Method and Theory in American Archaeology

GORDON R. WILLEY and PHILIP PHILLIPS

A rchaeology and

Anthropology meet

in thi&stud^

of the prehi$iQTy\

of the Americas^

P88 $1.50 (U.K. 10/6 n.et»

PLEASE HANDLE WITH CARE

University of Connecticut Libraries

•Q* «^0 f^» «^9 <^»

3 =1153 013523bl D

GAYLORD RG

Digitized by the Internet Archive

in 2011 with funding from

LYRASIS members and Sloan Foundation

http://www.archive.org/details/methodtheoryinam1958will

S < C oO

Method and Theory in American A rchaeology

Method and Theory

BY GORDON R. WILLEY

AND PHILIP PHILLIPS

in American Archaeology

PHOENIX BOOKS

4£fc

THE UNIVERSITY OF CHICAGO PRESS

CHICAGO & LONDON

Library of Congress Catalog Card Number: 57-11215

The University of Chicago Press, Chicago & London The University of Toronto Press, Toronto 5, Canada

© 1958 by The University of Chicago. Published 1958

First Phoenix Edition 1962

Third Impression 1963

Composed and printed by

The University of Chicago Press,

Chicago, Illinois, U.S.A.

To JAMES ALFRED FORD

Preface

In the summer of 1952, as a result of numerous discussions, we de- cided to set down our thoughts on certain methodological and the- oretical questions in American archaeology. The original plan was an article in two parts : the first, a statement of what we believed to be the minimal aims of archaeology and the basic operations di- rected toward the achievement of these aims, and the second, some theoretical formulations about New World prehistory. The first part was subsequently published under the title "Method and Theory in American Archaeology: An Operational Basis for Cul- ture-Historical Integration."1 The second part followed a year and a half later as "Method and Theory in American Archaeology II: Historical-Developmental Interpretation."2 The comments and criticism which these papers drew from colleagues and students have kept us interested in the subject, and, as a result, we have re- written both original papers and combined them, along with an introduction, originally published as a brief journal article,3 in the present volume.

A good many of the revisions and additions we have made to the original papers are the result of second thoughts, which we trust are better than the first ones. Others, by no means the least sub- stantial, are the direct result of critical comments and suggestions on the part of colleagues in archaeology and anthropology. To list them all would provide a roster of impressive proportions. We would, however, like to single out Albert C. Spaulding and Irving Rouse for the time, interest, and advice they have expended in our behalf.

1. Phillips and Willey, 1953.

2. Willey and Phillips, 1955.

3. Phillips, 1955.

Vll

Finally, a word about the restrictive connotation of the qualify- ing term "American" as used in our title and throughout the book. Obviously, the methods, theories, and ideas in general which are propounded and discussed in this work are not limited to the Ameri- cas any more than archaeology is so limited. Many, or most, of them have originated in the Old World with Americanists as late borrowers. We have used the qualification inasmuch as our own experience is in the American field and all the examples and subject matter are so confined. There is another reason for using the term. In no other large part of the world does archaeology stand so com- pletely on its own feet as in the New World. The historic con- tinuities and documentation binding past to present are infinitely weaker here than in the European, Middle Eastern, and Asiatic areas where archaeology has been carried forward. American archaeology complements, but is in no sense an adjunct of, history; hence its methodology stands in somewhat sharper relief than the methodology in many parts of the Old World. This is not a denial of the vital importance of historical and ethnological data in inter- preting the American past, but such considerations lie outside the scope of this book.

Gordon R. Willey Philip Phillips

vm

Contents

Introduction: American Archaeology and General

Anthropological Theory /

PART I. AN OPERATIONAL BASIS FOR CULTURE- HISTORICAL INTEGRATION

1, Archaeological Unit Concepts 11

2. Archaeological Integration 44

Summary 57

PART II. HISTORICAL-DEVELOPMENTAL INTERPRETATION

3. The Historical-Developmental Approach

in American Archaeology 61

4. Lithic Stage	79
5. Archaic Stage	104
6. Formative Stage	144
7. Classic Stage	182
8. Postclassic Stage	193
Summary	200
BIBLIOGRAPHY
Bibliography	209

Index 259

IX

Introduction

American Archaeology and General Anthropological Theory

It has been said that archaeology, while providing data and general- izations in such fields as history and general anthropology, lacks a systematic body of concepts and premises constituting archaeological theory. According to this view, the archaeologist must borrow his theoretical underpinning from the field of study his work happens to serve, or do without. Whether the latter alternative be an admis- sible one does not seem to be an arguable point. Acceptable field work can perhaps be done in a theoretical vacuum, but integration and interpretation without theory are inconceivable.

The above remarks apply to archaeology in general, but the sole concern of this study is American archaeology. It seems to us that American archaeology stands in a particularly close and, so far as theory is concerned, dependent relationship to anthropology. Its service to history in the narrower sense, i.e., as the record of events in the past with the interest centered on those events, is extremely limited, because for pre-Columbian America there is in effect no such history. The use of traditions derived from native informants and other documentary sources of the contact period as starting points for pushing back into the unrecorded past — the "direct his- torical approach" — is not archaeology serving history, but the re- verse. As a technique of investigation, American archaeology, like archaeology generally, provides useful data for geology, paleon- tology, climatology, etc., and it recovers valuable material for art museums and the study of aesthetics, but it is not involved theo-

INTRODUCTION

retically with any of these subjects. To paraphrase Maitland's famous dictum: American archaeology is anthropology or it is nothing. The American archaeologist, unless he thinks he can dis- pense with theory altogether, is therefore obliged to take a stand on some of the basic questions of general anthropological theory. This we shall do briefly in the following pages.

The methods outlined in this study, and our arguments in their behalf, are predicated on two general theoretical assumptions: (1) that anthropology is more science than history, and (2) that the subject matter of anthropology is both society and culture. The first part of this statement appears to settle out of hand the position of anthropology in respect to the dichotomy science-history, a question that has vexed philosophers ever since the emergence of anthropology as a field of study. It seems to us that the force of this antithesis is largely spent. There is now considerable agree- ment among theorists that the world of anthropology is a mixture of recurrent and unique events acting and reacting upon each other in a tremendously complex fashion. The only serious disagreements are in respect to the role and importance of the two components of the mixture. Our view is that the part played by recurrent events, though it may be the smaller, is the more significant; and that this is just as true for an archaeology devoted to the service of anthro- pology as it is for anthropology itself. Archaeology, in the service of anthropology, concerns itself necessarily with the nature and position of unique events in space and time but has for its ultimate purpose the discovery of regularities that are in a sense spaceless and timeless.1 And, since it appears that a comparative method will be most likely to disclose such regularities, it follows that the archaeologist is faced with the responsibility of finding, in the seemingly endless flow of cultural and social events, forms and

1. This we hope will not be taken to mean that the events referred to take place outside space and time. In this and all subsequent references to space and time in this study, it is of course geographical space and chronological time that are denoted.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

systems of forms that are not only comparable to each other but also comparable to, or at least compatible with, the forms and sys- tems of forms of cultural or social anthropology.2 We shall return to this point later.

The second article of belief referred to above is that the subject matter of anthropology is both society and culture, another polarity that is not standing up under analysis. The interpenetration of social and cultural facts now seems to be taken as axiomatic. Following Kroeber and others, we have chosen to regard them here as aspects of the same basic reality. Definition of this basic reality is fortu- nately outside the scope of the present inquiry. It is sufficient for our purposes to characterize it loosely as patterned human be- havior. Archaeology, of necessity, deals very largely with pat- terned human behavior in its cultural aspect. In American archaeol- ogy especially, we have tended to suppress the social aspect alto- gether. Some Americanists have been drawn into the extreme posi- tion that sees in culture an independent order of phenomena, in- telligible in terms of itself alone — the "cultural superorganic." Most of us, without subscribing to the superorganic view of cul- ture, have nevertheless operated "as if" it were a fact. In our opin- ion even this moderate position, though operationally expedient and to a certain extent inevitable, is ultimately detrimental to the main task of archaeology, which is to organize its data in terms of a real world, a world in which cultural and social phenomena (to name only these) are inextricably mingled.

The reader will have noted by this time that we are driving toward an accommodation between the seemingly opposed methods and outlook of archaeology and cultural anthropology. Compari-

2. "Social anthropology" in England, "cultural anthropology" in the United States — these are not precise equivalents but are closer, it seems to us, than practitioners in the two countries appear to believe. From the detached point of view of the archaeologist, at any rate, they are practically synonymous terms. In this study we follow the American usage but without any convic- tions regarding the predominance of the cultural over the social aspect of our subject matter.

INTRODUCTION

son may be facilitated by considering the operations of the two disciplines on three levels of organization that are generally appli- cable to all scientific analysis: observation, description, and explana- tion. The accompanying diagram is a crude attempt to show how the operations of archaeology and cultural anthropology can be con- sidered as converging toward a synthesis from one level to the next.

Explanation Description Observation

Processual interpretation

Culture-historical integration

Field work

Archaeology

Ethnology

Ethnography

Field work Cultural Anthropology

On the observational level, archaeological and cultural anthro- pological field work are placed far apart on the diagram because of wide differences in the phenomena observed. These differences, however, can be too easily overemphasized. Cultural anthropology observes group behavior and the products of group behavior in their twofold aspects, social and cultural. Its primary concern is with the social aspect, but certain categories of behavior, notably those which are symbolized in language, art, myth, etc., may be studied very largely in their cultural aspect. Archaeology observes primarily the materialized products of group behavior but has con- siderable opportunity to observe symbolized behavior in the forms of art, iconography, and (rarely) written language, and occasional- ly touches social behavior through inferences, as in the interpreta- tion of burial practices, house plans, settlement patterns, roads, ir- rigation systems, and the like. Thus it appears that the raw mate- rials of the two disciplines are not so different after all; what is dif- ferent is that archaeology is obliged to view its material almost en- tirely in the cultural aspect. It has sometimes attempted to turn this limitation into an asset by embracing the cultural superorganic, as already noted.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

The term "culture-historical integration," as used here, covers almost everything the archaeologist does in the way of organizing his primary data : typology, taxonomy, formulation of archaeologi- cal "units," investigation of their relationships in the contexts of function and natural environment, and determination of their inter- nal dimensions and external relationships in space and time. How- ever high-sounding these terms, it appears that the activities repre- sented by them remain essentially on the descriptive level. Explana- tory concepts, such as acculturation, diffusion, and stimulus diffu- sion, are utilized, but the aim is primarily to describe what hap- pened to specific cultural units at specific times and places; no at- tempt is made (on this level) to draw generalizations from these observations and descriptions. Culture-historical integration is thus comparable to ethnography with the time dimension added, but we dare not push this analogy too far, because the archaeologist's de- scriptive formulations, like his observations, lie mainly in the cul- tural aspect of his subject matter. Later in this book we make a plea for unit concepts that are intelligible in the social aspect as well, but we are under no illusion that any except the very smallest of them can be precisely equated with correspondent units of social structure. Nevertheless, we have placed culture-historical integra- tion and ethnography closer together on the diagram than their re- spective field operations, in the belief that archaeological unit con- cepts can and should make more sense in terms of the social aspect than is generally supposed.

So little work has been done in American archaeology on the explanatory level that it is difficult to find a name for it. It might have been left blank on the diagram to emphasize this lack. The term "functional interpretation," which has gained a certain amount of currency in American studies, was used in the original version of this diagram but is not entirely satisfactory, since it implies that the functional is the only explanatory principle involved. We have substituted here the broader "processual interpretation," which might conceivably cover any explanatory principle that might be invoked. In the context of archaeology, processual interpretation

INTRODUCTION

is the study of the nature of what is vaguely referred to as the cul- ture-historical process. Practically speaking, it implies an attempt to discover regularities in the relationships given by the methods of culture-historical integration. Whatever we choose to call it, the important consideration is that, on this explanatory level of organization where we are no longer asking merely what but also how and even why, our formulations must be viewed in both their cultural and their social aspects.3 It is not possible to go about in- vestigating culture-historical processes and causality without ref- erence to the efficient causes of cultural change, which are people or groups of people, and therefore lie in the social aspect of reality. Perhaps it is fair to say that there has been a lack of progress in processual interpretation in American archaeology to date precisely because unit formulations have been put together with so little reference to their social aspect. In the same vein of optimism al- ready displayed, we have put processual interpretation and ethnol- ogy (which includes among its many meanings the operations of cultural anthropology on the explanatory level) side by side on the diagram to suggest a further convergence of aims, if not of prac- tice. At this point, the archaeologist is in effect a cultural anthropol- ogist,4 but it is well to remember that his activities on this level are conditioned by his formulations on the descriptive level and that

3. To name only two of the important factors in a complex equation. Geo- graphical and ecological factors, already present on the descriptive level, carry over with increased importance onto the explanatory level. These have been deliberately ignored in our diagram, which is focused on the special rela- tionships between the cultural and social aspects of anthropology. The same neglect of physiological and psychological factors should be noted.

4. This point has been very well put by Walter Taylor, who also rational- izes the operations of archaeology on a series of levels that differ in detail from ours but can be reconciled with them, as in the following passage: "When the archaeologist collects his data [observational level], constructs his cultural con- texts [descriptive level] and on the basis of these contexts proceeds to make a comparative study of the nature and workings of culture in its formal, func- tional, and/ or developmental aspects [explanatory level], then he is 'doing' cultural anthropology and can be considered an anthropologist who works in archaeological materials" (1948, p. 43).

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

these in turn have special characteristics which it is our purpose to describe.

Diagrams and models have the happy faculty of proving what- ever they are designed to prove, and ours is no exception. Never- theless, we think that this model, in spite of the crude simplification inherent in any system of "levels," represents a pattern that is not wholly fictitious. As archaeology, in the service of anthropology, moves from one operational level to the next, it is compelled to pay more attention to the social aspect of its subject matter, until there takes place on the explanatory level an actual convergence with cultural anthropology and the possibility of an eventual synthesis in a common search for sociocultural causality and law.

INTRODUCTION

Part I

An Operational Basis for Culture-Historical Integration

Chapter 1

Archaeological Unit Concepts

"Culture-historical integration" is the term we have chosen to designate what we regard as the primary task of archaeology on the descriptive level of organization. The procedural objectives of cul- ture-historical integration have tended to be divided, in theoretical writings on American archaeology, between the reconstruction of spatial-temporal relationships, on the one hand, and what may be called contextual relationships, on the other.1 Operationally, neither is attainable without the other. The reconstruction of meaningful human history needs both structure and content.

Cultural forms may be plotted to demonstrate geographical con- tinuity and contemporaneity, but, when we move to establish his- torical relationships between them, we immediately invoke proc- esses like diffusion, trade, conquest, or migration and in so doing shift the problem from the bare frame of space and time into the realm of context and function. Conversely, the processes named have no historical applicability without control of the spatial and temporal media in which they operate. Taylor was undoubtedly correct in stating that American archaeologists have placed heavy

1. Taylor, in the work already cited (1948), puts these procedures on two distinct levels of interpretation, which he calls "chronicle" and "historiogra- phy." See also Willey's (1953a) use of the terms "historical" and "proces- sual." The latter term was used by Willey in reference to the description of the way in which specific cultures function in specific times and places, not as we are using it in the present study in reference to the attempt to draw general- izations from culture-historical data. All four terms, "chronicle," "historiogra- phy," "historical," and "processual," in the writings cited, refer to operations on the descriptive level of organization as defined in the present study.

ARCHAEOLOGICAL UNIT CONCEPTS

11

emphasis on the skeletal chronicle at the expense of the recovery of what he calls "cultural context," but a review of the recent lit- erature indicates a strong trend in the contrary direction. We sub- mit that this is now an area of agreement for American archaeol- ogy : culture-historical integration is both the spatial and temporal scales and the content and relationships which they measure. The essence of this study's departure, if it may be called a departure, is that these objectives are not regarded as being on different and unequally significant levels of interpretation or as even being capable of effective separation operationally. It seems to us that the apprehen- sion and formulation of archaeological unit concepts involve the simultaneous investigation of contextual and spatial-temporal rela- tionships.

A method basic to archaeology on the descriptive level is taxon- omy. Under this general heading, the archaeologist deals with two sorts of concepts: types, and cultures or, as we prefer to say, archaeological units.2 The former usually pertain to artifacts or other products of technology but may be used in connection with other categories of cultural behavior such as burial types. Some archaeologists also apply the type concept to full archaeological assemblages, or units as we would call them, using the designation "culture type." While this usage has the apparent virtue of econo- my through elimination of the conceptual difference between type and unit, we prefer to keep them distinct, because we think that unit concepts have certain characteristics not shared by artifact types and that these in turn have important methodological bear- ings. Our interest here is centered on unit concepts, but a few re-

2. There is a good deal to be said for and against the prevailing use of "cul- ture" to denote every conceivable kind of archaeological unit. It is very con- venient in many cases where we do not really know what kind of unit we are dealing with, but it is conducive to sloppy thinking. We shall be trying here to confine its use to units of a certain magnitude, maximum units in fact, but are not enthusiastic about the cumbersome "archaeological unit" as an alternative. When the context is clearly archaeological, perhaps the term "unit" will suffice.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

12

marks on the concept of type as applied to artifacts will serve to get us into the subject.3

There is, happily, a general working agreement among American archaeologists about what constitutes an artifact type, though there is still some dispute about what it signifies in terms of the basic reality that we have postulated as the subject matter of archae- ology. The principal difference of opinion may be crudely stated as opposition between those who believe that types are arbitrarily "designed" by the classifier and those who think that types exist in nature and that the classifier "discovers" them. According to the first view, types are simply analytical tools that are to be judged solely on the basis of their usefulness; the second maintains that they have, or should have, behavioral reality in the sense that they would be recognized as norms, the "right way," in the societies that produced the objects being typed. Our attitude is that these opposing views are not completely antagonistic. We maintain that all types are likely to possess some degree of correspondence to this kind of reality and that increase of such correspondence must be the constant aim of typology.4 The actual procedure of segregat- ing types is therefore a more complex operation than is suggested simply by such words as "design" or "discovery," and is in effect a painstaking combination of both.

No less laborious than artifact typology are the procedures in- volved in the formation of archaeological units, in which we have to consider not only the relationship of forms as such but their spatial and temporal relationships as well. Disregarding the latter for the moment, we may begin our consideration of the nature of

3. Another valid objection to the use of "culture type" as a synonym for "archaeological unit" is that we need such a generalizing term when we want to talk about a type of culture without reference to any specific time or place. A "theocratic irrigation state" (Steward et al., 1955, p. 65), for example, is a culture type that may in theory occur wherever the necessary cultural and natural preconditions are present.

4. Cf. Phillips, Ford, and Griffin, 1951, pp. 63-64, for a discussion of the problem of the "empirical" versus the "cultural" type as applied to ceramics. Those inclined to take sides on this question are advised to see Evans, 1954; Ford, 1954a, 1954£, 1954c; and Spaulding, 1953a, 1953£, 1954a, 1954£.

ARCHAEOLOGICAL UNIT CONCEPTS

13

unit concepts by examining their strictly formal characteristics. We are seeking a broad definition that will cover all sorts of units, from those whose content may consist of a very small number of highly specialized forms to units represented by the fullest com- pendium of cultural data that can be recovered by the most refined techniques of investigation and the most imaginative interpretation. It is plain, therefore, that the size or comprehensiveness of the as- semblage is no criterion. What is required rather is that the con- stituent forms be physiognomic, recurrent, and internally consistent. To borrow a phrase from V. Gordon Childe, they must relate to one another in a way that permits us to assume them to be "the concrete expressions of the common social traditions that bind to- gether a people."6

So much for the formal content of an archaeological unit of whatever magnitude. Now the particular interests of archaeology also require that this content be placed in the contexts of geo- graphical space and time. It is often maintained that this is some- thing the archaeologist does after the unit has been defined. How- ever sensible in theory this may appear, in practice it does not work that way. The working procedure, as every archaeologist knows, is initial formulation, investigation of spatial and temporal dimensions, reformulation, reinvestigation of spatial and temporal dimensions, and so on indefinitely. The operation is immensely complicated by the fact that the fixing of these internal dimensions is, more often than not, dependent on external relationships. We have only to recall certain essential conditions of our unit's exist- ence. In the same place, but before or after it in time, were similar units whose contents intergrade with its content in so-called transi- tional periods that are almost impossible to establish with precision; beside it in space were other contemporaneous units with similar intergrading at frontiers equally difficult to draw. Yet we maintain that the fixing of these spatial and temporal boundaries, however difficult, is an essential part of the definition of the unit. Small wonder that the unavoidably arbitrary nature of such operations

5. Childe, 1950, p. 2.

METHOD AND THEORY ~IA IN AMERICAN ARCHAEOLOGY ll

has led many archaeologists to believe that an archaeological unit is nothing more than a fragment excised from a spatial-temporal continuum, a fragment that could not be said to have existed as a unit before it was named and defined.

The same problem confronts us here that we have already dis- cussed in connection with artifact typology. An archaeological unit, as described above, may appear to be a rational construct in terms of the observed facts of cultural continuity and cultural relation- ships, but what are the chances that it corresponds in any real sense to an intelligible unit of culture-history? In our original paper we took the position that such correspondences may eventually be possible but that "the archaeologist is on firmer footing at present with the conception of an archaeological culture as an arbitrarily defined segment of the total continuum." In a long and exceedingly astringent letter to the authors, Albert C. Spaulding attacked this position with such force that we have obtained his permission to quote a portion of it here.

