual practice of anthropologists.
Needham in all probability knows how to use the notion of kinship or
descent. It is only when, as in his article, he attempts an explicit account of
the concepts that he goes off the rails.
Kinship structure means the manner in which a pattern of physical relationships
is made use of for social purposes-the way in which a physical criterion
is used for the selection of members for a group and the ascriptions of rights,
duties etc. Of course, the available physical facts are used selectively, distorted
(but systematically), with some irregularity, etc. But: the elements of the
physical pattern are essentially simple and universal, whilst the social patterns
imposed on it are highly diversified and complex. And it is just this, the
existence of the universal and simple physical substrate, which makes it
possible to describe descent systems with some precision and to compare
them meaningfully.
Once this central point about kinship notions is grasped, many of Needham's
other numerous errors fall into place. I should add that, in one way, I have
found Needham's article greatly reassuring. After I wrote and published the
article which Needham attempts to criticise, I was somewhat worried by the
fact that the point about the logic of kinship was something so obvious that
it did not need stating. I knew that some anthropologists spoke as if they
subscribed to something in contradiction with it, but it seemed charitable
194 ERNEST GELLNER
and reasonable to assume that this was merely a manner of speaking: and when
discussing the matter with some of them, I found that this was indeed so.
Needham, however, by coming out explicitly, and indeed with astonishing
self-assurance, to assert the contrary viewpoint, has put me in his debt by
showing that the point did need stating . . .
This brings me to Needham's error (4), concerning the motivation
anthropologists have for adopting the misleading slogans about the alleged
independence of physical and social kinship. (To Needham, of course, they
appear not misleading but wholly true.) Needham: "It has nothing to do with
the alleged anxiety of social anthropologists about the social nature of their
subject . . . but derives from the very nature of descent systems." Needham
gives no reason whatever for supporting this ex cathedra announcement;
presumably, as he holds the misleading slogan to be true, no further explanation
of why it is made is required. But in view of the fact that, as shown, the
very opposite follows from "the very nature of descent systems", one does
need an explanation.
It is indeed true that what Needham says does sometimes, as he puts it,
"form an early part of elementary instruction in social anthropology". (Italics
and irony mine.) The valid insight which may be conveyed by means of this
type of erroneous formulation, such as Needham's formula about physical
and social kinship being of different orders, is that these two are not identical.
It is necessary to train the student not to take genealogies at their face value,
to put it roughly, and secondly not to accept myths of his own society, such as
the "natural" affinity of brothers etc. (What people say about their own kin
connections is not necessarily or generally true; and if there is an affinity or
any other behavioural relationship between kin of given categories, then this
is never a consequence of the physical kin relationship as such, but of the
social arrangements sustaining it.) All this is true, and in order to stress it,
slogans about the independence of social and physical kinship are used. But
these slogans, or rather the true element contained in them, are not in conflict
with the analysis of "kinship" given above; Needham, however, unlike other
anthropologists, takes them too literally and is misled.
Indeed Needham is right about this being the pofzs asinorurn of the matter,
if not about the identity of the asinus. It is true that those who have difficulty
in seeing even the crude point-that physical and social kinship are not
identical-will not see, and might as well not bother with, the slightly subtler
point concerning their essential connection. One might add that those who
see only the crude point, and not the slightly less crude one, may still perhaps
do useful ethnographic work: at that level, conceptual self-knowledge may
not matter, (whilst a failure to see even the crude point, and thus to be unable
to operate the concept, would indeed be disastruous). But if, like Needham,
one also aspires to give an analysis or account of the anthropological notion
or kinship, it does matter a great deal.
The next cluster of Pu'eedham's mistakes depends only in part on his misunderstanding
of the concept of kinship or descent, springing also from his
195THE CONCEPT OF KINSHIP
failure to understand the notion of an ideal language. The feature of an ideal
language which is relevant to the argument is this: a logically ideal notation
is one such that, when a statement is made in it, it is easy to see what is
contributed by the rules of the notation itself, and what depends on the
particular object of the statement. The elements "contributed by the notation
itself" may be said, roughly, to be those features which are true of the subject
matter of a statement simply in virtue of it being described, or being describable,
in the notation at all. T o give a simple example: suppose one describes
an area made up of regular squares by giving the coordinates of each
square, plus some additional property of each square, say its colour. "4.7;
yellow" would then be a statement in a language devised for describing the
area and its constituent squares. This language would have the desired
logical property. The fact, for instance, that each square is bounded by
four other squares, that it is characterised by two numbers, and no more than
two, and that no pair of numbers can designate more than one square, is
clearly seen to be not so much a property of any particular square, but of the
notation. It really specifies the conditions which a "world" must satisfy
before it can be properly described by this notation. (For instance, a more
than two-dimensional continuum would not be describable by this notation.)
