You are on page 1of 40
~ SEE ILLiad TN: 5316! 56 IA Journal Title: Ceramic Production And Call #: GN799.P6 C47 1992 Distribution: An Integrated Approach Date: 7/20/2018 12:08:47 PM Location: General Stacks, Red Volume: Issue: Month/Year: 1992 Pages: 275-313 Requested by: Article Author: Pool, Christopher A. Article Title: Integrating Ceramic Production Alejandro Figueroa and Distribution ‘SMU 1D: (38033402) Phone: 813-417-5528 E-mail: atigueroa@smu.edu Status. Graduate Doc Del Article/Book Chapter Pull Slip searched for Staff/Student Assistant Initial: ILLiad TN: Date: 12 Integrating Ceramic Production and Distribution Christopher A. Pool Introduction This volume has a simple theme: production and distribution are interacting components of economic systems and should be studied as such. Surely, the necessary interrelation of production and distribution is universally recognized, yet my coeditor, George Bey, and I have been impressed by the scarcity of archaeological and ethnoarchaeological studies that explicitly treat ceramic production and distribution in an integrated fashion. As Prudence Rice has observed, "the distribution sphere of pottery economics has received considerably more archaeological study than has production” (Rice 1987b:197), and "surprisingly little ethnographic research has addressed pottery economics" (Rice 1987b:204). Why do these gaps in research exist? For archaeology, the most obvious answer is that studies are constrained by the nature of their data. The integrated study of production and distribution systems is greatly facilitated by the identification of production loci. This is a relatively straightforward matter where large-scale pottery producers constructed permanent facilities and produced voluminous refuse, or where potteries are historically documented. It is no accident, then, that many of the more successful reconstructions of pottery production and distribution have been described for Roman and post-Roman Europe (eg. C. J. Arnold 1981; Brisbane 1981; Hodder 1974a, 1974b, 1974c; Peacock 1982; Peacock and Williams 1986). On the other hand, archacological investigations of many complex societies, which have traditionally focused on site centers, may have missed 275 276 Christopher A. Pool even large-scale industries located on the margins or in the hinterlands of sites (€.g., in Mesoamerica [Kneebone and Pool 1985]). When production loci have been identified, reports have tended to describe the production entity and its manufacturing technology, while giving little consideration to its integration into broader economic and social systems (e.g. Bordaz 1964; Munera Bermtidez 1985; Redmond 1979; Winter and Payne 1976). These descriptions, while valuable, are only the first step in reconstructing pottery economic systems. In most prehistoric societies, however, pottery production appears to have been carried out at a much smaller scale in relatively unspecialized contexts. In such cases, identifying the precise locus of pottery production on the basis of overt categories of evidence is difficult or impossible (Stark 1985). The difficulty of identifying the precise location of pottery manufacture and the importance placed on exchange in models of sociopolitical evolution have conspired to focus more archaeological attention on the distribution of pottery (Rice 1987b:197). Without identified production loci, reconstructions of exchange systems are forced to assume that pottery was manufactured in the region where it is most commonly found (Bishop, Rands and Holley 1982:300-302; Rice 1987b: 177). While this “criterion of abundance" is a perfectly reasonable assumption, it limits the resolution with which intraregional exchange can be identified. Hence archaeological studies of ceramic distribution tend to focus on long-distance exchange (but see Fry {ed.] 1980). In contrast to archaeological studies, ethnographic and ethno- archaeological research has documented the methods and organization of pottery manufacture and the social status of the potter but has largely ignored the means by which vessels are distributed from potters to consumers (Rice 1987b:168-191, but see Howry 1976; Papousek 1981; Reina and Hill 1978). Even less well-documented are the effects of the variable demands of different groups of consumers on the choices potters make in selecting particular forms, designs, and manufacturing techniques (cf. Hardin 1977). This book advocates approaches to ceramic research that explicitly recognize and seek to elucidate the interrelations between distribution (conceived broadly to include mechanisms of exchange as well as the demands of consumers and their distribution across the landscape), and production, including such elements as resource procurement, manufactur- ing technology, organization of production, standardization of products and aesthetic choices of potters. Toward this integrative end, the editors have invited the contributors to approach their data on production, distribution, or both, with an eye toward the interactive effects of the two subsystems. Their chapters serve as case studies that illustrate a variety Integrating Ceramic Production and Distribution 27 of concerns, methods, and problems of integrated approaches to ceramic production and distribution. No single chapter in this volume has thoroughly integrated all aspects of pottery economics, though Will’s comes close. We can hardly expect that they would. All of the contributions employ data from projects that in one sense or another are in progress. Rather, these chapters must be viewed as steps along the path toward understanding the economic role of pottery in nonindustrial contexts. In this final chapter, I point out some of the larger obstacles along that path and some of the ways they may be surmounted. Assessing Production and Distribution The first obstacle to be avoided is a conceptual one. It might be surmised that an integrated approach would simultaneously consider production, consumption, and distribution in constructing typologies of economic behavior. In fact, there already exist classifications that incor- porate elements of distribution in the description of categories of ceramic production (e.g. van der Leeuw 1977, 1984; Peacock 1981, 1982:8-10; Rice 1987b:184-186). The categories described in these classifications provide a framework that many, including myself, have found valuable for organizing discussions of specialization in ceramic production (¢.g. Pool and Santley this volume). Specialization is, however, a complex phenom- enon that incorporates many different behaviors, each reacting to different sets of influences. As is true for any complex phenomenon, one cost of employing such classifications is that they obscure variation by lumping diverse cases into a handful of categories. Furthermore, to the extent that modes of distribution become a defining characteristic of the categories, they also obscure the nature of interaction between production and distribution (Rice 1987b:186). ‘An integrated approach, however, need not deny the separability of production and distribution. Quite the contrary, the explicit consideration of interaction between production and distribution requires that they be analytically distinguished. The problem is similar to that which political anthropologists face in describing and explaining the multitudinous forms of political organization that exist. Some critics of political typologies (e.g. Easton 1959) have favored the alternative approach of scaling phenomena along multiple "continua," or dimensions of variation, to avoid masking interesting variability