It is true that any assemblage represents a segment of a continuous stream of cultural tradition extending back into time, but once you grant the purpose of a scientific exposition of culture history, the process of classifying the time stream with respect to its cultural characteristics is anything but arbitrary. It should be classified in terms of events which are themselves associated with a cluster of other new events so as to yield a succession of distinct culture types. A possible exception to this generali- zation would be a segment of a continuum in which culture change did not occur, or in which change occurred at a uniform rate so that it could be represented as a straight line on a graph. In the hypothetical case of no change, classificatory subdivision would serve no useful purpose that I can think of and would be arbitrary; in the case of a slanting straight line, sub- division would in a sense also be arbitrary, since a complete description would require no more than a statement of the point of origin, slope, and point of termination (if any) of the line. In fact, the two cases are basically the same case. However, widely accepted cultural theory indicates that the normal pattern is one of relative stability, then rapid growth through the introduction of a critical new element followed very quickly by a num- ber of other new elements, then a period of relative stability, and so on. Plotted on a time against culture change graph, this results in an ogival growth curve, and the recognition of the sharp curves is a scientific obliga-

ARCHAEOLOGICAL UNIT CONCEPTS

15

tion of the archaeologist. The segments so recognized are certainly not the result of arbitrary classification; the changes in slope of the line are just as characteristic of it as is its continuous nature. So time itself is con- tinuous and proceeds at an unvarying rate, but culture change in relation to time probably never proceeds at an unvarying rate, and useful archaeo- logical classifications of chronology are those which have sharp rate changes as their limiting points. A good chronological classification yields a number of periods, each of which is characterized by a distinctive "cul- ture" ("culture type" in my terminology), and the proper job is to pin- point the critical element so that the time of its invention can be made to serve as the starting point of the new period.

To consider a second implication of "space-time-cultural" continuum, space itself is a continuum, but culture is not uniformly distributed in that continuum. To get off a lofty crusher first, there are no archaeological sites in the middle of the Atlantic Ocean, either on the surface or on the bottom. On a more sensible plane, archaeological materials characteristically occur in discrete clusters [we call them "sites"] in space, and the task of describ- ing them includes discovery of their boundaries — there isn't any arbitrary slicing up of a uniform continuum involved in the problem. To turn now to what I think you actually have in mind, it is quite conceivable that the cultural characteristics of components might vary uniformly or continu- ously (i.e. in infinitely small steps, or at a uniform slope, so that the spatial position against cultural content graph, time controlled, shows a curve or straight line instead of a noticeably jagged line), but I am uncertain as to how important this is in fact. I think that ethnographic evidence strongly indicates that if any considerable number of components over any sub- stantial area were considered, a classification of culture type with respect to space would reveal fairly neat clustering, not simple polarity or radia- tion. Certainly you would concede this if sharp ecological boundaries were involved. Discovery of clustering at the multicomponent level would elimi- nate the claims of arbitrariness effectively. In summary, then, your vague and sweeping talk about arbitrary slicing up of a continuum results in the confounding of several important cultural problems and leads to methodo- logical obfuscation, not clarification.

Without yielding entirely before Spaulding's assault, we are prepared to admit that the assumption of a more or less unvarying rate of cultural change in a spatial-temporal continuum has been overdone, by us as well as by others. We now prefer to say that the archaeologist is on a firmer footing with the concept of an ar- chaeological unit as a provisionally defined segment of the total

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

16

continuum, whose ultimate validation will depend on the degree to which its internal spatial and temporal dimensions can be shown to coincide with significant variations in the nature and rate of cul- tural change in that continuum. In simpler terms, it is the archaeolo- gist's job to be aware of the arbitrary nature of his unit concepts and, at the same time, alert to the possibility of making them less arbitrary.

Absorbing as this question of cultural continuity and change must be to all archaeologists, it is really not an issue in the present discussion. Our object here is to show that a fundamental, unvary- ing characteristic of all archaeological unit formulations of what- ever magnitude is that they are arrived at by combining three sorts of data: formal content, distribution in geographical space, and duration in time. These three ingredients are present, though not always explicit, in all unit concepts but may differ significantly in the part they play in the formulation. Variations in content, general- ly but not always a function of the limits of available information and the archaeologist's ability to draw inferences therefrom, need not concern us here. Our interest centers rather on variations in spatial and temporal dimensions and particularly on the fact that there is no regular (or constant) relationship between variations in these two dimensions. Probably a large share of our classificatory difficulties and the ensuing arguments could be avoided by the gen- eral recognition of this fact. It becomes essential, therefore, in the definition and use of archaeological unit concepts of whatever na- ture to understand precisely what quantities of space and time are involved in the formulation.

In the search for practicable units of study, American archaeolo- gists have invented a large number of unit concepts and designated them by an even larger number of names. Our initial requirement, therefore, is a comprehensive nomenclature by means of which existing schemes and their working parts can be roughly equated. We have committed ourselves to the proposition that all unit con- cepts, whatever the outlook and intentions of their originators, have implicit spatial and temporal dimensions. Therefore, they can

ARCHAEOLOGICAL UNIT CONCEPTS

17

be differentiated by reference to the amounts of space and time they are thought to involve. If these premises be granted, it seems a reasonable approach to bring one of these two variables under control before considering it in combination with the other. The spatial factor is easier to deal with; so we may begin by setting forth a series of geographical categories that we have found useful in this connection.

SPATIAL DIVISIONS

A site is the smallest unit of space dealt with by the archaeologist and the most difficult to define. Its physical limits, which may vary from a few square yards to as many square miles, are often impos- sible to fix. About the only requirement ordinarily demanded of the site is that it be fairly continuously covered by remains of for- mer occupation, and the general idea is that these pertain to a single unit of settlement, which may be anything from a small camp to a large city. Upon excavation, of course, it rarely turns out to be that simple. The site is the basic unit for stratigraphic studies; it is an almost certain assumption that cultural changes here can only be the result of the passage of time. It is in effect the minimum opera- tional unit of geographical space.

A locality is a slightly larger spatial unit, varying in size from a single site to a district of uncertain dimensions; it is generally not larger than the space that might be occupied by a single community or local group. It is hardly necessary to add that such limits as are implied in this qualification have the variability found in the size and settlement patterns of local groups from one sort of society to another. In strictly archaeological terms, the locality is a geo- graphical space small enough to permit the working assumption of complete cultural homogeneity at any given time. This is not to say that two or more discrete archaeological units might not, under special conditions, simultaneously occupy the same locality, or even the same site. For example, it has long been thought that people carrying an intrusive Salado culture occupied certain sites of the Gila-Salt region of south-central Arizona and coexisted

METHOD AND THEORV IN AMERICAN ARCHAEOLOGY

18

peacefully with the indigenous Hohokam populations.6 If this is the correct interpretation of the archaeology of the sites in ques- tion, it reflects a rather uncommon situation.

Examples of localities well known in the literature of American archaeology are given below in connection with the discussion of local sequences.

A region is a considerably larger unit of geographical space usual- ly determined by the vagaries of archaeological history. Quite often it is simply the result of concentrated research by an individu- al or group. Rightly or wrongly, such a region comes to be thought of as having problems of its own that set it apart from other regions. Regional terms are those most often found in the titles of archaeo- logical papers of wider scope than site reports. Through constant reiteration they become fixed in the literature and achieve a kind of independent existence. Regions are not altogether without refer- ence to the facts of geography, however. In stressing the accidental factor in their formation, we must not overlook the tendency for environmental considerations to assert themselves. In portions of the New World where physical conditions of sharp diversity pre- vail, archaeological regions are likely to coincide with minor physiographic subdivisions. An excellent example is furnished by the Glades region comprising the southernmost portion of the Florida peninsula. Here the relationship between culture and a highly characterized environment has been particularly close throughout the entire span of the archaeological record.7 Many other similar examples could be cited, and the effect would be to show that, of the various spatial units considered here, the region offers the most favorable field for the detailed study of the rela- tionships between culture and environment.

In terms of the social aspect of culture — and here we must tread warily— the region is roughly equivalent to the space that might be occupied by a social unit larger than the community, a unit to which we may with extreme trepidation apply the term "tribe" or

6. Gladwin et al., 1937, p. 13.

7. Goggin, 1949, pp. 28-32.

ARCHAEOLOGICAL UNIT CONCEPTS

19

"society." This rough equation is based on what we know of Amer- ican tribal distributions in early historic times and must be accorded the same flexibility that we see in the size of those distributions. The same caution is required in attempting to characterize the re- gion in terms of the cultural aspect. Generally speaking, it is a geographical space in which, at a given time, a high degree of cul- tural homogeneity may be expected but not counted on. As we shall see later, it is quite possible for more than one archaeological phase to occupy a region at the same time.

An area is a geographical unit very considerably larger than a region; it corresponds roughly to the culture area of the ethnogra- pher. Archaeological areas, like regions, have come into existence by common consent, but the element of historical accident is re- duced somewhat by the fact that many individuals and institutions are likely to have been involved in their investigation. They tend to coincide with major physiographic divisions. That the North American Southwest, for example, has maintained its identity as an archaeological area through more than a half-century of intensive investigation is certainly due in large part to culture-environment correlations of a positive nature. It is hardly necessary to add that, although the area as defined here may have general physiographic integrity, its limits are not so easy to draw on a map as those of the smaller region. The southeastern United States is a case in point; it has to be defined afresh every time anyone writes about it. The problem is a familiar one in culture-area studies.

It often happens that there are territories of geographical extent intermediate between the region and the area which possess quali- ties and degrees of cultural unity that give them a definite useful- ness in archaeological or ethnographical studies. We refer to such spatial units as subareas. A case in point might be taken with refer- ence to Middle America (or Mesoamerica) , which has been defined quite specifically as a culture area.8 Embracing southern Mexico and upper Central America, Mesoamerica has a distinctive culture content and patterning; yet within this sphere of common cultural

8. Kirchhoff, 1943.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

20

likenesses there are differences of traditionally recognized signifi- cance. As but one example, the cultures of the lowland Maya coun- try stand in sharp contrast to those that are adjacent to them. Their art, architecture, ceramics, calendrics, and writing bind the Maya lowland territory into an obvious unit of history. At the same time, these Maya lowlands are too complex and diversified to be or- ganized as a region, as that cultural-geographical term is defined here. Workable Maya regions might be the Peten, the Usumacinta, the Motagua-Chamelecon, or the several divisions of Yucatan, etc. In consequence, the useful term "Maya lowlands" signifies a sub- area.

BASIC ARCHAEOLOGICAL UNITS

The concepts about to be discussed under this heading were re- ferred to in our original paper as "formal or content units" to ex- press the fact that the element of content is more important in their formulation than the spatial and temporal dimensions. Use of such a term, however, suggests that they are different in kind from what we later describe as "integrative" and "maximum" units, which is not strictly true. It seems preferable, therefore, to call them simply "basic units," which indeed they are.

The component, a useful term which has achieved nearly univer- sal currency in eastern North American archaeology, has been de- fined by W. C. McKern as the manifestation of a given archaeologi- cal "focus" at a specific site.9 Strictly speaking, in the McKern system, the component is not a taxonomic unit. In theory the basic unit of classification is the focus, comprising a number of com- ponents, and the same may be said of what we designate as a "phase." It is a working assumption that no phase worthy of the

9. "The manifestation of any given focus at a specific site is termed a com- ponent of that focus. This is in no sense an additional type of culture manifes- tation, one of the five class types; rather, it is the focus as represented at a site, and serves to distinguish between a site, which may bear evidence of sev- eral cultural occupations, each foreign to the other, and a single specified manifestation at a site. In many instances several components, each at cultural variance with the other, may be found to occur at a single site" (McKern, 1939, p. 308).

ARCHAEOLOGICAL UNIT CONCEPTS

21

name will fail to manifest itself in more than one component. In practice, of course, it often happens that a phase is initially defined on the strength of a single component, i.e., a site or a level within a site, but the expectation is implicit that other components will be found and the original definition modified accordingly. It will be noted later, however, in the discussion of the social implications of the phase, that it is theoretically and actually possible for a phase to consist of a single component.

The phase is, in our opinion, the practicable and intelligible unit of archaeological study. Choice of the term accords with prevailing usage in a preponderance of New World areas, including the South- west, most of South America, and all of Middle America. Kidder has defined the phase in the following terms: "A cultural complex possessing traits sufficiently characteristic to distinguish it for pur- poses of preliminary archaeological classification, from earlier and later manifestations of the cultural development of which it formed a part, and from other contemporaneous complexes."10

Like Kidder, we prefer "phase" to the approximately equiva- lent "focus" because of its stronger temporal implication. The emphasis cannot be placed entirely on time, however. Modifying Kidder's definition slightly, we would prefer to describe the con- cept as an archaeological unit possessing traits sufficiently characteristic to distinguish it from all other units similarly conceived, whether of the same or other cultures or civilizations, spatially limited to the order of magnitude of a locality or region and chronologically limited to a rela- tively brief interval of time. It must be acknowledged that this defini- tion gives a specious impression of uniformity. It would be fine if phases could be standardized as to the amount of space and time they occupy. Unfortunately, there are so many variable conditions entering into the formulation that it is neither possible nor desirable to define the scope except within rather broad limits. A phase may be anything from a thin level in a site reflecting no more than a brief encampment to a prolonged occupation of a large number of sites distributed over a region of very elastic proportions.

10. Kidder, Jennings, and Shook, 1946, p. 9.

METHOD AND THEORY ~J ~\ IN AMERICAN ARCHAEOLOGY Z.Z.

It will be noted that Kidder's definition of phase lays more em- phasis on cultural continuity than ours, since it implies necessary relations to what goes before and what comes after. We have freed it of this requirement to provide for the many instances in which we simply do not know what goes before or comes after, or for those less frequent occasions when a new phase appears as an in- trusion without apparent relationship to any precedent continuity. In any case, whether as an instance of continuity or of discontinuity, the phase most often appears as one member of a series that will be referred to hereinafter as a "local" or "regional sequence." These terms will be defined presently, but let us first examine a little more closely the spatial and temporal implications of this basic archaeo- logical unit.

We have already alluded briefly to the impossibility of close de- limitation of the phase in respect to the dimensions of time and space. It may help to clarify the problem to consider it in relation to various levels of cultural development. In Part II of this book we will describe a developmental scheme for New World archaeology with five hypothetical stages : Lithic, Archaic, Formative, Classic and Postclassic. It is not necessary to anticipate the definitions of these stages to point out here that the spatial and temporal dimen- sions of phases are not likely to be the same on all stages. For ex- ample, in the Lithic stage, in which a migratory, hunting-gathering economy is postulated, phases can be expected to occupy more geographical space than in the sedentary Formative stage. There is no regular reduction from stage to stage, however; in the Classic and Postclassic stages the spatial dimensions of phases may also be larger than in the Formative but for a different reason: the socio- political groups are larger. Temporal dimensions, on the other hand, may actually exhibit a regular diminution from stage to stage, if the commonly held assumption is corjrect that the rate of cultural change accelerates with increased advancement and complexity. Without elaborating this point or further refining the definition, we wish simply to emphasize that the concept of phase has no ap- propriate scale independent of the cultural situation in which it is

ARCHAEOLOGICAL UNIT CONCEPTS

23

applied. This is not so great a deficiency as might appear. Looked at internally, so to speak, phases may have very considerable and highly variable spatial and temporal dimensions; looked at from the total range of New World culture-history, they are very small quantities indeed, and it is from this point of view that they assume a sort of rough equivalence, enabling us to use the concept of phase as an operational tool regardless of the developmental stage in- volved.

As typological and stratigraphic analyses become more refined, it often becomes desirable to subdivide phases into smaller (primari- ly temporal) units, and it seems best to regard these as sub-phases and to give them numbers instead of names. It also sometimes hap- pens that two or more phases in the same locality or region, origi- nally set up as independent units, subsequently appear to be more intelligible as subphases of a single unit, though they continue to be operationally useful in sequences and area correlations. It is clearly impossible to lay down any precise rules governing the formation of subphases. In general, their use seems appropriate in cases where differences apply only to a few specific items of content or where such differences are expressible only in variations in frequency. In other words, if it is impossible to present a sensible account of the culture of a unit except in terms of what went before or came after, it is probably better regarded as a subphase. It is hardly necessary to add that subphases and components are entirely different kinds of subdivisions of the phase.

TEMPORAL SERIES

A local sequence in its purest form is a series of components found in vertical stratigraphic succession in a single site. It may also, how- ever, be a composite series made by combining shorter stratigraphic "runs" from various portions of a site or from several sites within a locality, or it may be derived from seriating components by vari- ous means without benefit of stratigraphy at all. However ob- tained, the local sequence has this important feature: it is local. The spatial dimension, by definition, is small enough to permit the

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

24

working assumption that differences between components reflect differences in time.

We have already referred to the fact that members of a local se- quence, though technically regarded as components, are often re- ferred to as phases on the ground that they are local manifestations of the larger units and, also, that it is theoretically possible for a phase to be represented by a single component. The local sequence may, therefore, be defined as a chronological series of components, phases, or subphases, within the geographical limits of a locality as de- fined in this study.

Local sequences, which are the very stuff of archaeology, abound. A famous example of the kind of sequence based upon a single site is at the ceremonial Maya center of Uaxactun, in the Guatemalan Peten. The Uaxactun sequence was determined by combinations of "dirt" and architectural stratigraphy revealed in long and intensive excavations.11 The sequence runs the gamut of the Middle American Formative and Classic stages and probably represents some two thousand years of more or less continuous human occupation. Continuity throughout all phases of the sequence is provided by ceramics which, in spite of several marked innova- tions, show a recognizable local evolution. The earliest Formative stage phases are the Mamom and Chicanel, in that order, and these are followed by Matzanel, generally considered Proto-Classic. The Classic phases, well represented in temple and palace construc- tion, are the Tzakol and Tepeu. R. E. Smith, who defined and named these phases, has recently subdivided the latter two into subphases: Tzakol 1, 2, 3, and Tepeu 1, 2, 3.12 A comparable, but slightly more complex, local sequence is the one at Kaminaljuyu in the Guatemalan highlands, near Guatemala City. This is an exten- sive ceremonial and dwelling center, and a long Formative-through- Classic sequence of Maya occupation and building was worked out

11. Ricketson and Ricketson, 1937; A. L. Smith, 1950.

12. R. E. Smith, 1955, pp. 21-25.

ARCHAEOLOGICAL UNIT CONCEPTS

25

here by the cross-matching of numerous mound, tomb, and refuse excavations over a very large site zone.13

Kroeber's pioneer seriation of seventeen Zuni ruins by means of surface pottery alone was limited to sites that could be reached in an hour's walk from the modern pueblo. This classic study thus affords a striking illustration of the local sequence and the advan- tages of keeping it local. "Particularly does the necessity of con- centration apply geographically. A promising site here and another a hundred miles away may show striking differences in innumer- able respects. But in the present chaos of knowledge who can say which of these differences are due to age and which to locality and environment?"14

We have spoken of local sequences built up by comparing stra- tigraphies and seriations from a number of sites within a relatively small, circumscribed territory. The Viru Valley, on the north coast of Peru, is an example.15 The valley is about fifteen miles long and three miles wide at its widest point. In prehistoric times it was densely occupied. Although the valley is not a single site, or site zone, there are no contemporaneous cultural differences from one part of Vim to another. Its ceramic and architectural series are, indeed, local. Quite likely, when more sequential information is available from neighboring north coast valleys, it will be seen that Viru joins conveniently with several others to form a regional sequence. Such a regional sequence is suggested by the frequent ref- erences to the "North Coast" on Peruvian culture sequence charts,16 though it must be pointed out that the regional integrity of the Peruvian "North Coast" is still to be demonstrated.

A regional sequence is not merely a local sequence with larger spa- tial dimensions. The difference can best be approached from the

13. Kidder, Jennings, and Shook, 1946; Shook, 1951; Shook and Kidder, 1952.

14. Kroeber, 1916, p. 21.

15. Ford and Willey, 1949; W. C. Bennett, 1950; Strong and Evans, 1952; Willey, 1953c; Collier, 1955.

16. Bennett and Bird, 1949, p. 112.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

26

operational standpoint. In the normal extension of archaeological information, components, subphases, phases, and local sequences multiply, and questions of wider relationships come to the fore. Ideally, the archaeologists of a region come together in a harmoni- ous session where a careful matching of local sequences produces a new sequence of larger scope. Actually this happy event occurs but rarely. What more often happens is that phases and local sequences gain in scope by a sort of osmosis. They flow outward, so to speak, often propelled by their originators, uniting to themselves their weaker correlates over a widening circle. The process is necessarily accompanied by a progressive generalization of definition until much of their original usefulness to research is impaired.

Nevertheless, we will assume for the sake of argument that local sequences remain local and that regional sequences are the result of correlating them — not combining them, be it noted, because in the process the original formulations are retailored to fit the wider spa- tial and (sometimes) deeper temporal dimensions. The phase now appears in its widest practicable extension and at its farthest remove from the primary data; at any rate, it is our contention that the con- cept of phase cannot safely be extended beyond the limits here de- scribed. With these operational conditions in mind, we may define the regional sequence as a chronological series of phases or subphases within the geographical limits of a region as defined in this study.

One of the firmest sequences in North American archaeology is that established by James A. Ford and his associates in the south- ern portion of the Mississippi alluvial valley, centering approxi- mately about the mouth of the Red River in Louisiana.17 This is an instructive example in which stratigraphy in a small number of closely related sites at the center (Marksville locality) and seriation of a large number of sites extending out from the center have been skilfully combined in a regional sequence of great strength but sufficient flexibility to permit the incorporation of new formula- tions from time to time. The unusual vigor of the "Lower Valley"

17. Ford, 1936; Ford and Willey, 1940, 1941; Ford and Quimby, 1945; Quimby, 1942, 1951; Ford, 1951, 1952.

ARCHAEOLOGICAL UNIT CONCEPTS 7 7

sequence is manifested by a tendency to dominate in correlations with other sequences in neighboring regions, but it is important to point out that its strength derives from the hard core of stratigraphy at the center. If there is any valid criticism of this sequence, it is that it tends to take in too much territory. Workers in the delta of the Mississippi and adjacent coastal regions, for example, have tried to fit their data into it, not without considerable strain, which is scarcely to be wondered at, considering the sharp environmental differences involved in such an extension.18 This is a case of the "osmotic" tendency noted above. It also exemplifies another source of difficulty, to wit, the confusion that inheres in practically all archaeological sequence formulations between culture and chronology. As soon as we begin to rank cultural phases in order of time, they tend to become "periods." As periods, of course, they are theoretically not spatially limited; they may be extended in- definitely. If the Marksville phase of the Lower Valley sequence is merely the interval between points E and F on a continuous time band (as in Ford's recent writings), then anything that can be es- tablished as lying within that interval, in the delta, or anywhere else for that matter, can also be called Marksville. The catch is that Marksville is and will remain more than a mere chronological period. The interval marked off by points E and F is determined by cultural criteria in the first instance; the identification of other material as belonging to that interval (in the absence of independent calendrical dating) is determined by those same cultural criteria. When the latter are not sufficiently in conformity with the cultural criteria for comfortable identification, it is a sure sign that the limits of the particular phase in question, either spatial or tem- poral or both, have been exceeded. We shall have considerably more to say about this ever present problem in the sections devoted to area syntheses.

