THE CROSS-CULTURALSURVEY

GEORGE PETER MURDOCK

Yale University
F OR A NUMBER of years, the Institute of Human Relations at Yale
University has been conducting a general program of research in the
social sciences, with particular reference to the areas common to, and
marginal between, the special sciences of sociology, anthropology, psy-
chology, and psychiatry. In I937, as one of the specific research projects
on the anthropological and sociological side of this program, the Cross-
Cultural Survey was organized.'
A year of previous experience in collaborating with other social scientists
in research and discussion had made it clear to the anthropologists associ-
ated with the Institute that the rich resources of ethnography, potentially
of inestimable value to workers in adjacent fields, were practically inacces-
sible to them. Working in the laboratory, the clinic, or the community, the
psychologists, sociologists, and others made frequent requests of the cul-
tural anthropologists for comparative data on various aspects of behavior
among primitive peoples. Sometimes they wanted perspective, sometimes
suggestions, sometimes a check on their own scientific formulations. In try-
ing to assist them, the anthropologists found that they could usually cite a
limited number of cases from their own knowledge and give an impression-
istic judgment as to the general status of ethnography on the question. For
scientists, however, this was often not enough. What guarantee was there
that the remembered cases were representative, or the impressions valid?
What was needed was access to a dependable and objective sample of the
ethnographic evidence. Only rarely was it possible to refer the seeker to an
adequate summary of the evidence; in the great majority of instances, he
could satisfy his scientific curiosity only by resorting to the vast descriptive
literature itself and embarking on a research task of discouraging magnitude.
An actual example will illustrate the difficulty. Several years ago, a group
of physiologists, working in the laboratory, had come to a series of conclu-
sions with respect to the relationship between periodicity of eating and
bodily health as reflected in measurements of weight, stature, etc. It oc-
curred to them that the literature of anthropology should contain data by
which their conclusions might be independently tested, and they referred
to the author for advice. He was able to tell them that ethnographers cus-
tomarily report the relevant data on eating habits-the number of meals
1 Based on a paper presented to the American Anthropological Association in Chicagos
Dec. 28, I939. For further information on the research program of the Institute of Human
Relations, and upon the relation of the Cross-Cultural Survey thereto, see M. A. May,
"Report of the Director of the Institute of Human Relations for the Academic Years I937-
I938, I938-i939," Bulletinof Yale University,series35, XXVII (I939), I-35.

36i
362 AMERICAN SOCIOLOGICALREVIEW
per day, their temporal spacing, the degree of regularity or irregularity in
eating, etc.-and that physical anthropologists present the pertinent so-
matological information. Since the material had been gathered, it could be
assembled and the crucial correlations drawn. To have done so, however,
would have required several months of research, since the data had nowhere
been summarized and it would have been necessary to ransack an immense
amount of descriptive literature to assemble it. Understandably enough,
the physiologists were discouraged from undertaking this promising but
formidable task.
Other sciences have systems of abstracts, bibliographical aids, and quan-
tities of secondary collections, by means of which the researcher can quickly
track down the pertinent data and acquaint himself with previous research
on any subject. With a few notable exceptions,2 anthropology lacks such
aids. Its materials are widely scattered in descriptive reports, an immense
number of which must be scanned if adequate information is desired on any
particular topic. The factual data of sociology are in a similarly chaotic
condition. It became apparent, therefore, that if these sciences were to be
brought to bear effectively in the cooperative research program of the
Institute, a representative sample of the cultural materials on the various
societies of the world needed to be organized for ready accessibility on any
subject. The Cross-Cultural Survey was developed, in part to fill this need,
in part to facilitate a distinctive type of scientific research which will be
described below.
The first problem was to devise a standard system of classification for
the arrangement and use of the collected materials. After six months of
preliminary research, with the aid of helpful suggestions from about a
hundred anthropologists, sociologists, and other specialists, the author and
five collaborators published the Outline of Cultural Materials. Although
this manual has proved of some incidental utility in field research, it was in
no sense designed for such a purpose. It was written solely as a guide for
organizing and filing our abstracted cultural materials, and for facilitating
reference to the data already classified and filed.
Since the publication of the manual, in I938, the staff of the Cross-
Cultural Survey has been engaged in the actual assembling of materials.
To date, the descriptive data on nearly a hundred cultures have been ab-
stracted, classified, and filed. It is hoped ultimately to assemble andor-
ganize all the available cultural information on several hundred peoples,
2 Useful for special purposes are the massive collections of Frazer, Sumner and Keller,
Thurnwald, and Westermarck, such classic monographic studies as those of Hahn on domestic
animals, Nieboer on slavery, Schurtz on age groupings, and Steinmetz on punishment, and
such recent special treatises as that of Clements on theories of disease. These compilations,
however, do not lend themselves to the determination of "adhesions" in Tylor's sense, i.e.,
correlations within a culture indicative of functional relationships, and thus have but limited
use in the testing of scientific hypotheses.
 THE CROSS-CULTURAL SURVEY 363

