Articles

The Concept of a Society

DAVID COPP University of California, Davis

The concept of a society is central to several areas of philosophy, including

social and political philosophy, philosophy of social science and moral phi-
losophy. Yet little attention has been paid to the concept and we do not have
an adequate philosophical account of it.1 It is a concept that is difficult to
explain systematically, and it is subject to distortion or simple-minded at-
tacks whenever it plays a major role in a philosophical theory. Methodologi-
cal individualists have raised metaphysical or ontological concerns about
the idea that there are such things as societies, and other philosophers have
found the concept of a society to be unacceptably vague. For these reasons,
I believe it worthwhile to devote this essay to an attempt to explain the con-
cept.
A society is a kind of collective entity, as is a nation, a state, an organiza-
tion, and a family. Therefore, a satisfactory account of societies must mesh
with a general theory of the ontology of collectives (see Copp 1984). A
plausible account must not treat societies as sets, or as any other kind of ab-
stract entity, because it is logically impossible for such a thing to have the
kind of properties that a society can have. For instance, a society can oc-
cupy a territory, but an abstract entity, such as a set, necessarily has no spa-
tial location at all.2 Societies, therefore, are not abstract entities, and an ade-
quate ontology of societies must not treat them as such. However, I shall not
concern myself with ontological issues, for although individualists do have
worries about the idea that there are societies, they would have the same or
similar worries about all other kinds of collective entity. An account of the

Dialogue XXXI (1992), 183-212

184 Dialogue

concept of a society does presuppose an ontology of societies, but my task

in this paper does not require agreement on ontological issues. My task is to
explain the distinction between societies and other types of collective entity.
Some theorists who use the concept of a society seem not themselves to
be clear about it. Richard Brandt once proposed a form of rule utilitarianism
that involved defining the "Tightness of an act" in terms of "conformity
with the prescriptions of that moral code, the recognition of which as mor-
ally binding by people in the agent's society would maximize intrinsic
good" (Brandt 1963, p. 179). And in later work, Brandt retained the idea of
defining "morally wrong" with reference to "the agent's society" (1979,
p. 194). Yet he worried that "it is difficult to assign a definite meaning to
the phrase 'in the agent's society' " (1963, p. 179). He asked, "What is a
Columbia University professor who lives in the suburbs to count as his 'so-
ciety'? The faculty club? His suburb? New York City? The state of New
York? Any choice seems a bit arbitrary" (1963, p. 183).3 I believe on the
contrary that none of the items on Brandt's list is a society: a club is not a
society, nor is a city or a state. It is possible, of course, that Brandt did not
intend his use of the word "society" to be taken literally, or that his view
would be more plausible if the word were not taken literally. Yet, in order to
decide these issues, we need to know what societie's are.
Lack of attention to the concept of a society can hamper assessment of
any theory in which the concept is pivotal. Consider, for example, a putative
counter-example to John Rawls's theory of justice (Rawls 1971). David
Gauthier discusses a situation in which there are sixteen Robinson Crusoes,
each living alone on a different island (Gauthier 1986, p. 218-19).4 There is
a clever and energetic Crusoe on a well-supplied island and a lazy Crusoe
on a poorly supplied island. "In between are fourteen other Crusoes." They
do not co-operate with each other, but they are in communication by radio.
Furfrier, given the ocean currents, each could send provisions from his is-
land to certain of the other islands by building a small raft and setting it
adrift. The best-off Crusoe could send provisions by raft to the worst-off,
for example. Gauthier claims that Rawls's theory has the consequence that
the extra benefits of the better-off Crusoe are undeserved and that the over-
all situation is unjust (Gauthier 1986, p. 219-20). Gauthier thinks this conse-
quence is implausible and he may be correct. For my purposes, however,
the relevant point is that Rawls intends his principles of justice to govern the
distribution of the results of co-operation in society, not to govern every sit-
uation in which some have more and others have less (Rawls 1971,
p. 60-65). Yet Gauthier does not consider whether the sixteen Crusoes in the
example constitute a society. If they do not, as seems clear, the example is
of questionable relevance.5 Of course, there remain the questions of how
Rawls would justify restricting the application of his principles to societies,
and how he would justify his well-known claim that "the primary subject of
justice is the basic structure of society" (1971, p. 7; see also Rawls 1977).
 Society 185

Yet in order to be clear about these questions, we need to be able to distin-

guish between societies and other kinds of collective entity.
My project is partly one of conceptual analysis, for I aim to provide an
account that is sensitive to our intuitive understanding of a society as a
"comprehensive social community." 6 Concepts of society have a theoreti-
cal role to play, however (in sociology, for instance); and so our intuitive
concept may need to be refined in order to achieve explanatory or other the-
oretical gains. The concept of society that I will develop should not be
evaluated simply on the basis of its degree of success in capturing our intui-
tive understanding of what societies are. Its final test is its utility in philo-
sophical theory, chiefly moral theory. But I do want my account to be recog-
nizable as an account of the concept of society, and so I shall not ignore
either sociological theory or the pull of intuition.

1. What Is a Society?
Given that the concept of a society is central to Rawls's theory of justice,
we might hope for some guidance from him. Yet he says little about it. He
distinguishes society from private associations and other "less comprehen-
sive social groups" (1971, p. 8) and he assumes that a society is "a more or
less self-sufficient association of persons who in their relations to one an-
other recognize certain rules of conduct as binding and who for the most
part act in accordance with with them." He also supposes that these rules
"specify a system of cooperation" (1971, p. 4). This account is reasonable,
as far as it goes, and its main elements correspond to elements of my ac-
count. Yet it is not fully accurate, for it would imply that an isolated agricul-
tural commune is a society in the relevant sense, which it is not. Rawls
makes the idealizing assumptions that "the boundaries of [a society] are
given by the notion of a self-contained national community" (1971, p. 457),
and that a society is "a closed system isolated from other societies" (1971,
p. 8). But these are oversimplifications, for societies are not "closed" or
"self-contained," and they are not paired one to one with nations or with
states. For instance, there is a North American society that contains at least
two other societies as parts, and these societies are obviously interdepend-
ent.
Rawls does remark that the rules of a society that specify a "system of
cooperation" are "designed to advance the good of those taking part in it,"
and that "a society is a cooperative venture for mutual advantage" (1971,
p. 4). However, neither of these comments is literally true, for the rules of a
society are unlikely to have been designed, in any straightforward sense;
and, as Rawls points out, societies are marked by conflict as well as by
shared interests (1971, p. 4, 520). South African society is hardly a co-oper-
ative venture for the mutual advantage of all members. These comments
may describe how the parties to the Rawlsian contract should view society,
but they are not accurate as an account of what societies are in fact. Rawls
186 Dialogue

realizes this, for he says, "In justice as fairness society is interpreted as a

co-operative venture for mutual advantage" (1971, p. 84, my emphasis).
The sociological literature is more helpful than the philosophical, but it
also fails to provide a definitive account. Leon Mayhew summarized as fol-
lows the situation as he perceived it in 1968: "Analytical definitions usually
treat a society as a relatively independent or self-sufficient population
characterized by internal organization, territoriality, cultural distinctiveness
and sexual recruitment. Specific definitions vary considerably in regard to
which of these elements is emphasized" (Mayhew 1968, p. 577). As far as I
can tell, however, there has been even less agreement than Mayhew sug-
gests.
For instance, Aberle et al. argue that cultural distinctiveness is not re-
quired. They say, "Two or more societies may have the same culture, or
similar cultures." Hence, they contend, the Greek city states were separate
societies even though their cultures were quite similar. Moreover, they say,
"One society may be composed of groups with some marked differences in
culture" (1950, p. 102). American culture testifies to the truth of the latter
claim.
Talcott Parsons attempts to simplify by identifying a kind of "self-suffi-
ciency" as necessary and sufficient for a group to "be a society. He says, "A
society is a type of social system, in any universe of social systems, which
attains the highest level of self-sufficiency as a system in relation to its envi-
ronment" (1966, p. 9). And he explains that a self-sufficient group that is a
society must contain cultural materials and "role opportunities" which are
"sufficient for individuals to meet their fundamental personal exigencies at
all stages of the life cycle without going outside the society, and for the so-
ciety to meet its own exigencies" (1966, p. 17). Unfortunately, Parsons's
claim is too strong. It would rule out as societies all groups with a shortage
of m&n, Or with a shortage, of food, because some.menjbers would then have
to look outside in order to find a mate or to find food. However, the French
after World War I constituted a society despite a shortage of men, and the
many African tribes that face starvation from time to time may nevertheless
be societies.
In an earlier book, Parsons suggested a somewhat weaker requirement:
" A social system . . . which meets all the essential functional prerequisites
of long term persistence from within its own resources will be called a soci-
ety" (1951, p. 19). France had enough men to produce a new generation,
and the African groups have enough food to carry on, even though with
much starvation. Unfortunately, this new definition incorrectly implies that
there could not be a society that had to import all of its food. And it
wrongly implies that a man and a woman living alone on a subsistence farm
in the middle of Illinois would constitute a society, for they constitute a
minimal "social system," their farm supplies their needs, and they have the
 Society 187