My application of this to kinship systems in anthropology was based on
the observation that the subject matter of the anthropological treatment
of kinship could be made in some measure to satisfy such conditions.
The subject-matter in question is people plus, as shown above, the nature,
overlap, interdependence and divergence of their physical and social kinship
and not social relations alone, as Needham argues). But the physical kinship
relations of people are, ultimately, very simple: these "elements" of the
system, i.e. people, are either not related at all, or related by one and only
one relation, a triadic one (A and B begat C). They may be related by this
relationship either directly, or indirectly by its reiteration (in the course of
which, naturally, the same individual may occupy different positions).
This provides us with, as it were, the substrate of the co-ordinates,
of physical kinship whose basic pattern is, for the purposes of social sciences,
given.
Superimposed on this pattern and corresponding to the colours in our
simple examples above, are the social predicates, i.e. the kinship terms actually
used by the society under investigation, the rights and duties allocated, etc.
The subject matter of an investigation into a society's kinship structure or
"descent system" is a specification and understanding of these terms and
roles, etc., and the nature of their overlap with physical kinship: the manner,
and extent, in which they are selected out of the physical kin substrate of the
society. The type of notation suggested would, if devised, bring out more
clearly the elements which are being related.
Strictly speaking, the situation is rather complex, for three levels are
involved:
(i)The basic pattern is determined by the given fact (given to social studies
196 ERNEST GELLNER
by biology) that every person has two physical parents, one of each sex. The
same point of course again applies to these parents, and so on. All this is, for
social studies, "necessary" and not a result of some findings. (If it were not
so in some cases, it would not, ex hypothesi, be social anthropology that had
found it out.)
Furthermore, it is a part of this basic pattern that other people may and do
in fact exist, who will either be unrelated, or related to the initial person
either by his mating, or the mating of one of his ancestors, either as a partner
or as offspring; and again, by a kind of transitive "contagion", if, for any of
the kinds of reason stated above, A and B are related, and B and C, then of
course so are A and C.
This and only this gives us the basic physical map, so to speak. In as far
as the notation of our proposed language would have to mirror it, the
"structure" of all names, at this level, would be identical. (The great practical
difficulty for devising the notation would of course be in connection with
specifying the type of relations indicated in the second paragraph. The
necessity of ancestors makes a notational device for their formal specification
easy; the contingency of descendants, mates and collaterals makes it difficult.)
(ii) Within this universal form of every person's name, there could and
would be similarities and divergences of structure, in another but still physical
sense. Suppose, to take a simple case, a person were the child of parents who
were themselves full siblings: this would manifest itself in his "name" by
the repetition of the same (incorporated) name in the place reserved for
paternal and for maternel grandfather, and similarly for paternal and maternal
grandmother. In this sense, the "structure" of a name would not be invariable.
(iii) Having this "map", the ascription of social kinship terms, roles, duties,
etc. to points on it can be expressed in terms of the additional, "synthetic"
predicates ascribed to each type of name, relatively to the name of "Ego".
("Ego" would not have to be supplied with a name, but the ego-name would
be a kind of permanent "origin" of the system.) If, for instance, Ego's name
is A, a rule of our notation would enable us immediately to infer, or rather
construct, the "name" of his physical father, say B. The fact that in our society
the physical father is also the social one might be expressed by saying that,
for the society, F(B).
The occasional occurrence of adoption could be conveyed by saying that,
in some cases, "G(X) leads to F(X)", where the variable X shows that this
social ascription is independent of the physical kin position, and where G,
like F, is a social predicate, and in this case means something like "performed
acts necessary for adoption".
The job of describing further each of these levels is at present generally
done-satisfactorily enough for practical purposes-by the diagrams used
by anthropologists to sketch kinship systems and of course by the accompanying
prose.
