plog - style in artifacts

19
Analysis of Style in Artifacts Author(s): Stephen Plog Reviewed work(s): Source: Annual Review of Anthropology, Vol. 12 (1983), pp. 125-142 Published by: Annual Reviews Stable URL: http://www.jstor.org/stable/2155643 . Accessed: 06/07/2012 20:20 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. . Annual Reviews is collaborating with JSTOR to digitize, preserve and extend access to Annual Review of Anthropology. http://www.jstor.org

Upload: atarip

Post on 11-Dec-2015

55 views

Category:

Documents


3 download

DESCRIPTION

Plog - Style in Artifacts

TRANSCRIPT

Page 1: Plog - Style in Artifacts

Analysis of Style in ArtifactsAuthor(s): Stephen PlogReviewed work(s):Source: Annual Review of Anthropology, Vol. 12 (1983), pp. 125-142Published by: Annual ReviewsStable URL: http://www.jstor.org/stable/2155643 .Accessed: 06/07/2012 20:20

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Annual Reviews is collaborating with JSTOR to digitize, preserve and extend access to Annual Review ofAnthropology.

http://www.jstor.org

Page 2: Plog - Style in Artifacts

Ann. Rev. Anthropol. 1983. 12:125-42 Copyright ? 1983 by Annual Reviews Inc. All rights reserved

ANALYSIS OF STYLE IN ARTIFACTS

Stephen Plog

Department of Anthropology, University of Virginia, Charlottesville, Virginia 22903

An immense amount of information on patterns of stylistic variation in prehis- toric artifacts has been produced by studies done throughout the history of archaeological research. The construction of space-time frameworks for most regions on the basis of such patterns has produced detailed information on stylistic sequences through time and stylistic distributions across space. Much of that critical information is encoded in complex sets of artifact types which have been developed. When questions concerning stylistic variation move beyond general patterns of variation across space and through time, however, the available information decreases drastically despite the decades of research on stylistic variation. However, significant improvements have been made during the last 20 years. Beginning at least as early as the innovative ceramic sociology studies of the 1960s (11, 27, 41-43, 65), when initial attempts were made to examine the relationship between stylistic variation and aspects of social organization, many new questions about stylistic variation began to be addressed. Can spatial clusters of stylistic attributes smaller than the previ- ously defined culture areas be isolated that might be the result of residence groups, lineages, marriage networks, or clusters of communities cooperating in economic activities? Can studies of stylistic variation lead to more ac- curate and precise estimates of site occupation dates? Why do rates of stylistic change and degrees of stylistic variation fluctuate so much through time or across space?

As the questions about the causes and nature of stylistic variation have increased, so have the number and types of studies. Within the last five years, for example, at least a dozen doctoral dissertations have been written that have focused almost exclusively on issues concerning stylistic variation. In addition, an increasing number of ethnographic studies of patterns of stylistic variation

125

0066-4294/83/1015-0125$02.00

Page 3: Plog - Style in Artifacts

126 PLOG

have been initiated (20, 22, 30, 32, 33, 44) to complement the information which can be obtained from analyses of prehistoric artifacts.

While it might be expected that the increasing number of studies would have resulted in some generally accepted answers to the kinds of questions listed above, such has not been the case. Nowhere is this clearer than in the eth- nographic or "ethnoarchaeological" studies noted above, studies which some believed would provide many of those answers because of the possibil- ity of talking to and observing the makers of the artifacts and obtaining clear-cut information about variables such as marriage patterns. One need only compare studies of the Kalinga in the Phillipines (22, 44) and of several popu- lations in Africa (30, 32, 33) to see that some of the conclusions reached are contradictory. If there is any inference that is common to most, if not all, recent studies, it is the important but nevertheless somewhat unfulfil- ling conclusion that the causes of stylistic variation are complex. There is, therefore, considerable difference of opinion concerning the degree of sup- port or lack of support for various theories which have been developed to explain stylistic variation. These theories will be described briefly in the following section, and I will then discuss some of the types of studies which have been suggested as necessary to resolve some of the current differences of opinion.

THEORIES OF STYLISTIC VARIATION

As noted above, many recent studies of stylistic variation have led to differing conclusions concerning the reasons similarities or differences in style occur or do not occur among different individuals, social groups, villages, or regions. One theory which continues to be advocated is the fundamental tenet of the ceramic sociology studies: the degree of stylistic similarity between indi- viduals, residence groups, or villages is directly related to the amount of social interaction between those individuals, groups, or villages. As Flannery (19) has noted, this argument is an extension to smaller social groups of some of the basic assumptions which archaeologists always have used in defining culture areas. The isolation of culture areas was based on the assumption that stylistic distributions would be marked by spatial discontinuities with homogeneous distributions within individual areas. Thus, stylistic similarities or differences could be used to identify membership in regional social groups but not directly to estimate levels of interaction intensity. In contrast, many of the studies based on the social interaction theory explicitly assumed that levels of stylistic similarity or agreement could be used as direct estimates of interaction intensi- ties. This is exemplified by the following statement: "It is assumed that if a pair of contemporaneous sites share a higher coefficient of agreement than another contemporaneous pair, there has been greater contact between the former pair" (16).

Page 4: Plog - Style in Artifacts

STYLE IN ARTIFACTS 127

In addition to the social interaction theory, it also has been suggested that one of the major sources of variation in some stylistic characteristics is differences among individuals in motor habits (28, 29). It has been proposed that this variation in the execution of particular stylistic attributes is subconscious and thus unrelated to levels of social interaction.

In contrast to the above theories which view aspects of stylistic variation simply as a by-product or passive reflection of culture systems or individual motor habits, others have proposed that style may play a very active role. Several individuals (2, 3, 8, 9, 51, 68, 69) have argued that stylistic variation may be used in certain contexts to signal or communicate information "not only about the identity of the maker or user, but also potentially about his social group membership, status, wealth, religious beliefs, and political ideology (2, p. 118). Wobst (69, p. 323) suggests, however, that the utility of stylistic messages decreases with increasing social distance between the sender and receiver, inasmuch as there would be fewer messages that could not be transmitted more cheaply using another mode of communication. Thus, stylis- tic variation would be expected to be used in information exchange only under certain social conditions which have been outlined by several archaeologists (2-4, 26, 51, 62).

