SOME MISCONCEPTIONS OF MULTILINEAR EVOLUTION1

William C. Smith

It is the object of this paper to consider certain conceptual difficulties
in Julian Steward's theory of multillnear evolution. The particular concepts

with which we shall be concerned may be grouped into four pairs, reflecting the
order of discussion to be followed: these concepts are (1) "evolution" and

"evolutionism"; (2) "culture change" and "cultural evolution"; (3) "unilinear"
and "multilinear" evolution; and (4) "causality" and "law."

In the published works of Julian Steward (as elsewhere) the term "evolu-
tion" is often used to denote two concepts which, while not unrelated, should
be distinguished if mi sunderstanding is to be avoided. These two concepts may
be referred to by the terms "evolution" and "evolutionism-" '"Evolution" here
denotes "developuiental process":  it is this meaning which Steward wishes to
convey when he says that certain nineteenth century thinkers held that every-
where "cultural evolution must be governed by the same principles and follow

the same line, and [that] all mankind would progress toward a civilization like
that of Europe" (1956:69).  "Evolutionism" on the other hand refers to a point
of view, a body of theory, a methodology: thus Steward refers to "the evolu-

tionism of White and Childe"f (1953.317). But evolution (as process) and evolu-
tionism (as methodology) are not always distinguished by Steward: he says,
for example, that "The facts now accumulated indicate that human culture

evolved along a number of different lines; we must think of cultural evolution
not as unilinear but as mrultilinear. * O" (1956:73)--clearly speaking of evol-
ution as a process; but elsewhere he states that "Multilinear evolution is
essentially a methodology.   ott (1956:318).

In any evernt, it is with evolution as methodology that Steward is primarily
concerned. He contrasts "evolution" and "culture change,," for example, not as

distinct processes to, be' studied, but as basically different theoretical orien-
tations or methodologies. Thus he seems to equate "culture change" with "cul-

tural relativism and historical particularism" as a methodological view differ-
ent from that of "cultural evolution," which he defines as "a quest for

cultural regularities or laws. * o" (9153315).  Now it might be argued that
even If Steward does often use the term "evolution" to refer to methodology

rather than to process, he may nevertheless intend to imply something of proc-
ess. This may indeed be the case; but inasmuch as "cultural evolution" refers
to a process to be studied, it should be possible to distinguish this process
from other cultural processes--such as, for example, cultural change. How-

ever, we have seen that these processes are defined in Steward's writing only
by implication; and the implication is that "cultural evolution" is a process
which lends itself to study by the method of empirical scientific inquiry;

while "culture change" is a process more amenable to treatment by what Steward
calls the methods of "cultural relativism and historical particularism" (5953:
3l5). Now the notion that some processes can be studied scientifically while
others (namlely the "historical") cannot, is not a new one. In speaking of

some of the earlier manifestations of this idea, Kenneth Bock has said: "Adher-
ence to this view can confuse the issue. . . The belief that some things come
about by a series of happenings singular to each of them and other things come
about by comparable or like processes, or even the belief that among common

processes there will be some unique aspects, are conclusions that can be reached
only through comparison; they are not warranted or serviceable as assumptions
from which inquiry takes its departure' (l956:113). The processes of culture
change and cultural evolution can be distinguished in these terms only as a

result of inquiry--thus the distinction between them cannot reasonably be used
in defining the methods by which they are to be studied. In short, if we at-
tempt to regard Steward's term "evolution" as referring to a process, we find

that the process remains quite ambiguously defined. We shall therefore abandon
this contention, and accept the view that by "evolution" Steward does in fact
mean (as he has declared) "methodology."

Steward has distinguished three forms of the methodology which he calls
"evolution7': these are unilinear evolution, universal evolution, and multi-

linear evolution. Unilinear evolution deals with particular cultures, placing
them in stages of a universal sequence; but, according to Steward, ethnographic
data have demonstrated that there are no stages through which all cultures have
passed or must pass. Universal evolution deals with "culture" rather than with
"cultures"; but to Steward its assertions are so general and trivial as to be

of little interest. Multilinear evolution deals with particular cultures, and
asserts that particular cases of parallel development follow from the operation
of particular causes; it seeks empirically verifiable generalizations, but not

universal ones. Thus multilinear evolution retains the generalizing, scientific
character which Steward sees in. all evolutionary schemes, but attempts to tie
these closer to empirical reality (1953:315-318).