Another strong regional sequence, though expressed largely in ceramics, is exemplified by the Peten region of the central part of the lowland Maya subarea. The key local sequence is that at

18. Mclntire, 1954.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

28

Uaxactun, already referred to, but stratigraphies at Benque Viejo and San Jose can be correlated with ease.19 There are also indica- tions from other sites in the same general vicinity, in both Guatema- la and British Honduras, that ceramic types and their sequences are in close harmony. In sum, an archaeological region of at least fifty miles in diameter, and probably considerably larger, can be established around Uaxactun, and the local sequence at that center can be expanded into an equally valid sequence for the region.

Before we leave the subject, it may be well to emphasize the artificiality of the relationship between phase and region in a re- gional sequence. We have said that the maximum practicable spa- tial dimension of the phase is comparable to that of the region, but no actual geographical coextension is implied. Such a one-to-one relationship may occur fortuitously, because it often happens that a region comes into existence on the heels of a phase, so to speak; but there is no reason to expect that earlier or later phases will also coincide with that region. In fact, the chances are that they will not do so. As an example, there is a neat coincidence of region and phase on the Florida Gulf Coast at the time of the Weeden Island phase. The latter is relatively uniform in its entire extent from the vicinity of Pensacola (if not farther west) south and east to Charlotte Harbor below Tampa Bay.20 This correlation breaks down, however, if we move either backward or forward in time. On an earlier level, the Santa Rosa-Swift Creek phase seems much more restricted in its distribution, and at a later time the old Wee- den Island region is divided between the Fort Walton phase in the north and the Safety Harbor phase in the south.

INTEGRATIVE UNITS

In our first paper we designated the large-scale unit concepts horizon and tradition as "integrative devices," but we now prefer to call them "units" in tardy realization of the fact thrft they are not radically different in kind from other unit concepts described in

19. Thompson, 1939, 1940.

20. Willey, 1949, chap. vi.

ARCHAEOLOGICAL UNIT CONCEPTS

29

this study but, rather, different in the uses to which they are put. We call them "integrative" because, in our opinion, they constitute the most practicable means for effecting culture-historical inte- gration on a geographical scale larger than that of the region. They are not the only means, however, and, before we describe them further, it will be necessary to consider the problem of large-scale integration in more general terms.

One good result of our first presentation of the nomenclature advocated here was a paper by Irving Rouse21 in which, and in subsequent correspondence with the writers, he looks with favor on our basic units (component and phase) and their use in local and regional sequences but expresses some misgivings in regard to these larger integrative units (horizon and tradition) and the way we proposed to use them. Rouse maintains that it is necessary to distinguish three "levels of interpretation" and three correspond- ing methods of correlating phases in large-scale area comparisons : "descriptive," "distributional," and "genetic." The first has to do with formal comparisons of the sort employed in the McKern taxonomic system and does not apply to the problem of culture- historical integration as we understand it. His "distributional cor- relation" is a strictly spatial-temporal ordering of phases in which the unifying criteria are theoretically independent and extracultur- al — nothing more or less than position in geographical space and time — in contrast to his "genetic correlation," which does not scruple to use culturally determined criteria. We agree that our in- tegrative units operate mainly, if not entirely, on Rouse's genetic level of interpretation. He defines "genetic" very broadly, apply- ing it, if we understand him correctly, to any relationship between discrete units resulting from some form of historical contact. We prefer "integrative," which we have been using in practically the same sense, because it carries no implications of phylogeny. This is a minor verbal difference, but, when it comes to a general evalua- tion of the role of genetic or integrative concepts in archaeology,

21. Rouse, 1955.

METHOD AND THEORY ") (\ IN AMERICAN ARCHAEOLOGY J \J

we find ourselves in disagreement. Rouse takes what seems to us an overcautious view :

For these reasons, I would suggest that the genetic approach is not in itself an efficient method of ordering the phases, i.e. of making broad-scale syntheses of cultural relationships, although it may be useful as a check upon syntheses established by other means. It would seem prudent to make one's basic synthesis by means of descriptive or distributional correlation and then to point out the genetic relationships only in the instances where sufficient data are available and they indicate that this type of relationship has occurred.22

In our view, the only synthesis worthy of the name is a culture- historical synthesis. We have Rouse's own word for it that descrip- tive and distributional correlations tell us nothing about cultural or historical relationships; the first tells us whether cultures are like or unlike, the second places them in space and time. How are we to achieve by either or both of these means a "broad-scale synthesis of cultural relationships"? It would seem axiomatic that cultural relationships can be revealed and expressed only by means of inte- grative concepts that are culturally determined. This may be un- fortunate and is certainly productive of circular reasoning, as Rouse points out, but we do not see how it can be helped. The culturally determined integrative units that we are about to consider are ad- mittedly inefficient and hard to handle, but they constitute, in our opinion, the principal means for the realization of what we have repeatedly defined as archaeology's primary task — culture-historical integration.

Failure in our original paper to distinguish between the terms horizon and horizon style, which we used interchangeably, drew some well-merited criticism, thanks to which we here make that distinction and offer what we hope is a more sensible discussion of the whole subject.

The horizon-style concept, introduced into Andean archaeology by Max Uhle23 and later formalized by A. L. Kroeber,24 has amply

22. Rouse, 1955, pp. 719-20.

23. Uhle, 1913. 24. Kroeber, 1944, pp. 108-11.

ARCHAEOLOGICAL UNIT CONCEPTS 3 1

proved its utility in that area. It differs from other unit concepts described in this study in that the content is of a limited and special- ized nature and the spatial and temporal dimensions are combined in very unequal proportions. In theory, a horizon style, as the name implies, occupies a great deal of space but very little time. It may be roughly defined as a specialized cultural continuum represented by the wide distribution of a recognizable art style. On the assumption of historical uniqueness of stylistic pattern, coupled with the further assumption that styles normally change with con- siderable rapidity, the temporal dimension is theoretically reduced to a point where the horizon style becomes useful in equating phases or larger units of culture in time that are widely separated in space. As one of the present authors has already observed: "The horizon styles are the horizontal stringers by which the upright columns of specialized regional development are tied together in the time chart."25

Thus the horizon style is one culturally determined means of establishing a horizon, but the latter must be defined in broader terms. This is the step we neglected to take in our first paper. The horizon-style concept has a limited application, since it presupposes a level of aesthetic development that many archaeological cultures in the New World failed to reach. It is conceivable, however, that other kinds of cultural data might serve equally well to mark horizons, although this is a proposition that has not been sufficiently investigated. We have in mind such items as highly specialized artifact types, widely traded objects, new technologies, unusual modes of burial, or peculiar ritual assemblages — in other words, any kind of archaeological evidence that indicates a rapid spread of new ideas over a wide geographic space.

Rouse recommends a stricter definition of the horizon. In a letter to the authors, written subsequently to his paper from which we have just quoted, he makes it clear that his conception of the hori- zon is on his distributional level and operates ideally with inde- pendent extracultural dating criteria, whereas ours, he points out,

25. Willey, 1945a, p. 55.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

32

is on the genetic level and is therefore unsuitable for use in distribu- tional correlations and should be called by another name. If we could be convinced that it is possible at all times to maintain a clear cut distinction between distributional and genetic methods of corre- lation, we would be inclined to accept this suggestion. However, it seems to us that certain horizon styles — or horizons, to use the more inclusive term — can be looked upon as both distributional and genetic in Rouse's terminology. For example, the presence of Southern Cult horizon-style elements in various far-flung cultures of the southeastern United States links those cultures genetically, in the very broad sense that they reflect some kind of contact, but the cultures so linked are quite different from one another and, so far as we can see, have very different genetic relationships in all other respects. Though the genetic implications cannot be ignored, the primary concern of the archaeologist with the South- ern Cult style is often purely distributional : he thinks of it as a horizon marker whose appearance in various regional sequences is useful in cross-dating. However, we are prepared to waive this point and agree that it is best to use only independent, extracultural dating in distributional correlations or, as we call them here, area chronologies. We adhere, therefore, to our original notion of the horizon, not quite agreeing that it cannot be used for purely tem- poral correlations, but freely admitting its kinship with the concept of tradition on the genetic level of interpretation.26 The horizon, then, may be defined as a primarily spatial continuity represented by cultural traits and assemblages whose nature and mode of occurrence permit the assumption of a broad and rapid spread. The archaeological units linked by a horizon are thus assumed to be approximately contemporaneous. The word is italicized because it is recognized that horizons based on cultural criteria unsupported by independent dating may have considerable temporal depth and that the assumed

26. We trust that the reference to Rouse's "levels of interpretation" is not producing confusion in respect to our "levels of organization" as described in the introductory section of this study. All his levels fall within the definition of our descriptive level.

ARCHAEOLOGICAL UNIT CONCEPTS

33

correlation is not necessarily horizontal but may, and probably does, have a "slope" depending on the amount of time required for the spread of the elements used as horizon markers. Notwithstand- ing these limitations, the horizon is an invaluable integrative unit for the investigation and expression of external relationships over wide geographical areas. If we have agreed that it is of dubious value in establishing purely temporal equations, it is merely to avoid blurring the useful distinction between Rouse's distributional and genetic methods of correlation in the formation of area chronol- ogies, a point we shall return to later.

Tradition is another integrative concept that seems to have origi- nated in South American archaeology. A familiar, not to say indis- pensable, word in any historical context, "tradition" has of late acquired a special meaning in archaeology, or rather a number of meanings, for it is still in an early stage of formulation and does not mean the same thing to all who use it. Owing to the fact that the concept is designated by a term that has long been used in archaeo- logical writings, it is difficult to say just when it began to be a methodological tool. In the Andean area it came into use as a counterpoise to the horizon style and was at first applied only to pottery. Once it became apparent that the utility of the horizon style depended upon the combination of broad spatial and short temporal dimensions, it was clear that some other formulation was needed to express a different kind of ceramic unity in which these proportions were reversed. This gave rise to the term "pottery tradition," certainly not a verbal innovation, but perhaps the first use of the term "tradition" with a definite methodological sense.

The relationship of the two concepts, horizon style and pottery tradition, proved to be so important in Peruvian archaeology that one of us may be permitted to repeat himself, as follows :

These speculations concerning the relationships of the later White-on- red styles to the earlier component styles of the White-on-red horizon lead us to wonder if there are not other widely inclusive historical units of an order different from that of the horizon style. It appears certain that the Peruvian Andes and coast were a unified culture area in that the important

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

34

culture developments were essentially local and basically inter-related for at least a thousand years. This fundamental cultural unity justifies seeing ceramic developments in terms of long-time traditions as well as coeval phenomena. The concept of a pottery tradition, as used here, includes broad descriptive categories of ceramic decoration which undoubtedly have value in expressing historical relationships when the relationships are confined to the geographical boundaries of Peruvian- Andean cultures. The pottery tradition lacks the specific quality of the localized pottery style, and it differs from the horizon style in that it is not an integration of artistic elements which has been widely diffused at a given time period. A pottery tradition comprises a line, or a number of lines, of pottery development through time within the confines of a certain technique or decorative con- stant. In successive time periods through which the history of ceramic de- velopment can be traced, certain styles arose within the tradition. Trans- mission of some of these styles during particular periods resulted in the formation of a horizon style, other styles in the continuum of the tradition remained strictly localized. The distinction between a horizon style and a pottery tradition should be kept in mind as the two are opposable concepts in archaeological reconstruction.27

Shortly after this idea of tradition had first been injected into Peruvian studies, Wendell Bennett enlarged the concept very con- siderably under the name "area co-tradition." This formulation he defined as "the over-all unit of cultural history of an area within which the component cultures have been inter-related over a period of time."28 The emphasis implied in the "co-" is on the linkage of whole cultures, each with its own history and persistent traditions, and on the area in which this linkage takes place. Thus the area co-tradition is a vast enlargement over the simple tradition in terms of content, since it is no longer confined to a single technological development but becomes a broad coalescent cultural continuum. At the same time, it introduces a restriction in that stable geo- graphical boundaries are implied.

The subsequent history of the area co-tradition and the contro- versies that grew out of attempts to apply it in other areas29 need

27. Willey, 1945a, p. 53. 28. W. C. Bennett, 1948, p. 1.

29. Martin and Rinaldo, 1951; Willey, 1953a, pp. 373-74; Rouse, 1954; Cotter, 1954.

ARCHAEOLOGICAL UNIT CONCEPTS

35

not detain us, since we are not proposing to incorporate the con- cept in the nomenclature described here.

About the same time that the Peruvianists were beginning to talk about pottery traditions, McGregor introduced the term into the archaeology of the Southwest with somewhat different connota- tions. He defined tradition very broadly as "more or less deeply rooted human characteristics — persistent attitudes or ways of do- ing things — which are passed on from one generation to another,"30 thus emphasizing the non-material and configurational aspects of traditions and the social and behavioral continuities reflected in their persistence. He maintained that characteristic attitudes can be inferred from material traits and that the determination of tra- ditions (as defined by him) is not only possible but essential in making broad archaeological comparisons. McGregor's traditions are for the most part technologically oriented: house types, pottery, ground stone, etc.; he merely advocates that they be formulated in terms of the preferences and attitudes they reflect. There is nothing revolutionary in this, but it is a point of view that cannot be too often stated.

The first significant use of the tradition concept in eastern North American studies, so far as the authors are aware, was by John Goggin, whose definition follows :

My concept of Florida cultural traditions is similar in theory but more inclusive in content than a ceramic tradition. A cultural tradition is a dis- tinctive way of life, reflected in various aspects of the culture; perhaps ex- tending through some period of time and exhibiting normal internal cultural changes, but nevertheless throughout this period showing a basic consistent unity. In the whole history of a tradition certain persistent themes domi- nate the life of the people. These give distinctiveness to the configura- tions.31

Goggin recognizes ten cultural traditions in Florida archaeology, allowing to them a great deal of latitude in spatial and temporal di- mensions. It seems to us that here he has discovered the outstand-

30. McGregor, 1950. (Paper submitted in 1946.)

31. Goggin, 1949, p. 17.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

36

ing merit of the tradition as a methodological tool, namely, its flexibility. Goggin also, rather more than others who have used the concept, emphasizes the importance of environmental factors in the shaping and conserving of traditions. Here again he has put his finger on another virtue of the concept. It offers a most effective means for giving expression to culture-environment correlations.

From the examples just given, it may be seen that the concept of tradition has been expanded considerably since its first use in Peruvian archaeology. In our original paper we expressed satisfac- tion with this growth and gave a definition of our own that was as broad as, if not broader than, any previous ones. It may be asked, however, whether by such radical expansion the tradition is not deprived of some of its utility as an integrative unit. In the blunt- ing that is inseparable from increased generalization it tends to lose its primary significance of long temporal continuity as a counter- poise to the broad spatial continuity represented by the horizon. Furthermore, since the publication of our first paper, we have come to realize that, for expressing the maximum segmentation of culture-history, the old standbys "culture" and "civilization" will not be easily supplanted by such cumbersome expressions as "full cultural tradition." We are therefore now proposing a return to something like the original meaning of tradition in Peruvian ar- chaeology, with emphasis put back on single technologies or other unified systems of forms rather than on whole cultures. The tradi- tion thereby takes its former place as a means to integration, with the culture or civilization standing as the product of integration. Our first definition is accordingly amended to read : an archaeological tradition is a {primarily) temporal continuity represented by persistent configurations in single technologies or other systems of related forms. The lack of specification in respect to the spatial dimension may be supplied by the use of qualifying terms, as in "regional tradition," "areal tradition," and so on.

This definition of tradition takes us not quite all the way back to the original pottery tradition of Peruvian archaeology. Some expan- sion is left, in that traditions may be based on more complex sys-

ARCHAEOLOGICAL UNIT CONCEPTS 3 7

terns of forms than that represented by a single technology. We have in mind something like the functionally interrelated trait com- plexes of the ethnographer. About the same amount of expansion is involved in substituting the concept of horizon for the horizon style. The working relationship between horizon and tradition, therefore, remains the same as between horizon style and pottery tradition in the passage quoted above (pp. 34—35).

After making the decision to cut back the tradition, as outlined above, we received a manuscript copy of the report on the Society for American Archaeology's seminar on "Cultural Stability" held at Ann Arbor in August, 1955. At this meeting the concept of tradition was thoroughly anatomized. In defining it as "a socially transmitted form unit (or series of systematically related units) which persists in time," the conferees were careful to avoid any qualifications as to size, either in terms of content or in spatial and temporal dimensions, beyond the obvious implication of long dura- tion in the clause "which persists in time." They were evidently prepared to think of "whole cultural traditions," but most of the discussion, and nearly all the examples, applied to traditions of less than whole cultural scope. Our impression is that such limitation made the tradition a more useful tool in their investigation of cul- tural stability. "Thus, although it is logically necessary to entertain the possibility of traditions ranging from the smallest to the largest, it is also important to recognize that the cultural and historical significance of a tradition diminishes toward either size extreme."82

The reference in the Ann Arbor definition to social transmittal serves to emphasize the fact that traditions operate on Rouse's genetic level of interpretation, perhaps even more surely than hori- zons. Thus, if we may express the reciprocating relationship be- tween them without too flagrant a confusion of categories, the tradition gives depth, while the horizon gives breadth, to the genetic structure of culture-historical relationships on a broad geographic scale.

In seeking to polarize such essentially fluid concepts as horizon

32. Haury tf a/., 1956, p. 39.

METHOD AND THEORY "1 Q IN AMERICAN ARCHAEOLOGY J O

and tradition, we have unavoidably overstressed their salient fea- tures and produced an impression of profound difference between them that may be misleading. Looked at from the broader view of total archaeological method, horizon and tradition are not really very different. Both are unit concepts in the sense that their formu- lation involves the combination of cultural data and spatial-tem- poral dimensions. They differ considerably in the nature of the cultural data and, theoretically, in the quantities and proportions of the spatial and temporal dimensions; we have stated with tiresome repetition what those differences are supposed to be, but in practice such standards are difficult to apply. Our first apprehension of a major archaeological continuity, as of the proverbial elephant in the dark, is usually so tenuous that we do not know what it is or what to call it. The same continuity may be viewed by one investigator as a tradition, by another as a horizon. Even when the outlines are fairly clear, it may still be thought of either way, depending on the problem under consideration. For example, Hopewellian zoned rocker-stamped pottery, when looked at from the standpoint of its geographic extension in the eastern United States, may be quite legitimately thought of as a horizon phenomenon; on the other hand, when the interest is centered on the course of its develop- ment through successive phases in a particular region, it may equally well be regarded as a tradition. This is perhaps an extreme example of the fluidity we have praised in describing these integra- tive concepts.

There remains but to mention briefly the useful but largely neg- lected concept climax. In reference to the integrative units just de- scribed, the climax may be defined as the type or types of maximum intensity and individuality of an archaeological horizon or tradition. This is necessarily a value judgment, but only in relation to the horizon or tradition involved. In whole cultural terms the climax becomes the phase or phases of maximum intensity and individuality of a culture or civilization. So far as possible, in this wider context, the emphasis should be multilinear, not confined to exquisite de- velopments in aesthetics. Such developments may conceivably take

ARCHAEOLOGICAL UNIT CONCEPTS

39

place in periods of low cultural intensity. Theoretically, there should be a climax in every horizon, tradition, culture, or civiliza- tion, on all stages of development. Our choice of the term "classic" — which means about the same thing — to denote the climactic stage of New World culture is by no means intended to preclude its use in more humble contexts. Any horizon or tradition may have its classic type, as any culture or civilization may have its classic phase. The word makes equal sense whether we are talking about the classic black stirrup-mouth jar of the Chavin horizon, the classic Folsom point of the fluted-point tradition, the classic Pueblo III phase of the San Juan Anasazi culture, or the classic stage of New World culture as a whole.

OPERATIONAL RELATIONSHIPS OF BASIC AND INTEGRATIVE UNITS

Before we go on to a consideration of large-scale area schemes and maximum archaeological units, it may be well to summarize briefly the working relationships between the unit concepts so far described, because these constitute, in our opinion, the most effec- tive apparatus for the integration of the primary archaeological data. These relationships are crudely diagramed in Figure 1, which is designed to show the difference between the basic units, com- ponent and phase, and the integrative units, horizon and tradition, not in their nature as archaeological units but in the way they oper- ate in large-scale integration. The relationships in the component- phase system are predominantly formal and static; those in the horizon-tradition system are fluid and historical or, as Rouse would say, genetic. In the first pair, the spatial and temporal dimensions that inhere in all archaeological unit concepts are, in this larger context, reduced to negligibility. This is what permits us to refer to the phase as the "manageable" unit of archaeological study. It can be manipulated in large-scale area schemes as though its internal dimensions were non-existent.