who will be adequately distributed with regard to geography and fairly

representative of all major types and levels of culture. Although primitive
cultureswill preponderate numerically, because they reveal thewidestrange
of human behavioral variations, there will be a fair representation of the
historical civilizations of the past, of modern folk cultures, and of the
communities studied by contemporary sociologists.
For each of the cultures analyzed, the entire literature is covered, includ-
ing manuscript materials when available. In some instances, more than a
hundred books and articles have been combed for a single tribe or historical
period. All material in foreign languages has been translated into English.
The information, if of any conceivable cultural relevance, is transcribed in
full-in verbatim quotations or exact translations. The object has been to
record the data so completely that, save in rare instances, it will be entirely
unnecessary for a researcher using the files to consult the original sources.
Mere abstracts are deemed unsatisfactory and are resorted to only in ex-
ceptional cases, when the information is excessively detailed or technical.
The Outline of Cultural Materials is not a "trait list," nor are the files con-
fined to data on the items listed in it. These items are merely suggestions as
to the kinds of material to be filed-or sought-under a particular heading,
and they make no pretense of being exhaustive. Special pains are taken to
preserve intact the functional relationships of the data. Wherever division
according to the categories of the manual would be arbitrary, or would
destroy the context, the original account is preserved intact and is filed in
one place, with a carbon copy or a cross-reference slip under each other
category to which the information is pertinent. Each file, moreover, con-
tains a short synopsis of the total culture to which any note can be referred
for context.
The collection of organized and classified materials in the files of the
Cross-Cultural Survey should prove useful in nearly every type of research
which anthropologists and other social scientists have hitherto pursued. If
an investigator wishes to study a particular culture, he will find all the data,
from whatsoever source, organized conveniently for his use. If he is inter-
ested in a topic, he can run through the material under one or more head-
ings for as many cultures as he likes, and secure his information in a mere
fraction of the time required to comb the sources for himself. If he desires
to test an hypothesis, he can similarly examine the material under two or
more categories and obtain a quantitative check in the form of a correla-
tion. A cross-cultural test of the physiologists' hypothesis on the periodicity
of eating, alluded to above, could, for example, probably now be made with
not more than two days of research. Even regional or distributional studies
are possible for areas, like the Gran Chaco of South America, on which the
files approach completeness. The Cross-Cultural Survey, in short, should
prove useful in a wide variety of scientific researches for which ready access
364 AMERICAN SOCIOLOGICAL REVIEW