under a few rubrics. Consistent clustering of phenomena along these dimensions (which may be parallel, or "bundled") might well lead to the recognition of distinct classes (Easton 1959:239; de Montmollin 1989:17). Recently this approach to the study of political 278 Christopher A. Pool structure has been explicitly championed in mesoamerican archaeology by de Montmollin (1989), who acknowledges that similar "continuum oriented" approaches have been applied to the development of social and political organization in the Valley of Oaxaca by Blanton, Kowalewski, Steponaitis and others (Blanton et al. 1981, 1982; Kowalewski et al. 1983; Kowalewski and Finsten 1983; Steponaitis 1981). Assessment of variation along multiple dimensions may also be applied profitably to economic systems (e.g. Torrence 1986:82-90). The following discussion identifies dimensions of variation that have been assigned particular importance in classifications of ceramic production and distribution (¢.g. Peacock 1981, 1982; Stark 1985; van der Leeuw 1977, 1984), although they have not always been explicitly separated. Some are ‘clearly highly correlated, but their analytical separation allows the empirical demonstration of those correlations.! Dimensions of Production Scale, Intensity, and Efficiency. With respect to ceramic production, the words scale and intensity are often used interchangeably. The two words convey different shades of meaning, however, which should be made explicit, but rarely are. Scale carries the connotation of gross size or amount. It may be applied equally well to the inputs of a production ‘system (energy or labor, capital, materials) as to the outputs of a system (products and by-products). The phrase “large-scale production,” for example, implies the consumption of large amounts of energy, capital, or materials by potters to produce many vessels (and incidentally, large amounts of by-products). “Intensity” implies a comparison, or ratio, of one factor to another. In the study of craft production, though, there appears to be some disagree- ment as to what comparisons are to be included within the concept of intensity. For Rice, "intensification increases the output per unit of time or labor" (1987b:190). Discussing lithic production, however, ‘Torrence argues, "it is crucial to distinguish between behavior which is labor intensive (high levels of inputs) and that which is efficient (outputs outweigh inputs Tegardless of gross quantities)" (1986:89 emphasis in original). In other words, intensive production involves high levels of inputs per production entity. Indeed, Rice appears to include this usage within her concept of intensity, for she concludes the sentence quoted above with the statement that intensification "may also be achieved by changing production from part time to full time, increasing the number of producers, or using more efficient techniques" (1987b:190). While both usages are well established in the archaeological and economic literature, Torrence’s distinction Integrating Ceramic Production and Distribution 279 between intensity and efficiency has the advantage of greater terminological precision? ‘The scale (levels of inputs and outputs) and efficiency (ratio of outputs to inputs) of production both respond to and effect changes in the demand for products, the number of production entitities (ic, households, workshops, or factories) filling that demand, the availability and organiza- tion of labor, the availability of resources, and manufacturing technology. ‘Thus, while hardly "prime movers," scale and efficiency are factors of fundamental importance in ceramic production and distribution. Archae- ological measures of scale for outputs and material inputs may be derived from the frequency, volume, mass, or number of vessels represented by discarded sherds at a production entity, given assumptions regarding breakage patterns and the proportion of vessels damaged in firing. The size of the labor force in a production entity may be estimated from the number and volume of facilities such as kilns, trampling floors, tanks, drying and storage sheds, and waster dumps, or the number of potting households nucleated in a portion of the site. The assessment of efficiency is less tractable because it requires that inputs of materials, time, or labor be assessed independently of outputs. Size of the Production Entity. The size (i.e. the spatial extent) of the production entity is clearly and strongly influenced by the scale of production, though it is also conditioned by the way activities are organized in space. The latter point is especially important when compar- ing the size of entities at differing scales of production. In large-scale entities the existence of permanent facilities rigidifies the use of space. The overall extent of such entities may therefore be an adequate measure of their size when they are compared with one another. In small-scale entities that lack permanent facilities, however, the sporadically conducted activities associated with ceramic production may be moved from place to place within a house lot, creating the impression of a larger production locus than actually existed at any one time. On the other hand, the lack of permanent facilities in small-scale production entities tend to restrict archaeological evidence for activity areas to the residues of one activity— firing—which may lead to the underestimation of the total space devoted to ceramic production. ion of Activities. A number of studies have demonstrated that production activities are segregated from domestic activities and sequential Stages of manufacture are segregated within the production entity in Tesponse to increasing production scale and specialization. Assessment of the degree of activity segregation therefore requires the, identification of the location of the production entity with respect to houselots and the identification of activity loci within the production entity. The latter may be facilitated by the presence of permanent facilities. 280 Christopher A. Pool Location of Production. Potters position themselves on the landscape in response to numerous influences. An important dimension of the geographical pattern that is created is the degree to which production entities are dispersed or nucleated. Depending upon the distribution of the consuming population, dispersed production may reflect some combination of widely available resources, inefficient exchange mechanisms, household self-sufficiency, or lack of administrative control. Conversely, production may cluster in response to restricted resources, nucleation of the consum- ing population, location of distribution nodes, or administrative regulation. In either case, and especially for relatively unspecialized, low intensity production, archaeologists must be alert for geographical patterns that Suggest the influence of economic and social considerations other than those directly related to the production and exchange of pottery? Variability of Products. The relationship between the number of vessels produced at an entity and the number of discarded sherds is by no means straightforward. Nevertheless, if the ceramic assemblage at a production entity can be taken as representative of the production output, or if differential rates and patterns of breakage can be controlled for different wares and forms, then the variability of the production entity's output may be assessed. Variability is not, however, a single dimension, and different kinds of variability have different causes and implications. The number of different vessel classes represented in the assemblage, that