Up to this point most archaeologists would probably go along, though many, depending on the areas in which they have worked,

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

40

would prefer to use other terms in place of "phase." Where our point of view differs is in the conviction that the phase, as defined in these pages, is the largest archaeological unit that can be so manipulated. Our later discussion of some of the characteristics of the maximum units, culture and civilization, we need anticipate

C = Component P = Phase TT= Tradition HH = Horizon

Fig. 1. — Diagrammatic integration of components, phase, horizon, and tradition.

here only to the extent of pointing out that these are in a very real sense the ends, not the means, of culture-historical integration. Their vastly greater internal spatial and temporal dimensions, par- ticularly the latter, render them completely unmanageable as units of study in both formal and spatial-temporal comparisons. Forms are fluid, changing constantly through space and time; with ex- pansion of these dimensions, the changes in forms within the unit

ARCHAEOLOGICAL UNIT CONCEPTS

41

are such that the unit can only be described in progressively more general terms — the familiar phenomenon of shorter trait lists for larger taxonomic subdivisions. For example, at the Woodland Con- ference in 1941,33 eastern North American archaeologists laid down a formal definition of the Woodland cultural pattern couched in such general terms that, as later discovered, but for the inclusion of pottery, it fitted the Archaic pattern just as well.34 In a similar manner, temporal relationships between large units become im- possible to describe through excess of internal time dimensions and the consequent possibilities of overlap. Failure to recognize this is responsible for many fictitious problems. The Adena-Hopewell problem, a long-standing subject of debate, might be cited as an example. The sequential arrangement of these two cultures in the Ohio Valley and their correlation with other units of comparable dimensions in eastern North America became faintly ridiculous when radiocarbon dates indicated a probable overlap of several centuries in their time spans. This is not to maintain that spatial and temporal relationships of cultures and civilizations are not legitimate problems for the archaeologist, but to point out simply that in this wider arena the variables of space and time are more difficult to control. Such questions certainly cannot be investigated and expressed by the crude diagrammatic methods ordinarily em- ployed in large-scale area schemes.

To return to the cabalistic diagram in Figure 1 : if it be granted that the component-phase system operates mainly with relation- ships of a formal nature, it will certainly be allowed that horizons and traditions belong to a different system — a system in which the element of formal content is reduced in importance, while the vari- ables of space and time have the dominant roles. The real point, however, is not how like or unlike these two pairs of units may be, conceptually or operationally, but that there is no built-in taxonom- ic relationship between them. Components and phases enter into horizons and traditions; their external relationships are expressed

33. Woodland Conference, 1943.

34. Sears, 1948.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

42

by these units, but they are not combined to form them. In fact, the opposite is more nearly the case. A single phase may conceivably enter into more than one horizon. For example, the Esperanza phase in the highland Maya sequence at Kaminaljuyu, Guatemala, par- ticipates in both the basal-flanged bowl horizon and the Teotihua- can tripod-jar horizon.38 The basal-flanged bowl, of presumably lowland Maya inspiration, is a widespread horizon marker for the period of the Early Classic cultures in southern Middle America. The tripod cylinder jar, deriving from Teotihuacan about the same time, has an even more extensive distribution and imitation. As for traditions, there is certainly no question that a number of these may, and usually do, converge in a single phase. This is clearly seen, for example, in the Mochica phase of the north coast of Peru. Mochica red-and-white modeled pottery represents a convergence of the old Cupisnique, or north Peruvian Plastic and Incised, traditions, with the north Peruvian White-on-red tradition that first made its ap- pearance in the Salinar phase.36 In sum, the effectiveness of this apparatus, as we see it, depends on the free interplay of basic and integrative units without rigid limitations of a systematic nature.

35. Kidder, Jennings, and Shook, 1946, p. 250.

36. Willey, 1945«.

ARCHAEOLOGICAL UNIT CONCEPTS

43

Chapter 2

Archaeological Integration

AREA SYNTHESIS

The integrative units described in the preceding section are well adapted to express the extraordinarily complex and fluid relation- ships of archaeological phases over wide geographical areas, but they cannot furnish the rigid spatial-temporal frames that archaeolo- gists seem to find so reassuring. For these we generally turn to some system of matching phases from region to region, expressing the results by means of a diagram in which the regional sequences are arranged in parallel columns with their phases stacked up like boxes, scaled against a real or imaginary vertical time band at one side. Illustrations are available in current archaeological literature in great profusion and variety. The appearance of an archaeological report without such a diagram would in fact be something of a curi- osity today. This is what Rouse, in the paper already cited, has called "distributional correlation of phases," and the resulting dia- gram, of which he presents an excellent model,1 is often referred to as an "area chronology."

We have expressed agreement with Rouse that so far as possible such a scheme should be effected by means of independent, extra- cultural cross-dating, e.g., radiocarbon dating, dendrochronology, river-channel and river-terrace sequences, discrete geological and meteorological events (volcanic eruptions, droughts, etc.) — in short, any available techniques of dating that do not involve as- sumptions about culture. Unfortunately, in the present stage of

1. Rouse, 1955, Fig. 2, p. 716.

44

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

archaeological development in the Americas we rarely, if ever, have sufficient dates of this kind; consequently it is difficult to find in the literature examples conforming to Rouse's model. Even in the Southwest, where tree-ring dating has made such remarkable prog- ress, it appears that dates are still insufficient for interregional cor- relation.2 Therefore, present area chronologies in the New World do not scruple to employ any sort of correlation data that comes to hand — descriptive, distributional, or genetic. They are often dressed out to look like real chronologies by the addition of a few calendrical dates, usually guesses, marking off the "periods" on one side of the chart. These seldom date more than one of the phases that are correlated to form the "period," the correlation being really based on cultural similarity, horizon phenomena, trade objects, etc., and the well-worn assumption that these reflect con- temporaneity. In other words, the correlations are not based on in- dependent dating so much as the dating is based on the correla- tions. A casual glance through the literature of American archaeol- ogy of the past twenty-five years will reveal numerous examples of such charts. The Peru-Bolivian, or central Andean, area is especial- ly well represented.8 For Middle America it is also easy to cite similar chart presentations.4

Outstanding examples of area schemes that are virtually without benefit of dating are to be found in the southeastern United States. Here, owing to the failure of independent dating techniques to de- velop as rapidly as in some other areas, it has become a habit to equate widely separated archaeological units on strictly formal principles, and, whether as cause or effect, there prevails a sweep- ing assumption of synchrony of culture change throughout the area. Does culture A have plain fiber-tempered pottery? If so, it equates in time with culture B, which has plain fiber-tempered pot-

2. Wheat, 1954, p. 577.

3. Kroeber, 1944, p. 112; Willey, 1948a, p. 9; Bennett and Bird, 1949, p. 112. These are but a few samples.

4. Armillas, in Kroeber, 1948, p. 116; Kroeber, 1940, pp. 484-85; Mac- Neish, 1954£, p. 623; Wauchope, 1950; Sorenson, 1955, pp. 54-56.

ARCHAEOLOGICAL INTEGRATION

45

tery. The possibility that the relationship may be genetic, and there- fore not necessarily synchronous, is not entertained. We are so fond of this method of reasoning in the Southeast that we tend to ignore the few independent (radiocarbon) dates that we have. As a recent example, William H. Sears's excellent final report on excavations at the Kolomoki site in southwestern Georgia contains an elabo- rate correlation of southeastern sequences against a calendrical scale, but no radiocarbon dates appear on the chart, nor are they mentioned in the text.6

There is an unfortunate tendency in area chronologies of this culturally correlated variety for the periods to take on some of the characteristics of developmental stages. The larger the area, the more likely is this to happen. The best example in recent literature is the scheme for "Eastern United States" as set forth in the monu- mental Cole Anniversary Volume under the editorship of James B. Griffin.6 The area, which takes in everything from the Plains to the Atlantic, is too large, in our opinion, for any practical archaeologi- cal purposes. Nevertheless, Griffin addresses himself to the heroic task of correlating hundreds of phases and cultures in this vast area into a succession of "culture periods," defined, as the term implies, both chronologically and culturally but necessarily in such broad terms as to be little more than general stages in the development of American culture. In some of the contributions to the volume, the terms "period," "culture," and "stage" are used interchangeably, depending on the exigencies of the moment. Deliverance from this kind of semantic ambiguity will come when current techniques of absolute dating have reached a point of such dependability that we can place a given unit within a temporal frame, on the one hand, and in a developmental sequence, on the other, without confusing the two operations.

We are forced to conclude that the pure area chronology based on distributional criteria alone is, in the New World at least, a

5. Sears, 1956, p. 80.

6. Griffin, 1952a.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

46

presently unattainable ideal. But, with the rapid accumulation of radiocarbon and other types of independent dating now taking place, it is a certain possibility in the future. In the meantime we must be content with the sort of hybrid area schemes now in vogue. There is no harm in them so long as we are clear about how they are made and what they mean, but, when they lead us to think that we have discovered something about the way culture "works," it is time to be on guard. In any case, it will be entertaining to compare them with true area chronologies when these are available.

MAXIMUM UNITS: CULTURE AND CIVILIZATION

We have tried to show that area chronologies express the spatial- temporal relationships between archaeological phases, while hori- zons and traditions express their culture-historical relationships. These are unquestionably their primary functions, but, on a more abstract level of interpretation, all these units and devices may be considered means by which the archaeologist gropes for the major segmentations of culture-history. There is clearly an irresistible urge to relate cultural forms to their largest possible contexts, ex- pressed in the terms "culture" or "civilization" as used by students of culture-history when they are referring to events on a world- wide scale. We did not sufficiently take account of this urge in our first paper, which was conceived in terms of practicable archaeologi- cal methods and was not particularly concerned with ultimate ends. We found in the phase an operationally manageable unit, i.e., a unit small enough in spatial and temporal dimensions to be manipu- lated in area correlations and sequential schemes. Phases were to be related in larger culture-historical organizations by means of con- cepts such as the horizon and the tradition, but it was ruled that these were strictly integrative units and that the phases were not to lose their identity in these larger formulations. But the urge for totality was at work, with the result that the definition of one of these concepts, the tradition, was so broadly stated as to include the "full cultural tradition," which is, in effect, a self-sufficient

ARCHAEOLOGICAL INTEGRATION

47

archaeological unit more like the phase, with greatly amplified spatial and temporal dimensions, than like a mere device for ex- pressing relationships, as advertised. In short, here was a serious contradiction, which we now find it necessary to remove.

The solution we have chosen has already been described : to cut back the concept of tradition to something like its original scope, taking from it the expanded connotation of "full cultural tradition" and returning that connotation to the unit concept of "culture" or "civilization," where it has always lodged. Thus, the phase re- mains the manageable unit; horizon and tradition remain the inte- grative units for expressing relationships between phases; culture and civilization, the maximum units reflecting the major segmenta- tions of culture-history. In a strictly methodological context, and subject to the developmental relativism which applies to all unit concepts, culture and civilization can be treated as equivalents. As a somewhat arbitrary specification, however, applicable to the New World only, we suggest that "culture" be used to denote maximum units on all stages up to and including the Formative, reserving the term "civilization" for such units on the Classic and Postclassic stages.

THE SOCIAL ASPECTS OF ARCHAE- OLOGICAL UNITS

We have described a number of archaeological unit concepts and discussed their working relationships, but we have allowed ques- tions of their social meaning to accumulate in the belief that these can be more sensibly dealt with under a single heading. The task of finding social equivalents for archaeological units is beset by the most formidable difficulties, most of which stem from the fact that the kinds of data archaeology depends on are precisely those ele- ments of culture that diffuse most readily across social and political boundaries. Consequently, we seldom experience the satisfaction of feeling that our units are coextensive, either spatially or temporally, with corresponding social units, even in the simplest and most ex- plicit of archaeological situations. However, we must remind the

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

48

reader that, according to our theoretical position, archaeological units are formulated on the descriptive level of organization with a view to their use on the explanatory level and, to qualify for the latter, must be intelligible in both the cultural and the social aspects of the behavior that is our subject matter. So we are obliged to keep in mind the possibility of social equivalents to our unit formulations, even when we cannot say with any degree of assurance what they are.

In the case of archaeological units of the smallest magnitude, however, we are on fairly solid ground. The social equivalent of the component is the "community," as defined by Murdock and others : "the maximal group of persons who normally reside together in a face-to-face association."7 Murdock's three types of community — band, neighborhood, and village — manifest themselves archaeologi- cally in the component, and it is even possible sometimes to tell which type is represented. So far, so good. The equivalent of phase, then, ought to be "society," and in a good many cases it probably is. The fact that, in practice, phases often consist of a single component need not disturb us; on the lower levels of cul- tural development the society likewise may consist of a single com- munity. At the other end of the developmental scale, however, so- ciety becomes a larger concept, spatially at least, than phase. We shall return to this point presently in discussing the social aspect of our maximum units, culture and civilization. For the purposes of the immediate discussion, let us think of society in its minimal sense, as "a group of people acknowledging a single political au- thority, obedient to a single system of law, and in some degree or- ganized to resist attack from other such societies."8 How does this relate to the concept of phase? Logically the correspondence is reasonable. Such a society comprises a number of communities ; the phase comprises a number of components; component equals com- munity; therefore, phase equals society. Unfortunately, in practice it does not work. We have no means of knowing whether the com-

7. Murdock, 1949, p. 79.

8. M. A. Smith, 1955, p. 4.

ARCHAEOLOGICAL INTEGRATION

49

ponents we group together into a phase are the same communities an ethnographer (supposing such a person happened to be on hand) would group into a society. We cannot be sure that the individual members of these communities would recognize themselves as be- longing to the same "people." They might not even speak the same language. Ethnography offers abundant examples of different so- cieties sharing a material culture that would be impossible to differ- entiate archaeologically. It would be only slightly more difficult to find examples in which the material culture of individual communi- ties within a society diverged sufficiently to cause them to be classi- fied archaeologically in separate phases. A frontier garrison com- munity, organized specially for defense, might be a case in point.

More vexing, perhaps, are questions having to do with the stabili- ty of material and social culture through time, a dimension happily ignored by the ethnographer. We have laid down the rule that the temporal dimension of the phase must be kept within manageable limits, but this is admittedly in terms of the somewhat prodigal view of time held by the archaeologist. We are not forgetting that the life-span of our phases is ordinarily determined by the persistence of material traits which can be remarkably stable. Within such a span it is conceivable that social changes might be sufficient to en- able our hypothetical ethnographer to speak of several societies. Conversely, under special conditions, even a primitive population may exhibit revolutionary changes in material culture without losing its identity as a society. We have abundant examples of this in recent colonial history.

In sum, it looks as though the present chances are against archaeological phases having much, if any, social reality, but this does not prevent us from maintaining that they can have and that in the meantime we may act as if they did have. We have already ex- pressed a similar attitude in connection with artifact typology. Just as, with the inevitable refinement of archaeological techniques, it will become increasingly possible to define types in terms of so- cial behavior, it will become likewise increasingly possible to de- fine phases in terms of social structure. This possibility must be

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

50

reckoned with, but it is not really the point of the present discus- sion. We do not maintain that every, or even any, specific phase is the archaeological expression of an extinct society. We simply call attention to the fact that there is a certain conceptual agree- ment between phase and society. Both are intelligible units of their respective fields of study. They have similar roles and similar scales, and in this crucial matter of scale both exhibit the same rela- tivism with respect to the level of cultural development. This con- gruence, we contend, qualifies the phase as the intelligible unit of comparative study and, thus, offers the best hope of incorporating archaeology into general anthropological science.

The integrative units, horizon and tradition, because of their incomplete cultural content, cannot be regarded as self-sufficient in the cultural aspect; so there is an a priori case against their intelli- gibility in the social aspect. This is not equivalent to denying that they are socially transmitted. These units are the archaeological expressions of the processes of diffusion. They have come into exist- ence in response to an awareness that particular forms and systems of forms — as distinct from whole, functioning, cultural units — flow through geographical space and time in a manner seemingly inde- pendent of the cultural matrices in which they are found. It is be- side our purpose to investigate the mechanics of diffusion, or the question of how they may be detected in archaeology — a subject for a lengthy treatise in itself — but we suppose that in the main they operate through the agency of individuals or organized groups, such as trading companies, religious bodies, armies, and migrating popu- lations, which, with the exception of the last, are not complete and self-sufficient social units. In general, therefore, horizons and tra- ditions fail to meet the test of intelligibility in the social aspect, and no more need be said about them here.

When it comes to the social aspect of the maximum units, cul- ture and civilization, we are on more difficult terrain. Here the problem of stage-relativism becomes acute. We have arbitrarily de- fined the culture as the maximum unit on all developmental stages up to and including the Formative. Even with this specification,

ARCHAEOLOGICAL INTEGRATION

51

there is still a tremendous variability in its connotations, both cul- tural and social. On the lowest stage, the Lithic, the term "culture" usually refers to single technologies or "assemblages" reflecting a similar economic adjustment shared by a large number of social groups. The content of such a "culture" is seldom sufficiently com- plete or physiognomic to suggest that a single homogeneous society is responsible. Put the other way around: on this level we are un- able to infer the existence of social units large enough to be coex- tensive with these "cultures." It would perhaps be preferable to organize these incomplete data — all archaeological data are in- complete, but Lithic data are more so — in terms of phases and tra- ditions, eschewing the term "culture" altogether. We do not really expect this wholesome suggestion to be followed, but, if we could at least eliminate "cultures" represented by a single type of pro- jectile point, it would be progress in the right direction.

On the Archaic stages of development, the situation is a little different. Archaic phases are generally richer in cultural content; spatial and temporal dimensions tend to be smaller. It seems to be possible in some cases to organize the data into larger units than phases, for which the term culture stands without benefit of quota- tion marks. It is difficult to say, however, what the social equiva- lents of such cultures might have been. We have inferred elsewhere that social organization in the Archaic stage had not generally pro- ceeded beyond the level of complexity represented by the tribe, but this is a gratuitous assumption that might not stand up under investi- gation. In short, it remains to be shown whether or not the unit, culture, is intelligible in the social aspect on the Archaic stage of development.

The Formative is by definition the stage of the appearance of new economic patterns, and these we assume to have been accompanied by the formation of societies of greater scale and complexity than existed theretofore. Ethnohistory furnishes examples in the numer- ous confederacies, and we have abundant evidence of the existence on this stage of religious and ceremonial organizations far exceed- ing in scope anything that could be represented by a phase in ar-

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

52

chaeology. This would seem to be the stage on which the unit, cul- ture, becomes intelligible in the social aspect. However, it is diffi- cult, even in an ethnohistoric context, to make satisfactory equa- tions between archaeological and social units on this level of cul- tural development. In the case of the Iroquois Confederacy, for ex- ample, we can equate specific archaeological phases with the indi- vidual member tribes, but the unit "Iroquois culture" seems to have been shared by tribes that were not only outside the famous con- federacy but in some cases not even affiliated linguistically. In a context that is largely prehistoric the difficulties are even greater. As a trial example we may consider the Coles Creek-Plaquemine culture, a Formative continuity in the lower Mississippi Valley lasting from about a.d. 900 to a.d. 1731 (date of the final dispersal of the Natchez tribe) . We regard this as a unit because the elements of continuity from phase to phase seem to be stronger than the ele- ments of change. Now what would be the social equivalent of such a continuity? Projecting back from ethnohistoric data, we know that there were several phases of this culture in the seventeenth cen- tury and that these phases pertain to discrete tribal groups, such as the Natchez, Taensa, Houma, Tunica, or Yazoo. There is no evi- dence that these tribes were united into a political federation. In fully prehistoric times it is possible that some sort of hegemony may have been imposed on these groups, but, if the pattern of oppor- tunist warfare and ephemeral alliances that we see in historic times is any index to what happened earlier, it is inconceivable that any such hegemony could have embraced the full geographical range, or time span, of the Coles Creek-Plaquemine culture.

So far, in the discussion of possible equations between archaeo- logical cultures and larger social units, our implicit definition of a social unit has been essentially a political one. We have also seen that it is extremely difficult to identify the correlates of political structure in an archaeological record. Leaving aside for the mo- ment, then, the question of political forms, let us simply ask what kind of archaeological evidence might be indicative of larger social units. This, we feel, is the most fruitful starting point in the search

ARCHAEOLOGICAL INTEGRATION

53

for archaeological and social equations; and, in seeking evidence of this order, it is also our opinion that the Classic and Postclassic cul- tures of the New World offer more potentialities than the Forma- tive.

One of the characteristics of Classic cultures in both Middle America and Peru is the possession of great art styles. These styles are symbolic systems. Their content and the contexts in which they are presented assure us that they permeated much of their cultural setting and that they were a part of the conscious awareness of the peoples of these cultures. For example, the recognition and appre- ciation of an art style like that of the lowland Maya imply a realiza- tion upon the part of the beholders that they are affiliated with other peoples who also recognize and believe in the symbolism of this same style. Such a sense of affiliation must necessarily have had so- cial significance. It is difficult, perhaps impossible, to define the exact nature of the bonds giving coherence to such a social order or society in the large. They may, or may not, have been political. We have no definite knowledge about the ancient Maya political structure. Maya ceremonial centers from the Usumacinta to the Motagua may have been held together under a single authority, or each may have maintained a completely independent sovereignty. Religious ideology was undoubtedly one of the ties that held Maya society together. Not only does the art style suggest this, but wide- spread diffusion of uniform calendrical and hieroglyphic lore sup- ports the interpretation. Yet the Maya bonds appear to have been more than the interchange of religious and astronomical ideas among a few priestly elite from the great centers or cities. Such may have been the mechanism of their original dissemination, but there are numerous signs that the ordinary Maya villagers who sustained the centers and the theocratic leadership also participated in, and had at least a peasant's understanding of, the ideas of the upper classes. These ideas, in so far as they were generally accepted, probably became much more than purely religious dogma. It is likely that they were assimilated into the values and beliefs of Maya peasant culture and that these values and beliefs were shared by Maya

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

54

society in the same way that the art expressing these values was also shared by that society.

If we argue that stylistic unity may demonstrate social as well as cultural cohesion, for what other prehistoric American art styles can we build a case similar to that for the Maya? Teotihuacan? Zapotec (Monte Alban)? Mochica? Nazca? These four have the requisite qualities of greatness and pervasiveness. For each the im- print of the style is unmistakable, whether rendered in stone, pot- tery, wall painting, or textiles. Also, all four have the relatively tight territorial integrity which suggests a unity of a social kind. All four lack, however, those continuities into the ethnohistoric pe- riod that the Maya case possesses in some degree, and this makes it more difficult to be assured of common language, custom, and belief. What of the still older styles, Olmec and Chavfn? For these the circumstances of remote antiquity and great geographic range lessen the possibility that the social dimension can ever be approxi- mated. It may be, as has often been surmised, that Olmec and Cha- vfn style occurrences, in their respective Middle American and Peruvian spheres, signify only religious, ideological, or cultist diffusions and that the ideas behind these styles never penetrated the receiving cultures to the extent that general values and beliefs were substantially modified. If so, there would be less reason to be- lieve in the coherence and homogeneity of an Olmec or Chavfn so- ciety. Perhaps these two styles represent an earlier stage in the civilizing process of social enlargement and homogenization that we have inferred for lowland Maya culture and society. To con- tinue this same line of speculation, it is possible that such a horizon style as the Southern Cult of the southeastern United States is de- velopmentally prototypical to Middle and South American great styles in the same way that southeastern Formative cultures are prototypical to the Mexican and Peruvian Classic civilizations.