material for the long-term persistence of a family. However, they are just a
family; they are not a society.
In a third work, Parsons suggested a third definition:7 a society is "a col-
lectivity . . . which is the primary bearer of a distinctive institutionalized
culture and which cannot be said to be a differentiated subsystem of a
higher-order collectivity oriented to most of the functional exigencies of a
social system" (1961, p. 44). He explained that a society's "distinctive
institutionalized culture" would be a system of norms governing co-ordina-
tion and co-operation and embedded in the society's culture or its institu-
tions. I agree that societies are characterized by such norms, but the second
part of his account is incorrect, for it seems to imply that a society cannot be
contained within a larger society. Yet we know that French society is con-
tained in European society.
Mayhew explains thejdiffjculties quite well: • -
the search for a universally valid definition of the nature and boundaries of a soci-
ety as a self-contained unit may obscure the complexity of social life.... Eco-
nomic, religious, political, educational and other types of activity come to cohere
into partially independent systems with unity, boundaries and mechanisms of their
own. These systems overlap; and when a relatively broad range of such systems
cohere around a common population, we may speak of a society. There is no rea-
son to suppose, however, that this society will be self-contained, that it will not
overlap with other societies, or that its boundaries will be uniform across its con-
stituent systems. (1968, p. 583)

2. The Concept of a Society

A satisfactory account must distinguish societies from states and from na-
tions (since some societies are neither states nor nations), while also ex-
plaining that the population of any state or nation will typically comprise a
society. Collectives such as families, business organizations and monaster-
ies are not societies (Parsons 1966, p. 17), and neither are typical North
American towns (Aberle et al. 1950, p. 102). On the other hand, ancient
Athens and Sparta, the United States, Europe, the early Mormon settlement
in Utah, and many of the small-scale social groups discussed by social an-
thropologists are societies, or, at least, their populations comprise societies
(Johnson 1960, p. 11-12; Benedict 1968; Aberle et al. 1950, p. 102). The
same was true of the Jews during the Diaspora.
A proper account should capture certain key characteristics of societies,
first of which is that they are multi-generational and extended through time.
Second, membership in a society is not a matter of choice, at least not ini-
tially, for one's society is inherited at the beginning of one's life, along with
one's family. Third, a society need not be organized, hierarchically or by
means of any institutional structure, and it need not have a complex social
system of roles and statuses. Of course, our society's culture does in fact de-
fine a system of statuses and roles, such as class and social positions —
188 Dialogue

worker, intellectual, debtor, professional, manager, and so on — but a system

of this sort is not necessary.8 What is required, and this is the fourth charac-
teristic, is that the members interact among themselves in activities directed
to securing the material necessities and priorities of life, or to securing
priorities identified in their culture; and that their interaction be governed by
a system of rules that is at least implicitly accepted in the group as defining
the norm for interaction, and perhaps incorporated into the group's culture.
Finally, a society provides the framework for its members' lives, embracing
the bulk of their friends and socially most important acquaintances. These
last two characteristics ensure that a society has a degree of social independ-
ence and isolation.
A satisfactory account must recognize that the concept of a society is
vague along at least two dimensions. First, the factors that enter into deter-
mining whether a given collective is a society can vary in degree, and the
concept of society does not establish a determinate threshold regarding
these factors. For instance, there must be some degree of instrumental inter-
action in a group that is a society, but we cannot say how much is required.
Second, the boundaries of a society are permeable. People can sometimes
leave and join other societies. Even if we can establish with confidence that
some group is a society, the concept of a society will not always determine
the exact membership. There are no formal membership criteria in a society,
and so we will often be unable to determine whether some person has, for
instance, successfully joined a society.
Let me compare the concept of a society to the concept of a beach. A
beach is an expanse of relatively fine-grained and evenly sized material
along a shoreline; but these factors vary in degree. How large must an ex-
panse of sand or gravel be, if it is a beach? And how fine-grained must the
material be? Also, the boundaries of a beach are vague. Are the sand dunes
a part of "the beach? The concept of a beach is vague,; and so is that of a so-
ciety.
I propose a two-stage account, according to which a society is a kind of
societal population.9 I shall simply state my proposal and then proceed to
explain the terminology and the details.
Dl: P is a societal population if and only if P is a group of people
characterized by a quasi-closed, multi-generational, temporally
extended network of social relationships, including relationships
of friendship, affection and kinship (as appropriate), and relation-
ships productive of new generations of members.
D2: P is a society if and only if
1. P is a societal population within which there is interaction in
behaviour directed to securing the material necessities or prior-
ities of life, or to securing priorities identified by a culture
characterizing P, where this interaction is governed by a set of
behavioural standards that is generally followed and shared by
 Society 189

the members of P, and used by them as a standard for criticiz-

ing behaviour, and
2. there is no P* such that
a. P is a proper part of P*, and
b. P* is a societal population that satisfies clause 1, and
c. the kinds of interaction in P, in virtue of which P satisfies
clause 1, whether directed to securing material priorities or
cultural priorities or both, also occur with roughly compara-
ble frequency within (P*-P) and between the members of
(P*-P) and the members of P, and
d. from the point of view of the members of P, there are no
alienating differences between (i) the most salient behav-
ioural standards governing such interaction in P, (ii) the
most salient-standards governing .such interaction between
the members of (P*-P) and the members of P, and (iii)the
most salient of the standards believed by the members of P
to govern such interaction in (P*-P), salience in each case
being assessed relative to the members of P.
Roughly, a society is a multi-generational, temporally extended population
of persons, embracing a relatively closed network of relationships of friend-
ship, affection, kinship, and co-operation in reproduction, and limited by the
widest boundary of a distinctive system of instrumental interaction. Simpli-
fying greatly, a society's borders are to be found where the borders of a so-
cial network coincide with the borders of a system of interaction that ap-
pears salient to the people embraced by the network.

3. Societal Populations
It is Mayhew's idea to define a society in terms of a societal population, and
I follow him in this, for I define a society as a kind of societal population.10
Mayhew's definition of a societal population has several shortcomings,
however. For him, a societal population consists of the inhabitants of a
"maximal" territorial area within which mating is common and residence is
relatively permanent, where membership is conferred by birth, and where it
is relatively rare that territorial boundaries are crossed for purposes of mat-
ing or to establish a new residence (1968, p. 585).
The first problem is that his account requires a societal population to con-
sist of the inhabitants of a maximal territorial area with certain characteris-
tics. Because of this, it does not permit one societal population to be a
proper part of another, in the way that French society is a part of the larger
European society. An adequate definition must permit societal populations
to be nested within one another. Second, Mayhew requires that a society be
territorial, and I assume that he has in mind continuous geographical territo-
ries, but Jewish society did not occupy a continuous territory during the
Diaspora, and a society may continue to exist even if its territory is con-
190 Dialogue

quered and occupied by foreigners. In both of these cases, the group's "ter-
ritory" would be scattered and broken up by areas resided in by outsiders.
Third, it need not be rare for members of a societal population to cross hith-
erto established societal boundaries in order to find new residences. An ex-
ample is American society during the eighteenth and early nineteenth centu-
ries. Finally, I shall argue that although a societal population must be multi-
generational and capable of perpetuating itself, it is not logically necessary
that new generations be produced by ordinary sexual mating. This is a con-
ceptual point with, perhaps, little practical significance, but I shall attempt
to account for it in my definition.
A society is a group that provides a framework for its members' more im-
portant interpersonal relationships and transactions. It is this idea that
Mayhew was attempting to capture by identifying societal populations in
terms of patterns of mating and residence. The point is simply that people
tend to find or to seek their friends and other objects of affection and associ-
ation, including their mates and their kin, from among fellow members of
their society. As a result, a society exhibits a network of social relationships
whereby members are bound to other members of the same generation, and
to members of earlier and later generations. My suggestion, therefore, is
that a societal population be defined as a group that is characterized by a
quasi-closed, multi-generational, temporally extended, network of social re-
lationships. In the simplest case, where a society is hermetically sealed, and
where exotic technologies of reproduction are merely philosophical fan-
tasies, a societal population would be closed under the relations ancestor of,
descendent of, friend of and mate of Every friend, mate, ancestor and de-
scendent of any member would also be a member, and this characteristic
would distinguish societal populations from families, corporations, discus-
sion.§roups, mobs, and so on. Unfortunately, societies are not closed in this
way, and there are cultural differences in the nature of social relationships.
Therefore, we need a more sophisticated understanding of the kind of social
network that unifies societal populations. It will be useful to begin by dis-
cussing certain out-of-the-way examples.
An exogamous society would illustrate both that a society can admit
members from outside, and that culture can affect the nature of social rela-
tionships.11 Just as there are taboos against incest, there could be a society
with a taboo against sex between people who are both born into the group.
We can imagine a pair of exogamous societies living in a kind of symbiosis
in which each takes mates for its boys (or girls) from the other, and sends
mates to the other group for that group's boys (or girls). In an exogamous
society, one supposes, if a child were born to a couple, both of whom were
born into the society, the child might not be recognized as a member of the
society at all. At a minimum, the child's social status, or that of its parents,
would be tarnished. An exogamous society would include both "hereditary
members," who inherited membership by birth to a pair of members, and
 Society 191