is, its taxonomic richness, reflects the elaboration of ceramic manufacture at the production entity (Rice 1981; 1989). The relative frequency of vessels in each class, the evenness of the assemblage, indicates the degree of product specialization at the production entity. Finally, the variability of attributes within vessel classes provides measures of their standardization. Dimensions of Consumption Although for the purposes of this volume we have subsumed consump- tion under distribution, it is, of course, an analytically separable component of economic systems. Like production, consumption may be assessed along dimensions of scale, intensity, size of constituent segments, segregation of activities, dispersion of consumers, and variability of assemblages. Comparison of consumption and production along these dimensions may, in turn, suggest forms of distribution and the overall character of economic systems. Scale and Intensity of Consumption. Scale and intensity are used here in the same sense as for production, with scale referring to the overall consumption of pots in the site or region and intensity referring to the consumption of pots per person or household, The relation between sherds Integrating Ceramic Production and Distribution 281 recovered and pots consumed, though not entirely straightforward (Schiffer 1976), is somewhat more direct than that between pots manufactured and sherds recovered from production contexts. Given a statistically representa- tive sample of disposal contexts, the scale and intensity of consumption may be inferred from the same measures of frequency, volume, mass or number of individual pots represented by discarded sherds. It is important to recognize that the scale of consumption and the scale of production within a site or region need not always coincide. Discrepancies between estimates of the number or mass of vessels consumed and produced may suggest the importation or export of vessels. Segregation of Consumption. Particularly in large and complex social systems, all vessel categories are not consumed equally by all segments of the population. The degree to which vessel categories are segregated in particular contexts within the site and region is an important dimension of variation, which may indicate the division of the society into different consuming segments along class, ethnic, or occupational lines. Size of Consuming Segments. A third dimension of interest is the number of persons represented by contexts of consumption. Like the size of production entities, the size of consuming assemblages may be expected to correlate with the scale and diversity of activities they represent, but the relationship is confounded by breakage and replacement rates of ferent vessel categories. It is preferable, therefore, to isolate this dimension of variation, and assess it independently through the methods of demographic archaeology, using such data as the number of residential structures containing each vessel class (see Hassan 1978; 1981). Nucleation of Consumers. As with production, a vital dimension of consumption is the degree to which consumers of particular vessel categories are nucleated or dispersed across the landscape. This axis of variation adds a spatial dimension to the segregation and size of consuming segments of the population. It is, however, analytically separable, since more or fewer consumers may be spatially segregated in larger or smaller areas. Spatial concordances and discordances between production and consumption may then be used to infer possible mechanisms of distribution (c.g. the chapters in this volume by Feinman ct al., Pool and Santley, and Stark). Variability of Consumer Assemblages. As with production assemblages, measures of richness, evenness, and quantitative dispersion may be applied to evaluate elaboration, specialized use, and standardization of ceramics in consumption contexts. Such measures obtained from consumption contexts have been employed to infer characteristics of production (e.g. Rice 1981, 1989), but the analytical separation of consumption and production may permit more precise characterization of the economic system, including the 282 Christopher A. Poo! number of producers patronized by individual households or population segments. Dimensions of Distribution Distribution, of course, is the act that links producers and consumers, Despite occasional discoveries of marketplaces, storage facilities, and dumps for vessels broken in transshipment (e.g. Will, this volume), it is an act that leaves few direct traces in the archaeological record. Though some characteristics of distribution patterns and mechanisms may be inferred from production or consumption contexts alone, it is the comparison of the dimensions of variability described above for both kinds of contexts (and doubtless others as well) that the richest information may be derived. These comparisons may then allow the distribution systems of a site or tegion to be scaled along the following dimensions (cf. Plog 1977:129; Rice 1987b:197-200). ‘Range and Direction of Distribution. The spatial extent over which a ceramic category occurs defines its range of distribution. Comparison of its occurrence in consumption and production contexts may allow finer estimation of the minimum distances and directions over which pots traveled. The caveat, expressed by Blinman and Wilson in this volume, that the endpoints of production and distribution do not necessarily delineate the intervening path must be heeded, however. Scale of Distribution. The total amount of pottery exchanged within a system may be estimated from consumption contexts alone, but again, if the volume of ceramics represented in consumption contexts are compared with the volume of manufacture inferred from production contexts, then a more precise characterization of the distribution system may be achieved. Interactions in a distribution system in which 1,000 consumers are supplied with 10,000 vessels by 100 producers are of a very different scale from one that involves only ten producers. So are interactions that supply 10,000 vessels to 1,000 consumers over a ycar of a different intensity from those that occur over a span of ten years. Centralization of Distribution. The degree to which distribution is centralized and/or regulated is a fundamentally important dimension of variability because it relates directly to the mechanics and social contexts of distribution as described ethnographically (e.g. Polanyi 1957; Renfrew 1977). Evidence that may be brought to bear on these questions include the spatial relationships between producers, consumers and administrative centers (c.g. Feinman 1980 and chapters in this volume by Feinman et al., Pool and Santley, and Stark), as well as the comparative diversity and Integrating Ceramic Production and Distribution 283 standardization of production and consumption assemblages (cf. Pires- Ferreira 1976). Discussiz The foregoing identifies some, but certainly not all, of the dimensions along which production, consumption and distribution subsystems may be scaled and compared. The value of analytically separating dimensions of variability is that it enhances methodological rigor by explicitly delineating the ways in which pottery economic systems vary. To stop with dissecting the living system into its component tissues, though, would hardly constitute an integrative excercise. It is only a prelude to a consideration of the interaction of these components. Points of Interaction Between Production and Distribution An integrated approach to production and distribution is one that seeks to