The theory which Needham finds so surprising in my article-that I should
start my analysis with a reference to names and not categories-corresponds
197THE CONCEPT OF KINSHIP
to the fact that anthropologists' diagrams of kinship also start with an individual
(typical) "ego".
The ordinary anthropological manner of doing all this is of course satisfactory
for practical purposes. But it fails to bring out the difference in
logical status of truths of the three levels, and the manner in which they are
related. The anthropologist, his intuitions trained and formed by familiarity
with his subject matter and his habitual techniques for handling it, knows
how to interpret the information despite the logical untidiness of its presentation.
An "ideal notation" would however distinguish explicitly between
information of differing logical status. This might or might not be of practical
importance. Perhaps not. But, in any case, it would assist anyone interested
in the logic of anthropology to avoid confusions such as Needham's.
As indicated, the central figure of an ideal notation is that it separates,
in any statement, that which is provided by the notation itself and that which
reflects something in the particular object described. The central feature of
"kinship structure", on the other hand, is that it is an account of how social
ascriptions-kinship terms, roles, etc.,-are superimposed on to, or recruited
in terms of, a pattern of physical relationships which are biologically given.
once these two facts are grasped (and both evade Needham) the relationships
of the two subjects become clear. The basic rules, co-ordinates of an ideal
kinship notation, would utilise the universality and basic simplicity of physical
kinship; what would be said about each society's own and possibly idiosyncratic
manner of arranging its affairs on this basis, would be conveyed by
means of the further predicates, contingently attributed to various relative
positions or "names" in the scheme.
Needham's specific errors can now be brought out. He believes (p. 1)
mistakenly (error 5) that it constitutes a refutation of my argument to say,
as he does, that "descent systems do not simply regulate marriage, but are
as c c
importantly concerned" with other matters. Quite so; but the remark
connecting kinship structure with the specification of possible mates was,
in the context, bringing out that the relationships making up the physical
substrate of kinship consist only of the presence, reiteration, or absence of one
and only one relationship (the triadic one of mating-procreation), and that
the only choice which exists at this level6 concerns the identity of the partner.
Hence, at this level, the kinship structure is exhaustively determined by the
rules governing mating. All else on this level follows. The social relationships,
the consideration of which become relevant later on, are superimposed on a
pattern which is moulded, exclusively, by the pattern of mating, for there
is nothing else which can affect it. The social predicates which are then
imposed on the pattern, and which also (to complicate matters) enter into
the rules determining that pattern itself, were discussed later. This criticism
of Needham's is a misinterpretation arising out of his failure to consider
the context of the remark.
Disregarding for the moment things such as infant exposure, abortion, etc., which affect
the quantity of surviving offspring.
198 ERNEST GELLNER
Error 6: Needham credits me with the view or assumption that kinship
terms are names, individual designations of single human beings. He rightly
finds such a view "almost incomprehensible", and one may only wish he had
found it wholly so. But I am aware, oddly enough, that "uncle" is not a name,
like Harry S. Truman. But an ideal language of the kind envisaged would
indeed have to begin with names, i.e. with unambiguous devices for picking
out single individuals. It does not however follow in the least-nor was it
supposed, nor is there any evidence in the relevant article to make it seem sothat
these names were to replace kinship terms. On the contrary, kinship
terms were to be defined in terms of the relation between types and classes of
types) of names in the ideal notation, just as in normal anthropological
practice kinship terms are defined in terms of the relation between types
of position (and groups of positions) on the diagrams. (Needham himself later
notices conclusive evidence in the article itself against his interpretation, when
he notes that all names in the scheme would have had the same structure.)
Kinship terms are indeed classifications relative to an individual, and the
intention of my scheme was that a kinship term of a given society would be
definable in terms of the kinds of names that were to be related by it. (For
instance: suppose the names were constructed by including a list of ancestors'
names in a certain prescribed order. Then the concept of "cousin", as occurring
in our society, would be defined as the relationship ascribed to people whose
names contain the identical sub-name at certain places, i.e. the places in which
grandparents are named.)