Hodder (32, 33) also has emphasized an active, symbolic role for stylistic variation, stating (33, p. 36) that "material symbols can actively justify the actions and intentions of human groups." He argues, however, that the manner in which material culture relates to society depends on factors such as ideolo- gical structures and symbolic codes (33, p. 210). Thus, "styles may well express and justify ethnic differentiation, but the manner in which they do this can only be understood by examining structures of symbolic meaning" (33, p. 205).

Unfortunately, attempts to test these theories have not always led to consis- tent conclusions. On the one hand, for example, a variety of evidence has been presented in several studies (2, 3, 30, 33, 49-52) suggesting that the rela- tionship between stylistic variation and social interaction proposed by some of the ceramic sociologists is not supported. On the other hand, a number of individuals (22, 62-64) have defended the validity of that relationship. There are a variety of reasons for these differences of opinion. One problem has been that critical concepts often have been ambiguously defined. Rubertone (58, p. 100) has noted that the use of "the term 'interaction' indiscriminately as a concept to describe the sum of all interaction between units masks the variabil- ity in social relations that is essential to understanding complex societies, e.g. the nature of class affiliations, the existence of ethnic groups, and neighbor- hoods." In addition, tests of the different theories are far from complete and frequently lack the methodological rigor that ideally we would like to see. While some of my own research (51) has been regarded as supporting the information exchange theory, I have noted that, for a number of different

Page 5: Plog - Style in Artifacts

128 PLOG

reasons, including the lack of adequate stylistic analyses of broad areas, my evaluations of the theories had to be based on a synthesis of the subjective impressions of a number of archaeologists, rather than on an objective analysis of a large set of stylistic data. Many analyses, including some of the early ceramic sociology studies as well as some very recent research, also have been based on inadequate stylistic classifications and have failed to control a number of basic factors which can affect variation in the distributions of stylistic attributes (51). Finally, views concerning the complexity of the relationship between stylistic variation and social, religious, economic, and demographic factors often have been overly simplistic.

It is significant, however, that an increasing number of studies of style are being conducted and many have been directed toward solving some of the above problems. While some have attempted to develop a synthetic view of style which integrates some of the alternative theories (22, 62), for example, others have attempted to expand upon aspects of the theories (1-3, 26, 31-33, 66, 67). Thus, despite some of the differences of opinion that exist, I am optimistic about the results that may be achieved in the future. Archaeologists too often have been willing to accept proposals concerning stylistic variation that had not been adequately tested and had minimal empirical support. The current debates over various issues that are characteristic of a number of recent studies are not without their problems, but they are a good sign that more rigorous testing of ideas is being demanded and ultimately they should lead to some definite progress in our understanding of stylistic variation. Also, while there may not be widespread agreement concerning the factors that cause similarities and differences in stylistic distributions, many studies have reached similar conclusions concerning some of the steps that must be taken if future analyses are to increase our understanding of style. These steps include more adequate consideration of artifact production and exchange, expansion of studies to focus on stylistic variation on more than one type of material or artifact, better studies of variation over large geographical areas rather than concentration on a few sites or a small valley, and recognition that variation in different structural or hierarchical levels of style or different types of stylistic attributes may be caused by different factors. These are the steps on which I will concentrate in this paper.

CLASSIFICATION AND LEVELS OF STYLISTIC VARIATION

One of the most important changes in stylistic analyses that has occurred has been greater emphasis on classification methods. Classifications used in many of the ceramic sociology studies, as well as some developed during earlier

Page 6: Plog - Style in Artifacts

STYLE IN ARTIFACTS 129

decades (7, pp. 47-48), suffered from a number of different problems includ- ing lack of replicability (51, p. 42; 60) and a lack of equivalency of the analytical units such as design elements (51, 54). However, it should be noted that these problems occurred primarily with studies of materials on which stylistic elaboration was complex, such as ceramic vessels with painted designs or textiles, as opposed to analyses of other materials such as stone tools. The deficiencies were a result to a large degree of the inexplicit, intuitive criteria for isolating the analytical units (cf 27, p. 23). In recent years, however, the issue of classifying more complex types of stylistic elaboration has received con- siderable discussion (2, 3, 20, 22, 51, 54, 55, 62).

While questions concerning classification procedures still remain and adequ- ate tests to demonstrate the degree of replicability of suggested approaches must still be conducted, progress is being made. One important example is the increasing recognition that stylistic elaboration is not a single, indivisible entity but can be described in terms of a number of different attributes ranging from the symmetry patterns of designs to the width of individual painted lines or the location of notches on projectile points. This point was made 15 years ago by Whallon (65, p. 223), who noted that "style has many aspects and levels of behavior which may be analytically distinguished and measured," and was underscored by Friedrich's (20) study of Tarascan potters and vessels. It has not had a significant impact on analyses of complex stylistic elaborations until recently, however.

In addition to the multidimensional nature of style, it also has been empha- sized that stylistic attributes may be regarded as having a hierarchically orga- nized structure (20). In discussing the structure of designs on ceramic vessels from the Tarascan village of San Jose, for example, Friedrich (20, p. 333) distinguished three hierarchical levels: spatial divisions, design configurations, and design elements. The highest hierarchical level was the system of spatial divisions. Within those divisions were placed designs which had two levels of organization: design elements and configurations. Design elements were "de- fined as the smallest self-contained unit," while design configurations were "defined as arrangements of design elements that are of sufficient complexity to fill a spatial division" (20, p. 335).