understand the nature of interaction between producer and consumer. From the perspective of the potter, three aspects of this relationship appear to be particularly salient. These are the social context of the telationsip, the physical mechanisms by which exchange is accomplished, and the volume of demand, both overall and for particular classes of pottery. The simplest possible relationship between producer and consumer occurs when the producer and consumer are the same person (Howard 1981:5). Such cases are classified by Peacock (1982:13-17) as "household production” in which a houschold makes pottery purely for its own consumption (as opposed to "household industries," which make pottery for consumption beyond the producing household [see Pool and Santley, this volume for discussion of modes of ceramic production]). The household that makes pottery only for itself, with no exchange (e.g. through gift- giving) to other households is an ideal type, and is probably rare or nonexistant. It is, however, approximated by egalitarian societies like the Iroquois, described by Allen in this volume. Even in the ideal case, though, additional consumers exist as members of the household. It is therefore useful to regard distribution in the context of household production as severely attenuated, rather than absent. By doing so a baseline is provided against which more complex systems may be measured. The characteristics of this baseline case then include: 284 Christopher A. Pool 1. Demand is set solely by consumption rates within a single house- hold. 2. Limits of variation acceptable to the consumer are perfectly understood and reproduced by the potter. 3. As in more complex systems, distribution involves moving vessels from their location of manufacture to their location of use, but this distance is measured in meters rather than kilometers. . Middlemen are eliminated from the distribution system. Again as in more complex systems, the location of discard does not perfectly reflect the location of use. Nevertheless, secondary discard is typically rare as compared to more complex systems. wa In contrast, specialist production unites more consumers with propor- tionately fewer producers, hence the volume of demand filled by each producer increases. In addition, the greater distances and the larger number of participants involved necessitate additional mechanisms of distribution. These mechanisms involve transport of the vessels by one or both parties (ie., the producer to the consumer, the consumer to the producer, or both to a third location), and/or the intervention of a third agent (a wholesaler, middleman, or central agency) (Redman 1977:9-10). As the chapters by Arnold and Nieves, Chavez, and Zubrow illustrate, variations in the overall level of demand and the requirements of particular groups of consumers directly affect potters’ decisions regarding manufactur- ing techniques, vessel form and decoration, and the organization and location of production. In turn, these decisions condition archaeological patterns of assemblage variation and the spatial organization of production localities. In the broadest terms, the variation observed within a ceramic complex occurs within bounds set by the natural and cultural environment. Because specialist producers satisfy the demands of many households, they work within an aggregate range of acceptable variation in materials, form and decoration. Differences in the range of variation exhibited among potters should be expected as a result of differences in skill, labor investment, access to resources, and the size and composition of the consuming Population. Producers may further elect to restrict manufacture to particular classes of pottery (‘product specialization” c.f. Rice 1987b:187- 191), creating a system of complementary specialization. Such complemen- tary specialization may be organized by sex or by entire communities, as Chavez documents for highland Peru. Efforts to meet increasing demand also strongly affect variation in manufacturing techniques. Reductions in the labor invested in each vessel are often ‘accomplished by standardizing vessel classes through the routinization of motor habits, by the use of devices like molds and the Integrating Ceramic Production and Distribution 285 potter’s wheel, or by simplifying decorative motifs (Rice 1987b:203, Longacre, Kvamme, and Kobayashi 1987, 1988; cf. chapters in this volume by Arnold and Nieves and by Allen). A surprising result of Zubrow’s formal analysis is the complexity of variation in manufacturing techniques that might result purely from economic considerations of cost and benefit. Despite the evident labor advantages associated with manufacturing techniques that favor standardization, it must be recognized that specializa- tion and standardization are separate processes, and the correlation between them is subject to many confounding influences.’ These include variation in the physical properties of selected raw materials, the technology of production, the number and organization of laborers in a production facility, and the perception of and insistence on standardization among producers and consumers alike (c.g. Arnold and Nieves). Among tradi- tional, non-industrial communities that most archaeologists study, an intricrate web of local customs relating to the manufacture and use of pottery may strongly constrain decisions at each of these levels. The most detailed demonstration of this phenomenon is found in the discussion of the effects of costumbre on traditional pottery production in Highland Guatemala by Reina and Hill (1978:231-251; see also Reina 1963 and Foster 1948:80, 1988:300-301). The obvious general effect of strong sanctions against experimentation with labor-saving techniques will be a tendency for vessels to remain relatively unstandardized even as production volume increases in response to increasing demand or concentration of production in the hands of fewer specialists. The specific effects of the opposed forces of tradition and efficiency on ceramic products, however, are less predictable (e.g. Pool 1990:249-305). Here archaeologists must consider the nexus of perceptions among producers, consumers, and the archaeologists themselves (Arnold and Nieves, this volume). Innovative techniques favoring standardization may find application in vessels intended for particular consumer segments, particular classes of vessels, or particular attributes on individual vessels. Thus, while a correlation between standardization and specialization can be observed cross-culturally and over long time periods, the association is neither simple nor inevitable within individual societies. Levels of demand are determined by several variables, including rates of breakage, the nucleation and overall density of population (cf. Stark, this volume), and ease of transport. In addition, the mechanics of interaction between producer and consumer both reflect and condition levels of demand. Renfrew (1977:9-10) describes five arrangements by which goods are exchanged. These include travel of the consumer to the producer, travel of the producer to the consumer, travel of both producer and consumer to a third location where they exchange goods directly, transport of goods between producer and consumer by a third party, and 286 Christopher A. Pool travel of the producer to a central agency that assigns goods in exchange (redistribution). Generally speaking, these five arrangements correlate with increasing demand and extent of distribution. Therefore, systems in which the producer travels to the consumer or vice-versa appear to be most