Error 7: Needham supposes he can refute the contention that "for the
purposes of social science it is a logical truth that a man has a man for a
father" by pointing to phenomena such as a woman becoming, socially, the
"father" of a man, etc. I am not unaware of these phenomena. He has, of
course, completely misunderstood the assertion: the point is that the presupposed
biological truth is given to social anthropology (and is a "logical
truth" for it), and is used by it, and not that there is some common universal
pattern of social paternity (which would indeed be absurd). Needham, who
fails to see that a sociological account of social relationships such as marriage
involves plotting their relationships against the existing physical facts (or,
in marginal cases, against the lack of physical facts), consequently also fails
to see why and how the truth imported from biology plays a part in
anthropology. (The same error reappears later in connection with mating.)
Needham's confusion here, as above, springs from the failure to see that the
very identification, description and explanation of these phenomena requires
one to see them for what they are, i.e. as involving a disparity between a
social term x and physical kinship category, and to be able to say what the
disparity is and how it is sustained. If one were not aware of the disparity
with physical fact, about which we are informed by the given truths of biology,
we should be unable either to note or to categorize these phenomena. Physical
kinship categories are universal, as human biology does not vary sufficiently
to make this otherwise; the social recognition of physical kinship, the distinc-
199THE CONCEPT OF KINSHIP
tion and classifications made, the kinds of physical relationships utilised for
group-membership recruitment, and the social attributes predicated of the
recruited classified persons, may and do vary from society to society. But
Needham supposes (error 8, p. 98) that the existence of these social diversities,
rather pompously indicated by him by the ill-understood phrase
(borrowed from yesteryear's philosophic fad) that each "has to some extent
its own logic", constitutes a further objection to my argument. On the contrary,
the purpose of my scheme was to have an orderly manner of examining and
comparing those diversities, which do indeed form the subject matter of
anthropological studies of kinship. (Shapes vary; to express comparisons
among them, one needs a constant system of co-ordinates. The universal
facts of physical kinship were to be used to provide these co-ordinates in
terms of which the diversity was to expressed.)
Needham's complaints that a system of naming which remains the same
cannot capture the distinctive characteristics of varying kinship systems again
misses the point. (Error 9): a system of co-ordinates is not incapable of expressing
different areas or shapes just because the system of co-ordinates remains
the same and consists of points whose structure is similar. On the contrary,
just this is the condition of its doing the job of expressing the diversity of
patterns within it.
Needham misinterprets (error 10) the difficulty raised in connection with
the existence of two sexes. The difficulty arises because whereas, say, a system
of co-ordinates defines a homogeneous set of points, the "points" of my system
would have to contain individuals of two types (male and female), whose
differentiation is relevant to what is subsequently said of them. This difficulty
arises only at the initial level of specifying the points on the "logical space",
and not in connection with what is subsequently said about the points on it,
and thus has nothing whatever to do with whether there are "unilineal,
cognatic or bilineal" societies.
A similar misconception underlies error 11, expressed by Needham's
jibe about Victorian anthropologists, to the effect that no one since then has
argued for the primacy of matriliny. The point that it is harder to be ignorant
of the mother's identity than the father's, as indeed it is, (maternity is not in
doubt, as the deceived husband remarked), was, unambiguously, not a premise
for any kind of conclusion about what kind of kinship systems actually
occur in simple societies, but only for deciding what kind of logical framework
would be most useful (assuming we want one at all) for anyone attempting
to construct a formal scheme for the study of kinship systems.
Needham's next error (12) is in connection with what he calls the third
misapprehension. He credits me with a failure to realise that what is important
about members of kinship groups is what they do. Far from being unaware
of it, the scheme as expounded in the original article is intended to bring
this out, and is quite unambiguous in this respect. After kinship categories
had been picked out from the biological logical space, the social truths about
them would be conveyed by "sociological" predicates, synthetically applied.
200 ERNEST GELLNER
(Necessarily so: for what people do is not entailed in their physical kinship
position as such. The social, and not logical or biological, compulsion making
them do what people of the given kinship category habitually do in the
society, is a central part of the subject matter of social anthropology.) Needham
grotesquely misinterprets a phrase referring to social characteristics, which
he quotes from me: "only synthetic factual truths" (italics Needham's). The
force of the "only" was not "merely"-I do not consider these sociological
attributes to be unimportant, on the contrary; the force of "only" was
"exclusively". The whole point of the exercise was to separate clearly, and not
to disparage, what is logically distinct, i.e. the physical kinship position and
the social predicates (including social kin terms) attributed to it. There is no
excuse whatever for Needham's complete misinterpretation of my meaning.