to a body of organized cultural data is needed. It is intended, of course, to

make the material generally available on a cooperative basis. Recent users
of the files include-to cite but a few examples-a sociologist analyzing
social classes, a psychologist interested in adolescent problems, and a
psychiatrist seeking a cultural definition of insanity.
In addition to its practical objective of facilitating diverse forms of
social science research, the Cross-Cultural Survey has a special theoretical
objective. It is organized so as to make possible the formulation and varn-
fication, on a large scale and by quantitative methods, of scientific generali-
zations of a universally human or cross-cultural character. Sociologists and
most other social scientists regard the establishment of generalizations or
"laws," i.e., verified statements of correlations between phenomena, as
their primary aim, but anthropologists tend to shy away from theory, as
Kluckhohn3 has pointed out, and to confine themselves to historical rather
than scientific interpretations of their subject matter. Nevertheless, it
seems premature to conclude that anthropology cannot be made a science
until, using all known safeguards, we have made at least one serious and
systematic attempt to formulate scientific generalizations about man and
culture which will withstand a quantitative test. Anthropology has many
objectives. That envisaged by the Cross-Cultural Survey is not intended
to supplant the others, nor does it lay claim to greater importance. It is
simply regarded as legitimate, promising, afid opposed by no insuperable
theoretical obstacles.
The plan rests, at bottom, on the conviction that all human cultures,
despite their diversity, have fundamentally a great deal in common, and
that these common aspects are susceptible to scientific analysis. Its theo-
retical orientation may be expressed in a series of seven basic assumptions.
These are not claimed to be original, since many of them are shared by all
social scientists, and all of them by many.
i. Culture Is Learned. Culture is not instinctive, or innate, or trans-
mitted biologically, but is composed of habits, i.e., learned tendencies to
react, acquired by each individual through his own life experience after
birth. This assumption, of course, is shared by all anthropologists outside
of the totalitarian states, but it has a corollary which is not always so clearly
recognized. If culture is learned, it must obey the laws of learning, which
the psychologists have by now worked out in considerable detail. The prin-
ciples of learning are known to be essentially the same, not only for all
mankind but also for most mammalian species. Hence, we should expect all
cultures, being learned, to reveal certain uniformities reflecting this uni-
versal common factor.
2. Culture Is Inculcated. All animals are capable of learning, but man