What, then, is characteristic and distinctive in the methodology of multi-
linear evolution as contrasted with that of unilinear and universal evolution?
Multilinear evolution is a more modest approach, we are told: flit is distinc-
tive in searching for parallels of limited occurrence instead of universals"

(1953:315). It seems clear that to Steward this is a point of methodological

virtue, but one may question both its necessity and its desirability. -Surely,
generalizations (or "parallels") pertaining to all societies are as valuable

as those pertaining only to some societies; and the range of the generalizations
we seek from a given body of data should be regarded as a matter to be decided
by empirical inquiry--not as an a priori limitation upon inquiry. While the

universal assertions of the ninetee    century evolutionists have proven empiri-
cally false, this does not imply that making such assertions or asking such

questions is methodologically undesirable. A methodology which rules out such
questions does so at its own risk; and, one may suggest, limits itself unneces-
sarily. But it is by just this limitation that Steward distinguishes the

methodology of multilinear evolution from other evolutionary schemes. To the

extent that it is nothing more than an a priori limitation upon inquiry, it may
with reason be distrusted as a methodology. But is it really nothing more? A
brief glance at the methodology of multilinear evolution in action may throw
some light upon the matter.

Steward has defined "multilinear evolution"1 as -"a methodology based on the
assumption that si gnificant regular ities in cultural change occur, and . .

concerned with the determination of cultural laws" (1953:318). In the essay,
"Cultural Causality and Law," he sets forth the following program: "'(1)

There must be a typology of cultures, patterns, and institutions. . . (2)

Causal interrelationship of types must be established in sequential,or syn-

chronic terms, or both. . E (3) The formulation of the independent recurrence
of synchronic and/or sequential interrelationships of cultural phenomena is a
scientific statement of cause and effect, regularities, or laws" (1955:180-

181). Due to lack of space the matter of typology cannot be discussed here;

but even assuming the availability of an adequate typology, there are certain
difficulties in Steward's concept of causality which invite examination.

There are, in Steward's tems, two aspects to the formulation of cultural
regularities or laws: (a) the establishment of causal interrelationships of
types, and (b) the recognition of independent (cross-cultural) recurrence of
such relationships. Now it would seem that (b) would necessarily presuppose
(a)--that is, that causal relationships must be established initially before
they can be recognized cross-culturally.  But it is by no means clear how

causal relationships are to be established in the first place. We find, for
example, that in Mesopotamia, settled village agriculture preceded urban

settlement; but can we therefore simply conclude that the former caused the
latter? Not without committing the fallacy of post hoc, ergo propter hoc

And if we cannot establish causal relationships initially, how can we hope to
recognize their cross-cultural recurrence?

The dilemma is not really as serious as this, however. Steward simply
does not start with the determination of causal relationships; he starts,

rather, with the observation of parallel, recurrent cultural sequences (1953:
323; 1955:185-198). He begins, for example, by recognizing the recurrence of
several "eras" (each characterized by a particular set of economic and social
institutions) in the development of early civilization in five major areas of
the world.  The mere recognition of such recurrence of sequential development,
he suggests (1955:199), could be considered as a formulation of cross-cultural

regularity or law. But such a formulation would be superficial, Steward feels,
in failing to provide "a satisfactory and generally valid functional explana-
tion of cause-and-effect relationshipsO v *t. He then proceeds to suggest an
explanation for these recurring sequences. Clearly then, the method employed
is to start with parallel sequential developments, then attempt to construct
a suitable explanation which accounts for all the data. But a curious thing
has happened here:  in the program which Steward has previously outlined, no

mention was made of the need for both formulation of cross-cultural regulari-
ties and explanation of such regularities; and while it would readily be

granted that both are desirable, it is legitimate to wonder which is to be

taken as a statement of cultural law--regularity, or explanation of regularity?
While the point at issue here is a subtle and important one (cf. Popper 1957:
124), it may be assumed that regularities must be established before they can
be explained; and that in Steward's usage "regularity" and "law" are inter-
changeable (Steward 1953:315; 1955:181-185).

Further difficulties in Steward's notion of causality appear in his dis-
cussion of the patrilineal And. Having noted that certain characteristics

were present among a numzber   widely separated primitive societies, he then

succeeded in relating these dparacterlstics functionally; from this he concludes

that they are causally related, and that the patrilineal band is the result of
a (sequentially unspecified) "line of evolution" (Steward 1955:122-142; 1956;
74-75). Granting that certain patterns of subsistence technology, settlement
pattern, and social organization can be seen as functionally related--that is,
as contributing to the persistence of the total socio-cultural system of which
they form a part--one may still ask how it is possible to jump from this to a
statement of causal relationship. Steward asserts that causal relationships
can be synchronic and functional, as well as diachronic and sequential (1955:
180-181).  It seems ambiguous to say that one event can "cause" another event
with which it occurs simultaneously; but to press the matter further would

involve us in a morass of epistemological analysis. The difficulty of deriving
causal relations from functional, synchronic ones is clearly indicated by H. M.
Blalock, Jr. in a recent article in the American Anthropologist (1960:624-631).
Cohen and Nagel (193h:245-249), in a standard handbook on scientific method,
dismiss the whole issue by speaking not of causal relations but of invariant
relations, uniformities, regularities--of order, in short.