He proceeds to speculate about what else I might suppose or mean in order
to make sense of my alleged disregard of the social characteristics: needless
to say, these speculations are as false as they are unnecessary, as the problem
they are invoked to solve (why I should underestimate social facts) does not
arise.
The curious thing about some of Needham's errors and conclusions is that
they are not even mutually consistent. His opening charge is that I fail to
distinguish biology and social function; where he does come across evidence
that in fact I do distinguish them, instead of seeing that this shows his earlier
charge to be wholly misguided, he proceeds instead to invent a further one:
he treats my distinction of the physical from the social as evidence for another
crime of his own unwarranted invention, my alleged disregard of the
importance of the social . . .
Needham is of course right in supposing that the social characteristics are
analytically contained in the very meaning of the autochtonous kinship terms;
and they are of course not so contained in the physical kin positions of the
individuals as such7 (as he realises). It is the connection between the former
(i.e. roles plus local kinship terms) on the one hand, and the physical kin
position on the other, which constitutes much of the study of "kinship
structure". The category used by the anthropologist to describe the kinship
structure of the society is itself, in effect, a report on the manner of this
(synthetic) connection, and hence both the physical and the social kinship
positions are (analytically) implicit in the anthropologist's classification.
Needham's confusion is complex: seeing (as in a glass very darkly) this
connectedness, which however he cannot state as he does not consciously
believe in the relevance of one of the terms related (i.e. the physical kin
position), he stresses instead the near-vacuous connection between native
kinship term and social role. But that connection, though indeed it obtains,
doesn't amount to much: naturally, a term means what it means. Naturally,
"brothers" are expected to be "fraternal": (and this tells us little unless we
know which people are expected to behave "fraternally", i.e. that they are
'Although, conversely, (to complicate matters), a view of the physical kinship position correct
or otherwise - is contained in the meaning of the autochtonous kinship term.
201THE CONCEPT OF KINSHIP
also "brothers" in some other-e.g. physical-sense, and which.) In any
society, (terms such as) "being-an-X" and "behaving X-wise" are to a
considerable extent defined in terms of each other, and we do not get much
further by connecting them. We do get somewhere when we discover how
(for example) those whom we call "brothers" and expect to be "fraternal",
are selected, relatively to a given individual, from the total population. This
gives us the connection between a role and its recruitment. The true and
important generalisation that in simple societies the identification of important
roles is "in terms of kinship" only escapes vacuity because we can specify
just how in each society "kinsmen" are picked out in terms of physical
kinship in general. But "brothers act fraternally" or "kinsmen are kinned
(kind)" says very little, if "brother" or < c
kinsmen" are emptied of their
physical meanings.
Misunderstanding as he does the nature of anthropological studies of kinship
(at any rate, when he attempts to give an account of them, though perhaps not
when he actually practises it), Needham has absurd and self-contradictory
expectations of what a systematic notation for this study would involve. He
seems to expect it to be just one further (native-type) kinship terminology,
doing the same kind of work as one of them, and yet serving to compare them
all, and to fuse rather than distinguish the social and the physical as they
do in their various ways, and yet at the same time to serve for comparing
their different ways of doing it . . .
At one stage he complains that the ideal language could not be used by any
society as its kinship terminology. Of course it could not, nor was it so intended.
When however, he comes across evidence that it was not intended for such
a role (the fact that the general structure of the "names" remains constant),
he treats this as one further complaint against it, rather than, as he should,
as a corrective of his first interpretation . . .
There is a further set of heterogeneous errors of Needham's, not all of them
specifically connected with his failure to understand what is involved in the
notion of a descent system or of what is involved in an ideal language.
There is, for instance, the issue of tangibility. Needham denies (error 13)
that kinship relationships are more tangible and stateable with accuracy than
most aspects of a simple society. On the contrary, for the very reasons which
led me to suggest the schema for a formalised language, they are stateable
with considerable precision and are tangible: the connotation of an autochtonous
kinship term can be ascertained just because we can plot its denotation
against the physical kinship map of the people in question. T o take another
of Needham's own examples, his account of the Kuki term tu: it "covers
members of three generations, two distinct descent lines, and both sexes".