appropriate to spatially restricted systems of exchange within communities or between adjacent communities. Arrangements by which both producer and consumer travel to a third location tend to be associated most frequently with local markets, as, for example, Reina and Hill (1978: 207-229) describe for highland Guatemala. More complex mechanisms that involve itinerant middlemen or central agencies of redistribution are capable of integrating producers and consumers over a much larger region (Rice 1987b:192-195). Producers, who must have access to their consumers directly or through intermediaries, must balance these arrangements against access to ceramic resources or their suppliers in choosing the location of their operations. The importance of such considerations is particularly evident in the spatial patterns observed by Stark in the Mixtequilla and by Feinman and his colleagues in the Valley of Oaxaca. ‘The expansion of production to accomodate increasing demand for pottery also calls for the reorganization of activities, which creates spatial effects within production localities. These effects have been described in cultural contexts as diverse as prehispanic Mesoamerica and Roman Europe (P. Arnold 1985, 1987; Peacock 1982; Pool 1990; van der Leeuw 1976, 1977). They include the building of permanent constructions to facilitate storage and manufacturing activities; the increasing segregation of activities within the production locality; and the segregation of production activities from houschold activities. P. Arnold (1985, 1987) has therefore suggested that the identification of facilities and activity segregation may provide more definitive evidence for intensified production than assemblage composition or the recovery of manufacturing tools (cf. Stark 1985), ‘To summarize, ceramic production and distribution involve a tangled web of interactions among producers and consumers. Their accomodations to one another's needs affects all stages of production as well as the overall character of the ceramic assemblage and of the exchange system. Unfortunately, though it is a fertile field of study, the identification of the effects of these interactions is fraught with methodological difficulty. The remainder of this chapter elucidates these difficulties by identifying and discussing the major classes of data that comprise the basis for integrated studies of production and distribution. Only after these classes of data are clearly delineated will discussion focus on the interpretation of the interaction between the productive and distributive subsystems. In the'most general terms, these data (like all archaeological data) derive from two principal sources: the content of assemblages and the Integrating Ceramic Production and Distribution 287 spatial relationships within and between assemblages. For the most part, the assemblages of concern are those associated with contexts of produc- tion and contexts of consumption, encompassing both use and disposal. By comparison, the precise loci of economic transactions are much more difficult to identify. Archaeologically visible exchange loci include shops, as have been reported from Roman Europe (Peacock 1982:156), market- places like the Roman forum (Peacock and Williams 1986:60) or the Great Compound of Teotihuacan (Millon 1973, 1981), and transshipment points where breakage was common, as in the ports of the Mediterranean (Will, this volume). All of these cases reflect market exchange or governmental redistribution in urbanized societies. Such loci, however, represent only a portion of the range of economic transactions in urban societes. It must be remembered that even in urban societies, exchange also occurs in multiple-use plazas, on the steps of churches and temples, on street corners, from the backs of wagons, and on the doorsteps of consumers and potters alike. Transactions in these less specific locations would leave few, if any, indications of the exchange act itself. Reconstruction of economic systems therefore relies most heavily on comparisons among assemblages from contexts of production and consumption, and it is to these I will address my discussion. Specifically, I focus on the identification of production loci, the sources and interpreta- tion of variability in ceramic assemblages, and the spatial arrangements within and among production and consumption loci. Identifying Locations of Ceramic Production The archaeological integration of ceramic production and distribution first requires that two questions be answered: "Where was the pottery made?" and "Where was the pottery used?" Though the answer to neither of these questions is entirely straightforward (Schiffer 1976; Binford 1981, 1983), the identification of the locations of pottery production has been a particularly troublesome issue for many regions and times, and deserves special consideration. The precision of the answer to "Where was the pottery made?" may vary depending upon the specificity required by a particular research question. At its least specific, the answer may be “nearby” versus “else- where." That is, a distinction may be made between local pottery and pottery produced outside the area of interest. At more specific levels, regions or sites of production may be assigned to particular pottery types. Such attributions have frequently been made on the strength of stylistic as well as petrographic and chemical grounds. In either case they typically depend on the assumption that a pottery type is most likely to have been 288 Christopher A. Pool made in those areas where it is most abundant (Rands and Bishop 1980:19-20; Bishop Rands and Holley 1982:300-302). This involves the characterization of ceramic assemblages, which is considered in a later section. At the most specific level the question of where pottery was produced involves the identification of individual production loci. In contrast to the less specific assignments of pottery to regions and sites of origin, which link ceramic distributions to assumptions regarding disposal, use, exchange and production, the identification of particular production loci involves the evaluation of evidence specific to the production process itself. This evidence includes raw materials, manufacturing implements and facilities, unintentional by-products, and intentional products (c.f. Rice 1987b:177-180; Santley, Arnold and Pool 1989; Stark 1985). Raw Materials Raw materials such as clay, temper, and pigment are among the more elusive indicators of ceramic production (Pool 1990:93), though they have ‘on occasion been found in association with artifacts or facilities suggestive of production (e.g. Munera Bermtdez 1985; Pool ct al. 1987; Rattray 1990; Sheets 1979). This is hardly surprising, since ceramic manufacture is an additive process, and unused materials may be recycled in the next production effort. The recovery of large amounts of raw materials should only be expected when production localities were abandoned rapidly. Furthermore, small amounts of raw clay, temper and pigments are easily dispersed and incorporated in the soil matrix of a site, and may therefore escape detection. Production Implements Implements employed in the various stages of ceramic manufacture are notoriously difficult to identify archacologically (Stark 1985). Contemporary potters use a variety of perishable implements to thin, shape, impress, incise, and paint vessels. These include wooden sticks, reeds, gourds, basketry, netting, fabric, corn cobs, and brushes of plant fibers or hair (Rice 1987b:136-152). Furthermore, many of the imperishable implements used in preparing clays and