3 C. Kluckhohn, "The Place of Theory in Anthropological Studies." Philosophy of Science,

VI (I939), 328-344.
 THE CROSS-CULTURALSURVEY 365
alone seems able, in any considerable measure, to pass on his acquired
habits to his offspring. We can housebreak a dog, teach him tricks, and im-
plant in him other germs of culture, but he will not transmit them to his
puppies. They will receive only the biological inheritance of their species,
to which they in turn will add habits on the basis of their own experience.
The factor of language presumably accounts for man's preeminence in this
respect. At any rate, many of the habits learned by human beings are trans-
mitted from parent to child over successive generations, and, through re-
peated inculcation, acquire that persistency over time, that relative
independence of individual bearers, which justifies classifying them col-
lectively as "culture." This assumption, too, is generally accepted by an-
thropologists, but again there is an underestimated corollary. If culture is
inculcated, then all cultures should show certain common effects of the
inculcation process. Inculcation involves not only the imparting of tech-
niques and knowledge but also the disciplining of the child's animal im-
pulses to adjust him to social life. That there are regularities in behavior
reflecting the ways in which these impulses are thwarted and redirected
during the formative years of life, seems clear from the evidence of psycho-
analysis, e.g., the apparent universality of intrafamily incest taboos.
3. CultureIs Social. Habits of the cultural order are not only inculcated
and thus transmitted over time; they are also social, that is, shared by
human beings living in organized aggregates or societies and kept relatively
uniform by social pressure. They are, in short, group habits. The habits
which the members of a social group share with one another constitute the
culture of that group. This assumption is accepted by most anthropolo-
gists, but not by all. Lowie,4 for example, insists that"a culture is invariably
an artificial unit segregated for purposes of expediency.... There is only
one natural unit for the ethnologist-the culture of all humanity at all
periods and in all places . .." The author finds it quite impossible to
.
accept this statement. To him, the collective or shared habits of a social
group-no matter whether it be a family, a village, a class, or a tribe-con-
stitute, not "an artificial unit" but a natural unit-a culture or subculture.
To deny this is, in his opinion, to repudiate the most substantial contribu-
tion which sociology has made to anthropology. If culture is social, then the
fate of a culture depends on the fate of the society which bears it, and all
cultures which have survived to be studied should reveal certain similarities
because they have all had to provide for societal survival. Among these
cultural universals, we can probably list such things as sentiments of group
cohesion, mechanisms of social control, organization for defense against hos-
tile neighbors, and provision for the perpetuation of the population.
4. Culture Is Ideational. To a considerable extent, the group habits ot
which culture consists are conceptualized (or verbalized) as ideal norms or
4R. H. Lowie, The History of Ethnological Theory, 235-236, New York, I937.
366 AMERICAN SOCIOLOGICALREVIEW
patterns of behavior. There are, of course, exceptions; grammatical rules,
for example, though they represent collective linguistic habits and are thus
cultural, are only in small part consciously formulated. Nevertheless, as
every field ethnographer knows, most people show in marked degree an
awareness of their own cultural norms, an ability to differentiate them from
purely individual habits, and a facility in conceptualizing and reporting
them in detail, including the circumstances where each is considered appro-
priate and the sanctions to be expected for nonconformity. Within limits,
therefore, it is useful to conceive of culture as ideational, and of an element
of culture as a traditionally accepted idea,5 held by the members of a group
or subgroup, that a particular kind of behavior (overt, verbal, or implicit)
should conform to an established precedent. These ideal norms should not
be confused with actual behavior. In any particular instance, an individual
behaves in response to the state of his organism (his drives) at the moment,
and to his perception of the total situation in which he finds himself. In so
doing, he naturally tends to follow his established habits, including his cul-
ture, but either his impulses or the nature of the circumstances may lead
him to deviate therefrom to a greater or lesser degree. Behavior, therefore,
does not automatically follow culture, which is only one of its determinants.
There are norms of behavior, of course, as well as of culture, but, unlike the
latter, they can be established only by statistical means. Confusion often
arises between anthropologists and sociologists on this point. The former,
until recently, have been primarily preoccupied with ideal norms or pat-
terns, whereas sociologists, belonging to the same society as both their
subjects and their audience, assume general familiarity with the culture
and commonly report only the statistical norms of actual behavior. A typical
community study like Middletown and an ethnographic monograph, though
often compared, are thus in reality poles apart. To the extent that culture
is ideational, we may conclude, all cultures should reveal certain similari-
ties, flowing from the universal laws governing the symbolic mental proc-
esses, e.g., the worldwide parallels in the principles of magic.
5. Culture Is Gratifying. Culture always, and necessarily, satisfies basic
biological needs and secondary needs derived therefrom. Its elements are
tested habitual techniques for gratifying human impulses in man's inter-
action with the external world of nature and fellow man.6 This assumption
is an inescapable conclusion from modern stimulus-response psychology.
Culture consists of habits, and psychology has demonstrated that habits
I From the point of view of behavioristic psychology, of course, an idea is merely a habit
of a special sort, a tendency to react with implicit linguistic or symbolic behavior rather than
with overt muscular responses. The underlying mechanisms, e.g., of learning, are similar if not
identical. Fundamentally, therefore, our fourth assumption should be subsumed under our
first-that culture is learned-as a special case thereof. In view of the importance of symbolic,
especially linguistic, behavior in man, however, it has seemed advisable to segregate the
ideational point for separate exposition.
6 The only exceptions are partial and temporary ones, with respect to elements of culture
in the process of dying out or being supplanted.
 THE CROSS-CULTURALSURVEY 367
persist only so long as they bring satisfaction. Gratification reinforces
habits, strengthens and perpetuates them, while lack of gratification in-
evitably results in their extinction or disappearance. Elements of culture,
therefore, can continue to exist only when they yield to the individuals of a
society a margin of satisfaction, a favorable balance of pleasure over pain.7
Malinowski has been insisting on this point for years, but the majority of
anthropologists have either rejected the assumption or have paid it but
inadequate lip service. To them, the fact that culture persists has seemed
to raise no problem; it has been blithely taken for granted. Psychologists,
however, have seen the problem, and have given it a definitive answer, which
anthropologists can ignore at their peril. If culture is gratifying, widespread
similarities should exist in all cultures, owing to the fact that basic human
impulses, which are universally the same, demand similar forms of satis-
faction. The "universal culture pattern" propounded by Wissler8 would
seem to rest on this foundation.
6. Culture Is Adaptive. Culture changes; and the process of change ap-
pears to be an adaptive one, comparable to evolution in the organic realm
but of a different order.9 Cultures tend, through periods of time, to become
adjusted to the geographic environment, as the anthropogeographers have
shown, although environmental influences are no longer conceived as de-
terminative of cultural development. Cultures also adapt, through borrow-
ing and organization, to the social environment of neighboring peoples.
Finally, cultures unquestionably tend to become adjusted to the biological
and psychological demands of the human organism. As life conditions
change, traditional forms cease to provide a margin of satisfaction and are
eliminated; new needs arise or are perceived, and new cultural adjustments
are made to them. The assumption that culture is adaptive by no means
commits one to an idea of progress, or to a theory of evolutionary stages of
development, or to a rigid determinism of any sort. On the contrary, one
can agree with Opler,'0 who has pointed out on the basis of his Apache
material, that different cultural forms may represent adjustments to like
problems, and similar cultural forms to different problems. It is probable,
nevertheless, that a certain proportion of the parallels in different cultures
represent independent adjustments to comparable conditions.
The conception of cultural change as an adaptive process seems to many
anthropologists inconsistent with, and contradictory to, the conception of
7 Culture is gratifying, of course, not in an absolute but in a relative sense. To a slave, for
example, the submission and drudgery demanded by his status are not actually pleasant;
relative, however, to the painful alternative of punishment or death for rebellious behavior,
observance of the cultural requirements of his status is gratifying or "reinforcing." Agricul-
tural labor, again, may not be enjoyable in itself, but it is gratifying because it brings rewards,
e.g., in food.
C. Wissler, Man and Culture, 73-79, New York, I923.
See A. G. Keller, Societal Evolution, New York, i9i5.
10 M. E. Opler, "Apache Data concerning the Relation of Kinship Terminology to Social
Classification," Amer. Anthropol., n.s., XXXIX (0937), 207-208.
368 AMERICAN SOCIOLOGICAL REVIEW