Recent studies have argued that the recognition of the multiple attribute, hierarchical nature of complex stylistic forms is important in evaluating theories of stylistic variation because variation in specific aspects, levels, or attributes of style may be explained by different factors (20, 22, 51, 62). In one of the most comprehensive studies done, for example, using sherds from the neolithic period in northwestern Europe, Voss (62) argued that variation in some stylistic attributes is likely to be the result of variation in levels of social interaction, while other design attributes are more likely to have been used for the type of information exchange discussed by Wobst. These proposals were to

Page 7: Plog - Style in Artifacts

130 PLOG

a large extent based on conclusions reached in the study by Friedrich noted above. Voss (62) suggested that determination of which attributes would be affected by different factors cannot be determined a priori (62, p. 112), but he suggests that it is likely that studies of interaction intensity should focus on design "features which involve the relative use of space and those which deal with variability within design configurations, particularly continuous or inter- val scale measurements such as design element repetitions and the width of particular lines" (62, p. 105). Such attributes may be good measures of interaction because they are not recognized as components of the formal design structure and thus are not regarded as part of the design by other potters (62, p. 106). Thus, "the nuances of style, those design attributes which are often identified with individual variability, appear to be reflective of interaction" (62, p. 106). In contrast, Voss (62, p. 112, 205) suggests that common, discrete or nominal scale design attributes which are highly visible and part of the formal design structure are more likely to be used for information exchange. Thus, "it would be expected that stylistic variability should incorporate at least one dimension related to social group affiliation and group linkages, and one dimension related to the specifics of social group interaction" (62, p. 274).

Although Voss suggests the patterns of stylistic variation on neolithic cera- mic assemblages from northwestern Europe are consistent with his proposals, consideration of other data provides only mixed support for his specific argu- ments. Friedrich's (20, p. 338) analysis of Tarascan potters, for example, suggested three possible indicators of communication intensity. Two of the three are attributes at lower hierarchical levels of the design structure, but none of the indicators are the continuous or interval scale attributes which would be expected from Voss's proposal, although some are characteristics which vary within design configurations. Rather, they are structural characteristics of the designs, such as the organization of spatial divisions, or discrete characteris- tics, such as the shapes of areas that may be filled with particular design characteristics. Also, as noted above, Hill (28, 29) has argued on the basis of his studies that variation in a number of continuous stylistic attributes would not be affected by the intensity of interaction between individuals. In support of Voss's proposal, however, Newton's (47) analysis of differences in hammocks between two Brazilian tribes indicated that two discrete attributes of the twining were statistically significant indicators of tribal boundaries while several continuous twining measures were not. Also, analysis of design attri- butes on ceramic vessels from a small region of the American Southwest has shown that factors such as exchange, stylistic change through time, and vessel forms tend to have a more significant impact on continuous attributes than on nominal or discrete attributes. Seventy percent of tests involving continuous attributes revealed statistically significant differences while only 27 percent of the tests with nominal attributes produced statistically significant results (51, p.

Page 8: Plog - Style in Artifacts

STYLE IN ARTIFACTS 131

113). This pattern may be related to Graves' (22, p. 296) proposal that "The lower the level of decoration in the design hierarchy and in motor performance, the greater the likelihood of design variation or change."

In addition to Voss's proposals, other studies also have suggested rela- tionships between specific design characteristics or hierarchical levels and factors which would be expected to cause variation in those characteristics. Similarities and differences in the frequency of symmetry classes have been argued to be a result of variation in interaction intensities (63, 64). Analysis of Kalinga ceramic design suggested that "easily substituted design attributes," attributes that can be interchanged without affecting the coherence of the entire design system, "least reflect kin group affiliation" (22, p. 293). In contrast, structural classes of design tend to be associated with individual settlements or a community made up of a group of settlements (22, p. 297). None of these proposals should be regarded as established facts, but all of them are important steps toward increasing our understanding of stylistic variation.

Whether or not these specific proposals prove to be correct, it is clear that certain types of attributes do vary in a different manner than other attribute types and thus may be affected by different factors. This simple conclusion has several important implications. First, factor, regression, or other multivariate analyses of large sets of stylistic attributes are unlikely to produce any clear-cut spatial patterns since variation in the different attributes may have been pro- duced by a variety of different factors. Thus, it is not surprising that reanalysis of the long design element lists used in some of the early ceramic sociology studies (27, 43) has failed to reveal any spatial clusters of stylistic attributes that could correspond to residence groups (50).

Second, not only our attempts to understand patterns of stylistic variation but also our use of such variation to understand the archaeological record will be improved by focusing on individual attributes or at least small sets of similar types of attributes. One such improvement is exemplified by several recent studies which were conducted in an attempt to improve dating accuracy in different areas. As noted above, artifact types overwhelmingly have been the analytical unit which archaeologists employ in most areas to date sites. While in part based on technological attributes, types also are defined to a large extent by a variety of stylistic attributes of different hierarchical or structural levels. Although the use of types to date sites undoubtedly has been successful in most cases, the recent studies referred to above suggest that further refinements in dating accuracy can be achieved by focusing on individual attributes rather than artifact types (14, 24, 40, 53).

This can best be illustrated by considering a study in the American South- west, an area where ceramic types have been studied and redefined for over 50 years and have been dated almost throughout that period through associations with tree-ring samples (5). Types in the area are defined by a number of

Page 9: Plog - Style in Artifacts

132 PLOG

different technological and stylistic characteristics, including both nominal and interval scale stylistic attributes of different hierarchical levels, but are largely distinguished from each other by a variety of decorative characteristics (45, p. 252). For example, one ceramic type, Kana-a Black-on-white has been char- acterized as having the following design features: thin lines, zigzag lines, secondary ticks as opposed to dots, irregular lines, equilateral triangles with appended flags, and overlapping lines. Some have argued that prehistoric sites in the American Southwest can be dated to periods as short as 25 years by considering the relative frequencies of ceramic types. However, a recent multiple regression analysis of tree-ring dates and ceramic type frequencies from sites on Black Mesa in northeastern Arizona indicated that the standard error of the dates estimated from the type frequencies was +44 years (39).