"non-hereditary members," who gained membership by way of the custom

that accords membership status to people recruited or acquired as sexual
partners from the outside. The norm, for an "hereditary member," would
be birth to a pair of members, one of whom was not born into the society.
Sociological accounts generally assume that societies maintain them-
selves by sexual reproduction; yet, while this may be true in fact, it does not
seem to be logically necessary. A society could exist which lacked the ca-
pacity for sexual reproduction. A large group of homosexuals could isolate
itself and develop the characteristics of a society, or a society could lose the
capacity for sexual reproduction — perhaps sterility is caused by a local ab-
erration in the earth's magnetic field.12 A society without the capacity for
sexual reproduction would obviously have to perpetuate itself in some other
way, such as by kidnapping or ideological recruitment or by means of an
exotic new reproductive technology such as cloning. All -the new members
of such a society would be non-hereditary members, in that they would not
inherit membership by birth to a pair of members.
Ideological recruitment into a society is not impossible; it can be one way
for a society to garner new members. However, a complete reliance on ideo-
logical recruitment would typically involve a great deal of interaction be-
tween members of the group and potential recruits in the outside world, and
it would typically mean that a large proportion of the group's members
would be new members with continuing ties of affection to people outside
the group. For these reasons, groups that recruit entirely by ideological
means — such as monasteries and sects with a rule of celibacy, such as the
Shakers — often do not have a sufficiently independent social network and
pattern of co-ordinative and co-operative relationships to count as so-
cieties.13 Sufficient independence for such a group to qualify as a society
could only be gained by the members' physical removal, or emotional alien-
ation from the outside. This may have been achieved by certain sects or
cults of recent history.
Similar factors would tend to undermine the social independence of exog-
amous groups which exchange mates. Families are exogamous, because of
incest taboos, and this is one reason why families and kinship lineages come
to be grouped together into larger societal populations. Ties of affection and
kinship are social relationships that constitute linkages between families. In
the same way, the exogamous clans of the Baringo and Nuer are unified into
larger societal populations by the social relationships that follow both logi-
cally and naturally from the linkages between exogamous groups that trade
in mates for their children.14 In order to realize my fantasy of a pair of sym-
biotic exogamous groups each of which qualifies as a society, the groups'
social independence would have to be established by a sufficient physical or
cultural alienation or separation to counteract the unifying links of marriage.
Any society will lose some members: a member may leave it to join an-
other society, or may be ostracized, or may be traded as a mate into another
192 Dialogue

society. And, almost any society will have some non-hereditary members
who have joined the society either voluntarily, or as a result of being traded
as a mate, or for some other reason. There must be some characteristics
which mark the fact that an erstwhile member has left the society, and some
which mark the fact that a former non-member has become a member. The
relevant characteristic often will be culturally defined: for instance, a former
member may have been ostracized, or a non-hereditary member may have
been selected in a culturally sanctioned way as a mate for a member. But on
other occasions, the characteristic may be external to the culture. A non-
hereditary member, like an hereditary member, will typically be disposed, to
some significant degree and in a non-temporary way, to choose friends and
companions from within the group, and to choose mates either from within
the group or from within groups sanctioned by the culture of the group. A
person's joining a society will typically be marked by his acquiring such a
disposition, and a person's leaving will typically be marked by his acquiring
a non-temporary disposition to choose friends or mates from outside. Mem-
bers of a society are connected with other members by the friendship, kin-
ship and sexual relationships characterizing the societal population: heredi-
tary members are linked to other members of the social network both in the
past and in the future; non-hereditary members are linked to other members
only or mainly at times after their times of joining; and erstwhile members
are linked to members in the past but to relatively few members in the fu-
ture.
The very early period of the North American settlements was one of tran-
sition in which aboriginal societies shared a single land mass with segments
of several European societies. American society developed gradually, and it
has always included a high proportion of new members, especially during
its formation and the periods of heavy immigration. The mark of a new
member was the disposition to seek any new friends and mates from within
the group. This explains the respect in which many of the Irish immigrants
to America effectively ceased to be members of Irish society. They identi-
fied with their adoptive communities in that they would have preferred to
find new friends and mates from this new, larger community. Other immi-
grants effectively remained members of Irish society, despite their physical
separation from the bulk of the society. They continued to identify them-
selves as Irish, and they lived in Irish communities where possible. These
cases are clear enough. However, the many immigrants who were am-
bivalent would be borderline cases who would neither have abandoned Irish
society nor have joined the adoptive society.
A person's disposition to seek any new friends and mates within a certain
group may reflect his preferences, but it may also reflect geographical, so-
cial, cultural, and political factors. "Identification" with a group is one
thing, but the availability or non-availability of people as potential friends
and mates is also a factor. For there may be political borders or social or
 Society 193

geographical barriers that prevent one from meeting anyone who is not a
member of a certain group, or barriers that make it very difficult to meet
anyone from outside. The result may be that one comes to be disposed to
seek any new friends or mates from inside the group. So, one can acquire
this disposition just as a result of coming to live in the territory occupied by
a societal population, if the move makes it difficult to meet anyone from
outside. On the other hand, cultural and social barriers can make it very dif-
ficult for one to meet anyone who is a member of a group, even if one lives
in the group's territory. One can be ostracized from a society, and if the os-
tracism is effective, one may lose membership, or fail to gain it, despite a
desire to belong.
The idea, then, is that a societal population is characterized by a quasi-
closed, multi-generational, temporally extended network of social relation-
ships, including, as appropriate, relationships of friendship, affection and
kinship, as well as relationships concerned with the production of new gen-
erations of members. I need to explain this idea systematically, and in suffi-
ciently general terms, that societies of all the various types I have been dis-
cussing could qualify as societal populations.
First, the network of social relationships must be multi-generational, in
the sense that it contains members who are or could be ancestors or de-
scendents of one another, and it must have a substantial extension through
time, and exist over several generations. Of course, in a non-sexually repro-
ducing society, there may be no one who is a biological ancestor or de-
scendent of any other member, but in most cases, a society's generations are
biologically related. And, in every case, the social network includes rela-
tionships "productive of new generations of members." For the most part,
these will be sexual relationships, of course, but there remains the logical
possibility of a non-sexually reproducing society which includes a group of
recruiters, or kidnappers, or technologists, actively engaged in the business
of securing new members.
Second, a network of social relationships, in the sense I have in mind,
would typically include a network of biological kinship relationships and of
natural affective relationships, such as relationships of mate-of and off-
spring-of, but it need not include a culturally defined kinship system. By a
kinship system, I mean a network of relationships governed by a shared set
of behavioural standards regulating such things as responsibility for parent-
ing. In typical cases, the culture of a society defines expectations governing
relationships such as the parenting relation, and these expectations can be
different in different cultures. Where a kinship system exists, there is a cul-
ture, and the societal population is typically a society, satisfying the defini-
tion of society given above as D2. But societal populations are not necessar-
ily organized by a culture, so they are not necessarily characterized by sys-
tems of kinship relationships.
194 Dialogue

Third, there are temporal limits to the persistence of societal populations.

For example, English society today may be a different society from medi-
eval English society, and not merely a later part of the very same society,
even though there were medieval ancestors of present-day English people.15
In Section 7, I shall discuss problems of identity over time in detail, but I
need to make two points here.16 There are two kinds of case in which a
crude temporal limit can be identified. First, as in the English example,
there may be sufficient change in the interactive standards characterizing a
societal population at different times that we would not regard temporally
distant parts of the population as parts of the same society. I shall soon ex-
plain this idea. Second, there may be discontinuities in networks of social
relationships, and they may indicate the temporal limits of societal popula-
tions. For example, the emergence of an American society was marked by
discontinuities between the previously existing European social networks
and the later American social network. The European networks had previ-
ously been relatively independent, but parts of them became linked with one
another by social relations developed in North America, and the resulting
intersection of these networks developed a significant degree of independ-
ence from later stages of the European networks. The result was the emer-
gence of a North American societal population.
For the rest, the basic idea is that a multi-generational group which is a
societal population is closed by and large under the key social relationships.
If we were to trace relationships of friendship, affection, companionship
and mating, beginning with any member, we would find that we remained
among the group's members, on the whole. By and large, the friends and
mates of a member belong, as do those friends' and mates' friends and
mates, their friends and mates, and so on. The exceptions would be rare. We
can also trace relationships of kinship, if they are defined by the culture of
the group. By and large, within the temporal limits m/yitioned above, and at
least in established endogamous societies, the ancestors, descendents and
other kin of a member are also members, as are the ancestors, descendents
and kin of a member's friends and mates, the ancestors, descendents and kin
of those people, and so on. Exceptions occur when our attempt to trace kin-
ship within a group leads us to a non-hereditary member, or to someone
who was a member of the group, but is no longer. The exceptions will be
frequent in exogamous or non-sexually reproducing societies, or in forma-
tive societies, or, more generally, in societies that contain a large proportion
of non-hereditary members, or where relatively large numbers of former
members are now members of other societies. However, as we saw, non-
hereditary members must have characteristics that set them apart from non-
members, and people who are no longer members must be distinguished in
some way from members.
If we could ignore the exceptions and the temporal limits, we could say
that a societal population is closed under kinship relations and under rela-
 Society 195