pigments, forming vessels, and executing designs are morphologically identical to implements used in other activities, and indeed often began their use life in other activites. Among these ambiguous tools are grinding stones, stone flakes, and bone or stone knives, points, perforators, and polishers. Identification of such items as Integrating Ceramic Production and Distribution 289 ceramic manufacturing implements must rely on use wear analysis, residues of pigments or clay, and their presence in production contexts. The last criterion is, of course, useless as an independent criterion of production, and is particularly problematic when production occurred within a household context. Kiln furniture (vessels and sherds used to protect pots from drafts, smoke, humidity, and contact with one another) is also difficult to identify (Stark 1985:173-174). Nevertheless, distinctive patterns of oxidation and reduction on joined sherds has occasionally suggested their use as kiln furniture (e.g. Bordaz 1964:83-103; Winter and Payne 1976:39). Production Facilit Production facilities are relatively immobile constructions employed in various stages of ceramic production. Such facilities include bins and pits for storing raw materials, tanks for slaking and levigating clays, trampling floors for preparing pastes, potter’s wheels and benches on which vessels are formed, kilns and ovens for firing, and sheds to house the forming operations, drying vessels awaiting firing, and fired vessels awaiting transportation (see D. Arnold 1985; Peacock 1982; Rice 1987b; Stark 1985; and Nicholson and Patterson, this volume, for descriptions). In contrast to most manufacturing implements, permanent and semi-permanent facilities may provide strong evidence of ceramic production, particularly when they are found in association with one another. ‘True kilns with separate chambers for fuel and pots are the most readily identifiable of such facilities, by virtue of the effects of high temperatures on their interior surfaces, as well the details of their construction (see, e.g., Adams 1988; Nicholson and Patterson 1985 and this volume, Peacock 1982; Pool 1990). Kiln-like ovens, which do not separate fuel and pots, may range in form from shallow depressions in the earth to the beehive ovens used by the Maya of Ticul (Thompson 1958). In their simpler forms these are less securely identified as ceramic production facilities, but the presence of other production indicators may suggest their function. Areas of baked earth resulting from open firings are the least readily identified firing facilities. Their preservation depends in part on the hardness and thickness of the baked layer, which in turn depends upon the length of time the surface is exposed to high temperature. Ethnographi- cally recorded open firings, however, do not usually attain high tempera- tures for extended periods (Rye 1981; Shepard [1956] 1980), and are therefore likely to be obliterated. This is particularly true when, as is ethnographically common, they are associated with small-scale production 290 Christopher A. Pool in household contexts where the same space may be used serially for a variety of activities (2, Arnold 1987). Furthermore, it is difficult to say with certainty whether the activity conducted over a patch of burnt ground involved firing pots, cooking food, burning trash, or all three (Allen, this volume; Stark 1985:165). As this description of firing facilities illustrates, the construction of permanent facilities is strongly dependent on the scale and specialization of production. P. Arnold (1985) has cogently observed that such facilities represent not only an investment of labor, but also of space. Thus, in household contexts where production is carried out infrequently, invest- ment in facilities that permanently remove space from domestic use is - unlikely. Under conditions of more intensive production removed from household contexts and employing more laborers, space is likely to be devoted to the simultaneous performance of different stages of production, and permanent facilities may be constructed to facilitate those activities. By-products Ceramic production creates a variety of unintentional residues, principally in the course of firing, Among these are accumulations of ash and carbon, fragments of cracked or otherwise damaged vessels ("wasters"), vitreous pieces of slag and fused sherds, and amorphous lumps of fired clay. In the absence of kilns or other permanent facilities, such residues may provide the best class of evidence for ceramic manufacture, but even they are not completely unambiguous. One ambiguity derives from the definition of wasters. Ethno- archaeologists have the advantage of knowing which sherds discarded within production localities were broken in the course of firing, and typically refer to all such sherds as "wasters," whether or not they exhibit firing defects (Stark 1985:165). Archaeologists, who do not have this luxury, must rely on additional evidence to distinguish wasters from vessels broken in use and discarded in the same contexts as those broken or deformed in firing. When the loci of different activities are near one another, as in household production contexts, mixing of wasters and other ceramics is certain to occur. Even the identification of firing defects may be problematic, For example, some paste compositions tend to exhibit fewer firing defects than others. In particular, vitrification causing vessels to bloat, sag, and warp depends upon paste composition as well as firing temperatures that vary with the fuel and firing technique employed (Rye 1981; Stark 1985). In addition, ‘the accidental refiring of sherds may produce a variety of discolorations and other firing defects. Furthermore, low-temperature firing Integrating Ceramic Production and Distribution 291 mistakes like spalling and surface fire-cracking do not necessarily render vessels unserviceable, and may therefore occur in consumer contexts (Stark 1985:175). Considering the difficulties of identifying waster sherds, it would appear best to limit the term to sherds that exhibit severe firing defects that are likely to have rendered a vessel unusable. These may include bloating, warping, and fusing of overfired sherds, as well as a variety of characteristic cracks that result from overly rapid heating and cooling (see Rye 1981:108-115 for descriptions and illustrations of these defects). It should be noted, however, that the latter will only render the vessel unusable if they extend through the vessel wall, and even then they may be repaired. If established usage must be maintained, however, archaeologists should employ Stark’s (1985; 1989) felicitous distinction between "severely deformed wasters" and "de facto wasters." Products The final products of ceramic manufacture record a great deal of information regarding their production. Individually, each pot encodes in its composition, fabric, form, surface marks, and surface coatings a partial record of the choices made and the labor invested in different stages of its manufacture (Feinman, Upham and Lightfoot 1981). In aggregate, the vessels and sherds comprising a ceramic assemblage record variation in the selection and execution of such choices by the potters of a production locality, a site, or a region. From these data it is possible to infer aspects of the organization and scale of ceramic production. The acquisition of these data constitute the characterization of assemblages, discussed below. The occurrence of unusual concentrations of particular wares or forms has frequently been used to identify the location of specialized ceramic production, however, and therefore deserves separate consideration here (cg. Feinman et al; Santley, Arnold and Pool 1989; Stark 1985 and this volume). It should first be recognized that the identification of ceramics discarded at production localities as the products of manufacture intro- duces a degree of terminological ambiguity. Strictly speaking, outside of unspecialized production ‘for the replacement of a household’s pottery inventory, many or most of the ceramics discarded in production contexts are by-products—de facto wasters accidentally broken during firing or shortly thereafter. Referring to them as products, then, is really a terminological convenience justified by the assumption that the properties and proportions of ceramic categories in production contexts reflect those of the producer’s output. 