In a subsequent study, a similar analysis was done by using frequencies of individual stylistic attributes rather than the frequency of types (53). The attributes considered, however, were those used to define ceramic types in the area. Of the 52 attributes considered, the frequencies of only 26 (50 percent) were correlated with the tree-ring dates at higher than the 0.50 level. Thus, less than 25 percent of the variation in the frequency of half the attributes used to define the traditional ceramic types could be accounted for by factors causing stylistic change through time. More importantly, when multiple regression analysis was used to measure the extent to which the tree-ring dates could be estimated from the attribute frequencies, it was found that estimated dates with a standard error of ? 19 years could be generated using an equation based on the frequencies of only three attributes. All three of the attributes selected by the step-wise multiple regression analysis were characteristics at lower hierarchical levels in the design structure, a finding that is consistent with Graves' (22) conclusion that there is a greater likelihood of change or design variation as the level of decoration in the design hierarchy or in motor performance decreases. A subsequent study using data from other areas of the American Southwest has produced results similar to the Black Mesa study (24). These analyses thus support the argument that some stylistic attributes vary in a different manner from other attributes. It also demonstrates that important improvements can be made in areas such as dating accuracy by focusing on individual attribute frequencies, as Rowe (57) suggested, rather than by using analytical units such as types or analyzing stylistic units such as the design elements (27, 44, 60) that are defined by similarities and differences in a number of different stylistic attributes. Although the reliability of style as an indicator of chronology has sometimes been questioned (10), the problem is not that style is unreliable, but that some dating studies have focused on the wrong aspects of style or have emphasized polythetic analytical units such as types. Because variation in the multiple attributes used to define types may be caused by different factors, variation in type frequencies also will be caused by a number of factors rather

Page 10: Plog - Style in Artifacts

STYLE IN ARTIFACTS 133

than simply those that cause stylistic change through time. Polythetic type definitions thus in some cases may obscure the dating value of individual stylistic attributes.

Archaeological studies such as those noted above and ethnoarchaeological analyses such as Graves' (22) examination of the relationship of stylistic variation and birth cohorts are likely to increase our ability to date sites accurately using stylistic attribute frequencies. The selection of attributes for different types of archaeological studies often has been intuitive (65, p. 224). As our understanding of the factors that cause variation in different attributes increases, however, more explicit selections can be made based on the topic of interest. As a result, our ability to describe and understand the archaeological record should improve.

EXCHANGE AND STYLISTIC VARIATION

The relationship between spatial patterns in style distributions and the produc- tion, exchange, and consumption of material items is another topic upon which future studies should concentrate. Studies of stylistic variation too often have been based on the assumption that the artifacts analyzed were locally produced (26). As more and more studies of exchange patterns have been conducted using mineralogical or chemical characterization, however, it has become clear that exchange was much more common prehistorically than many studies, particularly those focusing on stylistic analysis, have recognized. Thus, the extent to which patterns of spatial varation in stylistic attributes can be ex- plained by the spatial distribution of exchange networks must be considered more frequently in future studies.

While this point is a simple one, its importance cannot be underemphasized because without an understanding of artifact production and consumption, the conclusions of many analyses of stylistic variation must be questioned. In one recent study of ceramic vessels, for example, Crown (10) drew several conclu- sions about the nature and causes of decorative variation on the vessels, conclusions that were based on the assumption that the vessels were all locally produced. However, although some mineralogical and chemical tests were made in an effort to determine whether or not the vessels studied had been locally produced, the results of those analyses were ambiguous. If, in fact, some of the vessels were not produced locally as assumed, many of the conclusions which were drawn concerning the causes and patterns of stylistic variation would not be supported. Thus, the absence of adequate answers to questions concerning the location of artifact production leaves the conclusions of many stylistic analyses questionable.

Beyond the simple point that exchange networks can affect patterns of stylistic variation, there are other issues concerning the relationship between

Page 11: Plog - Style in Artifacts

134 PLOG

exchange and stylistic variation that have been raised by recent studies. First, the nature of artifact production and consumption has implications for the extent to which conclusions derived from ethnographic or ethnoarchaeological studies are relevant to any prehistoric time period or geographical area or the extent to which the patterns of stylistic variation discovered in ethnographic studies will coincide. Longacre (44), for example, chose the Kalinga area for his ethnoarchaeological study because it provided a context in which "variabil- ity in ceramics and other cultural materials as well as aspects of behavior and organization that they might reflect" could be controlled. For example, pottery was produced by each household, for the use of that household (44, p. 51). Longacre notes (44, pp. 50-51) that if the ceramics had been produced by specialists for distribution in a market, then different "selective pressures" would influence the manufacture of the vessels. Thus, while the Kalinga provided an excellent context for Longacre's study, the extent to which the conclusions of that study are relevant to other situations will depend on several factors including the degree to which the production and consumption of artifacts is comparable to the Kalinga.

Second, too often it is assumed that there is a simple relationship between stylistic patterns, the nature of craft production, and the degree of exchange. Two examples can be used to illustrate this problem. Hodder (33) has noted that we often take it for granted that the extent of centralized production can be inferred from the spatial distribution of stylistic characteristics. That is, we assume that local production results in localized style distributions and that centralized production and abundant exchange will produce widespread, homogeneous style distributions. His ethnoarchaeological study in Africa did not support that assumption, however, and Van der Leeuw (61, p. 400) has noted other studies which revealed similar problems. Some archaeological studies of the distribution of stylistic attributes also suggest that artifact produc- tion actually may have been more localized during periods of widespread style zones than during periods of localized style zones (26, 51).

In addition to the fact that stylistic distributions across space cannot be used to infer production characteristics, patterns of stylistic change through time cannot be tied to particular modes of production. Similar patterns of stylistic change over time may be a result of different modes of production. Braun's (3) analysis of Woodland period ceramics from Illinois, for example, demon- strated a decrease in both stylistic variation and vessel wall thickness on locally produced pots during a portion of the period studied. These changes were shown to be a result of related changes in social networks and subsistence. A similar trend of decreasing decorative diversity and vessel wall thickness was discovered by Irwin (35, pp. 300, 315;36) in a study of ceramics from the Mailu area of southeast coastal Papua New Guinea. In that case, however, it was argued that both trends were a result of increasing specialization and centraliza-

Page 12: Plog - Style in Artifacts

STYLE IN ARTIFACTS 135

tion in the production of pottery, a conclusion that was reached after mineralo- gical and chemical analyses of sherds and clays. Thus, Braun's and Irwin's studies, as well as a somewhat similar case discussed by Krause & Thorne (38, p. 253), illustrate the possibility of similar stylistic trends resulting from different modes of artifact production.