Civilization has been characterized, both qualitatively and quan- titatively, by Childe, Redfield, and others.9 The chief requisites are city life with its concomitants of large population size and

9. Childe, 1946, pp. 82-105; Redfield, 1953; Braidwood, 1952.

ARCHAEOLOGICAL INTEGRATION

55

density, formalized religions, class systems, craft specializations, the beginnings of codified learning and science, and great artistic traditions. Inferentially, it is our opinion that a consciousness of a larger social order is also a feature of civilization. In other words, the societal basis of a civilization differs from that of a culture, both quantitatively and qualitatively. In quantity, it is larger, it en- compasses more people, it overrides community and tribal barriers. In quality, it demands beliefs in, and allegiances to, ideas and values that are abstract and remote from the individual and his tribal hearth.

The foregoing review of the various archaeological unit con- cepts in their social aspect has brought out several interesting but highly tentative possibilities. First, it appears that the basic units, component and phase, are theoretically capable of intelligibility in the social aspect on all the lower stages of development up through the Formative, but the difficulties of close equations are not to be minimized. Though the question of just what the social equivalents of these smaller units might be in the more highly organized Classic and Postclassic societies has not been investigated, it is nevertheless easy to foresee that the difficulties of doing so would be formidable. Second, while the integrative units, horizon and tradition, have been rather summarily dismissed because of apparent lack of complete- ness and self-sufficiency in both the cultural and the social aspects, this is a rather arbitrary exclusion which, needless to add, requires further investigation. Third, in testing rather more thoroughly the maximum units, culture and civilization, we seem to have stumbled upon a valuable insight which is at the same time a useful criterion for distinguishing between these units. For, while it has proved to be difficult to envisage any social equivalent for the culture, it ap- pears that the really definitive characteristic of the civilization is that such a possible equivalent does exist.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

56

Summary

In so far as the foregoing definitions and stated relationships can be formalized as a program for the integration of New World archaeo- logical data on the descriptive level of organization, they may be summarized as follows :

1. The primary emphasis should continue to be placed on the for- mulation of basic units, component and phase, in local and regional sequences under stratigraphic control.

2. Phases should be studied intensively in their cultural and natu- ral contexts.

3 . Their external spatial and temporal dimensions should be kept within manageable limits of magnitude.

4. Their external relationships should be studied and expressed by means of the integrative units, horizon and tradition.

5. Large-scale integrating syntheses should be kept within the limits of the "area," as defined herein, and horizontal correlation of phases in such schemes should be effected so far as possible by means of independent extra cultural data.

6. On the basis of these integrative studies, phases should be combined when possible to form the maximum units, culture and civilization.

7 . Constant effort should be made to invest all units of what- ever magnitude with the greatest possible intelligibility in both the cultural and the social aspects.

SUMMARY

57

Part II

Historical-Developmental Interpretation

Chapter 3

The Historical-Developmental

Approach in American

Archaeology

TOWARD A SYNTHESIS OF NEW WORLD PREHISTORY

Culture-historical integration, as we have seen, is the descriptive process concerned with cultural forms, with plotting these forms in space and time, and with defining their relationships and inferred functions. There is, we think, common agreement that these are archaeology's primary tasks, on the descriptive level of organiza- tion. It is with these tasks in mind that we have attempted to formu- late intelligible archaeological "units" for study. The questions of how such units may be named and defined and how they may be related one to another are, we maintain, the foundation of theoreti- cal formulations on the explanatory level of organization, i.e., in the fields of culture-continuity and culture-change. Of the various trial examinations and hypotheses that might be built upon this foundation, a synthesis of New World archaeology as a whole is one that falls short of the higher level of organization, occupying a sort of gray borderland between description and explanation. The historical-developmental interpretation presented in the following pages is an attempt at such a synthesis.

We have already spoken of the tendency for large-scale area chronologies to take on the characteristics of developmental schemes. It might then be supposed that the difference between the

THE HISTORICAL-DEVELOPMENTAL APPROACH IN AMERICAN ARCHAEOLOGY

61

synthesis of an area and the synthesis of a continent, or a hemi- sphere, is just a matter of scale, but this does not seem to be the case. Judging by the record, there seems to be a tacit agreement among Americanists that, whereas archaeological reconstructions of localities, regions, and areas may be treated in unified systems, we are unable to extend these systems to continental or hemispheric dimensions. This fact, in itself, is a highly interesting aspect of New World culture-history. That the great American art styles rarely, if ever, pass beyond the geographical limits of the areas of their characterization is surely a key to cultural understanding. Similarly, the relative scarcity of specific objects, "trade items," native to one area hearth and transmitted to another is a clue to the areal isolation of New World cultures. What, then, is the nature of the historical evidence that links the various American areas?

Let us narrow the question somewhat by asking which, if any, of the large-scale unit concepts defined in the first part of this study may be used to integrate archaeological data on an interareal basis. Our so-called maximum units, culture and civilization, fail to qualify, for the reason stated in the paragraph above. Archaeologi- cal areas in most cases owe their existence to previous formulations of those maximum units with which their boundaries are, initially at least, thought to be coterminous. Often, room has subsequently to be made for other cultures and civilizations, and thus the areas tend to be subdivided. Only rarely are they combined into larger areas owing to a discovery of cultural distributions wider than first supposed, and this again is a significant commentary on the isolable nature of American cultures.

But what of the integrative units, horizon and tradition? The horizon is characterized by its relatively limited time dimension and its significant geographic spread. It is usually expressed by an art style or a very specific complex of features of unmistakable histori- cal uniqueness. As we have just noted, such phenomena do not fre- quently pass outside the major areas in which they originated. The horizon clearly has little utility in multiple-area syntheses. What

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

62

then of the tradition? We have defined the tradition as "a (primari- ly) temporal continuity represented by persistent configurations in single technologies or other systems of related forms." Attention is called to the fact that, apart from the temporal factor — the essen- tial characteristic of all traditions is persistence in time — this defi- nition is extremely flexible. We were purposely vague about the behavior of traditions in the spatial dimension — and still are. Transmission of traditions through generations of people living in the same place is so obvious as to require no further comment, but how are they transmitted to other people living in other places? This is a question that is surely in need of further investigation. Does a tradition become diffused widely enough and rapidly enough to be usable as an integrating device in culture-historical schemes of larger than areal scope? American maize agriculture may be taken as an example of the sort of tradition we are talking about. It has distinctive content (Zea mays and its hybrids) as well as technical skills and utensils associated with its cultivation and preparation. Its tremendous spatial and temporal range in the New W)rld makes it unsuitable for tracing out detailed interareal rela- tionships; nevertheless it has an important historical unity. It is an American tradition whose diffusion has overridden areal boundaries and whose presence establishes common ties. Pressure flint-chipping and pottery-making are traditions of even broader scope, so broad as to be almost without historical significance. There are other tra- ditions, however, whose range is more restricted and whose ap- pearance might, therefore, be taken to indicate a higher degree of historical intimacy. The metallurgical techniques which extend from the south Andes northward to the southwestern United States furnish an example. They cross at least five major archaeological areas but are almost certainly a historical unit. The resist-dye painting of pottery (negative painting) , the ceramic mold, curious and complex vessel forms such as the stirrup-mouth jar — all these are multiple-area phenomena, each presumably of single historical origin.

THE HISTORICAL-DEVELOPMENTAL APPROACH IN AMERICAN ARCHAEOLOGY

63

But can traditions, of whatever kind, be integrated into an effec- tive historical system for the Americas as a whole? Horizons, as we have shown, enable us to equate local and regional phases in a reasonably precise spatial-temporal scheme for a particular area. Horizontal equivalence on the space-time chart means approximate contemporaneity. It might be supposed that traditions of the wide- spread sort referred to above would have a similar usefulness on a scale larger than that of the area, would operate in effect as super- horizons, but such does not appear to be the case. Maize agriculture, for example, did not commence at the same time, or even at approxi- mately the same time, throughout its range of distribution; nor can the Peruvian and Mexican cire-perdue casting be established as contemporaneous. There is, in short, no device (except calendrical dating) by which cultures can be temporally equated on a continen- tal or hemispheric scale. Traditions and horizons are, we repeat, the most effective units for the presentation and interpretation of intra-areal relationships, but this is the limit of successful spatial- temporal integration.

For larger syntheses, another type of formulation must be re- sorted to, one that is free from strict limitations of space and time yet has a general historical validity in the widest sense. The only possible kind of scheme that meets these requirements, so far as we can see, is a series of cultural stages in what we have chosen to call a historical-developmental sequence. These stages will of course be founded on common participation in important historical- ly derived traditions, but their formulation is a procedure distinct from the methods of systematic historical (spatial-temporal) inte- gration as laid down in the first part of this book. The historical- developmental sequence remains, however, as we shall attempt to demonstrate, essentially on the descriptive level of organization.

Possibly because of the strong reaction in this country against what is disdainfully referred to as "nineteenth-century evolution- ism," overt developmental classifications are comparatively new in American archaeology. There were, however, perspicacious stu- dents who saw what was the true nature of area "chronologies"

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

64

then current.1 Such recognition is reflected in the increasing use of nomenclature emphasizing development, e.g., Frank Roberts' modification of the Pecos classification, using terms like "Develop- mental," "Great," and "Recessive Pueblo" in place of Roman nu- merals,2 but these were still conceived as chronological periods, and archaeologists felt obliged to define them in strict calendrical terms. The Ford and Willey scheme for the eastern United States, published in 1941, was a succession of "stages" rather than peri- ods, but these were based on "chronological profiles," and the style of the paper is narrative and historical.3 James B. Griffin's slightly later synthesis of the same area showed no concern to differentiate between cultural development and chronology. "Suc- cessive cultural stages throughout the eastern United States can be erected on the basis of local stratigraphy, the interchange of specific cultural items and the common possession of definite cultural con- cepts at specific chronological periods ." * The scheme put forward in the gigantic Cole Anniversary Volume, Archeology of Eastern United States, under Griffin's editorship, shows the same disinclina- tion to differentiate between "stage" and "period." The terms are in fact interchangeable, as in this passage: "The description of these periods will emphasize a generalized cultural picture of the major features which characterize each of these stages by and large throughout the entire area."5 About the same time, the authors of the first comprehensive textbook of North American archaeology organized the eastern data into four "chronological stages,"6

1. See Kluckhohn and Reiter, 1939, p. 159: "Probably the single fact of greatest general import which has emerged thus far from the Be 50-5 1 excava- tions is that the various stages recognized by the Pecos Classification (and very commonly referred to as 'periods') do not, necessarily, represent separate and clear-cut time periods, even in the same geographical locality."

2. Roberts, 1936*, 1937.

3. Ford and Willey, 1941.

4. Griffin, 1946, p. 39. (Italics ours.)

5. Griffin's own contribution, "Culture Periods in Eastern United States Archeology," in Griffin, 1952a, p. 352. (Italics ours.)

6. Martin, Quimby, and Collier, 1947, p. 232.

THE HISTORICAL-DEVELOPMENTAL APPROACH IN AMERICAN ARCHAEOLOGY

65

thereby having it both ways. More developmental in terminology, and to a certain extent in implication, are a number of Middle American and Peruvian syntheses of more recent appearance.7 With one exception, to which we shall presently return, these schemes are all on an areal scale and adhere to their supporting chronology more closely perhaps than the North American schemes just mentioned. Nevertheless, they show a new awareness of the possibility, not to say necessity, of distinguishing between stages in the development of cultures and archaeological periods. In his summary of the 1947 Conference on Peruvian Archaeology, at which a number of these syntheses were presented, Kroeber was impelled to say: "Another query that began to strike me as Bennett read his paper, and that kept recurring during our two days, was how far schemes of development like these also exemplify or imply general schemes of historic evolution of civilizations — the most famous and complete example, of course, being that of Toynbee."8 From the developmental standpoint, nearly all area schemes that have so far appeared are subject to the same difficulties. They have followed in the wake of archaeological chronologies and have found it impossible to break free of them. Their strength has derived largely from the supporting chronology, and this very fact has led to misunderstanding. Cultural development has a time dimension, but this dimension is not necessarily uniform for all localities or re- gions of an area. When stage and horizon (or period) lines are represented on a chart as straight and horizontal, confusion is in- evitable. A similar difficulty arises when we attempt to extend de- velopmental stages from one major area to another, except that within the area the time discrepancies probably will not be so great. Students of Andean and Middle American archaeology have ra- tionalized the problem by insisting upon the homogeneity of the "area co-tradition" and by believing that developmental change will be more or less uniform throughout the area, with no re-

7. Armillas, 1948; W. C. Bennett, 1948; LarcoHoyle, 1948; Steward, 1948; Strong, 1948a; Willey, 1948a; Caso, 1953.

8. Kroeber, 1948, pp. 115-16.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

66

gional retardations. This is an unlikely possibility, certainly not a demonstrated fact.9

The scheme for eastern North America, referred to above (Ford and Willey, 1941), employed stage lines rising obliquely from south to north, and, apart from the circumstance that some of these lines now appear to have been sloped the wrong way, it is likely that this conception better fits the facts of chronology and develop- ment. However, when all chronology is relative, as in this particu- lar scheme, there are no firm landmarks for establishing relative rates of cultural change from region to region. This will be one of the important roles of true area chronologies based on independent dating.

When we shift from an areal to an interareal consideration of the New World data, we note that only two comprehensive develop- mental schemes have been formulated with recent archaeological evidence.10 Julian Steward's "functional-developmental" classifica- tion of American high cultures is a scheme for equating the major developmental stages of Middle American and Andean civiliza- tions. These equations are made on the basis of technology for the earlier stages and upon what might be called general configuration in development for the later. Although Steward's primary interest in this particular paper was the areas of American high cultures (or civilizations, as we have chosen to call them), his classification has much wider applicability. He defines six "periods" which are in effect developmental stages: (1) "Pre-agricultural" ; (2) "Basic Agricultural Beginnings" ; (3) "Basic or Inter-areal Developmental or Formative"; (4) "Regional Developmental or Formative"; (5) "Regional Florescent"; and (6) "Empire and Conquest." In a later article Steward proposes a similar classification of world-wide

9. Extension of the concept of area co-tradition to the North American Southwest by Martin and Rinaldo (1951) ran into immediate difficulties on this account. See critiques in Rouse, 1954; Willey, 1954.

10. Steward, 1948, and Krieger, 1953. Pedro Armillas has recently prepared a synthesis of New World culture-history for the "Program of History of the Americas" of the Pan-American Institute of Geography and History (MS of September, 1954).

THE HISTORICAL-DEVELOPMENTAL APPROACH IN AMERICAN ARCHAEOLOGY

67

scope. With some changes in terminology, and with a considerable expansion of later "eras," this revised scheme is basically the same as the first.11 Steward, in effect, sees the following major stages (which he terms "eras") in the development of native American civilizations: first, a pre-agricultural hunting and gathering stage; second, the beginnings of agricultural experimentation; third, the establishment of agriculture and permanent village life with spe- cial ceremonial and political institutions; fourth, a climax of the agricultural pattern with resultant artistic and technical elaborations and the maximum development of "religious states"; and, fifth, the rise of urbanism, the secular state, and the formation of empires.

Alex Krieger has recently proposed a somewhat similar classifi- cation. At the International Symposium on Anthropology held by the Wenner-Gren Foundation in New York in 1952, he read an in- ventory paper on the archaeology of "Anglo-America," in which an immense amount of recent data was presented but without any formal chronological or developmental structure.12 The latter he supplied in the discussion period in a statement, subsequently re- vised for publication, which not only presents the first adequate developmental scheme for North America as a whole but also con- tains the clearest discrimination between the concepts of stage and period that we have yet seen in print :

For present purposes, I will consider a "stage" to be a segment of a his- torical sequence in a given area, characterized by a dominating pattern of economic existence. The general economic life and outlines of social struc- ture of past peoples can often be inferred from archaeological remains and can be related to similar phenomena, whether the dates are known or not.

11. Steward, 1949a. In this study the eras are listed as: (1) "Hunting and Gathering"; (2) "Incipient Agriculture"; (3) "Formative"; (4) "Regional Florescence"; (5) "Initial Conquest"; (6) "Dark Ages"; (7) "Cyclical Con- quests"; (8) "Iron Age Culture"; and (9) "Industrial Revolution." Era 3 com- bines Periods 3 and 4 from the first classification. Era 4 is the same as Period 5 in the first classification. Eras 5, 6, and 7 parallel Period 6 of the first classifica- tion. Eras 8 and 9 are beyond the developmental range of New World civiliza- tions.

12. Krieger, 1953.

68

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

The term "period," on the other hand, might be considered to depend upon chronology. Thus a stage may be recognized by content alone, and, in the event that accurate dates can be obtained for it in a given area, it could be said that the stage here existed during such-and-such a period. Further, the same stage may be said to appear at different times or periods in different areas and also to end at different times. A stage may also include several locally distinctive culture complexes and minor time divisions. A great deal of discussion is needed on these points.13

Krieger's scheme employs four major stage divisions: "Paleo-In- dian";14 "Food-gathering"; "Food-producing"; and "Urban Life." It will be noted that, although his main concern is still with "Anglo- America," i.e., North America north of Mexico, the classification is one that would accommodate Middle and South American data as well. His definitions parallel those of Steward rather closely. "Paleo-Indian" does not have any real counterpart in Steward's scheme, evidence of this order of antiquity in the high-culture areas being so slight at the time of Steward's formulation. Krieger's "Food-gathering" is approximately equal to Steward's "Pre-agri- cultural," with an overlap into "Basic Agricultural Beginnings" perhaps, depending on where one prefers to put the emphasis in cultures of mixed gathering-food-producing economy. Krieger's "Food-producing" would then take the rest of Steward's "Basic Agricultural Beginnings" and both his "Formative" stages, and Krieger himself has made it explicit that he would bracket Stew- ard's "Regional Florescent" and "Empire and Conquest" together in the stage he has designated as "Urban Life."

A comparison of the two classifications brings out the somewhat different interests of the two classifiers. Steward was focusing his attention on the areas of "high civilization" in the New World. Krieger, while allowing for data of this type, was thinking in terms of the simpler cultures of North America. The discrepancy be- tween his "Food-gathering" and Steward's "Basic Agricultural Be- ginnings" reflects this. Steward, with the Peruvian sequence data

13. Krieger, in Tax et al., 1953, p. 247.

14. Krieger has recently proposed the term "Paleo- American" in place of "Paleo-Indian" (Suhm and Krieger, 1954, p. 15).

THE HISTORICAL-DEVELOPMENTAL APPROACH IN AMERICAN ARCHAEOLOGY

69

on the Huaca Prieta15 premaize agriculturists in mind, conceived of this stage as a slow transition from hunting and gathering to agriculture. Krieger, to judge from his examples, was considering the California cultures and the Archaic cultures of the eastern United States as representative of the "Food-gathering" stage. Actually, either concept might accommodate both the South and the North American data. The Huaca Prieta culture of the Peruvian coast depended for food on gathering as well as on agriculture; and there is a reasonable possibility that some of the Archaic cultures of the eastern United States were experimenting with plant domestica- tion.

To date, neither the Steward nor the Krieger scheme has been applied to New World data in complete detail. These authors may eventually do so. We make no claim that our effort here is a substi- tute for the particular interpretations of either. The ideas they have proposed, however, are of such wide and general interest to Ameri- canists that additional points of view are desirable. In our previous paper we put forward our particular scheme and subjected it to a trial run by projecting a considerable amount of data against it. This we propose to repeat here with certain modifications that we will describe presently.

Before doing so, however, we think it necessary to enter the controversial domain of culture-evolutionary theory far enough at least to demonstrate our contention that ours is not an evolutionary scheme. In a paper written after the ones quoted above, Steward defined with considerable precision three evolutionary approaches in anthropology: unilinear, universal, and multilinear.16 It does not seem necessary to support with long-winded arguments the state- ment that the developmental classification proposed in these pages conforms to neither of the two first-named approaches. The skepti- cal reader is advised to read Steward's paper, and if he still thinks that what we are describing is unilinear or universal evolution (as therein defined) , then we have completely failed to explain what we

15. Bird, 1948a, 1948£.

16. Steward, 1953.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

70

are about. Whether it be multilinear evolution is a more difficult question. In the general sense that we are concerned with a line of cultural development in one part of the world, while recognizing that there are other quite unrelated lines of development in other parts of the world, this might be called multilinear evolution.

But, if we understand Steward correctly, particularly his applica- tion of the theory in recent publications,17 he is concerned with tracing multiple lines of evolutionary development leading to specific kinds of cultural features or whole cultures (his "culture types") wherever these may be found. His formulations are limited in scope, more so than ours perhaps, but they are universals in the sense that they are unrestricted geographically or chronologically. A "theocratic irrigation state" can be designated as such whether found in Peru or Mesopotamia, and the stages leading up to such a culture type — if truly parallel — are reflections of inherent cul- tural causality. Historical causality is important only in that it has to be eliminated from the equation. Thus, in terms of basic theory, as expounded in the introductory section of this book, Steward's multilinear evolution is on our explanatory level. We, on the other hand, as already hinted, are classifying cultures in a sort of theoreti- cal twilight. Cultures A and B are classified as Archaic because they possess certain common denominators that we have chosen as criteria for that stage. Their common possession of these features may be the result of historical contact, environmental determina- tion, homotaxis in a truly evolutionary sense, or any two or all three of these. In other words, the system, if it can be called a sys- tem, is not rigged to exclude any particular kind of explanation. This is why we have stated that, although it seems to lie outside the domain of culture-historical integration, above the descriptive level of organization, it cannot claim to operate on the explanatory level. This is also why we have chosen to call it "historical-develop- mental" interpretation, which sounds like a contradiction in terms but which expresses the fact that our stage concepts are not wholly abstracted from the historical matrix. Within the geographical

17. Steward etal., 1955.

HE HISTORICAL-DEVELOPMENTAL APPROACH IN AMERICAN ARCHAEOLOGY

71

frame, large as it is, contact and diffusion may have played a larger role than set deterministic relationships within the province of cul- ture alone. Man's biological and subcultural psychological and emo- tional reactions cannot be omitted from the equation, nor can the reality of the social groups in which he exists be denied its forma- tive role. Environment and increases or decreases in size and dis- tribution of population, reflected in cultural-ecological adaptations, are also significant factors exerting strong influences on the nature and direction of cultural change.

