In her introduction, Susan Kus emphasises culture as being meaningfully constituted within historical and intersubjective contexts, and describes the need for a greater dialogue between the social sciences and philosophy. The question of the legitimation of a sociopolitical order in complex societies is seen as concerning the relationship between the social order, a natural order and the individual. It is the ‘naturalisation’ of the social order which is taken up as the focus of the discussion. A distinction is made between legitimating strategies, a distinction between the confounding of the social and natural spheres of activity and representation on the one hand, and the bringing of order different from that inherent in physical nature on the other. Means for the constitution of a legitimating order in spatial relationships are examined in relation to ethnohistorical evidence from the Merina of Madagascar. The spatial layout of the regional settlement pattern and within settlement pattern is shown to play a role in a legitimation of the social order in which the natural and the socially constructed are both involved.

Donley introduces an ethnographic analysis of settlements and houses on the East African coast which aims to increase our knowledge of how settlement space is integrated into other aspects of life. An important component of the cultural context considered is the Islamic religion, but the particular relationships between Indian Ocean traders, middlemen and local tribes, and the use of material items in the social and economic strategies of these groups also play their parts. The specific models identified concern the use of house form and appearance in the relationships between trading groups and in the control of women. The high, blank, fortress-like exteriors of the houses protected the position of the Swahili middlemen and their control over ‘pure’ women who played an essential role in the trading system. The analysis also concerns the use of decoration to maintain the purity of women and to protect men and women from defiling activities within the house, in the innermost room and in the toilet. The use of ethnoarchaeology to study the location of material items within a symbolically organised settlement space is emphasised.

Traders have built coral settlements on the eastern coast of Africa from at least the ninth century AD until the present day. Archaeologists have excavated some of these Swahili sites and described the finds, house-plans, mosques, tombs and walls of the towns. The periods of occupation have been established by dating the imported Chinese and Islamic wares, by inscriptions and coins, and by noting the stylistic seriations (Kirkman 1964; Garlake 1966; Chittick 1967; Wilson 1979, 1980).

Despite the detailed studies of ceramic design variation that have been made by archaeologists, there has been little discussion of the methods that should be used in these analyses. In particular, insufficient attention has been given the question of how designs should be classified [the studies of Carlson (1970), Friedrich (1970), Redman (1977, 1978), and Washburn (1977, 1978) are important exceptions as will be noted later]. Analyses have focused on many different types of designs (or classificatory units) such as design elements or design motifs that have not been defined explicitly. The result has been an enormous amount of subjectivity in delimiting these units and, thus, a large amount of variation in what different individuals considered to be “design elements.” Even terminology in describing a particular design varies tremendously. In summary, few people have critically evaluated our classifications to ask if they are reasonable ones. This is important, for as Hill (1970:17), among others, has argued, there is no reason to analyze intrasite or intersite variation in the distribution of items unless the items have been classified using sound methods.

Because there has been so little discussion of design classifications, it is difficult to describe how designs have been classified in most analyses. Many studies simply present a table showing drawings of different designs and their frequencies.

The tests presented in Chapters 6, 7, and 8 support the hypotheses presented in Chapter 2. The tests show that ceramic exchange, temporal variation, and stylistic differences between different vessel forms all account for part of the ceramic design variation at the sites included in the analysis. In addition, arguments for the importance of considering settlement-subsistence systems in explanations of stylistic variation are presented.

The results of the tests are summarized in Table 8.1; the table indicates that of temporal variation, vessel form, and exchange, the latter variable accounts for the most significant differences in the design attribute state frequencies on ceramics from Chevelon sites. There are significant differences between two pottery classes, PGM and SShO, in the relative frequencies of different primary forms, types of design composition, and type of hatching; in the frequency of appended secondary forms and rim decoration; and in the size of primary lines, hatching lines, framing lines, and the spacing of hatching lines. The co-occurrence of these two classes on Chevelon sites is a result of the importation of SShO vessels, most likely from the area around Winslow. The two pottery classes are characteristic of two cultural traditions in the American Southwest, the Winslow branch of the Anasazi and the Mogollon.

In this chapter, I will discuss a series of hypotheses which may explain part, but not all, of ceramic design variation within or between sites. A basic premise that guided the formulation of these hypotheses is that design variation is a result of a number of different factors. That is, no one variable will explain a large percentage of the variation in design frequencies but, rather, there are multiple factors each of which will explain a portion of the variation. At a time when archaeologists have increasingly recognized the complex nature of cultural systems, it is surprising and unfortunate that many of those studying ceramic designs have assumed that design variation is not a complex phenomenon and thus could be described and explained very simply.

This oversimplification can be illustrated in several ways. First, there has been a highly selective use of ethnographic information. The primary ethnographic information that was emphasized by studies in the American Southwest, for example, was Bunzel's discovery that Pueblo potters of the 1920s primarily had learned ceramic designs from their mothers (Bunzel 1972:54). This information led to the proposal that design clusters within archaeological sites could be interpreted as the loci of matrilocal residence groups. However, other ethnographic data that would affect the probability that this proposal would be true were ignored (Stanislawski 1973).