This lack of a simple relationship between stylistic patterns, craft produc- tion, and exchange suggests that the frequent attempts to infer exchange patterns from stylistic distributions are often likely to result in incorrect infer- ences (26) and may lead to circular reasoning. For example, in her symmetry analysis of ceramic designs on a group of prehistoric vessels, Washburn (63, p. 172) assumed that a high degree of structural similarity in designs within sites indicated local production of the pottery. No mineralogical or chemical charac- terization was done to test that assumption, and the possibility that the structu- ral similarity could be the result of the importation of large numbers of vessels, a phenomenon that has been demonstrated for regions near Washburn's study area as well as in many other regions of the world, was not tested adequately. Also, while Washburn argues on the one hand that the vessels were produced locally because the designs have a high degree of structural similarity, on the other hand she explains the structural similarity by the proposal that the vessels were produced locally by a group of interacting potters (63, pp. 172-82). Escape from such circular reasoning can only be achieved if style and exchange are measured independently, rather than assuming that exchange can be infer- red from patterns of stylistic variation.

Such studies must not, however, oversimplify the complex dimensions of craft production. Hantman and I (26) have noted that "too often the extremes of household production or regional specialization are assumed to be the dominant modes of production in an area, although considerable variation exists between these extremes." The production and distribution of craft items can vary along a number of different dimensions, and it would be surprising if such variation did not seriously affect spatial and temporal variation in stylistic patterns in many areas. Also, Rubertone (58) has argued that patterns of consumption must be understood as well as patterns of production. Thus, in order to understand prehistoric patterns of stylistic variation, it is important that we direct greater attention to the relationship of style and aspects of production, exchange, and consumption. Several recent studies (e.g. 12, 13, 17, 56) indicate many analyses of this type are beginning to be conducted in several areas.

THE SIZE OF STUDY AREAS

One of the major types of stylistic data that is currently lacking in most areas is detailed information on the frequencies of individual stylistic attributes over broad areas. This type of information is necessary for a variety of reasons.

Page 13: Plog - Style in Artifacts

136 PLOG

First, it has been noted that different models of stylistic variation lead to somewhat different expectations concerning the distributions of stylistic attri- butes. Because of the lack of detailed regional stylistic information, it is difficult to evaluate these expectations. Second, it is clear from the minimal amount of data available that different stylistic attributes have somewhat different spatial distributions and any comprehensive theory of style must account for those differential distributions (48, p. 18). Third, many studies in the last few decades have stressed that prehistoric populations participated in widespread social networks. If the organizational and demographic characteris- tics of those networks are likely to influence aspects of stylistic variation and, in turn, if information on stylistic variation is useful in making inferences about social networks as some have argued (4, 26, 52), it is important that analyses of stylistic variation begin to encompass such broad geographical areas.

Although some stylistic analyses traditionally have considered regional style distributions in order to construct space-time frameworks, much of that in- formation is encoded in type descriptions, as noted above. In addition to the fact that such units are polythetic and mix stylistic attributes of different hierarchical levels, they also suffer from another serious weakness. Because of a variety of factors, such as the inexplicit nature of some type descriptions and the subjectivity involved in the identification of types, it appears that many differences exist among archaeologists in their definitions of types and, as a result, in their classification of artifacts (53). Fish (18), for example, has found discrepancies in type identifications as high as 30 percent among archaeologists trained at one university by a single individual. In another experiment, the ability of several archaeologists who work in the American Southwest to agree on the ceramic types represented by 27 whole vessels from the Cibola area of east-central Arizona and west-central New Mexico was tested (59a). The 20 archaeologists were selected for their knowledge of Cibola ceramics. The experiment indicated that the average agreement on type identifications was below 50 percent, and in one case as many as 13 different type names were suggested for one vessel. Type identifications thus may suffer from a strong lack of replicability. Therefore, despite the excavation of numerous sites within many regions and the publication of the information from that research, it is questionable whether or not reliable information on stylistic distributions across broad areas can be extracted from such reports.

Archaeological and ethnoarchaeological studies are needed in order to col- lect information on frequencies of stylistic attributes across space and some research of this type already has been completed (e.g. 3, 66, 67). Although published analyses of data from previous excavations or surveys may often be inadequate, in many instances reanalysis of the collections from such fieldwork may be an efficient means of obtaining the necessary geographic coverage while minimizing costs.

Page 14: Plog - Style in Artifacts

STYLE IN ARTIFACTS 137

COVARIATION OF STYLISTIC PATTERNS

In addition to studying the distribution of stylistic characteristics over larger areas, other changes in stylistic analyses are also needed. One of the consistent characteristics of many studies of patterns of stylistic attributes is that they are limited to attributes on only one type of material or artifact or even subset of an artifact type, such as ceramic bowls. Ceramic vessels, in particular, have been emphasized in such analyses. There are at least two reasons for this limitation. First, some materials such as ceramics are abundant in the archaeological record of many regions and time periods, while many other materials are scarce. Second, Green (23) has noted that there are basic differences in the processes by which different artifacts are manufactured. For example, she characterizes the manufacture of ceramic vessels and stone tools as additive and subtractive processes, respectively. That is, ceramic vessels are produced by combining a number of different materials, while the majority of stone tools found by archaeologists were manufactured by the removal of parts of a single type of material. In addition to the implications this difference has for determin- ing the area of production for tools made through subtractive versus additive processes, it is also likely that such differences can account in part for dissimi- larities in the degree of stylistic variation present on different types of artifacts, and thus the extent such artifacts are useful in developing space-time frameworks for different regions.