tionships of friendship affection and mating. However, we can only say that
societal populations are "quasi-closed" under these relations. The social re-
lations link members of a societal population to other people, mainly to
other members.
I do not believe that the vagueness in this account can be avoided. There
cannot be a non-arbitrary and precise conceptual boundary in a situation
where the underlying realities are matters of degree, and the realities under-
lying the coalescence of people into societal populations are matters of de-
gree: the degree to which members of a group are disposed to choose
friends and companions from within the group, for instance, and the propor-
tion in which the friends of a member belong to the group, and the friends
and mates of those friends, their friends and mates, etc. The greater the de-
gree in which these things are so, or the greater the relevant proportions, in
the case of a specified, group, the greater the degree to. which the group
should qualify as a societal population.
It is easy to see that French society constitutes a societal population by
my account, and that European society also constitutes a societal popula-
tion. Societal populations can be nested, one within another. Jewish society
during the Diaspora was also a societal population: it was characterized by a
network of friendships and families that knit the group together, and it was
relatively rare for a member to look outside for a mate or a friend, or for a
member's parents not to be members themselves. The members of Jewish
society were often also members of the national societies in which they
lived, such as French society. So societal populations can overlap, having
some but not all members in common.17
I hope that this is adequate to explain the notion of a societal population.
I shall soon consider some examples that may provide additional help.

4. Societies and Societal Populations

Not every societal population is a society, for, according to my account, a
society has two additional characteristics, which I shall now explain, begin-
ning with the first clause of D2: the members of a society interact among
themselves in instrumental activities and share a set of behavioural stand-
ards that govern their interaction.
A societal population can be socially fragmented to the degree that it is
not a society. For instance, the small family groups living on a tropical is-
land might be so widely scattered, and so well supplied with food, that the
members of different groups see no need of co-operation, and hardly ever
meet except to fight for sexual partners. The population as a whole would
not be a society even though it would be a societal population. It would not
become a society unless a pattern of interaction developed in the entire pop-
ulation in a way that could give rise to mutual co-ordinating expectations
about their interaction. For example, a set of trading rules might come to be
accepted on the island, with the result that the different families could meet
196 Dialogue

at accepted times and places to trade their products, and could reasonably
expect to be able to continue to do so. Or droughts might become common,
and, as a result, the families might come to follow shared rules about water
conservation. These sets of shared rules would tend to become part of a cul-
ture, and they would sustain patterns of co-operation that would contribute
to tying the population together, and to making it more akin to a society. Of
course, the population could also be tied together by a pattern of exploita-
tion, for a system of slavery could develop, or an exploitative kinship sys-
tem. The examples illustrate why I suggest that a society is a group whose
members interact in a way that is governed by a set of behavioural standards
that they generally follow and share.
The notion of interaction being "governed" by a set of behavioural
standards must be left vague. It is not required that people have formulated
a set of principles that they intentionally conform to in their behaviour, nor
is it required that conformity be voluntary. It is not required that these
standards have any justification of a sort that would interest the moral scep-
tic. However, (1) there must be certain regular patterns of interaction in the
group, such as patterns of co-operation and co-ordination; (2) behaviour
conforming to these patterns must be commonly expected, and commonly
known to be expected; (3) behaviour that does not conform must be ex-
pected to be criticized or sanctioned as a result of not conforming, and
known to be likely to attract such criticism; (4) it must be true and common
knowledge in the group that (a) the regularities are widely conformed to in
the group, (b) almost everyone expects widespread conformity, and (c) al-
most everyone, at least among the expected beneficiaries of the patterns of
interaction, desires that everyone else in the group conform, given that al-
most everyone conforms.18 Of course, in the exploitative cases, such as
slave societies or caste societies, those who are worst off may not share this
desire that everyone else conform to the established patterns of interaction,
but the desire would still be prevalent, one would expect, among the benefi-
ciaries of the exploitation. In typical cases, it would be a consequence of the
culture of the group that all of this is the case. The widespread conformity,
and the knowledge of the widespread conformity and of the fact that it is ex-
pected and desired, would encourage and enable co-ordination of future
plans for interaction.
It is important what kinds of behaviour are governed by the normative
standards of a group. In particular, I think, in any group that counts as a so-
ciety, the cultural standards or characteristic norms must govern interaction
among the members which is directed to securing the material necessities
and priorities of life, such as food and shelter, or to securing priorities speci-
fied in the culture, such as religious salvation; there must be rule-governed
economic relationships or kinship relationships or religious rites, etc. We
may refer to the kind of behaviour governed by such standards as "instru-
mental." Suppose, for example, that a societal population shares a Ian-
 Society 197

guage. This means that verbal interaction among its members is rule-
governed in the right kind of way. But it does not follow that the group is a
society, for it does not follow that instrumental interaction among its mem-
bers is rule-governed; it does not follow that there are rule-governed eco-
nomic relationships or kinship relationships or religious rites. Of course, the
members of a group would not be likely to share a language if they did not
interact in a variety of instrumentally relevant ways, and a language proba-
bly would not develop in a group if its members did not interact in a variety
of relevant ways. But the mere sharing of a language is not sufficient to
transform a societal population into a society. The following three examples
illustrate this point about the relevant kind of rule-governed interaction, as
well as a point about the degree of relevant interaction.
The degree to which there is rule-governed interaction among the mem-
bers of a societal population is as important as the kind-oi interaction, when
determining whether the group merits being regarded as a society, "or to
what degree it merits this. The Ik is a tribal group in Uganda which was de-
scribed by the anthropologist Colin Turnbull in his controversial book, The
Mountain People.19 Turnbull denied that the Ik co-operated among them-
selves and that they had any social system or social organization. He said
the "disturbing thing [about the Ik] was the loss of sociality, the reduction
of systematized human relationships to the bare minimum where the exped-
iency of the moment was the only system there was" (in Wilson et al. 1975,
p. 356). On this basis, Turnbull seemed to want to deny that the Ik consti-
tuted a society, and, in my view, he would have been correct in this, if his
description of the Ik as lacking relevant rule-governed, interactive social re-
lationships had been correct. The issue of Ik sociality was very controversial
among anthropologists in the years after Turnbull published his book. They
would have agreed that the Ik constituted a societal population, for the Ik
were united by social relationships such as kinship, friendship, affection, an
emotional attachment to the group, and a desire to live together;20 and, of
course, the issue was not the presence or absence of a language. Rather, the
disagreement concerned whether the rule-governed relationships implicit in
this degree of social unity, especially in kinship arrangements and friend-
ships, were sufficient and of the right kind for the Ik to constitute a society.
Turnbull denied this, in the face of the type of individualism he observed,
but Grant McCall claimed, for instance, that the Ik had a "cooperative
group identity" sufficient to make the group " a cohesive (though belea-
guered) society" (in Wilson et al. 1975, p. 348). He pointed out evidence
that the Ik were more co-operative among themselves before the onset of
the extreme social dislocation that caused the phenomena Turnbull ob-
served. Also the Ik seemed to regard their patterns of co-operative
behaviour as important to preserve, and they evidently engaged in group so-
cial activities (see Wilson et al. 1975, p. 348). In any event, the example of
 198 Dialogue

the Ik shows the importance of rule-governed interaction as a factor in our

conception of a society.
Aberle et al. suggested that a group of people in "the [Hobbesian] war of
all against all" would not be a society (1950, p. 103-4), and I think their
point was the same as the point illustrated by the case of the Ik. The
Hobbesian war of all against all is an imaginary situation in which there is a
scarcity of needed and desired things, in which people are of roughly equal
strength and power, in which there is no organized government, and in
which, as Aberle et al. put it, people "pursue their ends by means selected
only on the basis of instrumental efficiency" (1950, p. 103). There are no
effective cultural or moral restraints on a person's choice of the means he
will use to achieve his ends. It is not that there can be no language in this
kind of situation, and it is not that there can be no instrumental interaction,
but instrumental behaviour is not rule-governed in the relevant way. Co-
operative combinations of people are "subject to immediate dissolution if,
for example, exploitation or annihilation becomes more advantageous for
any one member." The problem is that no behavioural standards governing
instrumental interaction are subscribed to by the people, and none are en-
forced on them, and none are sufficiently supported by the culture.
Aberle et al. also suggested that a society would be destroyed by the dis-
persion or "apathy" of its members (1950, p. 103-4). The relevant issue
here is again the loss or absence of rule-governed interaction in instrumental
behaviour. The mere physical separation of people need not end a society,
for there have been societies of hunters and gatherers that scattered for long
periods of time over large territories. I believe that the dispersion of a group
is important only to the degree to which it interferes with the members'
•ability to sustain rule-governed instrumental interaction, by interfering with
their ability to sustain a culture, to communicate, to find mates within the
grouper otherwise to co-opefate or interact among themselves. Similarly, I
think that people living peacefully and in proximity are not a society if there
is no rule-governed, instrumental interaction among them on a regular basis.
Indolent people living in a condition of abundance on a tropical island
might never become a society. Indeed some anthropologists claim, as
Aberle et al. point out, that whole societies in Melanesia have withered
away from ennui. The critical factor in these cases would be the loss of suf-
ficient rule-governed interaction in instrumental behaviour.
The second characteristic that distinguishes a society from a mere societal
population is captured by the second clause of D2. Intuitively, the idea is
that instrumental interaction in a society is governed by standards of behav-
iour that seem relevantly different to its members from the standards gov-
erning interaction of the same kinds in larger societal populations of which
it is a part. This characteristic distinguishes societies from societal popula-
tions that are not societies in their own right, being merely sub-societal parts
of larger societies. Aberle et al. claimed that the "absorption" of one soci-
 Society 199