Studies of stylistic variation in prehistoric artifacts have played an important role in archaeological research since the beginning of the discipline. A variety of different types of studies have been done using a number of different methods and types of artifacts. The purpose of the majority of these studies, however, can be placed in one of two categories. First, the largest amount of research has concentrated on the discovery and description of stylistic change through time in order to date sites. Using sets of artifacts from stratigraphic sequences or from dated deposits for temporal control, it has been established in many areas that stylistic attributes changed through time. Early examples of such studies include the excavations of Kidder (1931, 1972) and Nelson (1916) in the American Southwest, Ford (1935) in the southeastern United States, Gamio (1913) and Vaillant (1930, 1931) in the Valley of Mexico, Bennett (1934) in Bolivia, and innumerable studies in Europe where stratigraphic excavations first were used as an important means of establishing temporal sequences. Such studies have continued to be important as a means of establishing new sequences, verifying old ones, or developing methods of making more precise estimates of occupation dates [e.g., the research of Drennan (1976) in the Valley of Oaxaca, Mexico, Snarkis (1976) in Costa Rica, and Thomas and others (1976) in Luxembourg].

The analysis of stylistic variation on prehistoric artifacts has been a major component of archaeological research for decades, and the amount of literature that has been generated is immense. It would be extremely difficult for a single monograph to discuss all of the different issues that have been raised concerning stylistic variation and this study does not attempt to do so. Rather, it is an attempt to deal with one primary topic, the explanation of patterns of stylistic variation, which I feel has not received enough attention in previous analyses, particularly many that have been carried out in the last two decades.

Many types of inferences about the past and several methods of measuring characteristics of prehistoric human adaptive systems are based on patterns of stylistic variation and on assumptions about the factors which cause that variation. However, these assumptions are oversimplified in many instances and have not been adequately tested. Given this problem and the importance of patterns of stylistic variation in archaeological inferences, we must begin to try to understand the full range of factors that determine degrees and patterns of stylistic variation rather than to assume that these factors are already known. This study is an attempt at an initial step in that direction.

A brief discussion of the possible causes of ceramic exchange and of the possible importance of exchange in explaining spatial variability in ceramic designs was presented in Chapter 2. In this chapter, I will discuss briefly methods for identifying the exchange of materials and then will discuss in greater detail ceramic exchange in the prehistoric Southwest. Two primary topics will be considered in the latter discussion: First, the evidence concerning the frequency of ceramic exchange in the Southwest as known from previous studies will be summarized. Second, a specific case will be examined–the significance of ceramic exchange as an explanation of design variability in the Chevelon drainage area.

Methods of studying ceramic exchange

Two primary types of data and several different methods of collecting that data are useful for the study of prehistoric ceramic exchange systems. Only recently, however, have some of these types of analyses been widely available. Shepard has argued that information on the mineral content of ceramic vessels obtained through optical mineralogy (petrographic analysis) or X-ray diffraction analysis is the most useful data (1971:x). She notes that such analyses provide information on the materials used by the potter and the methods used, and also may provide clues concerning the sources of the materials.

The need to study design change over short periods of time has been discussed earlier. I have argued that we presently have little information on this topic, either in the American Southwest or in other areas of the world. It was hoped that tree-ring dates obtained through limited excavation by CARP in 1973 and 1974 would enable me to date precisely several sites with large black-on-white ceramic collections in order to carry out a study of temporal change in design. However, none of the dates obtained was a cutting date (beams with evidence that the last growth ring is present and that thus can be used to date the death of the tree), so it is not possible to pinpoint the construction dates of the sites. In addition, the maximum number of noncutting dates from a single site was three and for most sites only a single date was obtained. It was, therefore, not possible to infer construction dates from the clustering of noncutting dates. [The clustering of noncutting dates has been the only accepted method for inferring construction dates when cutting dates are not available (Bannister 1962, Dean 1978:250).] It is generally argued that little can be said about the construction date of a site from which only one or two noncutting dates were obtained.

The utility of the hypotheses presented will be assessed by evaluating their importance in explaining design variation in one area of the American Southwest. The particular region to be considered is the southern part of the Chevelon Creek drainage in east-central Arizona and the Phoenix Park, Day, and Cottonwood Wash drainages to the east of Chevelon Creek. Inasmuch as these areas have only recently received the attention of archaeologists and little published material is available, I will summarize the work that has been done in the area and the knowledge of the area's prehistory. First, however, a brief outline of the prehistory of the northern part of the American Southwest will be included for those who are unfamiliar with the area.

The American Southwest

The Chevelon Canyon area generally is considered part of the northern half of the American Southwest in terms of prehistoric cultural traits and evolutionary patterns. For the purpose of this study, the latter area is defined roughly as the northeastern and east-central parts of Arizona, the western half of New Mexico, the extreme southwestern portion of Colorado, and the southeastern corner of Utah. Several cultural units or areas have been defined within that general area, of which the most inclusive for the period after A.D. 1 are the Mogollon and Anasazi.

In Chapter 3, it was proposed that the numerous small sites in the Purcell–Larson area might be seasonal farming villages used primarily for agricultural activities. The primary reasons for suggesting this hypothesis are the small size (in terms of the number of rooms) of Purcell–Larson sites relative to some nearby areas, such as the Hay Hollow Valley area of east-central Arizona, and the low densities and low total amounts of lithics and ceramics on these sites.

A comparative study of settlement data from Hay Hollow Valley and from the southern block survey in the Purcell–Larson area by F. Plog (1974) has illustrated the differences in site size and site density between these two areas. During the periods between A.D. 1125 to 1200 and A.D. 1200 to 1275, the Purcell–Larson area had an average number of rooms per site of 2.0 and 3.0, and the percentages of all rooms occurring on sites with five or more rooms were 38 percent and 46 percent, respectively. During these same periods, the Hay Hollow Valley had an average number of rooms per site of 22.0 and 14.5, respectively, and the percentage of all rooms occurring on sites with five or more rooms was 100 percent during both periods (F. Plog 1974: Tables 3 and 4).