STAGE DEFINITIONS AND FORMULATIONS

In selecting criteria with which to define the stages of New World culture-history, it is obvious, from the foregoing review of the Steward and Krieger schemes, that there are but two broad divisions of a fundamental technological and economic nature: hunters-gatherers and agriculturists. This is analogous to the broad division in the Old World between the Paleolithic-Mesolithic stages, on the one hand, and the Neolithic and later stages, on the other. As in the Old World, these two general patterns of life have a sequence relationship, with the hunters-gatherers preceding the farmers. And, as is also the case in the Old World, there is a con- siderable chronological overlap, with hunting-gathering cultures persisting in some regions into periods of contemporaneity with farming cultures. It should be recognized that no other differentia- tion between stages with which we will deal has the same pro- fundity and significance as this one. In this we are in agreement with Robert Braidwood18 and Robert Redfield,19 who see the Old World "urban revolution," or "dawn of civilization," as something that was made possible by the establishment of agriculture several millenniums earlier, but not as marking a technological and econom- ic shift as profound as the one from food-gathering to food produc- tion. This point of view is reflected in our stage criteria and termi- nology. The criteria for dividing pre-agricultural stages are essen-

18. Braidwood, 1952.

19. Redfield, 1953.

METHOD AND THEORY *7 ") IN AMERICAN ARCHAEOLOGY / C.

tially technological. They refer to artifact types and traditions in technology. The criteria for dividing stages above the threshold of agriculture take reference in much more complex data. They per- tain to social and political organization, religion, aesthetics — to the whole of what Redfield has termed the "moral order."

Our method of formulating stages was to review archaeological sequences from all parts of the New World in local and regional de- tail, on an area basis and with reference to cross-areal comparisons. Our first concern was to set up a series of trial stages by seeking for clues to generalization between areas. We drew generalizations, tested them by going back to primary sources on the regional and local levels, and thus came to recognize certain "common denomi- nator" criteria for each of the stages. Needless to add, with a pro- cedure of this kind, our trial stages and the criteria of definition changed constantly. This was true during preparation of the first published article,20 and there is testimony to the fact that changes have continued in this present version. In 1955 we proposed six historical-developmental stages. We are now proposing five stages, with some changes in terminology and in criteria of definition. A quick comparison of the two schemes may be set down as follows :

1955 1957

Postclassic stage Postclassic stage~

Classic stage Classic stage

Formative stage Formative stage Preformative stage

Archaic stage Archaic stage

Early Lithic stage Lithic stage

The present "Lithic" corresponds almost exactly to the former "Early Lithic"; reasons for dropping the "early" will be given in the next chapter. The "Archaic" stage remains about the same, except for some modifications resulting from the elimination of the "Preformative." Phases and cultures formerly assigned to the "Preformative" have been divided between the "Archaic" and the

20. Willey and Phillips, 1955.

THE HISTORICAL-DEVELOPMENTAL APPROACH "1 "7 IN AMERICAN ARCHAEOLOGY / J

"Formative." Otherwise the "Formative," "Classic," and "Post- classic" are essentially unchanged in so far as archaeological con- tent is concerned; in definition and emphasis upon criteria, there have been some changes.

Before we set forth our definitions of these stages, it may be well to consider briefly a difficulty that inheres in all developmental in- terpretation, to wit, the familiar problem of cultural lag. We have found it useful to differentiate between two kinds of lag situations : belated and marginal. The first is one in which a culture shows the essential characteristics of a given stage long after the time general- ly considered to be appropriate to that stage. This is in effect a temporal marginality, often referred to as a "late marginal situa- tion." The second kind of situation, to which the term "marginal" applies primarily in a cultural sense, is one in which a culture shows significant characteristics of a given developmental stage in a con- figuration which, as a whole, fails to measure up to the definition of that stage. In the first, it is primarily a matter of persistence without change; in the second, it is a matter of change that, from the standpoint of the developmental criteria employed, is one-sided or incomplete. This second type of cultural, as opposed to merely temporal, marginality can also under certain conditions be the re- sult of developmental regression, as we shall have many occasions to point out.

Marginal cultures are usually belated as well; so, in effect, the distinction we are trying to make is between cultures that are sim- ply belated and those that are both belated and marginal, in the sense in which we are using these terms. Figure 2 is a very crude expres- sion of the relationship, which may be clarified by the use of ex- amples. Culture A in this figure is a belated Lithic culture, some- what theoretical, it must be confessed, because our definition of Lithic is not so unrestricted temporally as some of the other stages. This would be a culture late enough to have received influences from Archaic (or later) stage cultures but sufficiently immune to such influences to permit us to classify it as Lithic without any qualifications beyond those implied in the term "belated." In

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

74

simpler language, some Archaic elements may be present, but these are not sufficient to raise any classificatory difficulties. Certain late Lithic cultures in the North American Great Plains might be used as examples. Culture B, on the other hand, is marginal Archaic in the sense that it has incorporated sufficient Archaic elements to take it out of the Lithic category but not enough to qualify fully as an Archaic stage culture. Many, if not all, Plains Archaic cul-

Postclassic

Classic

Formative

Archaic

Lithic

Fig. 2. — Belated and marginal cultures in relation to developmental stages

tures appear to be of this character. Culture C is a belated Archaic culture, and here we are on firm ground, for there are many ex- amples available, some coming right down to very recent times. Most of the ethnographic cultures of the peripheral areas of North and South America could be classified as belated Archaic. Culture D, marginal Formative, is well illustrated by many of the Middle Woodland cultures of eastern North America, which received sig- nificant elements of the Formative Adena-Hopewell culture with- out profound modification of their essentially Archaic pattern. There are many marginal Formative cultures of the regressive type. Since Formative culture presupposes a stable economic base, usually

THE HISTORICAL-DEVELOPMENTAL APPROACH IN AMERICAN ARCHAEOLOGY

75

agricultural, movement of cultures into areas unfavorable to that base or an unfavorable change of climate will often result in re- gression to an Archaic stage of development. Culture E on the dia- gram is belated Formative. This is the situation of many New World Formative cultures. Classic and Postclassic stage cultures are all so late that it does not add anything meaningful to distinguish some of them as belated, and so these are not shown on the dia- gram. Marginal Classic, F in the diagram, is well represented in lower Central America, in cultures such as Nicoya, Costa Rica, etc. Culture G is a rather special case intended to show that develop- mental marginality may, under certain conditions, "skip" a stage. An example is to be seen in the Quebrada de Humahuaca region of northwestern Argentina, where Inca influence operating on an es- sentially Formative level has produced a culture of marginal Post- classic type.

The point of this apparent hair-splitting is that a vast number of New World cultures are belated or marginal in the sense here given these terms, and in many cases their classification hinges on our ability to say which. There is probably no such thing as a pure be- lated culture, i.e., one that fails to show any elements of higher de- velopmental stages whatever, and there is probably no such thing as a marginal culture that is not belated to some degree. So it usual- ly comes down to the question of the number of developmentally significant elements and their relationships to the total configura- tion, which will in fact determine whether we classify a culture as a belated culture of a given stage or a marginal culture of a higher one. These qualifiers are designed to enable us to hedge on difficult classificatory decisions, and they will be used pretty freely in the pages to follow. This will perhaps contribute little in the way of culture-developmental understanding but will at least point up some of the difficulties of large-scale developmental interpretation.

The remaining pages of this book will be devoted to comprehen- sive definitions of the five stages enumerated above, to their identi- fication, examination, and evaluation in terms of archaeological se- quences and syntheses. Before we start this trial run, let us make it

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

76

very clear that we are under no illusion that this is anything re- motely resembling a natural system. There is nothing inevitable about five stages; there might as well be four, or even eight. Nor is there any law, evolutionary or otherwise, that says that all New World cultures must pass through these stages one after the other in their proper order. We cannot even say, for example, that an Archaic stage culture will "in the natural course of events" pass on to the Formative stage. Whether or not it will do so depends on a multiplicity of factors, not all of which are in the domain of culture. As we have previously stated, our basic theoretical position is that culture is not an independent order of phenomena intelligible in terms of itself alone (the "cultural superorganic") , and from this it follows that there are no universal, irreversible processes of cultural development. We are not, in short, attempting to impose an evolu- tionary or any other kind of determinism on the data of New World archaeology. Nor can we accept the criticism that ours are hot the "right" stages, if such criticism carries the implication that "right" stages are to be found. This attitude involves an important method- ological distinction between what we have called culture-historical integration and developmental interpretation, our present concern. The aim of the first was the organization of archaeological data in terms of a real world. However far short of attainment it falls, our object was the formulation of archaeological units that have (or rather had) a correspondent historical reality. It therefore seems possible that such units may be judged as to their "rightness." In developmental interpretation, on the other hand, we abstract, not only from the primary data but from the above-mentioned inte- grated units as well, certain characteristics that seem to have sig- nificance from the point of view of the general development of New World culture. In a general sense the sequence is historical as well as developmental, but the individual stages can have no correspond- ent historical unity or reality. Therefore the test of "rightness" is irrelevant. This also disposes of the idea that developmental stage formulations must not be "set" until we know all the archaeology — which will be never.

THE HISTORICAL-DEVELOPMENTAL APPROACH IN AMERICAN ARCHAEOLOGY

77

Finally, it is clearly impossible in the space available to survey the entire field of American archaeology. The most we can do is to sample liberally from all areas and periods in the hope that no sig- nificant block of data is overlooked altogether. It is not to be ex- pected that the specialist in any area will be satisfied. We may plead in extenuation, however, that the purpose of our inquiry is not to effect a tight classification of New World cultures but to explore the possibilities of this kind of approach.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

78

Chapter 4

Lithic Stage

DEFINITIONS

In our previous paper we designated and defined the earliest stage of American culture as "Early Lithic."1 This stage was conceived of as embracing two major categories of stone technology: (1) un- specialized and largely unformulated core and flake industries, with percussion the dominant and perhaps only technique employed, and (2) industries exhibiting more advanced "blade" techniques of stoneworking, with specialized fluted or unfluted lanceolate points the most characteristic artifact types. At that time we hesitated to use this division as a basis for setting up two separate stages in our classification, because there seemed to be insufficient evidence of a time differential between them. The evidence is still inconclusive, although the case for a distinct and earlier core and flake stage is somewhat stronger than it was. We still adhere to our original de- cision to consider all these early evidences of man in America as a single stage, but we recognize the good possibility of an eventual division along the lines suggested above. In view of this possibility, we have modified our terminology and will refer hereinafter to this earliest New W)rld stage simply as "Lithic," allowing for future separation into "upper" and "lower" Lithic stages, if such a course seems advisable.

"Lithic" is not entirely satisfactory as a name, but we have so far been unable to come up with anything better. Its sole merit is that the evidence on this stage is predominantly in the category of

1. Willey and Phillips, 1955, pp. 730-39.

LITHIC STAGE

79

stone technology (although there is an increasing number of early bone finds). The terms most commonly applied to this stage, "Paleo-Indian" or "Paleo-American," are open to a serious objec- tion, in our view, in that they imply a grand twofold division of New World prehistory, with a "Neo-Indian" or "Neo-American" stage as a counterpoise. There is a grand twofold division, as we have shown, but the hinge comes later in the sequence, viz., between the Archaic and the Formative.

The Lithic stage cannot be defined without reference to geo- chronological considerations. Illogical as it may appear in a de- velopmental scheme, we have to start with the flat assertion that the Lithic is the stage of adaptation by immigrant societies to the late glacial and early postglacial climatic and physiographic condi- tions in the New World. The effective working criteria are, there- fore, associations of artifacts and other evidences of man's activity in geological deposits, or with plant and animal remains, reflecting these times and conditions. When types so established are found without such associations, which is more often the case, the rela- tionships are inferred in the normal archaeological manner, with the result that an immense corpus of Lithic stage data is available in the literature but has not so far been subjected to effective inte- gration and synthesis.

The nature of these finds has led to the assumption that the pre- dominant economic activity in this stage was hunting, with major emphasis on large herbivores, including extinct Pleistocene forms, and that the general pattern of life, like that of the animals on which it depended, was migratory in the full sense of the word. The possi- bility of a measure of circularity entering into this assumption can- not be investigated here but certainly should not be ignored. In any case, upon this basis many American students have erected a simple historical and typological dichotomy: an early hunting stage fol- lowed by a gathering stage, each with its own characteristic tech- nological traditions. In the present classification, although we have not been able to avoid this attractive simplification altogether, we will try to maintain a critical attitude toward it. An early non-agri-

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

80

cultural society would presumably make the best possible use- — within the limits of its culture — of whatever animal and vegetal resources were available; consequently, the relative importance of hunting and gathering would be a function of the ecological and technological balance in a particular place and time. Quite pos- sibly, however, the natural conditions in the late glacial and early postglacial times postulated for the Lithic stage may actually have been more favorable generally to hunting than to gathering, and technological limitations (lack of milling techniques and imple- ments?) may have contributed to tip the balance further. We would say, then, for purposes of definition, that the Lithic is pre-emi- nently a hunting stage, though other economic patterns were cer- tainly present whose local dominance, under certain conditions, is not precluded.

Known details of culture in the Lithic stage are few. Lithic stone technology covers an immense range of rough- and chipped- stone traditions2 but does not include the practice of grinding and polishing. Work in bone and horn is assumed to have been impor- tant, but the evidence has largely disappeared.3 Settlement and habi- tation patterns were such as to leave few traces in the ground. Cor- related with the absence of house remains is the scarcity of deep refuse deposits. Sociopolitical inferences for this stage are hazard- ous in the extreme. A small-scale kinship type of organization is postulated, but, within this generalization, great variability and a high degree of specialization must be allowed for. The data do not support the view that because Lithic cultures are relatively simple they are also uniform. Let us not forget that the major portions of both American continents were "pioneered" on this stage.

2. "Rough" refers to artifacts shaped by use rather than design; "chipped," to artifacts shaped by techniques ranging from crude percussion to controlled pressure flaking and fine-edge retouching.

3. Sufficient evidence of early bone finds has been summarized by Krieger (1951a; 1953, p. 242) and by Cotter (1954, p. 65) to suggest that, if condi- tions were more favorable for preservation, traditions of bone technology might rank with stone in the definition of Lithic stage culture.

LITHIC STAGE

81

A POSSIBLE EARLY PERCUSSION STAGE OF STONE TECHNOLOGY

Before considering the better-known Lithic cultures of the New Wforld, we may as well first get the problem of the putative "lower" Lithic stage out of the way. There is considerable documentation for such a stage, but its evaluation is extremely difficult for the present writers, in whose fields of specialization stone technology plays a distinctly minor role. It seems to us that, if this is ever to be more than a theoretical stage in the development of American culture as a whole, certain questions have to be answered in the affirmative: (1) Have stone complexes of this nature been found unmixed in specific sites under controlled conditions? (2) Do they constitute true assemblages, i.e., can the possibility be ruled out that they represent only partial inventories of the total tool com- plex owing to special conditions? Tools from a quarry site or a butchering station, where only a segment of the total economic ac- tivity is represented, might be cases in point. (3) Finally, and most importantly, can it be shown that there is a time differential be- tween these percussion industries and the more specialized indus- tries that would be classed as "upper" Lithic if such a division were made? This does not mean, of course, that all percussion as- semblages must be early but that some of them are is a minimum requirement.

The outstanding proponent among Americanists for this early percussion stage is Alex Krieger, who referred to some of the evi- dence in its favor in his paper at the International Symposium on Anthropology in 1952. More recently he summarized the evidence in greater detail at an informal meeting of American archaeologists held at Andover, Massachusetts, in April, 1956. 4 The finds re- garded as significant by Krieger are briefly as follows: (1) Heavy percussion-flaked tools and sharpened bone splinters found at Te- quisquiac in the Valley of Mexico in 1870, deep down in the Up-

4. We are grateful to Alex D. Krieger for allowing us to use this informa- tion and to Frederick Johnson, of the R. S. Peabody Foundation, who supplied us with a transcript of the meeting.

METHOD AND THEORY Q *} IN AMERICAN ARCHAEOLOGY O <L

per Becerra formation only two feet above the caliche layer that separates it from the Lower Becerra — caliche estimated by geolo- gists to be of mid-Wisconsin date.6 (2) An obsidian flake, crude choppers and scrapers, and a split and pointed bone implement, from Tule Springs, Nevada.6 Charcoal thought to be of human origin from this site was dated as older than 23,800 years.7 Faunal associations included mammoth, horse, and (mostly) camel. (3) Worked bones including a bone tube from the famous Potter Creek Cave in north- ern California, excavated by paleontologists in 1870. The deposit contained an abundant middle to late Pleistocene fauna. After con- siderable controversy,8 the finds were rejected as "water-worn bones," but they have been seen recently by Krieger, who declares them to be artifacts.9 (4) Very crude chipped stones of possible hu- man workmanship and a bone tube found with a rich late Pleisto- cene fauna in Friesenhahn Cave in south-central Texas.10 The bone tube, found beneath the skeleton of a saber-toothed cat, is similar to the one from Potter Creek Cave and may, in Krieger' s opinion, help to authenticate the otherwise dubious human origin of the stone objects. (5) Three rude sandstone heads found underneath twenty-six feet of gravel near Malakoff, Henderson County, Texas,11 not generally accepted by archaeologists but believed by Krieger to be "probably" of human workmanship. If so, there is not much to be said for the skill of the workman. (6) Basin-shaped hearths on Santa Rosa Island, off the southern California coast, containing masses of split bones of dwarf mammoth but no un-

5. The artifacts are mentioned and illustrated in dc Terra, 1949, pp. 46, 66, Pis. 9/, 10*. Radiocarbon dates earlier than 6000 B.C. have been obtained from the top few inches of the Upper Becerra formation; according to Krieger, the artifacts must be considerably older than this date.

6. Harrington, 1955.

7. Sample No. C-914 (Libby, 1954*).

8. For a selected bibliography of the dispute over the Potter Creek finds see Sellards, 1952, p. 123.

9. Krieger, 1953, p. 240. 10. Sellards, 1952, p. 94, Fig. 43. 11. Sellards, 1941; 1952, pp. 99-105, Fig. 47.

LITHIC STAGE

83

mistakable artifacts, which yielded a very early radiocarbon age of about 30,000 years (not yet published). A number of other finds of this putative pre-projectile-point category were rejected by Krieger because of uncertainty of their status as artifacts, lack of dating, or both. These include : G. F. Carter's La Jolla and San Diego finds;12 the Imlay Channel finds in eastern Michigan;13 the Black's Fork culture of southwestern Wyoming;14 the Malpais industry of the Mohave Desert in southern California;15 and the Death Valley finds in the same region.16 This list might be extended to include the Albuquerque, Comanche Springs, and Rio Puerco finds in central New Mexico;17 the Tolchaco complex of the Little Colorado River in Arizona;18 and the Farmington complex in the foothills of the Sierra Nevada of central California.19 It is manifest- ly unfair to list all these finds together without comment, as though they were all of equal standing. Some of them, especially the two last named, appear to be valid assemblages that are typologically "lower" Lithic, but confirmative evidence of early date is lacking. Our timid conclusion is that none of the finds cited by Krieger, or those added by us, fully answers the requirements of the classifi- catory problem. Some of them indicate a strong probability of the presence of man on the North American continent at a considerably more remote period than the "end of the Pleistocene" usually at- tributed to the better-known "upper" Lithic cultures. But there is a possibility that these cultures, too, have to be pushed farther back into the past. Radiocarbon ages for the Sandia culture in excess of 20,000 years have been recently published though not generally credited.20 Clovis, the earliest known culture sharing the fluted- point tradition, has not yet been satisfactorily dated, so far as the

12. Carter, 1949, 1950, 1952, 1954.

13. Baggerly, 1954.

14. Renaud, 1938, 1940. 17. Hibben, 1951.

15. M. J. Rogers, 1939. 18. Bardett, 1942, 1943.

16. Clements and Clements, 1951. 19. Treganza, 1952. 20. Sample Nos. M-247, M-349 (Crane, 1956). Cf. p. 91, n. 47.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

84

present writers are aware. A partial overlapping contemporaneity with Sandia is not an unreasonable hypothesis.21 Even when this possibility is disregarded, however, and the temporal priority of the finds that Krieger has assembled is accepted, it nevertheless re- mains a fact that the evidence in the form of artifacts associated with these remote cultures is insufficient to support the assumption of a prior state of percussion technology. To be sure, there are no points or other artifacts that suggest a "higher" stage, but in no single instance among the datable occurrences is the artifact sample large enough to be conclusive.

One of the complications of this problem of dividing the Lithic into two stages is that, in the Greater Southwest and contiguous areas, where most of the early finds have been made, the percussive techniques and characteristic tool types seem to have persisted with little change (to our unpracticed eyes) and with continued dominance into cultures that we have classified on other grounds as Archaic. Thus, the precise classificatory difficulty is to distin- guish between Archaic cultures and Lithic cultures that would be "lower" rather than "upper," if such a division were to be made. It is this, perhaps as much as the uncertainties of dating, that has kept us from making such a division.

We conclude, then, about as in our 1955 paper: The possibility of a "lower" Lithic stage of technology in the Americas remains an intriguing hypothesis. The case for temporal priority has grown stronger but remains to be proved, whereas persistence of such a percussion technology into post-Lithic times is more certain than ever. The general impression, from a world-wide standpoint, is one of marginal survival rather than technological regression, and we still feel that Linton's suggestion that the industries in question represent a belated survival of very early technological traditions distantly related to the lower Paleolithic cultures of Southeast Asia is an exciting possibility. Nevertheless, we are both unwilling

21. Compare the finds at the Lucy site, in central New Mexico, where San- dia points, including fluted examples, were found in the same "blowouts" with typical Clovis points (Roosa, 1956).