292 Christopher A. Poo! The degree to which this assumption is justified varies with the intensity and context of production. While the assemblages of large-scale workshops removed from domestic contexts probably do reflect the products of manufacture with a high degree of accuracy, the same may not be true of small-scale domestic production contexts. In cases of complementary specialization like that documented by Chavez (this volume), where a household produces a portion, but not all of its inventory, the household must acquire vessels from another producer. On the other hand, a household might produce all of its own pottery but specialize in the production of certain items for trade. In view of such variation, it becomes difficult to distinguish between de facto wasters and acquired vessels on the basis of frequencies or proportions in the assemblage. This is particularly true when, as is often the case, baseline data on typical domestic inventories is lacking (Stark 1985:167). A further confounding factor is the possibility that because of their composition, thinness, or size, particular forms or wares might be more susceptible than others to breakage in production contexts. Such vessel classes would be over-represented in comparison to their representation among vessels actually exchanged. The most serious problem in using anomalous concentrations of particular vessel classes is the possibility that they may also result from the restricted use of specific wares or forms for certain activities. Salt- making stations (¢.g. Muller 1984; Sanders, Parsons and Santley 1979:174- 175; Santley, Ortiz and Kludt 1988) provide dramatic examples of skewed assemblages associated with the use of specific forms (salt pans and tecomates) in an extractive activity. The hill of amphora sherds at Rome’s Monte Testaccio is an equally dramatic example of a skewed assemblage resulting from the breakage of amphoras in the transshipment of liquids (Will, this volume). Summary For every class of evidence, problems of its initial absence, its lack of preservation, and ambiguity in its interpretation are most severe for small- scale, nonspecialized production. These problems diminish as the scale and specialization of production increase, because the amount of raw material involved, the number and specialized nature of tools, the segregation of simultaneously conducted activities, the investment of labor and space for the construction of permanent facilities, and the frequency of broken and discarded vessels all tend to increase as well. It is nevertheless wise for all kinds of production contexts that multiple criteria of identification be employed (as Feinman [1980], Santley, Arnold and Pool [1989], Stark Integrating Ceramic Production and Distribution 293 [1985], and others have observed). Furthermore, despite problems in interpreting high frequencies and skewed distributions of ceramic categories, distributional and frequency data rather than overt technological data will often constitute the most accessible evidence for production at lower levels of specialization (Stark 1985). Variability in Ceramic Assemblages Identifying the locus of ceramic production from the characteristics of assemblages is but one of many economic inferences that may be derived from the analysis of ceramic products. Indeed, as Rice (1989:111) notes, “rightly or wrongly, the emphasis in the study of production is most frequently and directly placed upon the products of ceramic production and only indirectly on the processes or actual organization of production relations." Ceramic assemblages, usually from consumer contexts, have been analyzed to infer such aspects of the pottery economic system as the organization of production, degrees of specialization and standardization, and the sources of exchanged pottery. Analysis of ceramic assemblages involves three steps: characterization of the composition, morphology, and design of individual sherds and vessels (Rice 1987b:309), assessment of the variability of these properties, and interpretation of that variability. The techniques employed in ceramic characterization have been thoroughly discussed by Rice (1987b) and Shepard (1980), and specific discussions of compositional and technological analysis are provided by Bishop, Rands, and Holling (1982) and Rye (1981). Here I focus on relationships between variability in particular classes of data and particular aspects of production and distribution. Variability in Ceramic Attributes The broadest limits on variability in a ceramic assemblage are set by the physical properties of the resources available to potters. Consumers further restrict variability by imposing demands for particular properties of texture, color, morphology, and decoration that correspond to individual and traditional preferences. Within these externally imposed bounds individual potters are potentially free to experiment with different materials and techniques, though sanctions to conform to traditional methods may be strong (Reina and Hill 1978:231-251). Hence, producers tend to generate variety, while consumers tend to reduce variety through their selections (Rice 1987b:201). Innovation will therefore most likely be 204 Christopher A. Pool attempted in those aspects of manufacture that increase the efficiency of production or the predictability of results, but that do not violate the expectations and values of consumers. Because culturally generated perceptions and concepts of value are involved, the precise attributes so affected may be difficult to predict. The essential point here is to recognize that different sources of limitations on ceramic variability affect different sets of attributes variously and to appreciate the consequences of those influences for interpreting variability. Rice (1989:113) has discussed this point with regard to four broad classes of attributes, or attribute systems, that relate to resources, manufacturing technology, form, and "decoration/style." This schema is intuitively satisfying because each attribute system is most strongly associated with either material properties, production techniques, use, or aesthetic and informa- tional content. Furthermore, each attribute system is characterized by different kinds of data, and by different methods for recovering and interpreting them. The correspondences between attribute systems and properties of materials or classes of behavior are not exclusive, however, for each attribute system is influenced to some degree by physical properties of Tesources, economic advantage to producers, and consumer preference. Hence, each of these sources of variability may have differential effects on individual attributes within each system (e.g. paste texture, paste color, visible manufacturing