Some archaeologists have argued that stylistic analyses conducted to test proposals such as the social interaction theory should be limited to ceramic vessels and to painted designs on those vessels in particular. I suggest, howev- er, that it is important that future studies should begin to examine when possible the degree of covariation of stylistic patterns on different materials, different types of artifacts, or different forms of an artifact type if an adequate under- standing of the complex causes of stylistic variation is to be achieved. First, in addition to the suggestion noted above that basic characteristics of the manufac- ture of different artifacts may affect degrees of stylistic variation, it also has been proposed that very specific aspects of the manufacture or decoration of artifacts such as ceramic vessels must be considered. In discussing Mississip- pian ceramics from the eastern United States, Brown (6, p. 128-29) states that "a distinction must be made between two independent systems-a ceramic decorative technology restricted to wet-paste manipulation (as exhibited in modeling, incising, punctating, etc) and dry-paste decoration best known through engraving and painting." He argues that it is not accidental that there are clear differences in the form and distribution of styles executed in the two technologies, as wet-paste decoration must be completed during the manufac- ture of the vessel, and thus must be done by the potter, while dry-paste decoration is free from such restrictions. Brown (6, p. 129) suggests that "from

Page 15: Plog - Style in Artifacts

138 PLOG

that simple difference ensue important potential differences in the sociology of ceramic decorative production."

In addition to their relevance to Brown's study of Mississippian ceramics, such factors could be important, for example, in accounting for some of the very different degrees of stylistic variation presented on corrugated (wet-paste technology) and black-on-white (dry-paste technology) painted vessels in the American Southwest, differences that are difficult to explain using any of the theories of style discussed above (51, p. 138).

Second, Saitta (59) has suggested that if stylistic variation is viewed as a means of communicating information such as social affiliation, it would be expected that such messages "should be associated with artifacts requiring little post-production maintenance or artifacts which have low turnover rates, so that message integrity and longevity is maximized." This proposal, for example, also could help account in part for the very different levels of stylistic variation in banded or corrugated and painted ceramic vessels from the American Southwest. Banded or corrugated vessels from the Southwest appear to be containers used for cooking food (51, pp. 82-96) and, from ethnographic studies (21, 44), would likely have relatively short use-lives. Painted vessels, however, were probably used as serving bowls and for storage, activities which result in longer use-lives for vessels (21, 44). Given Saitta's proposal, the longer use-lives of painted vessels could increase their utility for carrying social messages, and this could account in part for the greater degree of decorative elaboration on such vessels as opposed to banded and corrugated pottery from the Southwest.

Third, as noted above, one expectation of the information exchange theory is that items of differing visibility will vary in utility as mediums for stylistic messages. As a result, they would be expected to have dissimilar patterns of stylistic variation. For example, it has been suggested that domestic utensils are unlikely to be useful for carrying stylistic messages owing to their low visibility (69). In particular, it has been proposed that artifacts such as ceramic vessels would not be used in efforts to communicate information such as social group affiliation (22, pp. 309, 314; 59). A variety of ethnographic data, however, indicates that the cultural identity of individuals may be encoded on visible items or features such as lip plugs, earrings, and hair and beard styles, as well as much less visible items or features such as stools, pottery, the manicure of dogs, butchering methods or hearth location or other subtle physical arrange- ments (33, 34, 37, 46). Present evidence thus suggests that patterns of stylistic variation on different materials may be dissimilar for several reasons. In addition, patterns of stylistic variation on items with differing degrees of visibility are not, in all cases, consistent with some of the expectations derived from the information exchange theory.

Page 16: Plog - Style in Artifacts

STYLE IN ARTIFACTS 139

If the latter conclusion is correct, it has important implications for the validity of components of the information exchange theory. Wobst has empha- sized the visibility of items in part because of his proposal that stylistic messages are more likely "if the potential receivers have little opportunity to receive the message otherwise, but nevertheless are likely to encounter it and are able to encode it" (69, p. 322). As a result, the utility of stylistic messages decreases with decreasing social distance between the sender and the receiver inasmuch as there would be fewer messages that could not be sent more cheaply using another mode of communication (69, p. 323). Visible artifacts or features thus are emphasized because it is those artifacts which are more likely to be seen by socially distant individuals. However, if patterns of stylistic variation on artifacts of different visibility are similar, these proposals would not be supported and it might be more reasonable to propose that stylistic variation on many items, including pottery, functions "as a symbol of social group identity for members of a community" (22, p. 309). Such an interpretation would be consistent with Voss's (62, p. 111) suggestion that style should be viewed as an identity function, and with some of the proposals of Hodder (33) and Rubertone (58, p. 91).

CONCLUSIONS

The above discussion of recent research on stylistic analyses and the types of studies that are needed in the future indicates some of the areas where advances have been or should be made in our understanding of stylistic variation. It should be emphasized, however, that we are still a long way from adequately testing ideas concerning the causes of stylistic variation or developing a comprehensive theory of style. More appropriate and more rigorous tests of different hypotheses will move us in that direction, but we also must begin to focus more extensively on at least one area where efforts are only beginning to be made: the interaction between stylistic variation, and not only social but also economic, political, demographic, religious, and ideological variables. Wiess- ner (66, 67), for example, has conducted ethnographic studies of the rela- tionship between spatial patterns of stylistic variation and alternative social and economic strategies for reducing risks in social and natural resources. Some- what similar studies have been carried out using archaeological data (2-4, 51, 52, 62). Economic factors also have been stressed in some of the studies of Hodder (31, 33). Possible relationships among stylistic variation, population density, marriage systems, and the development of discrete communications networks have been considered in another series of studies (4, 26, 51), while aspects of social stratification, ideology, and age and sex differences in rela- tionship to style have been discussed by Hodder (33). Continued efforts in these

Page 17: Plog - Style in Artifacts

140 PLOG

directions are needed, however. Additional ethnoarchaeological studies, parti- cularly analyses such as Wiessner's (66, 67) which consider broad regions, should be an important part of such an effort. Ethnoarchaeological studies will not provide all of the answers, however. Archaeological studies which focus on the development of stylistic patterns over long periods of time should and will make an important contribution. Analyses of stylistic variation have a long history in archaeology and they also have a promising future.

Literature Cited

1. Bradley, R., Hodder, I. 1979. British prehistory: an integrated view. Man 14:93-104

2. Braun, D. P. 1977. Middle Woodland- (Early) Late Woodland social change in the prehistoric central midwestern U.S. PhD thesis. Univ. Mich., Ann Arbor. 456 pp.