LITHIC STAGE

85

and unable to set up the "lower" Lithic as an independent stage in the present classification.

NORTH AMERICA

We move onto surer ground with the cultures that would be "upper" if a division of the Lithic were made. The best-known areas for finds on this stage are the central and southern North American High Plains and adjacent eastern foothills of the Rocky Mountains, home of a series of early cultures sometimes collec- tively referred to as Folsom-Yuma, on account of the highly spe- cialized projectile points that bear those names. Available data consist of a few thin habitation sites, a larger number of game "kills" and butchering stations, and innumerable surface finds in "blowouts," where they are associated with other objects of all ages, and in isolated spots, without associations of any kind. The nature of the finds is such as to yield a fragmentary and incomplete view of the total archaeological culture and, except in one or two lucky instances, even of the total stone assemblage. The conse- quent emphasis on projectile points — the various types of which are used for identifying the "cultures" — and on knives, scrapers, and other tools supposedly used in butchering, skinning, and pre- paring skins, has undoubtedly resulted in a one-sided view expressed in the frequent designation "early hunting cultures."

The earliest cultures of this category are Clovis and Folsom, with their celebrated fluted projectile points and extinct faunal as- sociations.22 Clovis is not only stratigraphically older than Folsom23 but is consistently associated with remains of mammoth and other Pleistocene fauna,24 whereas Folsom is usually found with an ex-

22. The Folsom site in northeastern New Mexico will long be famous as the scene of the first association of human artifacts with remains of extinct fauna in an incontestable early postglacial geological context, probably the most im- portant "breakthrough" in the history of American archaeology. See Cook, 1927; Figgins, 1927; Roberts, 1935, 1936a.

23. At Blackwater No. 1 site near Clovis in eastern New Mexico (Sellards, 1952, pp. 29-31).

24. Figgins, 1933; E. B. Howard ctal., 1935; Sellards, 1952; Haury, 1953a; Roosa, 1956.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

86

tinct bison that differs only slightly from modern forms. Clovis points are more generalized in form and distribution; Folsom seems to represent a late climax of the fluted-point tradition that is con- fined to the Plains. The Folsom culture has been dated around 8000 b.c, a date which appears to be consistent with the geological and climatic conditions.25 Since there is a rather complete shift in faunal associations between the two cultures, Clovis must be very considerably older. Thus we have to reckon with the possibility of a very early beginning and an immensely long time span for the fluted-point tradition in North America.26

Considerably more recent, but still on the Lithic stage, is the tradition of fine parallel flaking on points that used to be called Yuma but now are more often referred to as Eden-Scottsbluff. This culture, or cultures — -it is difficult to apply consistent terminology to assemblages of which only the projectile points have been ade- quately described — is stratigraphically younger than Folsom,27 and this relationship is to a certain extent confirmed by radiocarbon dates of 4000-5000 b.c.28

The older dichotomy between fluted and parallel-flaked tradi- tions, expressed in the terms "Folsom" and (now largely super-

25. 7932 ± 350 B.C. at the Lubbock site in northwestern Texas (sample No. C-558 [Libby, 1951]).

26. An unexpectedly early radiocarbon date of 35000 b.c. plus for Clovis has been recently published {Time magazine, August 6, 1956, p. 42). This re- markable date was obtained by the Humble Oil Company laboratory on char- coal from a hearth at the Lewisville site, near Dallas, Texas. A typical Clovis point was found in close proximity to the hearth. Qualified archaeologists who have investigated the finds accept the association, but the date is altogether out of line with previous estimates, a consensus of which would give to Clovis an age of about 12,000 years.

27. At the Finley site in southwestern Wyoming (Moss et al., 1951) and the MacHaffie site near Helena, Montana (Forbis and Sperry, 1952).

28. At the Horner site in northwestern Wyoming. Actual dates are 4925 + 250 b.c. (sample No. C-302 [Arnold and Libby, 1951]) and 4966 ± 500 B.C. (sample No. C-795 [Libby, 1954a]). Scottsbluff-like points at the Ft. 41 and Ft. 50 sites in the Medicine Creek locality of Nebraska are dated considerably ear- lier, but the final report on the stratigraphy and association has not yet ap- peared (Schultz and Frankforter, 1948; Davis and Schultz, 1952; Davis, 1953).

LITHIC STAGE

87

seded) "Yuma," has given way before an increasing accumulation of typologically intermediate forms, which may have been inter- mediate in time also. We refer to various point types which at first were called "unfluted Folsom" because of their general re- semblance to Clovis and Folsom in shape but which exhibit a flaking technique more like a crude approach to the fine parallel flaking of Eden-Scott sbluff. The best-documented points and "cul- tures" of this general intermediate category are: San Jon,29 Plain- view,30 Long (or Angostura),31 Portales,32 and Milnesand.33 A general continuity of these cultures with Clovis-Folsom, on the one hand, and Eden-Scottsbluff, on the other, is indicated by the other, less characteristic implements associated with these point types, which further suggests an intermediate temporal position, and the faunal associations and few available radiocarbon dates are in agree- ment.34

Thus we appear to have in the Great Plains an immensely long continuity of culture based primarily on the hunting of large mam- mals— from wholly extinct Pleistocene forms down to the modern bison — marked off into archaeological units mainly by changes in projectile-point forms. This obviously oversimplified picture masks an actual situation of great complexity that is in need of further or- ganization.

Many of the point types referred to above, particularly those of the fluted-point tradition, have a near-continental distribution. The Clovis-like points have such general distribution that it is un- safe to specify any major unglaciated areas in North America where they are not found. They seem to be relatively scarce, how- ever, on the Pacific Coast and in the nearby intermontane plateau and desert areas, and only doubtful specimens have been reported from Mexico and farther south. They also tend to diminish north-

29. Roberts, 1942.

30. Sellards, Evans, and Meade, 1947. 32. Sellards, 1952, pp. 72-74.

31. Hughes, 1949. 33. Sellards, 1955.

34. 5122 ± 300 b.c. for Plainview and Long (Angostura) points at the Allen site in southwestern South Dakota (sample No. C-604 [Libby, 1951]).

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

88

ward, a circumstance unfavorable to any theory of specific Asiatic origin. Perhaps the greatest total of fluted points have come from the Mississippi Valley and the East, but until fairly recently very little could be said about associated complexes. Now sites are being reported, and we can begin to speak of phases and cultures. In the Enterline culture of the Middle Atlantic subarea, we have Clovis- like fluted points and a complete stone assemblage with some char- acteristics that are not found in western fluted-point cultures.35 The recently reported Bull Brook site near Ipswich, Massachusetts, shows relationships to both Enterline and Clovis along with indi- vidual features of its own.36. Some of the fluted points from Bull Brook are said to be practically indistinguishable from the points found inside the carcass of the Naco mammoth in faraway Ari- zona.37 Fluted points from another New England site, the Reagan site in northwestern Vermont, show considerable variance from western norms, however.38 An important center for surface finds, the lower Ohio Valley and Tennessee-Cumberland regions pro- duced the first eastern fluted-point site, the Parrish site in western Kentucky,39 and it seems possible that the well-known "Ohio" or "Cumberland" fluted point may be the marker type for another, as yet unformulated, culture which may also include recently dis- covered sites in northern Alabama.40 It is commonly held by west- ern archaeologists that the eastern distribution of the fluted-point tradition represents a belated extension of early western hunting societies forced out of the High Plains by the advent of arid condi- tions in the altithermal period. Since there are no extinct faunal as- sociations and practically no dating of any kind for eastern fluted- point cultures, this theory is not directly refutable. However, the present trend of dating in the East is pushing the later Archaic stage

35. McCary, 1951; Witthoft, 1952.

36. Eldridge and Vacaro, 1952; Byers, 1954, 1955.

37. Byers, 1956, p. 9, citing personal communication from Emil Haury.

38. Ritchie, 1953. 39. Webb, 1951a.

40. Lewis and Kneberg, 1951; Lewis, 1953; Kleine, 1953; Soday, 1954; Mahan, 1954-55.

LITHIC STAGE

89

cultures back to a point that leaves little if any time for such a move- ment to have taken place.

Some of the point types that we have tentatively placed in an "intermediate" category, especially Plainview, also have been re- ported widely in the East, but, since the forms are more generalized, the nature of the relationship to their western counterparts con- tinues to be questionable. Complaints of uncritical use of western point typology by eastern archaeologists usually have to do with this category of types. "Plainview" points, for example, have been reported from many sites and complexes in the East, often on the basis of superficial resemblances that may have no culture-historical significance whatever. The contexts are usually Archaic, which al- lows us to sidestep the problem momentarily.

Parallel-flaked points of Eden-ScottsblurT types are less often reported in the East. Their general distribution in North America seems to be more northerly than that of the earlier fluted-point types, which possibly reflects a northward drift of the early Plains hunting cultures and their quarry with the slow retreat of the late Wisconsin ice.

In the areas considered so far, most if not all of the early finds pertain to hunting cultures, and the relationship between hunters and gatherers is not a problem. On the other hand, when we turn to the Southwest and the contiguous Basin and California desert areas, we find an environment which, for a period that extends very far back into the past, fostered a gathering-small-game-hunting way of life, for which "Desert culture" is a commonly accepted term. For reasons that will appear later, we have relegated the Desert cultures en bloc to the Archaic stage, even though some of them show evidences, including radiocarbon dates, of comparable antiquity to Folsom and other Lithic cultures of the Plains. In the Southwest proper, the best-known Desert culture is the Cochise.41 In our first paper we classified the earliest phase of Cochise, the Sulphur Spring "stage," as Early Lithic because of its extinct faunal associations and what at that time we regarded as early radio-

41. Sayles and Antevs, 1941.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

90

carbon dates.42 These dates are no longer too early for Archaic, and there is even some question about the association of the ex- tinct fauna.43 Furthermore, projectile points, formerly thought to be non-existent in Sulphur Spring, have since turned up in subse- quent excavations, and these are the same Pinto-like points that oc- cur in the later Chiricahua phase.44 It appears that, notwithstand- ing the time gap indicated by carbon 14 dating between Sulphur Spring and Chiricahua — which is certainly Archaic — the con- tinuity in artifact types is more pronounced than the differences.

That there is an indisputable Lithic stage in the Southwest, how- ever, and that it is chronologically early is evidenced by the re- mains of hunting cultures similar to, and in some cases specifically related to, the "upper" Lithic cultures of the Plains and the East. Perhaps the earliest is the Sandia culture, first encountered in a cave in the Sandia Mountains near Albuquerque, New Mexico.45 This culture, with its characteristic asymmetrically stemmed (Solutrian- like) points, was associated in the lower levels of the cave with a rich Pleistocene fauna in a geological context estimated by an emi- nently qualified geologist to be of Wisconsin interstadial age,46 and was stratigraphically overlain by younger deposits containing a late Folsom or an "intermediate" stone complex that also had extinct faunal associations. Radiocarbon ages of over 20,000 years for mammoth ivory from the Sandia level have been recently pub- lished,47 but, notwithstanding these combined evidences of high antiquity, the temporal position of Sandia is still in question.48

42. Dates for Sulphur Spring Cochise are: 5805 ± 370 B.C. and 4259 + 450 b.c. (sample Nos. C-216 and C-511 [Arnold and Libby, 1951]).

43. Information by J. Charles Kelley given at the Radiocarbon Conference Andover, October, 1956.

44. E. B. Sayles (verbal communication, 1955).

45. Hibben, 1937, 1941. 46. Bryan, in Hibben, 1941.

47. Sample Nos. M-247, M-349 (Crane, 1956). For an impression of the uncertainty surrounding these important dates see Hibben, 1955, and Crane, 1955.

48. Johnson, n.d.

LITHIC STAGE

91

Recent evidence from the Lucy site in central New Mexico sug- gests Sandia's possible overlapping contemporaneity with Clovis, Sandia's principal rival for the distinction of being the oldest pro- jectile-point-bearing culture in the Americas.49 From the stand- point of continental distribution, however, Sandia has proved a disappointment. Unlike Clovis and Folsom points (particularly the former), which, once properly recognized and defined, turned up in great numbers all over the country, Sandia points are still ex- tremely rare, and occurrences outside the type locality are typologi- cally dubious.50

Other evidences that the Southwest was not lacking in early hunting cultures classifiable as Lithic were found in the lower level of Ventana Cave in south-central Arizona — a somewhat impover- ished Folsom-like culture with abundant extinct faunal associa- tions, in a "humid" deposit referable to one of the later substages of the Wisconsin glaciation51 — and in a large number of surface finds more specificially related to the Lithic cultures of the Plains.52 More spectacular was the discovery of typical Clovis points in- side the carcass of a mammoth at Naco in southwestern Arizona.53 It has been suggested that such finds may reflect a seasonal hunting aspect of cultures which in their entirety would conform to the prevailing Desert pattern.54 If true, this would be very embarrass- ing, because we would be found to have classified two activities of the same culture in separate developmental stages. As a hypothesis

49. Roosa, 1956.

50. Hibben, 1946; Wormington, 1949, p. 75. The difficulty of identifica- tion is that asymmetrically stemmed points or "knives" are found in many North American stone assemblages of various ages. Within the range of vari- ability of this class of artifacts, individual specimens might easily approximate the Sandia type without any culture-historical significance. See M. W. Hill, 1953, Ritchie, 1953, Kleine, 1953, and Lewis, 1954b, for examples that come close to Sandia forms. There is also an early (?) Archaic lozenge-shaped point that tends to look like Sandia.

51. Haury et al, 1950.

52. Hurt, 1942; Haynes, 1955.

53. Haury, 1953a. 54. Cressman, 1951, p. 294; Haury, 1953a, p. 13.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

92

it cannot be ignored; in a transitional phase like Sulphur Spring Cochise it might be a reasonable expectation. However, the present trend of dating appears to support the contrary view — that the finds really do pertain to an earlier, predominantly hunting stage of culture in the Southwest.

In a brief summary of our discussion of the Lithic stage thus far, the following points may be emphasized: (1) The principal stage criteria are rough- and chipped-stone artifacts. (2) The natural context is that of the late glacial and early postglacial environments of the New World. (3) The areas where these evidences are most complete are western North America, particularly the High Plains and the Southwest. (4) Two major technological traditions, or groups of traditions, are postulated: one characterized by pressure flaking and lanceolate blades, the other by percussion chipping and crude choppers and scrapers. (5) There are indications that the per- cussion chopper-scraper traditions may have earlier beginnings than the pressure-flaked-blade traditions, but this remains to be demon- strated as a fact. There is certainly good evidence that the two existed contemporaneously for a long time. (6) The pressure- flaked-blade traditions are clearly best adapted to the ancient grass- land environment of the Plains and the East and to the hunting of large mammals now extinct, whereas the percussion chopper- scraper traditions seem more at home in the semiarid environments of the Greater Southwest, where they are associated with gather- ing economies. This is not an iron-clad separation, however, and in some instances both may be exemplified in the archaeological as- semblage of a single culture. (7) It should be understood that both groups of traditions, the early forms of which are markers for the Lithic stage, show continuity into later cultures of the succeeding Archaic stage. This is particularly true of the percussion chopper- scraper traditions, which carry on into the Archaic Desert cultures of the Greater Southwest.

So far we have drawn examples from Lithic stage cultures in the Southwest and the Plains, with a brief side glance at the East. There are in the rest of North America, so far as we can see, no

LITHIC STAGE

93

other Lithic cultures of comparable ponderability to Clovis, Fol- som, Enterline, and Eden-Scottsbluff. A long catalogue of widely scattered finds of artifact types related to these better-known cul- tures could be compiled, but it would not add anything but dis- tribution to the picture already given. The balance of our survey of the North American Lithic can be taken at a swifter pace.

The Great Basin area, comprising western Utah, all of Nevada, south-central Oregon, and large portions of southern California, has produced as many chronologically early finds as any other area of comparable size in the New World, but, for reasons which we hope will appear to be sound, we are putting most of them into the Archaic. The precise difficulty is that many sites with early radio- carbon dates show a dependence on gathering from the lowest levels on, with hunting in a secondary role and no remains of ex- tinct Pleistocene fauna. If the dates are correct, the shift to modern conditions must have taken place earlier in the Great Basin than in other areas so far considered.55 Extinct faunal remains have been found, however, in sites on the peripheries of the area. On the northern periphery, in south-central Oregon, at Paisley Five Mile Point, food bones in the lowest level of a cave included horse and camel, but there were no diagnostic artifacts in association.56 At Lower Klamath Lake, in the same region but just over the Califor- nia line, horse, camel, and probably mammoth were found, the last in direct association with fragmentary obsidian tools of indetermi- nate type. Not in situ, but believed to have come from the same geological deposit, were several beveled-bone foreshafts similar to i the one found at Clovis, New Mexico, possibly a marker type of the Clovis culture. Unfortunately, manos and post-Lithic-type projectile points were also in this deposit.57 No fluted or other Lithic stone types were found in either of these sites. On the southern periphery of the Great Basin, at Gypsum Cave in south- eastern Nevada, ground sloth and possibly horse and camel were found in association with an artifact complex whose characteristic

55. Jennings and Norbeck, 1955, p. 4.

56. Cressman et al., 1942, p. 93. 57. Ibid., pp. 99-100.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

94

"Gypsum point" is one of the marker types for the Archaic.68 The problem of Gypsum Cave is aggravated by early radiocarbon dates well within the range of time assumed for the Lithic,69 but we can- not regard the culture as anything but Archaic in terms of the defi- nitions used here.

Other finds in the extreme southern part of the Great Basin and contiguous lower Colorado River and southern California desert and coastal regions confront the would-be classifier with even more vexing problems, but some of the cultures involved appear to be Lithic and cannot be overlooked. The San Dieguito-Playa (or San Dieguito I) culture seems to reflect a hunting pattern, not- withstanding its lack of characteristic point types,60 and its close relationship with the Ventana culture indicates a date that would satisfy the chronological requirements of the Lithic stage.61 In the Lake Mohave complex (which includes too many types for com- fort) the San Dieguito-Playa percussion choppers, planes, scrapers, etc., are associated with points vaguely reminiscent of Sandia, Clovis, San Jon, and Plainview.62 The Mohave and Silver Lake points have wide distribution but are not consistent or stable in their association.63 An almost complete lack of milling stones in the Lake Mohave sites indicates a primarily hunting culture, and there is a good possibility that some, but certainly not all, of the material is early enough to qualify as Lithic.64 A site subsequently reported by the same investigators, the location of which was not specified but was presumably in the same region, revealed a "pure" Folsom- like complex with rather striking similarities to Lindenmeier in

58. Harrington, 1933, 1934.

59. Dung of giant sloth from Gypsum Cave was dated at 8504 + 340 b.c. and 6576 ± 250 B.C. (sample Nos. C-221, C-222 [Arnold and Libby, 1951]), but the association of dung and artifacts is not beyond question (Heizer, 1951a, p. 24; Cressman, 1951, p. 306).

60. M. J. Rogers, 1939.

61. Haury et al., 1950, p. 531.

62. E. W. C. Campbell et a/., 1937.

63. Wormington, 1949, p. 85. 64. Brainerd, 1953.

LITHIC STAGE

95

artifacts other than points.65 This, on typological grounds alone, we would classify as Lithic.

The relative infrequency of remains attributable to the Lithic stage on the Pacific Coast of North America appears to confirm the widely held assumption that early migration southward from Alaska was mainly in the intermontane and High Plains "corri- dors" on either side of the Rocky Mountains. Estimates of con- siderable antiquity have been made for the earlier manifestations of the La Jolla culture66 and the Topanga culture,67 of the San Diego and Los Angeles regions, respectively. The lowest level in the se- quence at Malaga Cove, also in the Los Angeles region,68 has been compared to Rogers' Dieguito and is thought to be early, but it I includes a microlithic "flake drill" type that is remarkably similar to the characteristic "perforators" of the Poverty Point culture of the lower Mississippi Valley on a late Archaic or even Formative level.69 The enigmatic Oak Grove culture of the Santa Barbara re- gion70 is also difficult to fit into our definition of Lithic. The four last-named cultures have been recently gathered together, along with material from the Little Sycamore site in Ventura County, under the designation "Early Milling Stone Cultures of Southern California" and have been assigned a guess date of 3000-2500 b.c.71 This convenient generalization has encouraged us to put them all into the Archaic stage.

Farther north, in central California, well-developed sequences in the San Francisco Bay and Sacramento Delta regions have no Lithic phases; the earliest cultures are clearly Archaic in terms of the : definitions used here. The Farmington complex has already been i mentioned as a possible "lower" Lithic stage culture, but there is no dating to support the assumption. North of San Francisco Bay, ,

65. Campbell and Campbell, 1940.

66. M. J. Rogers, 1929, 1945; Harding, 1951; information given by Clement Meighan at the Andover Radiocarbon Conference of April, 1956.

67. Heizer and Lemert, 1947; Treganza and Malamud, 1950.

68. E. F. Walker, 1951. 70. D. B. Rogers, 1929.

69. Ford, Phillips, and Haag, 1955. 71. Wallace, 1954.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

96

at the controversial Borax Lake site,72 there seems to have been an early component that included rather dubious fluted points and Lindenmeier-like scrapers and gravers, but, if this existed as a complex, it was not clearly differentiated stratigraphically from a culture more like the milling-stone cultures referred to above. In view of this uncertainty, it would seem preferable to classify all of Borax Lake as Archaic.

From here on up the Pacific Coast, our sources of information on the Lithic diminish rapidly. In the northern California, Northwest Coast, and Plateau areas, finds attributable to the Lithic are very scarce, partly because of ignorance no doubt, partly because we are getting farther away from the bases of our criteria and are un- able to recognize Lithic materials when we see them. A few pos- sibly early sites have been reported but in no case fully published.73 North Pacific archaeology is clearly not well enough known to per- mit generalization; nevertheless, it may be suggested that this is an area quite unsuited in nature to the migratory hunting-and-col- lecting way of life characteristic of Lithic cultures elsewhere.