ety into another would cause it to lose "its identity" as a society (1950,
p. 103-4). The problem with this idea is to explain the notion of absorption,
which I am attempting to explain by means of the second clause of D2.
The population of a typical town in France, for example, does not consti-
tute a society in its own right: the most salient standards governing interac-
tion among the town's population are the same as the most salient standards
governing interaction in French society as a whole, and they are also the
same as those governing interaction between the rest of the French and the
townspeople, viz., most likely, the French legal system. However, from the
point of view of the French, the most salient standards governing interaction
in France are French rather than European, and the most salient standards
that govern interaction in the rest of Europe are not the same as those that
govern interaction in France. So, the French population is a society in its
own right, even though it 'is a part of the larger European society, but the
population of the town is not a society because of the way in which it is a
part of the larger French society.
According to clause 2 of D2, a societal population P i s a society only if it
is not a proper part of a larger societal population P* that has both of the
following characteristics. First, the relevant kinds of instrumental interac-
tion that occur in P, whether directed to securing material priorities or cul-
tural priorities or both, also occur with roughly comparable frequency
within (P*-P) and between the members of (P*-P) and the members of P.
For example, a Mormon community within a larger secular society would
count as a distinct society, assuming it had the other characteristics of a so-
ciety. More generally, a religious community that has all the other charac-
teristics of a society would qualify as a society in its own right if any larger
group that included it were entirely secular, for it would not then be part of a
larger group within which religious interaction took place with sufficient
frequency. Second, from the point of view of the members of P, there are no
alienating differences between (i) the most salient behavioural standards
governing relevant instrumental interaction in P, (ii) the most salient stand-
ards governing such interaction between the members of (P*-P) and the
members of P, and (iii) the most salient of the standards believed by the
members of P to govern such interaction in (P*-P). Salience in each case is
assessed relative to the members of P. Hence, for example, a Mormon com-
munity within a larger Mormon society presumably would not count as a
distinct society. More generally, if one religious community is part of a
larger religious community, then it counts as a society in its own right only
if the more prominent behavioural standards with currency in the larger
group seem to the members of the smaller group to be alien or different
from their own standards.
My proposal makes use of generalizations about kinds of interaction and
about "alienating differences" between "salient" behavioural standards. I
do not see any way to avoid the vagueness. In applying my proposal, one
200 Dialogue

should consider only fairly abstract kinds of interaction, such as worship or

trading in general, rather than, for example, trading in a specific commodity.
The relevant standards of behaviour are those that by and large govern in-
teraction in the relevant groups (or are believed by the members of P to
govern interaction by and large). By "alienating differences" between the
salient standards of P and those of the other relevant groups, I mean differ-
ences between the standards, or other factors, that would lead the members
of P to view the more salient standards of the other groups as not "their
own" —those are not "our" standards, we might say, if we were members
of P. Finally, "salient" standards are standards that seem especially promi-
nent or conspicuous to the members of P. This is admittedly vague, but I
think that legal rules are typically the most salient rules governing social in-
teraction in a group. If so, this proposal would imply that a state with a uni-
fied legal system typically is one society, and that two states with distinct le-
gal systems typically are each a single society, even if they have similar cul-
tures, and even if there is much interaction across their borders.
Consider an illustration. A conquered society can remain distinct from
the empire that conquered it, despite the ruthless imposition of an alien legal
system, for its members may continue to be motivated by an indigenous
system of behavioural standards, which may continue to be more salient
from their point of view than the imposed laws of the empire, even if none
of them dares to violate the laws. They likely would continue to regard the
indigenous system as their own, and they likely would tend to or want to
conform to it partly in order to conform, and they likely would regard it as
the more important to preserve. If so, it would be the more salient system
from their point of view. Even if the empire imposed its laws in a way that
«made them more prominent than the indigenous standards, the members of
the conquered society likely would continue to view the laws of the empire
as "afi&n,'" as not""-their own." According to my- account, then, the con-
quered society would continue to be a society in its own right as long as the
more salient standards in the rest of the empire continued to seem alien and
different from the most salient standards in the conquered society, or at least
continued to seem alien.

5. Implications and Examples

Collectives such as families, business organizations, monasteries and the
United Nations are not societal populations, and so they are not societies.
For instance, the members of a monastery would retain many ties of friend-
ship and affection with non-members, and they would typically not be dis-
posed to select friends specifically from within the monastery, as opposed to
selection from the rest of the church or outside society. However, the early
Mormon community in Utah was a society, if, as seems likely, it was
characterized by a quasi-closed network of social relationships of the re-
quired sort, and if, as seems likely, its religious code of behaviour was the
 Society 201

salient code from the point of view of its members. This code differed from
the standards of behaviour that were current and most prominent within the
larger American society of which it was a part. Similarly, the Jews during
the Diaspora constituted a society. And many of the small-scale groups dis-
cussed by social anthropologists would be deemed societies by my defini-
tions, as would the populations of ancient Athens and Sparta and of the
United States. Finally, the population of Quebec would constitute a distinct
society within the larger Canadian society, according to my account, assum-
ing that the culturally salient behavioural standards within Quebec are dif-
ferent from the relevant standards that are salient within the rest of Canada.
All of these implications seem appropriate.
Societies exist where the boundaries of quasi-closed, temporally extended
networks of relevant social relationships coincide with the boundaries of sa-
lient rule-governed patterns xff instrumental behayiour. These boundajie^s do
not always coincide. Some societal populations are not themselves societies
because of the way in which they share a pattern of interaction with a larger
group; the French town illustrates this phenomenon. Also, some groups that
have distinctive patterns of rule-governed interactive behaviour are not so-
cietal populations. This phenomenon is illustrated by tribal groups within
the Nuer and the Baringo in Africa.21 Among the Nuer, for instance, the
tribe is a quasi-legal and political unit. Evans-Pritchard describes it as the
largest group that combines for offence and defence, and whose members
think that disputes among members should be settled by arbitration rather
than fighting (1940b, p. 278-79). A tribe shares a system of co-operation
which does not govern interaction between members of the tribe and mem-
bers of different Nuer tribes. However, a tribe is not a societal population,
for the different tribes are unified by a lineage system in which the same
clans and kinship lineages exist in different tribes, and there is marriage and
trade across tribal lines. Evans-Pritchard argues that the Nuer as a whole
constitute a society: the "whole Nuer" form a single community with a
common culture and "feeling of exclusiveness" (1940b, p. 279). The situa-
tion seems to be similar among the Baringo, where ties of kinship are
spread across tribal borders, and there is intermarriage and migration be-
tween tribes (Hodder 1982, p. 16-35). Hodder refers to the Baringo tribes as
"ethnic groups" (1982, p. 35), and to the entire Baringo as a "society"
(1982, p. 85).
A group may become a society as, to a greater degree, it comes to be
characterized by a network of social relationships, as the members tend
more and more to look inside it for friends or mates, as it comes to be
governed by more salient and distinctive standards of instrumental interac-
tion, and so on. Societyhood is a matter of degree. It is associated with a set
of "society-making properties" such that a group is a society to a greater
degree as it possesses more of the relevant properties to a greater degree.
Elsewhere, I have recommended such a view of nationhood (Copp 1979),
 202 Dialogue

but we do not need to abandon the intuitive and convenient practice of

speaking as if certain groups simply are societies. We should think of soci-
etyhood as being associated with a set of society-making properties, and a
threshold for each, such that groups which possess the relevant properties to
greater degrees than those established by the corresponding thresholds
count as societies, while those which do not, do not count as societies. Un-
fortunately, the thresholds cannot be specified in any very useful way.
I said earlier that societies may be nested. For instance, the French and
the Germans constitute societies, and they are parts of a larger European so-
ciety that embraces them both. The legal systems of France and Germany
have greater salience from the point of view of their respective citizens than
do the common standards of the European Community. The standards of in-
teraction that are most salient from the point of view of the French are dif-
ferent from the standards that are most salient from the point of view of the
citizens of any other European country. Hence, the French and the Germans
constitute distinct societies. Nevertheless, the population of Western Europe
also constitutes a society, for it is an interacting societal population gov-
erned by a set of behavioural standards. We see that definition D2 permits
the nesting of societies.
The definition also permits the partial overlap of societies. The Jews dur-
ing the Diaspora constituted a society and so did the population of France.
The two overlapped, but were nevertheless distinct societies.
There may be cases where the definitions have counter-intuitive implica-
tions, and there is no doubt that they are vague. Nevertheless, I doubt that
the vagueness can be eliminated.