3. Braun, D. P. 1984. Ceramic decorative diversity and Illinois Woodland regional integration. In Measurement and ex- planation of ceramic variation: some current examples, ed. B. A. Nelson. Car- bondale: South. Ill. Univ. Press. In press

4. Braun, D. P., Plog, S. 1982. Evolution of "tribal" social networks: theory and prehistoric North American evidence. Am. Antiq. 47:504-25

5. Bretemitz, D. A. 1966. An appraisal of tree-ring dated pottery in the Southwest. Anthropol. Pap. Univ. Ariz. No. 10

6. Brown, J. A. 1976. The southern cult reconsidered. Midcont. J. Archaeol. 1: 115-35

7. Colton, H. S. 1953. Potsherds. Flagstaff: North. Ariz. Soc. Sci. Art. 86 pp.

8. Conkey, M. W. 1978. Style and informa- tion in cultural evolution: toward a pre- dictive model for the Paleolithic. In So- cial Archaeology: Beyond Subsistence and Dating, ed. C. L. Redman, M. J. Berman, E. V. Curtin, W. T. Langhome, N. M. Versaggi, J. F. Wanser, pp. 61- 85. New York: Academic. 471 pp.

9. Conkey, M. W. 1980. Context, struc- ture, and efficacy in Paleolithic art and design. In Symbols as Sense, ed. M. L. Foster, S. H. Brandes, pp. 225-48. New York: Academic. 432 pp.

10. Crown, P. L. 1981. Variability in cera- mic manufacture at the Chodistaas site, east-central Arizona. PhD thesis. Univ. Ariz., Tucson

11. Deetz, J. 1965. The dynamics of stylistic change in Ankara ceramics. Ill. Stud. Anthropol. No. 4

12. Deutchman, H. L. 1979. Intraregional interaction on Black Mesa and among

the Kayenta Anasazi: The chemical evi- dence for ceramic exchange. PhD thesis. South. Ill. Univ., Carbondale. 288 pp.

13. Deutchman, H. L. 1980. Chemical evi- dence of ceramic exchange on Black Mesa. In Models and Methods in Region- al Exchange, ed. R. E. Fry, pp. 119-33. SAA Papers No. 1

14. Drennan, R. D. 1976. Fabrica San Jose and Middle Formative society in the Val- ley of Oaxaca. Mem. Mus. Anthropol. Univ. Mich. No. 8

15. Deleted in proof 16. Englebrecht, W. 1974. The Iroquois:

archaeological patterning on the tribal level. World Archaeol. 6:52-65

17. Feinman, G. M. 1980. The relationship between administrative organization and ceramic production in the Valley of Oax- aca, Mexico. PhD thesis. City Univ. New York, NY. 537 pp.

18. Fish, P. R. 1978. Consistency in archaeological measurement and classi- fication. Am. Antiq. 43:86-89

19. Flannery, K. V. 1976. Analysis of stylis- tic variation within and between com- munities. In The Early Mesoamerican Village, ed. K. V. Flannery, pp. 251-54. New York: Academic. 373 pp.

20. Friedrich, M. H. 1970. Design structure and social interaction: archaeological im- plications of an ethnographic analysis. Am. Antiq. 35:332-43

21. Gill, M. N. 1981. The potter's mark: contemporary and archaeological pot- tery of the Kenyan southeastern high- lands. PhD thesis. Boston Univ., Bos- ton, Mass. 244 pp.

22. Graves, M. W. 1981. Ethnoarchaeology of Kalinga ceramic design. PhD thesis. Univ. Ariz., Tucson. 360 pp.

23. Green, M. 1982. Chipped stone raw materials and the study of interaction. PhD thesis. Ariz. State Univ., Tempe. 360 pp.

24. Hantman, J. L. 1983. A socioeconomic interpretation of ceramic style distribu-

Page 18: Plog - Style in Artifacts

STYLE IN ARTIFACTS 141

tions in the prehistoric Southwest. PhD thesis. Ariz. State Univ., Tempe

25. Hantman, J. L., Lightfoot, K. G. 1978. The analysis of ceramic design: a new method for chronological seriation. In An Analytical Approach to Cultural Re- source Management: The Little Colorado Planning Unit, ed. F. Plog, pp. 38-63. Ariz. State Univ. Anthropol. Res. Pap. No. 19

26. Hantman, J. L., Plog, S. 1982. The rela- tionship of stylistic similarity to patterns of material exchange. In Contexts for Prehistoric Exchange, ed. J. E. Ericson, T. K. Earle, pp. 237-63. New York: Academic. 321 pp.

27. Hill, J. N. 1970. Broken K Pueblo: pre- historic social organization in the Amer- ican Southwest. Anthropol. Pap. Univ. Ariz. No. 18

28. Hill, J. N. 1977. Individual variability in ceramics and the study of prehistoric so- cial organization. In The Individual in Prehistory: Studies of Variability in Style in Prehistoric Technologies, ed. J. N. Hill, J. Gunn, pp. 55-108. New York: Academic. 258 pp.

29. Hill, J. N. 1978. Individuals and their artifacts: an experimental study in archaeology. Am. Antiq. 43:245-57

30. Hodder, I. 1977. The distribution of material culture items in the Baringo district, western Kenya. Man 12:239- 69

31. Hodder, I. 1979. Economic and social stress and material culture patterning. Am. Antiq. 44:446-54

32. Hodder, I. 1981. Society, economy, and culture: an ethnographic case study amongst the Lozi. In Pattern of the Past, ed. I. Hodder, G. Isaac, N. Hammond, pp. 67-95. New York: Cambridge Univ. Press. 443 pp.

33. Hodder, 1 1982. Symbols in Action. New York: Cambridge Univ. Press. 244 pp.

34. Hole, F. 1978. Pastoral nomadism in western Iran. In Explorations in Ethnoar- chaeology, ed. R. A. Gould, pp. 127-67. Albuquerque: Univ. New Mexico Press. 329 pp.