cultural change as an historical process. To the author, there seems nothing

inconsistent or antagonistic in the two positions-the "functional" and the
"historical," as they are commonly labeled. On the contrary, he believes
that both are correct, that they supplement one another, and that the best
anthropological work emerges when the two are used in conjunction. Cul-
ture history is a succession of unique events, in which later events are con-
ditioned by earlier ones. From the point of view of culture, the events
which affect later ones in the same historical sequence are often, if not
usually, accidental, since they have their origin outside the continuum of
culture. They include natural events, like floods and droughts; biological
events, like epidemics and deaths; and psychological events, like emotional
outbursts and inventive intuitions. Such changes alter a society's life con-
ditions. They create new needs and render old cultural forms unsatisfac-
tory, stimulating trial and error behavior and cultural innovations. Perhaps
the most significant events, however, are historical contacts with peoples of
differing cultures, for men tend first to ransack the cultural resources of
their neighbors for solutions to their problems of living, and rely only sec-
ondarily upon their own inventive ingenuity. Full recognition of the his-
torical character of culture, and especially of the role of diffusion, is thus a
prime prerequisite if a search for cross-cultural generalizations is to have
any prospect of success. It is necessary to insist, however, that historical
events, like geographic factors, exert only a conditioning rather than a
determining influence on the course of culture. Man adjusts to them, and
draws selectively upon them to solve his problems and satisfy his needs.
7. Culture Is Integrative. As one product of the adaptive process, the
elements of a given culture tend to form a consistent and integrated whole.
We use the word "tend" advisedly, for we do not accept the position of
certain extreme functionalists that cultures actually are integrated sys-
tems, with their several parts in perfect equilibrium. We adhere, rather, to
the position of Sumner" that the folkways are "subject to a strain of con-
sistency with each other," but that actual integration is never achieved for
the obvious reason that historical events are constantly exerting a disturb-
ing influence. Integration takes time-there is always what Ogburn'2 has
called a "cultural lag"-and long before one process has been completed,
many others have been initiated. In our own culture, for example, the
changes wrought in habits of work, recreation, sex, and religion through
the introduction of the automobile are probably still incomplete. If culture
is integrative, then correspondences or correlations between similar traits
should repeatedly occur in unrelated cultures. Lowie,'3 for example, has
pointed out a number of such correlations.
W. G. Sumner, Folkways, 5-6, Boston, i906.
12 W. F. Ogburn, Social Change,200, New York, I 922.
13 R. H. Lowie, Primitive Society, New York, i920.
 THE CROSS-CULTURAL SURVEY 369
If the seven fundamental assumptions outlined above, or even any con-
siderable proportion of them, are valid, then it must necessarily follow that
human cultures in general, despite their historical diversity, will exhibit
certain regularities or recurrences which are susceptible to scientific analy-
sis, and which, under such analysis, should yield a body of scientific gener-
alizations. A primary objective of the Cross-Cultural Survey is to for-
mulate and test generalizations of this sort.
The first methodological step will be the logical elaboration of hypoth-
eses. From whatever source derived-from generalizations advanced by
anthropologists and sociologists, from psychological theory, or from leads
found in the material-the hypotheses will be subjected to rigorous logical
analysis and worked over into a series of basic postulates and testable theo-
rems. By this procedure, the most effective of scientific methods, all logical
or deductive operations are performed prior to the empirical test; there
remain no fallible logical steps to be taken after the inductive labor is
completed-a weakness which has vitiated much comparative anthropology.
The second step will be the verification of the theorems. A postulate can
stand only if every theorem derived from it checks with the facts; if even a
single one fails in this test, the postulate falls. The verification will be quan-
titative. In scientific anthropology, it would seem, there is safety in num-
bers. Only if one deals with a large number of cases can one expect to
encompass all the significant causal factors, occurring in various permuta-
tions and combinations, estimate by statistical means their relative efficacy,
and segregate by their quantitative preponderance the universal or cross-
cultural factors from the local or historical ones. In testing each theorem,
it is intended to use an adequate number of cases, preferably at least two
hundred tribes if possible, selected from the files as the fairest sample feasi-
ble of all known cultures. In so far as possible, they will be chosen in equal
numbers from all continents and all culture areas, including a representa-
tive selection of historical and modern civilizations. Each theorem will be
posed to all the cases in terms of an anticipated correlation between two
traits or aspects of culture, and the positive and negative results will be
tabulated and expressed in terms of some reliable statistical coefficient. If,
for each of a set of theorems, the coefficients obtained are positive in sign
and significant in quantity, the postulate in question will be regarded as
tentatively verified.
The third step will be a critical analysis of the results from an areal or
distributional point of view. A valid cross-cultural hypothesis should hold
true in any area. If, however, some areas are discovered to yield negative
coefficients, while other areas with a larger total number of cultures yield
positive coefficients, it must be concluded that the apparent statistical con-
firmation of the hypothesis is fictitious and accidental, and the hypothesis
must either be rejected entirely or modified and tested again. To survive,
370 AMERICAN SOCIOLOGICAL REVIEW