* 6. Global Society and Apartheid Society

Soc«4ie& are not. generally characterized by a sense of unity and solidarity,
or an "identification" with a common history ot tradition, or an absence of
social conflict. Otherwise, presumably there could not be a society of all
people in an apartheid country, or in a country that treated a portion of its
population as slaves or simply as instruments of industrial production,
depriving them of the rights and civil liberties of the rest of its people.
There would not be a South African society that embraced all of the people.
Also, there could not be a global society, insofar as the people of the world
are divided into factions by economic, cultural, political and social conflicts
and differences.
Neither psychological unity nor an absence of conflict is characteristic of
societies. Societies are not ideal communities; they are not necessarily
united by a common aim, interest, or moral ideal. Societies are "compre-
hensive social communities," 22 and they are typically rife with social, eco-
nomic, and political conflict that undermines psychological unity. Accord-
ing to my account, societies are unified, not by means of a sense of com-
monality or an absence of conflict, but by a unity forged from a network of
 Society 203

social relationships, including instrumental interaction governed by salient

standards.
A society with an apartheid system seeks to keep certain of its people
apart in a legal economic and social ghetto where they are deprived by law
of the human rights, civil liberties, and economic opportunities enjoyed by
the rest of the population. However, the society is the comprehensive social
community, and it includes the exploited class as well as the advantaged
class. The Whites of South Africa constitute "the society" in one sense,
perhaps, for they are the ruling class, and the socially and economically
most advantaged class in the country. Yet they are only a portion of the
comprehensive social community that is South African society in the sense
of the term we have been concerned to understand. Each racial community
in South Africa is perhaps a societal population, for as a result of the rigid
enforcement of the laws governing inter-racial relationships, each corrwgju-
nity is likely characterized by a separate network of social relationships. In
addition, some tribes and communities may be societies in their own right.
Yet it is clear that the most salient set of behavioural standards governing
instrumental interaction throughout the country as a whole is the set of legal
and conventional restrictions that have institutionalized apartheid. The sys-
tem defines the privileges of the White community as it defines the disad-
vantages of the Black community. This means, in effect, that as matters now
stand, the South African communities are part of the larger society of South
Africa. Despite the conflicts and inequities of that society, it is one compre-
hensive social community.
Humanity constitutes a societal population. Whether this societal popula-
tion is a society depends on whether there is within it instrumental interac-
tion governed by standards of behaviour; it depends on whether there are
regularities of interaction that are known to be widely conformed to, and
that are widely expected and desired to be conformed to in the ways
sketched above. Of course, there is instrumental interaction within each of
the societies that coexist in the global population. But, we are concerned
with the nature of the interaction among these societies and among the
members of different societies. If these societies were isolated and closed,
or if all of the interaction between societies were unrestricted, warlike
behaviour, then there would be no global society because there would be in-
sufficient intersocietal, rule-governed instrumental interaction. The societies
of the globe would be like the individuals on the imaginary tropical island
we discussed earlier, or like Gauthier's sixteen Crusoes, and there would be
no society embracing them all. The global societal population is not a soci-
ety unless a rule-governed pattern of interaction exists.
In fact, there are patterns of interaction among the societies and people of
the globe, patterns that exhibit widely accepted standards of interaction, and
that exhibit regularities of behaviour that are widely known, expected and
desired to exist. There is an international economic system. There are pat-
204 Dialogue

terns, and expectations regarding the patterns, of trading relationships.

There is an intersocietal system for communication. The United Nations is a
forum for intersocietal co-operation, though there is less co-operation under
its sponsorship than we would like. International law, weak though it be, is a
set of intersocietal co-operative standards. Tourism and business bring
people to corners of the globe where they and those they visit can witness
not only to the diversities, but also to the possibilities of communication and
co-operation. Educational and cultural exchanges and visits bring a few
people each year from the less-well-off societies to the better-off societies
and vice versa. There is an intersocietal scientific and intellectual commu-
nity. There is sufficient co-operation and instrumental interaction among the
peoples and societies of the world for it to be true that there is a global soci-
ety embracing all of the people of the earth.
Given the existence of global conflict and inequity and the existence of
conflict and inequity within apartheid societies, and indeed within almost all
societies, it is small comfort to realize that all of the people of the world are
embraced by a single global society, and that the people of any racially, eco-
nomically and socially segregated society are nevertheless members of a
single, comprehensive social community. It is small comfort, but it is worth
bearing in mind.

7. Identity and Change

Societies change as time passes: members die; new members are born or
join from outside; the behavioural standards accepted and followed by the
members at one time come to appear outmoded and are modified. Societies
survive changes of these kinds, but some changes would destroy a society,
and some would result in the replacement of one society by a distinct soci-
ety.~€oi*sequently, it can seem pressing to ask how much, and what sorts of
change can be tolerated, and what characteristics must be preserved over
time, if a society is to survive change.
For instance, one might think that societies are essentially characterized
by their moral standards. And one might conclude on this basis that any
change in the moral code that has currency in a society would result in the
replacement of one society by another. This would be a very extreme and
implausible view.
Lord Patrick Devlin expressed a similar, though less extreme view in a
famous lecture delivered in 1959:
Societies disintegrate from within more frequently than they are broken up by ex-
ternal pressures. There is disintegration when no common morality is observed and
history shows that the loosening of moral bonds is often the first stage of disinte-
gration, so that society is justified in taking the same steps to preserve its moral
code as it does to preserve its government and other essential institutions. (1959,
p. 13)
 Society 205

Devlin does not think that any change in a society's moral code would mean
that one society "has disappeared and another has taken its place."
Whether this would have happened, he says, "would depend on the extent
of the change" (1959, p. 13, n. 1). Nevertheless, he seems to think that the
wholesale replacement of one moral code by quite a different one would en-
tail the replacement of one society by another.
However, it is clear that the members of a society could, in principle,
choose to adopt a new moral code. I see no reason to suppose that this
would have to spell the replacement of their society by a new society. The
society that would exist after the change would have the same membership
as the society before the change. There would be historical continuity in the
membership of the society before and after the change, and we could
presumably explain the characteristics of the new morality in terms of the
characteristics that the society had before the change. The-society before^the
change would or could be identical to the society after the change.
Harry Johnson suggests that some continuity of social structure is neces-
sary to the continuing existence of a society (1960, p. 68-69). To be sure, he
says that no social "mechanism" is "indispensable" to a society. Any
given mechanism could, in principle, disappear and be replaced with an-
other mechanism that had the same, or a very similar, function. That is,
there are "functional alternatives" to any given mechanism. But there are
limits to this, in Johnson's view, for "every social structure imposes some
limitations on the structural innovations that would be compatible with it"
(1960, p. 69). He says,
It would, for instance, be absurd to speak of the effects of introducing polygamy
into a model of the English marriage system unless one included among the "ef-
fects" all the changes in attitude and behaviour that must occur before the intro-
duction of polygamy would be possible. And these changes would produce a new
society requiring a new model. (Johnson 1960, p. 69, quoting Marshall 1956, p. 64,
my emphasis)
If Johnson means by "a new society" a "numerically distinct society,"
then I disagree with him.
It is true that there are empirical limitations on the substitution of "func-
tional alternatives" for existing "social mechanisms." However, it is im-
portant to distinguish clearly between, on the one hand, changes in the prop-
erties of a society that are drastic enough to require "a new model" of the
society, or a new account of how the society functions; and, on the other
hand, changes that destroy a society and replace it with another society.
Changes of the latter sort produce a society that is not merely different in its
properties, but is a different society, in the sense of a distinct or a new soci-
ety. The introduction of polygamy into an erstwhile monogamous society is
logically possible, and is compatible in principle with the society's survival,
but it would obviously be a non-trivial change. A qualitatively different so-
ciety would result, but not necessarily a distinct society. There is no reason
206 Dialogue