35. Irwin, G. J. 1978. Pots and entrepots: a study of settlement, trade, and the de- velopment of economic specialization in Papuan prehistory. World Archaeol. 9:299-319

36. Irwin, G. J. 1983. Chieftainship, kula, and trade in Massim prehistory. In The Kula: New Perspectives on Massim Ex- change, ed. J. W. Leach, E. R. Leach. New York: Cambridge Univ. Press. In press

37. Jones, R. 1974. Tasmanian tribes. In Aboriginal Tribes of Australia, ed. N. B.

Tindale, pp. 319-53. Berkeley: Univ. Calif. Press. 404 pp.

38. Krause, R. A., Thorne, R. M. 1971. To- ward a theory of archaeological things. Plains Anthropol. 16:245-57

39. Layhe, R. W. 1977. A multivariate approach for estimating prehistoric population change. MA thesis. South. Ill. Univ., Carbondale, Ill. 95 pp.

40. LeBlanc, S. A. 1975. Micro-seriation: a method for fine chronological dif- ferentiation. Am. Antiq. 40:22-38

41. Leone, M. 1968. Neolithic autonomy and social distance. Science 162:1150- 51

42. Longacre, W. A. 1964. Sociological im- plications of the ceramic analysis. In Chapters in the Prehisotry of Eastern Arizona, ed. P. S. Martin, J. B. Rinaldo, W. A. Longacre, L. G. Freeman, J. A. Brown, et al, 2:155-67. Fieldiana: Anthropol. 53

43. Longacre, W. A. 1970. Archaeology as anthropology: a case study. Anthropol. Pap. Univ. Ariz. No. 17

44. Longacre, W. A. 1981. Kalinga pottery: an ethnoarchaeological study. See Ref. 32, pp. 49-66

45. Martin, P. S., Plog, F. 1973. The Archaeology ofArizona. New York: Am. Mus. Nat. Hist. 422 pp.

46. Myers, T. P. 1975. Isolation and ceramic change: a case from the Ucayali River, Peru. World Archaeol. 7:333-51

47. Newton, D. 1974. The Timbira ham- mock as a cultural indicator of social boundaries. In The Human Mirror, ed. M. Richardson. Baton Rouge: Louisiana State Univ. 366 pp.

48. Plog, F. 1977. Archaeology and the indi- vidual. See Ref. 28, pp. 13-21

49. Plog, S. 1976. Measurement of prehis- toric interaction between communities. See Ref. 19, pp. 255-72

50. Plog, S. 1978. Social interaction and sty- listic similarity. Adv. Archaeol. Method Theory 1: 143-82

51. Plog, S. 1980. Stylistic Variation in Pre- historic Ceramics. New York: Cam- bridge Univ. Press. 160 pp.

52. Plog, S. 1980. Village autonomy in the American Southwest: an evaluation of the evidence. See Ref. 13, pp. 135-46

53. Plog, S., Hantman, J. L. 1984. Multiple regression analysis as a dating method in the American Southwest. In Spatial Organization and Exchange: Archaeolo- gical Survey on Northern Black Mesa, ed. S. Plog. Carbondale: South. Ill. Univ. Press. In press

54. Redman, C. L. 1977. The "analytical in- dividual" and prehistoric style variabil- ity. See Ref. 28, pp. 41-53

Page 19: Plog - Style in Artifacts

142 PLOG

55. Redman, C. L. 1978. Multivariate arti- fact analysis: a basis for multidimension- al interpretations. See Ref. 8, pp. 159-92

56. Rice, P. M. 1980. Peten Postclassic pot- tery production and exchange: a view from Macanche. See Ref. 13, pp. 67-82

57. Rowe, J. H. 1959. Archaeological dating and culture process. Southwest. J. Anthropol. 15:317-24

58. Rubertone, P. E. 1978. Social organiza- tion in an Islamic town: a behavioral explanation of ceramic variability. PhD thesis. State Univ. New York, Bingham- ton. 188 pp.

59. Saitta, D. J. 1982. The explanation of change in egalitarian society: a critique. Presented at Ann. Meet. Soc. Am. Archaeol., 47th, Minneapolis

59a. Swarthout, J., Dulaney, A. P. 1978. The great Cibola typological consensus test. Presented at Cibola White Ware Conf., Flagstaff, Ariz.

60. Tuggle, H. D. 1970. Prehistoric com- munity relationships in east central Ari- zona. PhD thesis. Univ. Ariz. Tucson. 169 pp.

61. Van der Leeuw, S. E. 1976. Studies in the Technology of Ancient Pottery. Am- sterdam: Univ. Amsterdam. 424 pp.

62. Voss, J. A. 1980. Tribal emergence dur- ing the neolithic of northwestern Europe. PhD thesis. Univ. Mich. Ann Arbor. 371 pp.

63. Washbum, D. K. 1977. A symmetry

analysis of upper Gila area ceramic de- sign. Pap. Peabody Mus. Archaeol. Ethnol., Vol. 68

64. Washburn, D. K. 1978. A symmetry classification of pueblo ceramic design. In Discovering Past Behavior: Experi- ments in theArchaeology oftheAmerican Southwest, ed. P. Grebinger, pp. 102- 21. New York: Gordon & Breach. 279 pp.

65. Whallon, R. 1968. Investigations of late prehistoric social organization in New York State. In New Perspectives in Archeology, ed. S. R. Binford, L. R. Binford, pp. 223-44. Chicago: Aldine. 373 pp.

66. Wiessner, P. W. 1977. Hxaro: a regional system of reciprocity for reducing risk among the Kung San. PhD thesis. Univ. Mich. Ann Arbor. 404 pp.

67. Wiessner, P. W. 1982. Beyond willow smoke and dogs' tails: a comment on Binford's analysis of hunter-gatherer set- tlement systems. Am. Antiq. 47:171-78

68. Wilmsen, E. N. 1973. Interaction, spac- ing behavior, and the organization of hunting bands. J. Anthropol. Res. 29: 1- 31

69. Wobst, H. M. 1977. Stylistic behavior and information exchange. In Papers for the Director: Research Essays in Honor of James B. Griffin, ed. C. E. Cleland, pp. 317-42. Anthropol. Pap. Mus. Anthropol. Univ. Mich. No. 61