This, plus of course further specifications concerning how these tu are to be
picked out, (and what they do and what is done unto them), gives us much of
the meaning of the term. Some shadowiness may enter with regard to the
rights and duties of tu, but even there, matters such as their inheritance
claims, the relative disposition of their habitat etc. can be ascertained by the
202 ERNEST GELLNER
observer. Compare this with the practical difficulties of giving a concrete and
accurate account of a people's values or beliefs! Things such as values, important
though they clearly are for the understanding of a society, are very
intangible and hard to assess. With regard to them, the situation which
Needham mistakenly supposes to hold with regard to kinship, the possible
absence of an equivalent "thing" in our language, does hold. (There is no
third, common, universal and given factor, such as the physical facts of
kinship, to mediate and thus to facilitate comparisons.) Kinship terms may
also not be directly translatable, but the very fact that they are essentially
related to physical kinship (in a way which varies from society to societybut
the idiosyncratic way in which it varies is fairly accessible to observation)
which in turn is built up from universally identical elements, is what makes
them explicable in a language other than that of the people themselves.
Needham proceeds to offer some speculations as to why I "appear to think"
(mistakenly, according to him) that kinship systems are tangible. These
unnecessary speculations of course represent neither what I think nor what
I might be supposed to think on evidence found in the relevant article.
The point can also be seen from the fact that anthropologists avoid, in
their accounts, convenient shorthand terms such as "cousin" and use instead
long-hand expressions such as "father's brother's son". Apart from the
elimination of ambiguity, the relevant feature of this explicit longhand is
that it fixes the relative position of the individuals on the physical kinship map
so that, subsequently, a society's habit of attributing kinship terms, roles,
etc. to the individuals so related can be explained. (And the reason why this
anthropologist's longhand is unambiguous is not merely that it breaks up the
relationship into its links, so to speak, but also because in this usage the term
"brother", "father" and so on are used in a purely physical sense, purged
of any of the fraternal, paternal etc. role-connotations which of course vary
from society to society, and which can then, like the native terms, be ascribed
to the individuals previously identified by the physical, de-socialised terminology).
For instance, with regard to our society, it may be said that "cousin"
covers father's brother's, and father's sister's offspring, and similarly for the
mother. In a schematic notation this information about the range of people
or "names" covered by "cousin" would be conveyed in a more condensed
and perhaps manipulable way by simply giving or specifying the range of
< c
names" of people to whom cousin-hood, relative to Ego, is attributed.
It is equally wrong to say (error 14), as Needham proceeds to do, that the
term tu could not be dealt with in my scheme. If a naming scheme such as I
proposed were designed, then of course, given the schema for Ego's name,
the necessary and sufficient conditions of being Ego's tu could be expressed
in terms of what types of names (in the devised language, not their own)
the tu would have to have. (In their own language, t t ~are tu . . . and behave
tu-wise . . .) This would merely convey, in a systematic way, the kind of
information which Needham in fact gives us in English.
There are also (error 15) Needham's remarks about functionalism. My
203THE CONCEPT OF KINSHIP
own observations, to the effect that functionalism should bring home the
connection, including systematic disparity, between physical and social
kinship, a connection which is in any case already entailed by the very notion
of kinship as used by anthropologists, were primarily an a fortiori argument
directed at those who, like Malinowski, attempt to isolate specific needs and
directly relate them to features of a society. I am not unaware of the fact
that this extreme kind of functionalism is not shared by other anthropologists,
having elsewhere commented on this fact in print. But in any case, the
argument applies, in a different form, even to the functionalism which is
very widespread amongst anthropologists, and so Needham's objection
remains invalid. Given the fact that mating always is regulated and needs
to be, and given the fact societies have kinship structures, it is virtually
certain that the latter institution plays some part (though perhaps neither
exclusively nor exhaustively) in serving the former need. Given this, there
will, in any given society, be some regular connection ("function") between
the two, and Needham's crucial separation of the two cannot obtain. In any
case, functionalism as a widely accepted method (and not as a doubtful and
widely doubted doctrine) does commit the anthropologist to finding out the
extent to which social kinship arrangements do assist in the regulation of
procreation, of the assigning of offspring to groups, and so on-and all this
can only be ascertained if one has some knowledge of the physical kinship
facts. The remark in my first paper claimed no more.