to believe that the introduction of polygamy would entail the destruction of

one society and its replacement with another.
Earlier, I said that modern English society and medieval English society
may be stages of distinct societies, and not merely different stages of one
and the same society. Following Paul Teller, I suggested that the explana-
tion for this is the degree to which the interactive standards that character-
ized the medieval society are different from those that characterize the mod-
ern society. It may seem crucial, as Teller proposed to me, that an English
person of today could fit into 1920 English society without major "retool-
ing" of his behavioural standards. But "if transported to 1300" he would
have to start from scratch, at least to a considerable extent.23 That is, al-
though no given change in interactive standards, such as a change from mo-
nogamy to polygamy, would inevitably spell the end of one society and the
beginning of another, a global change of sufficient psychological impor-
tance can do so. Perhaps, then, two temporal stages of a temporally ex-
tended population linked by the social relations are parts of the same society
only if a member of one would have been able to fit into the other, without
serious psychological disorientation.
The position I have reached is vague, but it can be compared to the vague
identity conditions of families. A family is constituted by its membership,
and distinguished from other families by differences in membership. I do
not count every indefinitely distant biological ancestor as a member of my
family. Yet if we are given the membership of my family at one time, then
we can find its past and future membership by tracing the relations is a de-
scendent of and is an ancestor of. Every member of my family is related by
a string of these relations to every other member, and no person who is not
so related is a member. Similarly, it is not possible to have distinct societies
whose memberships are identical at all times. And the network of social re-
lationships that characterizes a societal population -traces out the past and
future membership of that population. We continue to deal with numerically
the same society only if there is the right kind of historical continuity in the
membership. That is, all the members of a given societal population are em-
braced in the right way by the network of social relationships characteristic
of that population, and no one who is not embraced by it in the right way is
a member. But, as I explained earlier, there are temporal limits marked ei-
ther by discontinuities in the social network, as in the American example, or
by significant change in the nature of the interactive standards characteriz-
ing the group, as in the English example.
The behavioural standards accepted in a society, and indeed the morality
accepted in a society, need not remain unchanged. A society must be
characterized at every time by the acceptance of a set of behavioural stand-
ards, but this does not mean that the set accepted at one time must be identi-
cal to, or even similar to, the set accepted at any other time. Even if the
morality accepted in a societal population at one time is different from the
 Society 207

morality accepted in a societal population at a later time, still, the popula-

tions may be identical, and the societies they constitute may be identical.
Yet the English example suggests that sufficient change in the interactive
standards accepted can lead to a "hiving off" of one society from an ances-
tor society.

8. Conclusion
My account aims to "construct" societies out of persons and their social re-
lationships.24 Accounts of this sort are sometimes criticized on the ground
that the characteristics of a person depend as much on those of the society
of which he is a member, as the characteristics of that society depend on the
characteristics of its members. Rawls asserts that "membership in our soci-
ety is given,... we cannot know what we would have been like had we not
belonged to it (perhaps'the thought itself lacks a" sense)" (1977, p.*16J): I
think that this is an overstatement. But it would be a misunderstanding of •
my account of society to suppose it to narrow our options in explaining hu-
man psychology. It seems obviously to be true that the nature of one's soci-
ety and the nature of one's position in it have profound effects on one's psy-
chology. My account has no implications to the contrary.
I have implied that a society exists only where there is a societal popula-
tion within which instrumental interaction is directed to securing the mate-
rial necessities and priorities of life (or the local cultural priorities) and
governed by standards of behaviour which are shared in the group. It fol-
lows that the concept of a society is to be understood in terms of the con-
cepts of a societal population, of standards of interactive behaviour, and of
needs and cultural values. Given the examples I used at the beginning of
this paper, it is worth mentioning that my account implies that the group of
sixteen Crusoes does not constitute a society nor do churches, clubs, subur-
ban communities and criminal organizations. This is because they are not
societal populations. This ought perhaps to have been obvious. Furthermore,
given my emphasis on moral philosophy, it is worth mentioning that a moral
code is one kind of standard of behaviour. Accordingly, a societal popula-
tion is a society if it shares a morality having appropriate properties25

Notes
1 Mario Bunge is one of the few philosophers to develop a systematic theory (1974 and
1979). See below.
2 For instance, Talcott Parsons is committed to the view that a society is a set of rules
(1966, p. 9). And Mario Bunge writes that a society is a "self-supporting" ordered
triple. The elements of such a triple are to be: a non-empty set of human beings; a dis-
tinct non-empty set of things that, intuitively, constitute the environment; and the union
of a set of transformation relations on the environment and a set of social relations, in-
cluding "man-transforming" relations such as teaching (Bunge 1979, p. 21, 23, and
Bunge 1974). An ordered triple is an abstract mathematical entity. It is logically impos-
sible for such a thing to have the kind of properties that a society can have. Bunge no-
208 Dialogue

tices this problem but does not solve it. He simply asserts that a society is "a concrete
totality" (1979, p. 19), "like a mountain" (1974, p. 183); he does not explain how an
ordered triple could be a concrete thing. In places, Bunge backs away from claiming
that a society is an ordered triple, and is satisfied with the claim that a society is "repre-
sentable by" such a thing (1979, p. 14, 19). But, if we take this to be his view, he has
no account of what a society is. Moreover, his reason for being cautious seems to be not
that he doubts that a society is an ordered triple, but that he realizes that he has not pro-
vided an account rich enough to enable us to determine which ordered triples are sup-
posed to be societies (1974, p. 214-15). So we see that Bunge's view is either false or
unhelpful.
3 Louis Pojman discusses a form of ethical relativism which relativizes moral right and
wrong to the culture of a society. But his argument is vitiated because he blurs the dis-
tinction between a society and such other collective entities as religious groups,
churches, clubs, families, communities, criminal organizations, and even ad hoc "cote-
ries." See Pojman (1989), p. 27-28.
4 The example is adapted from Nozick 1974, p. 185. Nozick uses it to question the role
assigned by Rawls to the concept of society in his theory of justice.
5 It is not that Gauthier has simply missed the fact that Rawls is concerned with what jus-
tice requires in societies, for he goes on to discuss "what one brings to society" and its
relation to "what one gets in society" (1986, p. 220). Perhaps he takes the example to
be relevant to Rawls's concerns because the Crusoes could form a society, even if they
are not already a society. And if they were to form a society, then, as against Rawls,
Gauthier believes that the inequality among them would not immediately become un-
just or unfair, for each would be entitled to what he would bring to the society. Yet even
on this construal, Gauthier would be operating with a very attenuated conception of
what a society is. A group of sixteen individuals living on separate islands would not
become a society simply as a result of their causing a few rafts to carry provisions from
island to island.
6 The Concise Oxford Dictionary of Current English, 4th ed., s.v. "Society." The dic-
tionary distinguishes between the use of "society" to refer to comprehensive social
communities, and three other uses: its use in referring to associations of persons united
by a common aim, interest, or principle, such as the Society of Friends, or the Royal
•fjocieiy; the use. of "society" to refer to "the upper classes of a community"; and its
use to refer to "the social mode of life" in general, <as we'rhay have in mind when we
speak of criminals as enemies of society.
7 Quoted by Stanley Benn (1967, p. 473).
8 Stanley Benn disagrees: "a society is an aggregate of interacting individuals whose re-
lations are governed by role-conferring rules and practices which give their actions
their characteristic significance" (1967, p. 471). I should concede that a system of sta-
tuses and roles is found in virtually every known society, but I deny that such a system
is necessary because I think there can be rudimentary and egalitarian societies without
significant social organization.
9 I am following suggestions of Mayhew (1968) and of Aberle et al. (1950).
10 Mayhew thinks that a society is not to be identified with a societal population; rather, a
society is a "system of action," a kind of abstract entity in which a societal population
may participate (Mayhew 1968, p. 585). Common sense suggests on the contrary that a
society is a population of people, and I propose to follow common sense rather than
Mayhew.
11 John Baker reminded me of exogamous groups.
12 These possibilities were suggested by Walter Edelberg. He and several other people
helped me to see the possibility of non-sexually reproducing societies.
 Society 209