therefore, any generalization will have to pass two tests-a quantitative

statistical one and an analytical historical one.
The fourth step will be a detailed examination of all exceptional or nega-
tive cases. To a valid scientific principle, there are no exceptions; apparent
exceptions are always due to the intrusion of another countervailing prin-
ciple. Thus, water always obeys the law of gravity, which causes it to flow
downhill. In all cases where water moves in the opposite direction, as in
osmosis, capillary attraction, evaporation, the siphon, and the hydraulic
ram, the law of gravity is still in operation but its influence is counteracted
by some other force or principle. Similarly, a valid cultural principle should
have no real exceptions. This makes it important to examine carefully all
seeming exceptions. If countervailing factors are not found, the principle
becomes suspect. One example may be cited from ethnography. The func-
tional association of an Omaha type of kinship system with patrilineal sibs
appears to be a valid cultural principle, yet Wagner'4 has reported an
Omaha system for the Yuchi, who are known to have had matrilineal sibs.
Eggan cites evidence, and Speck'5 agrees with him, that the Yuchi formerly
possessed a Crow type of kinship system, which is functionally associated
with matrilineal sibs, and that they shifted to an Omaha system only in
relatively recent times in consequence of close contacts with Central Al-
gonkian tribes like the Shawnee, who are patrilineal and possess kinship
systems of Omaha type. Presumably, the Yuchi have changed too recently
for the integrative process to have run its course. Thus, the apparent ex-
ception is not a real one, and the principle is not negated. An hypothesis,
all of whose seeming exceptions can be explained in some such fashion as
this, may be regarded as finally validated-subject always, of course, to
correction as new evidence comes in.
The Cross-Cultural Survey, it may be said in conclusion, is designed to
contribute in several ways to scientific research in the disciplines concerned
with cultural phenomena. It can answer specific questions of fact with a
minimum of time-wasting labor. It can reveal gaps in the ethnographical
record and thus suggest what groups should be restudied and what hitherto
unreported data should be gathered in the field. It can subject existing
theoretical hypotheses about collective human behavior to a quantitative
test, and can be used to formulate and verify new social science generaliza-
tions. In short, it should prove helpful in nearly every type of research
requiring an organized and classified body of cultural materials.
14 G. Wagner, "Yuchi," Handbookof American Indian Languages, 111, 339-340, New York,
'934.
15F. G. Speck, "Eggan's Yuchi Kinship Interpretations," Amer. Anthropol., n.s., XLI
(i939), I7I-I72.