This brings us to the Arctic. We can hardly expect to fit Arctic data into our scheme of classification without difficulty. As one of the outstanding Arctic specialists has recently said: "The pattern and development of Eskimo culture are the result of a combination of geographical, ecological and cultural factors that are truly unique."74 This would apply equally to any cultural pattern in the Arctic, since the unique geographical conditions can be projected indefinitely into the past, and with them the special cultural adjust-

72. Harrington, 1948£.

73. The Lind Coulee site in eastern Washington may have a Lithic phase ("Notes and News," American Antiquity, 16, 290; 17, 281; 18, 189, 297). A date of 6746 ± 400 b.c. (sample No. C-827 [Libby, 1954Z>]) certainly indi- cates that kind of occupation. Three fluted points found some time ago in Ore- gon and Washington have recently been identified as Clovis in type {American Antiquity, 21, 451). In British Columbia a very good Eden-ScottsblufF point was recently received by the Provincial Museum, purportedly from the vicinity of Lake Windermere (Duff and Borden, 1954).

74. Collins, 1953£, p. 201.

LITHIC STAGE

97

ments required to make life endurable in such an environment. Apart from marginal phases of Indian cultures whose centers lie farther south, no major cultural configuration other than Eskimo has been conclusively shown to have occupied the area. The only culture that might represent a pre-Eskimo configuration is the well- known Denbigh "flint complex" first encountered at the Iyatayet site on the north Bering Sea coast of Alaska.75 Denbigh shows con- nections of a tenuous nature with late Lithic stage cultures of the Plains, but its more significant affiliations appear to be with eastern Siberian cultures that are variously regarded as Mesolithic or Neo- lithic. The latter are assumed to be the source of the burin and microcore-and-blade ("polyhedral core and lamellar flake") tradi- tions so characteristic of the Denbigh complex. These Mesolithic traditions, particularly the microcore-and-blade technique, are widely but erratically distributed in northern and eastern North America on a time level that is by no means early. The single fluted point from the Iyatayet site is a poor example that cannot be classified with any established type, but the fine parallel-flaked points would seem to be typologically related to the Eden-Scotts- bluff culture of the Plains. Radiocarbon dates for the Denbigh level at Iyatayet, however, are considerably later than those for Eden- Scottsbluff.76 This lends support to Henry Collins' belief that fur- ther information about the non-lithic aspects will bring out the es- sential Eskimo character of the Denbigh culture.

No other complex that might be considered Lithic has yet been

75. Giddings, 1951, 1952.

76. Radiocarbon dates for the Denbigh level at Iyatayet are: 1546 + 230 b.c. and 2705 ± 220 b.c. (sample No. C-792 and C-793 [Libby, 1954a]). Giddings, who collected the samples, declined to accept the dates and adhered to his original estimate of 8,500 years ago (1955, p. 376). A good deal has been made of the date of 4042 ± 280 b.c. (sample No. C-560 [Libby, 1951]) obtained from a sample of twigs from the lower level of the Trail Creek Cave on Seward Peninsula which also contained Denbigh type material. According to Helge Larsen, who excavated the cave, these twigs could not be positively as- sociated with the Denbigh artifacts and might even have been brought into the cave by hibernating bears (statement made at the Andover Radiocarbon Con- ference, October, 1956).

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

98

reported in the Arctic area. A few scattered fluted points have bee reported from various localities in Alaska, but most are without associations and reliable dating criteria.77 The general impression is that they are relatively late and marginal to the main centers of the fluted-point tradition farther south. "In the Canadian subarctic, an area of extreme difficulty for archaeology, the Yukon and Mackenzie valleys are crucial because they are the most likely diffusion routes in the Lithic stage. A few significant finds have been made,78 but, in so far as they can be related to Lithic traditions elsewhere, they seem to be relatively late in the scheme of things. We have to conclude that remains pertaining to the earliest immi- grants from Asia either have not been found or, if found, have not yet been recognized. The origins of Lithic cultures in North Ameri- ca are still completely obscure.

MIDDLE AND SOUTH AMERICA

We can, therefore, hardly expect clarification of the problem of Lithic origins in Middle and South America. Any theory of popu- lation-spread into these areas must assume north-to-south move- ments of peoples and cultures in a Lithic stage of development. In Middle America there is now ample confirmation of the presence of man with extinct fauna, including mammoth, in upper Pleistocene formations in the Valley of Mexico, but the associated artifact as- semblages are incomplete and the relationships with North Ameri- can Lithic cultures remain to be worked out. The San Juan "in- dustry" of Helmut de Terra, consisting often artifacts from three sites,79 has received some slight reinforcement from finds asso- ciated with the two Iztapan mammoths, among them three classifi- able points, of the Eden-Scottsbluff, Angostura, and Lerma types, respectively.80 The fact that these types are relatively late and heretofore not associated with mammoth may indicate that the

77. Hibben, 1943; Solecki, 1951; Collins, 1953£.

78. Johnson, 1946; MacNeish, 1951, 1953, 1954a.

79. De Terra, 1949.

80. Aveleyra and Maldonado-Koerdell, 1953; Aveleyra, 1956.

LITHIC STAGE

99

mammoth survived in central Mexico long after its extinction in the United States. It may also give slight additional meaning to the fact that good fluted points have not so far been found south of Durango. The poor example from Costa Rica is of doubtful sig- nificance.81 Another important evidence of Lithic stage culture in Mexico is to be seen in the Canyon del Diablo sequence of southern Tamaulipas,82 of which the first two phases, Diablo and Lerma, are said to represent hunting cultures; but the material has not yet been thoroughly described, and comparisons with Lithic material north of the border cannot be made. The occurrence of a Lerma point with the second Iztapan mammoth, referred to above, is a strong indication that both Diablo and Lerma belong in the Lithic stage. From Mexico on south through the Maya area and lower Central America, scattered and isolated evidences of man's presence in late glacial or early postglacial times could be cited, but they do not add up to anything of cultural significance and may be omitted here. Identification of Lithic remains in South America is beset with difficulties. Technological similarities between North American Lithic traditions and those presumed to be of comparable date in South America are, for the most part, of a very general kind. An even greater source of trouble is the limited geological and paleon- tological criteria for temporal equation with North American Lithic cultures. Nevertheless, there are a number of finds in South America that may be considered relevant. One of the best-substantiated claims for a South American Lithic stage is to be found in the Magellanic sequence.83 The lowest level, Period 1, in the Palli Aike and Fell's caves on the north side of the Straits, contained, along with remains of horse and sloth, pressure-flaked points of ovate triangular shape with broad tapering stems, a single un- stemmed Plainview-like point, crude chopping tools, bone flakers and awls, and some rude disk-shaped pieces of lava. Bird's original

81. Swanger and Mayer-Oakes, 1952.

82. MacNeish, 1950.

83. Bird, 1938.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

100

modest estimate of 1000-3400 b.c. for this culture has been con- siderably extended by a radiocarbon date of 6688 + 450 b.c.84

Other researches into the problem of early man in the southern part of the South American continent may be compared and con- trasted with the Palli Aike and Fell's caves Period 1 discoveries. In some instances there are indications of even earlier materials. M. A. Vignati's stratigraphy on the Argentine coast of Tierra del Fuego85 reveals a three-part sequence in which the earliest phase has only retouched cores and flakes or bifacial percussion-flaked blades and choppers. In the second level, he recovered a "knife" or point similar to those of Bird's Period 3, and in the third, or top, cultural layer there was a point similar to those of Bird's Period 4.86 It is to be noted, however, that crude bolas stones are found even in the first period of the Vignati sequence.87 Along the Atlantic coast of Patagonia, Menghin has correlated site occupations with a series of rising coast lines and has defined an earlier Oliviense complex, characterized by retouched flakes and scrapers, and a later Sola- nense complex, featuring single-notched-base projectile points.88 Geological dating here suggests that both these complexes ante- date Bird's Period 1. Menghin has also described Patagonian cave stratigraphies the lowest levels of which (Toldense complex) ap- pear to correlate with Palli Aike-Fell Period 1, on the basis of lanceolate Plainview-like points, though they include also crude bolas stones.89 The upper levels in these same caves (Casapadrense complex) resemble Bird's Period 2 at Palli Aike in the absence of stone points and the presence of numerous bone implements along with crudely retouched scrapers and blades. This order, of course, tends to reverse the percussion chopper to pressure-flaked point

84. Sample No. C-485 (Libby, 1952). 85. Vignati, 1927.

86. This last identification was made by Bird (personal communication, 1956).

87. It is interesting to note that the rude percussion industries of southern and Baja California have "charmstones" which may also have been bolas stones, although the similarities to South American types are not close.

88. Menghin, 1952b. 89. Menghin, 1952a, 1952b.

LITHIC STAGE

101

("lower" to "upper" Lithic) sequence. This Casapadrense phase is dated, in relation to volcanic activity in the region, as subsequent to 6000 b.c. It appears to relate to a somewhat similar phase (Tan- dilense) in Buenos Aires Province.90

Possibly within the range of our Lithic definition is the Ayampi- tin phase of the Cordoba and San Luis hills in northwestern Argen- tina. Here A. R. Gonzalez has obtained a stratigraphy at the Gruta de Intihuasi in which the Ayampitin assemblage, with its lanceo- late points and crude grinding utensils, underlies the later, Archaic- like Ongamira phase. The Ayampitin projectile points are closest to Bird's Period 3, but the radiocarbon dating supports a Lithic assignment.91

We cannot leave Argentina without commenting on Ameghino's claims for the antiquity of man in Buenos Aires Province.92 The geological basis for these claims of what seems an excessive age is still in dispute, and the characteristic artifacts, battered and splintered pebbles suggestive of crude hand-ax or chopper forms, have uncertain associations. In some of Ameghino's sites they seem to have been found with pressure-flaked points, grinding stones, and even pottery. Bird has reported similar implements from northern Chile in contexts that are relatively late.93

In Brazil the artifacts associated with the longheaded Lagoa Santa skeletal type,94 also found in some of the sambaquis of the southern coast,95 have been considered relevant to the problem of early man's presence in South America, but we have relegated them to the Archaic stage for reasons given in the next chapter.

From the great corpus of archaeological data pertaining to Peru and Bolivia, few finds could conceivably be on the Lithic stage. The only possibilities that might be mentioned are: (1) the long

90. Menghin and Bormida, 1950.

91. Gonzalez, 1952. A recent radiocarbon date on the Ayampitin phase is in the neighborhood of 6000 b.c. (A. R. Gonzalez [personal communication, 1956].

92. Ameghino, 1911. 94. Walter, 1948; Evans, 1950.

93. Bird, 1943. 95. Serrano, 1946.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

102

tapered-stem points and other implements from the Chicama Valley, which seem to have no connection with later preceramic and ceram- ic phases in the region;96 (2) the seemingly unrelated complex featuring stemless and diamond-shaped points found in the non- ceramic refuse of certain inland rock shelters near Huancayo;97 (3) the chipped-stone complex which Strong has reported from the south coast;98 and (4) recent discoveries from Viscachanf in the Bolivian highlands.99 From the northern Andes there is even less. Discoveries have been claimed as evidences of man in geologically early periods, such as the Punin skull,100 or the Alangasi mastodon101 (this claim recently demolished by Gross),102 but no artifact com- plexes that can be measured against our criteria for the Lithic stage. In Venezuela, J. M. Cruxent and Irving Rouse have recently re- ported a quartzite assemblage from sites in the locality of El Jobo in the northwestern part of the country. This has not yet been formally described, but specimens from it have been examined by a number of North American authorities; the consensus is that the assemblage definitely represents an early culture — the speci- mens are closely related to artifacts found with the second Iztapan mammoth — with more distant but probably significant relation- ships to some of the later Lithic cultures of North America.103

In this resume of South American data two things stand out: first, the tendency for Lithic finds to be concentrated in the tem- perate (southern) and arid portions of the continent, perhaps the only favorable areas for cultures primarily based on hunting;104 second, the fact that there are few specific resemblances to early cultures in North America, and those that do appear are with pro- jectile-point forms of the later post-fluted-point cultures of the Lithic stage.

96. LarcoHoyle, 1948, pp. 11-12, PI. 1.

97. Tschopik, 1946. 101. Uhle, 1930.

98. Strong, 1954. 102. Gross, 1951, p. 104.

99. Menghin, 1953-54. 103. Cruxent and Rouse, 1956.

100. Sullivan and Hellman, 1925. 104. Bennett and Bird, 1949.

LITHIC STAGE

103

Chapter 5

Archaic Stage

DEFINITIONS

Our previous use of the concept "Archaic" for the second stage of New World historical-developmental interpretation is here re- tained in its essentials. The term was introduced into the archaeol- ogy of eastern North America by William A. Ritchie, who applied it to his Lamoka phase in central New York.1 Soon after, the rich preceramic culture of the northern Alabama and Kentucky shell middens was revealed by the extensive excavations of W S. Webb and his associates. Relationships to Lamoka were noted, and the term "Archaic" was extended accordingly, with a "pattern" sig- nificance in the Midwest taxonomy.2 About the same time, re- examination of earlier shell-midden investigations in the far South- east3 and excavations on the Georgia coast4 expanded the Archaic still further geographically and also slanted it toward the concept of a "Shellmound" culture. Similar extension was going on all over eastern North America, and there were suggestions that cultures as far away as California were somehow related.5 However, it was also becoming clear that the concept had outgrown its specific historical implications, and Griffin in his first synthesis of eastern

1. Ritchie, 1932. Alanson Skinner had previously designated certain pro- jectile points from shell middens in the vicinity of Manhattan as "archaic" (1919, 1920), but Ritchie was apparently first to use the term in a full cul- tural sense.

2. Webb and Dejarnette, 1942. 3. Wyman, 1868, 1875; Claflin, 1931.

4. Unpublished work by Preston Holder and A. J. Waring, Jr.

5. Haag, 1942; Beardsley, 1948.

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

104

archaeology made an attempt to jettison it.6 Failing in this, he and his co-workers reinstated it as a "period" in what might be re- ferred to as the new midwestern chronology.7 As it was our con- tention— and still is — that periods in area chronologies of wide geographical scope lose their temporal significance and become, in effect, developmental stages, we have chosen to use "Archaic" as a stage designation, making use of its broad typological implications and expanding the concept to its ultimate hemispherical limits.8

For some time North American archaeologists have found it convenient to speak of "early" and "late" Archaic, but there is a remarkable lack of agreement about what these terms mean. In preparing our first classification, we considered making a similar division into "upper" and "lower" Archaic stages but abandoned the idea for two reasons: (1) the difficulty of finding criteria that would hold for all major areas of New World archaeology and (2) the belief that the concept of a Preformative stage following the Archaic would take care of many of the cultures generally regarded as "late" Archaic. This did not work out quite as expected — the precise difficulties will be discussed later on — and, with the elimi- nation of Preformative in the present scheme, the need for sub- division of Archaic seemed more pressing than before. There had also taken piace in the interval the conference of North American archaeologists at Andover in April, 1956, at which a twofold sub- division of the Archaic was generally agreed upon. We tried very hard, therefore, in preparing the present version of our classifica- tion, to follow this lead — and failed again.

6. Griffin, 1946. 7. Griffin, 1952a.

8. "Archaic" was used much earlier in American studies by Boas (1913), Spinden (1928), and others to refer to the beginnings of settled agricultural vil- lage life in Mexico and Central America. This connotation is, of course, en- tirely different from the use and meaning here. Our "Formative" is the ap- proximate equivalent of the Boas and Spinden "Archaic." Our decision on this terminology was influenced by the current and firmly intrenched North Ameri- can usage of "Archaic" and by the fact that such names as "Formative" and "Preclassic" have now replaced "Archaic" as the designation for early New World village agriculture.

ARCHAIC STAGE

105

The Andover classification, if we may call it such, is of narrower scope than ours. The problem was limited culturally to stone technology and geographically to North America. Within this frame of reference it seemed possible to a majority of the conferees to distinguish two broad stages of Archaic technology, with the presence or absence of ground- and polished-stone artifacts as the governing criterion. We made a serious effort to expand these technological stages culturally and geographically but finally con- cluded that expansion was neither possible nor desirable. In the perspective of New World archaeology as a whole, polished-stone technology does not seem to have any consistent developmental or even temporal significance. One must conclude that polished-stone artifacts were desirable things to have, perhaps even essential in certain environments, but that many groups got on very well with- out them. For example, the first successful efforts toward agricul- tural food production seem to have been made in cultures lacking polished stone. To have to call such cultures "lower" Archaic would be disconcerting. On this point we were better off with our Preformative.

Even in the restricted frame of reference of North American stone technology, it may be questioned whether subdivision of the Archaic along the lines suggested at Andover is entirely feasible. It is, of course, a methodological principle that developmental classifications and chronological classifications do not of necessity march together, but, when we compare concepts of "early" and "late" Archaic in the literature with concepts of "lower" and "upper" Archaic proposed at Andover, we find scarcely any agreement at all. It seems best, therefore, to leave Archaic as a single stage allowing for the use of qualifiers, "lower" and "upper" (or "early" and "late"), by area specialists according to whatever criteria seem best to fit the case.

To define Archaic as a single stage, however, is not without its difficulties. So far as we can tell from the meager remains charac- teristic of most early Archaic cultures, there is no important shift in economic and social patterns from the previous Lithic stages. It

METHOD AND THEORY 1A/ IN AMERICAN ARCHAEOLOGY 1UO

would be convenient if, following the lead of some Americanists, we could simply designate the Lithic as the stage of early hunting and the Archaic as the stage of early gathering cultures. We have already inveighed against this simple division. For purposes of large-scale integration, concepts of hunting versus gathering cul- tures are useful only as convenient rubrics. We found that we had to accommodate both in our definition of Lithic, and we are obliged to do the same for the Archaic. Nevertheless, it does seem possible to see a shift in emphasis between hunting and gathering, in favor of the latter, and there certainly were differences in the plants col- lected and the animals hunted.

With these points in mind, we may briefly define the Archaic as the stage of migratory hunting and gathering cultures continuing into environmental conditions approximating those of the present. With the extinction of the large Pleistocene mammals hunted by Lithic stage peoples — and this extinction is assumed to have been virtually complete by the beginning of the altithermal period — there is now a dependence on smaller and perhaps more varied fauna. There is also an apparent increase in gathering; it is in this stage that sites begin to yield large numbers of stone implements and utensils that are assumed to be connected with the preparation of wild vegetable foods. In most Archaic cultures these are shaped by use rather than design and do not, therefore, fit into the category of ground and polished stone, which is one of the often-referred-to criteria of the Archaic stage. The specialized techniques of gather- ing and preparation of wild foods, especially in areas where these consisted mainly of hard-shelled seeded forms, suggest a medium in which early experimentation in plant domestication could take place. As a result, it is in some rather primitive, and surprisingly early, Archaic cultures that the first evidences of New World agriculture are to be found. These we designated as Preformative in our first classification. We have since come to the reluctant con- clusion that the mere presence of agriculture, though of enormous importance historically in terms of the growth of particular Ameri- can patterns of culture, is not of primary significance from a more

ARCHAIC STAGE

107

abstract, developmental point of view. It becomes so only when it can be seen as dominant in the economy and integrated socially to produce the stable settlement patterns that we have postulated as the sine qua non of the Formative stage. Furthermore, certain cul- tures that we classified as Archaic in our first paper because they had no agriculture appear to have achieved stable settlement pat- terns and other prerequisites of the Formative without that stimu- lus. Lack of agriculture, therefore, is no longer a negative criterion in our formulation of the Archaic stage.

The point of departure for the description of Archaic stage cul- ture in North America is commonly a level of stone technology marked by the addition of grinding and polishing to the earlier techniques of percussion and pressure flaking. This is no longer an adequate approach, for archaeologists tend increasingly to consider the possibility of an "early" Archaic prior to the adoption of ground and polished stone. Although we are not attempting to set up such a stage, for reasons already given, we shall refer to many cultures in the following pages that lack ground and polished stone alto- gether. Nevertheless, it is appropriate to begin our description with this category, which certainly represents a technological innovation of immense significance.

Of primary interest as stage criteria are the heavy ground-stone woodworking tools generally regarded as prerequisite to the suc- cessful occupation of forest environments — axes, adzes, wedges, gouges, etc. Highly characteristic, but less useful as criteria, are implements and utensils used in the preparation of vegetable foods — milling stones, metates, mortars, manos, pestles, pounders, etc. Some of these have already appeared in the Lithic but usually in the guise of rough-stone artifacts, i.e., objects fashioned principally by use rather than design. These carry over into the Archaic, but there is an increasing tendency toward more specialized forms and careful workmanship. Particularly characteristic of the Archaic in some areas are stone vessels, precursors of pottery. Their presence reflects the greater stability of occupation postulated for some of the later cultures of this stage. Migratory peoples are not partial to

METHOD AND THEORY IN AMERICAN ARCHAEOLOGY

108

stone vessels for obvious reasons. Other Archaic elements of sig- nificance in particular areas are ground slate points and knives; polished atlatl weights of various forms; plummet-like objects (variously referred to as "plummets," •"charmstones," "/z/W); stone tubes, which may in some cases have been smoking pipes; stone beads; and an array of other objects of purely ornamental or problematical function. Thus, the addition of a new way of handling stone resulted in an immense increase in variety and complexity of the archaeological inventory.

In the chipped-stone category it is noteworthy that the high standards characteristic of many Lithic cultures are not generally maintained in the Archaic. Many of the older forms, however, per- sisted with little modification, particularly in the chopper and scrap- er categories. A new tool that seems to have had little if any sig- nificance in the Lithic is the drill. Drills are found in some Archaic cultures in great profusion and variety, a matter doubtless related to the fact that many of the ground and polished forms mentioned above are perforated and to the emphasis on beads and pendants of stone, bone, shell, and other materials that is a common feature in this stage. In some areas the older lanceolate projectile forms of the Lithic are carried over into the Archaic, but in general there is a greater variety of points, with emphasis on stemmed, corner- notched, and side-notched forms,