13 On the Shakers, see Kern (1981), and the Encyclopedia Britannica, 15th ed., s.v.
"Shakers." The encyclopedia reports that Shaker membership declined from about
6,000 in 1840, grouped in about 20 communities (18 in 1826), to approximately 1,000
in 1905. By 1844, they were advertising for members. The first community was estab-
lished in the U.S. in 1777.
14 I am indebted to Alison Wylie for references to the anthropological literature: on the
Baringo, see Hodder 1982, p. 16-35, especially p. 16; on the Nuer, see Evans-Pritchard
1940a and 1940b, especially 1940b, p. 284.
15 This example was suggested by Paul Teller.
16 G. W. Fitch objected to my account that if there is one female who is the ultimate an-
cestor of every human, then my account implies that there is only one societal popula-
tion. In order to deal with this objection, and with certain other technical problems, I
shall introduce the notion of a minimal societal population. I shall say that a minimal
societal population extends into the future and the past, from a given time, to include
only those generations which people of the time generally take into account in thinking
about their ancestors and_ descendants, and about which they have specific knowledge
or expectations. For example, few English people today are likely to know 'anything
specific about their medieval ancestors (as opposed to medieval English people in gen- •
eral); few are likely to think of any specific medieval English people as their ancestors.
In our culture, the minimal societal population to which a person belongs likely extends
at most three generations into the past and one or two into the future. I shall also say
that the mereological sum of a "temporally dense, overlapping sequence" of minimal
societal populations may also be a societal population (where a "temporally dense,
overlapping sequence" is a temporally ordered sequence of minimal societal popula-
tions such that [a] every minimal population that could be ordered in time between any
two members of the sequence is also a member of the sequence, except that [b] every
pair of members of the sequejice, each of which has a membership at a given time, has
exactly the same membership at every such time). The existence of temporal limits on
the persistence of societal populations means that not every such mereological sum is a
societal population. I will not invoke the notion of a minimal societal population in the
text, and I will ordinarily have in mind "maximal" societal populations: a maximal so-
cietal population is a mereological sum of a temporally dense, overlapping sequence of
minimal societal populations that extends to temporal limits of the kinds explained in
the text. That is, either a maximal societal population is bounded by a relevant discon-
tinuity in a social network, or it is bounded by significant change of a relevant kind in
interactive standards. These notions are explained above and in Section 7.
17 A societal population cannot be artificially constructed simply by adding a randomly
chosen alien to the population of some genuine society. That is, one cannot be con-
structed simply by taking the mereological sum of a societal population and a randomly
chosen alien. The alien would be neither a new member nor an hereditary member: her
ancestors and descendents would not be part of the group, nor would her friends, nor
would it be true that she is disposed to find new friends or mates from within the group.
The situation would not be fundamentally different if we augmented the group by add-
ing our alien's ancestors, descendents, mate, and mate's ancestors. The resulting sub-
group still would not be part of the net of social relationships characteristic of the group
as a whole. Few if any of the friends of these people, their friends' friends and mates,
their friends and mates, and so on, would be part of the group. Moreover, with respect
to none of the people in the subgroup would it be true, nor would the probability be
high, that if they had selected someone else as a mate, it would have been someone
from the group as a whole (or from a culturally sanctioned group). Neither the alien
alone, nor the group that includes her ancestors and so on, would be a part of the net-
210 Dialogue

work of social relationships characteristic of the group as a whole. It is true, however,

that the mereological sum of two randomly selected societal populations is a societal
population. But this is not an embarrassment, for the notion of a societal population is
an artificial one anyway. Moreover, it does not follow that the mereological sum of two
randomly selected societal populations is itself a society.
18 Cf. David Lewis on conventions (1968, p. 78-79).
19 This example was brought to my attention by Alison Wylie. See Turnbull 1972 and
Wilson et al. 1975.
20 See remarks by McCall in Wilson et al. 1975, p. 345-48, and Geddes in Wilson et al.
1975. Turnbuil does not deny their claims (Wilson et al. 1975).
21 I owe these references to Alison Wylie.
22 The wording is based on The Concise Oxford Dictionary of Current English, 4th ed.,
s.v. "Society."
23 Teller suggested, in personal correspondence, that "the problems here have much in
common with the problems with identity over time of species. How much does a spe-
cies have to evolve before we say 'new species'?" He said, "One approach to this
problem in biology is to say that two temporal segments of an evolving interbreeding
population are (parts of) different species if a member of one would not have been able
to interbreed with a member of the other if the two organisms had coexisted. You might
be able to do something crudely similar, with [interactive] standards playing the role of
interbreeding." I owe the example of contemporary and medieval English society to
Teller.
24 Given that we have identified the members of a society" and the periods of their mem-
bership—and this is where social relationships are germane —the account I gave in
Copp 1984 asserts roughly that a society is the mereological sum of the temporal stages
of its members during the time that they are members.
25 I am grateful for the useful comments of the students who read versions of the manu-
script in seminars I gave at the University of Illinois at Chicago in 1987 and the Univer-
sity of California, Davis in 1989. I presented a condensed version in 1985 to the Con-
gress of the Canadian Philosophical Association, Montreal and to The Great Expecta-
tions Philosophy Colloquium, Evanston, Illinois. I am grateful for the helpful discus-
sion on both occasions, and for the comments of two anonymous referees for this jour-
nal. I owe special thanks to Jphn Baker, Walter Edelberg, G. W. Fitch, Paul Teller and
Alison Wylie. My research was assisted by the generous support of several institutions.
During the year 1983-84, I held a Research Grant and Research Time Stipend
(410-82-0640) from the Social Sciences and Humanities Research Council of Canada.
During the spring of 1986,1 benefited from a Short Research Leave which was awarded
me by the Campus Research Board of the University of Illinois at Chicago. Most re-
cently, the generosity of the Research Triangle Foundation enabled me to spend the
year, 1988-89, at the National Humanities Center, North Carolina. Simon Fraser Uni-
versity granted me the leaves of absence I required in order to take advantage of these
opportunities. I am most grateful to all of these institutions.

References
Aberle, D. F., A. K. Cohen, A. K. Davis, M. J. Levy, Jr., and F. X. Sutton
1950 "The Functional Prerequisites of a Society." Ethics, 60, 2 (January):
100-11.
Benedict, Burton
1968 "Societies, Small." In International Encyclopedia of the Social Sci-
 Society 211

ences. Edited by David L. Sills. New York: Crowell, Collier and Mac-
millan.
Benn, Stanley I.
1967 "Society." In The Encyclopedia of Philosophy. Edited by Paul Edwards.
Rpt. 1972. New York: Macmillan Publishing and The Free Press. Vol. 7,
p. 470-74.
Brandt, Richard B.
1963 "Toward a Credible Form of Utilitarianism." Rpt. 1968 in Contempo-
rary Utilitarianism. Edited by Michael D. Bayles. Garden City, NY: An-
chor Books. P. 143-86.
1979 A Theory of the Good and the Right. Oxford: Clarendon Press.
Bunge, Mario
1974 "The Concept of Social Structure." In Developments in the Methodolo-
gy of Social Science. Edited by W. Leinfellner and E. Kohler. Dordrecht,
Holland: D. Reidel. P.-.175-215. „ . • • - . ,
1979 " A Systems Concept of Society: Beyond Individualism and Holism*."
Theory and Decision, 10 (January): 13-30.
Copp, David
1979 "Do Nations Have the Right of Self-Determination?" In Philosophers
Look at Canadian Confederation. Edited by Stanley G. French. Mont-
real: Canadian Philosophical Association. P. 71-95.
1984 "What Collectives Are: Agency, Individualism and Legal Theory." Dia-
logue, 23 (June): 249-69.
Devlin, Lord Patrick
1959 "The Enforcement of Morals." Rpt. 1965 as "Morals and the Criminal
Law." In Patrick Devlin. The Enforcement of Morals. Oxford: Oxford
University Press. P. 1-25.
Evans-Pritchard, Edward E.
1940a TheNuer. Oxford: Clarendon Press.
1940b "The Nuer of the Southern Sudan." In African Political Systems. Edited
by M. Fortes and E. E. Evans-Pritchard. London: Oxford University
Press.
Gauthier, David
1986 Morals by Agreement. Oxford: Oxford University Press.
Hodder, Ian
1982 Symbols in Action. Cambridge: Cambridge University Press.
Johnson, Harry M.
1960 Sociology: A Systematic Introduction. New York: Harcourt, Brace and
World.
Kern, Louis J.
1981 An Ordered Love: Sex Roles and Sexuality in Victorian Utopias —The
Shakers, the Mormons and the Oneida Community. Chapel Hill, NC:
University of North Carolina Press.
Lewis, David
1969 Convention: A Philosophical Study. Cambridge, MA: Harvard University
Press.
Marshall, Thomas H.
1956 "Anthropology Today." British Journal of Sociology, 7: 59-64.
212 Dialogue

Mayhew, Leon H.
1968 "Society." In International Encyclopedia of the Social Sciences. Edited
by David L. Sills. New York: Crowell, Collier and Macmillan.
Nozick, Robert
1974 Anarchy, State, and Utopia. New York: Basic Books.
Parsons, Talcott
1951 The Social System. Glencoe, IL: The Free Press.
1961 "An Outline of the Social System." In Theories of Society. Edited by
Talcott Parsons, E. Shils, K. D. Naegele and J. R. Pitts. New York: Free
Press of Glencoe. Vol. 1, p. 30-79.
1966 Societies: Evolutionary and Comparative Perspectives. Englewood
Cliffs, NJ: Prentice-Hall.
Pojman, Louis
1989 " A Critique of Ethical Relativism." In Ethical Theory: Classical and
Contemporary Readings. Edited by Louis Pojman. Belmont, CA: Wads-
worth.
Rawls, John
1971 A Theory of Justice. Cambridge, MA: Harvard University Press.
1977 "The Basic Structure as Subject." American Philosophical Quarterly,
14, 2 (April): 159-65.
Turnbull, Colin M.
1972 The Mountain People. New York: Simon and Schuster.
Wilson, Peter J., Grant McCall, W. R. Geddes, A. K. Mark, John E. Pfeiffer, James
B. Boskey and Colin M. Turnbull
1975 "More Thoughts on the Ik and Anthropology." Current Anthropology,
16, 3 (September): 343-58.