To the extent that this questionable notion of causality is an essential

feature of the methodology of multilinear evolution, one would be justified in
questioning the validity of the whole method. But it might be argued that

causality is not essential to any of Stewardts substantive discussions--that

these discussions have as their aim the formulation of significant cross-cul*;
tural regularities; and that in this they are generally successful. With the

latter point I would agree. Let us assume, then, that the cultural laws which
are Steward's aim are to be formulated upon the basis of the cross-cultural

recurrence of regular, invariant relationships between various cultural elements
or patterns., We can say that the "methodology" which Steward advocates is sim-
ply that of searching for such regularities, of formulating such laws. It is

this methodology which Steward calls "multilinear evolution"; it is "multiline-
ar" because there are many such regularities which pertain only to some (not
all) societies; and it is "evolutione because Steward has defined "evolution"
as a quest for cultural regularities. But, after all, one could say that all

scientific endeavor has as its aim the discovery, formulation, and expnation
of regularities--that is, of recurrent invariant relationships (cf. Cohen and
Nagel 1934:245-249, 391-402). Perhaps Steward's methodology is simply the

method of empirical scientific inquiry, applied to culture.   If this is not the
case, then one might wonder why Steward has not bothered to point out the dif-
ference. But if "cultvral evolutione  as method is indistinguishable from

"science" as method, it seems reasonable to suggest that the latter term, more
comprehensive in scope as well as more widely accepted, would be preferable.

What I have tried to suggest may be summarized as follows: first, Steward
does not clearly distinguish between "evolution" as a process to be studied and
"evolutionism" as a body of theory or a point of view.  If we take his notion

of "evolution" to refer to a process, we find that it is distinguished from the
process of "culture change" in a rather misleading way--that is, by the assump-
tion that the process of "evolution" can be studied scientifically, while the
process of "culture change" can be treated only historically and particularis-
tically. This may be true; but if so it is a conclusion which should follow

from empirical inquiry, and can hardly serve as an assumption upon which inquiry
is to be based. If on the other hand we take Steward's notion of "evolution" to
refer not to process but rather to mcethodology ("the methodology of multilinear

evolution") we find, first, that it is distinguished from other "evolutionary"
methodologies (e.g., that of "unilinear evolution") only by the somewhat

arbitrary a priori decision not to seek for generalizations applying to all

societies. Apart from this matter of arbitrary limitation, Steward's method-

ology is unclear on the difficult matter of causality:  first, it is not clear
how statements of causal relationship are to be established or tested; second,
it is not clear how statements of causality are related to statements which

formulate crosscultural regularities and those which explain such regularities.
Finally, if we eliminate from the methodology of multilinear evolution both its
difficult notion of causality and its arbitrary limitations upon inquiry., we

are left with what amounts to the search for invariant relations, unifomities,
regularities, or order in cultural phenomena. If the search for order in

empirical phenomena is, as some philosophers of science have suggested, the

essential characteristic of scientific inquiry, it is difficult to see how the
tmethodology of multilinear evolution" differs from "scientific inquiry into
cultural phenomena."

It is just such inquiry, I think, that Steward means to encourage; it is
undoubtedly well represented in his own substantive work. In constructing a

methodology with which to advance such inquiry, however, Steward makes certain
assumptions which I feel are neither necessary nor fully Justified. It has

been the purpose of this paper to call attention to these questionable points

in Steward's methodology, in the hope that such continued discussion of method-
ology will further the work with which Steward has long been concerned.

ENDNOTE

(1) The writer is indebted to Professor John H. Rowe for critical reading of

this manuscript; but for errors of interpretation the writer alone is
responsible.

BIBLIOGRAPHY
Blalock, H. M., Jr.

1960   Correlational analysis and causal inference.  American Anthropolo-

gist 62:624-631.
Bock, Kenneth

1956   The acceptance of histories:  toward a perspective for social

science.

Cohen, Morris R., and Ernest Nagel

1934   An introduction to logic and scientific method.
Popper, Karl R.

1957   The poverty of historicism.

1959   The logic of scientific discovery.
Steward, Julian H.

1953   Evolution and process.  In Anthropology Today, A. L. Kroeber (ed.),

pp. 313-326.

1955   Theor-y of culture change: the methodology of multilinear evolu-

tion.

1956   Cultural evolution. Scientific American 194:69-800

54~