Needham also accuses my system of being cumbersome (error 16). Though
normally shy of openly claiming merit, I assert categorically that my system
is not cumbersome. My absolute confidence springs from the fact, which
inexplicably escaped Needham, that there was no such system in the article,
which explicitly restricted itself to trying to show that such a system should
be possible, and what general conditions it would have to satisfy. A system
which lacks existence cannot be endowed with cumbersomeness . . . In this
way, as in the others indicated, Needham completely and unnecessarily misunderstood
what was stated in the text.
Finally, there are Needham's expressed doubts about the usefulness of
rhe intellectual exercice which I was recommending and which he misunderstood;
in these doubts he is quite possibly justified, and I am sometimes
inclined to agree. But nothing he says actually supports his doubt or sheds
light on the issue. There are, I imagine, many criticisms which may justly
be made of my paper (not least that it was pompous): but none of those
actually made by Needham are amongst them, or have any validity.
There are also certain criticisms of Needham's which are wholly unanswerable-namely,
the unspecific ones. He darkly hints at "other erroneous
or questionable ~oints",and later at mistakes in another work of mine. (For
my own part, I do not wish to suggest that Needham committed any errors
other than the sixteen which I have explicitly indicated.)
In one society which I know, a rule is known which calls upon a man to pay
half the usual blood money if he fires at another and misses, but to pay the
204 ERNEST GELLNER
full "dia" if he takes aim and does not shoot. The reason given is that, as
one does not know where the bullet would gave gone, the amount of the
compensation must be determined by assuming the worst. Given the quality
of Needham's marksmanship, I should gladly forgive him the half-price
for his misfired actual shots, but feel some sympathy for the rule which
would penalise him for waving his gun about in this manner.
Needham's sixteen mistakes in a fairly brief article, virtually co-extensive
with it, are no mean achievement. The grammar and spelling of the article
however are in order. Needham's error when analysing the anthropological
notion of kinship is important, in as far as it springs from an anthropological
manner of speaking which is indeed liable to lead to this kind of error (though
I do not think others commit it when they think about it explicitly, as
Needham has). Needham's other errors are partly corollaries of it, partly
unwarranted attribution of positions to the article he attempts to criticise.
Failing to see that anthropological kinship terms are classifications of the
relation between social and physical facts, he interprets my remarks about
the universal physical relation as though they were about the social (thereby
of course turning them into absurdities). Of the three charges presented
in his summary, (b), the confusion of individuals and categories, and (c),
the confusion of specification and function, are simply misrepresentations,
whilst (a), concerning the relationship of biology and descent, does show
a genuine disagreement, but it is he who is in error.
There is however one further error, diffused throughout Needham's article
and implicit in his manner and obiter dicta and this is the tone of authority
which he assumes. His rather touchings self-assurance is in sad contrast
with the quality of his argument, and in direct proportion to the uncritical
and careless disregard of evidence with which he is ready to ascribe the most
unwarranted and preposterous views to others. From his article, two facts
emerge: he has a high regard for his own authoritativeness, and he is wrong.
Ernest Gellner.
NEEDHAM:"For my part, I am fairly familiar with the topic of an ideal or logical language..."
You have printed the following article:
The Concept of Kinship: With Special Reference to Mr. Needham's "Descent Systems and
Ideal Language"
Ernest Gellner
Philosophy of Science, Vol. 27, No. 2. (Apr., 1960), pp. 187-204.
Stable URL:
http://links.jstor.org/sici?sici=0031-8248%28196004%2927%3A2%3C187%3ATCOKWS%3E2.0.CO%3B2-Q
This article references the following linked citations. If you are trying to access articles from an
off-campus location, you may be required to first logon via your library web site to access JSTOR. Please
visit your library's website or contact a librarian to learn about options for remote access to JSTOR.
[Footnotes]
1
Descent Systems and Ideal Language
Rodney Needham
Philosophy of Science, Vol. 27, No. 1. (Jan., 1960), pp. 96-101.
Stable URL:
http://links.jstor.org/sici?sici=0031-8248%28196001%2927%3A1%3C96%3ADSAIL%3E2.0.CO%3B2-F
http://www.jstor.org
LINKED CITATIONS
- Page 1 of 1 NOTE:
The reference numbering from the original has been maintained in this citation list.