marks, height, diameter, design elements, etc.). By collapsing the attributes systems of form and decoration/style these effects may be expressed schematically as in Figure 12.1. As in a geological ternary diagram, the displacement of the attribute systems from the pure influence of factors at the triangle’s vertices represents the relative contribution of the other two factors to the overall variability of the attribute system. If each factor exerted the same influence on all attributes in a particular system, then the attribute systems would be represented in the diagram by points. Instead, they are represented by ovals to reflect the possibility that individual attributes within attribute systems may respond variously to different influences, Thus producers’ appreciation of con- sumers’ desires and their own skills affect their selection of ceramic materials and manufacturing techniques from among those available. Similarly, consumers’ preferences for attributes of form or material properties are mediated to varying degrees by the producers’ technology and skills as well as the physical properties of naturally occurring materials. This is not to say that the character of an assemblage simply reflects the sum of accomodations of individual consumers and producers to one anothers’ whims. Consumers and producers alike are members of cultures, classes, ethnic groups, and so on, with canons of style and patterns of behavior that constrain their individual preferences. Integrating Ceramic Production and Distribution 295 Resources Producer Consumer FIGURE 121 Schematic representation of the influence of different factors on attribute systems. Materials. The attributes that most closely reflect variability in raw materials are those concerning the size, shape, mineralogy and chemical composition of the particles comprising ceramic pastes. Though all of these may be altered in the course of manufacture, they are nevertheless useful in distinguishing pottery from different regions and, under favorable circumstances, the geological sources of the raw materials. Mineralogical characterization of temper and natural aplastic inclusions in pottery is most often conducted by examining sectioned sherds with a binocular or petrographic microscope. Applications of these techniques are illustrated in this volume by Blinman and Wilson and by Feinman et al. (see also the ground-breaking studies of Shepard [1936, 1939, 1942a, 1942b, 1946] and Matson [1937, 1939]). Techniques for characterizing the mineralogical composition of ceramic bodies and clays that require more sophisticated instrumentation include X-ray diffraction analysis and a variety of thermal and gravimetric methods (Rice 1987b:375-389; Bishop, Rands, and Holley 1982). ‘Today the archaeologist seeking to identify the provenience of pottery also has available several analytical techniques capable of measuring elements in concentrations on the order of 100 or fewer parts per million (see Rice 1987b:391-392). Such techniques include optical emission spectroscopy, atomic absorption spectroscopy, proton-induced X-ray 296 Christopher A. Pool emission (PIXE), Méssbauer spectroscopy, and the techniques Pool and Santley employ in this volume, X-ray flourescence spectrometry and nuclear activation analysis. These vary greatly in their sensitivity, precision, and accuracy, in the elements they can measure, and in their cost (Rice 1987b:389-404, Table 13.1). The use of spectrographic and mineralogical techniques to discriminate among artifacts from different sources relies on the provenience postulate of Weigand et al. (1977:24) that "identifiable chemical differences exist between sources of a raw material, and the analytical approach can recognize these differences" (Bishop, Rands and Holley 1982:301). The identification of this concept as a postulate is apt, since it is less an assumption about the nature of source variability than it is a requirement for the selected approach to work. An additional assumption of composi- tional analysis is that variation in the composition of artifacts reflects variation in the raw material from which they are made (Rands and Bishop 1980:19). One implication of the provenience postulate and the assumption that artifacts and their raw material exhibit similar ranges of variability is that the nature, geological occurrence, and human manipulation of lithic and ceramic materials strongly condition the ability to distinguish artifacts from geographically distinct sources. In contrast to more easily sourced materials, like obsidian, the naturally occurring clays used by traditional and ancient potters are highly variable mixtures of small, plastic, clay minerals and diverse nonplastic mineral grains of varying size. Different combinations of these constituents create differences in the workability, color, firing behavior, and most importantly for provenience studies, the chemical composition of natural clays. The identification of raw material sources for a particular ware therefore requires a thorough knowledge of the natural distribution and variability of source deposits and of the transformations of natural clay in the preparation of the paste. Even the most precise analytical technique is limited in its application by these factors, and in many cases, fine scale ination among source deposits within a region may be impossible . Bishop and Rands 1982). Under favorable conditions, however, compositional analysis of ceramics, clays and tempers can provide invaluable information on resource procurement and intraregional exchange. Archaeologists attempting such research must be prepared to conduct detailed geological mapping and sampling of ceramic resources in addition to technological analysis of the ceramic pastes. With this information, it is sometimes possible to identify a limited number of clay or temper deposits as the most likely sources of the materials employed in ceramic manufacture. The number of likely exposures may then be Integrating Ceramic Production and Distribution 297 further reduced on the basis of their distance from production localities (eg. Pool and Santley, this volume). Given the problems of identifying natural clay sources, a frequent approach is to assume that chemical variation in locally produced ceramic pastes is dependent on local variation in clay and temper resources and to establish “resource procurement zones” through the analysis of ceramics alone (Bishop, Rands, and Holley 1982:275; Rands and Bishop 1980). In this approach the inference of local production rests on the provenience postulate and the assumption that nearby sources are more likely to be exploited than more distant sources, in accordance with principles of least effort (Rands and Bishop 1980:19-20; Bishop Rands, and Holley 1982:300- 302). Ethnographic support for this assumption is provided by D. Arnold (1985:35-52), who notes that 84% of the societies represented in his worldwide sample obtain their clay from within seven kilometers of the locus of production. Similarly, 97% of the societies in Arnold’s sample obtained their temper from less than nine kilometers away. It is also expected by the "criterion of abundance” and Renfrew’s "law of monotonic decrement” (1977:77-23) that pottery of a particular paste composition should have been manufactured in the locality where it is best represented. In contrast to the problems they introduce in the identification of source deposits, potters’ manipulations of raw materials may actually increase the precision with which regions or sites of manufacture can be identified (Bishop, Rands, and Holley 1982). The addition of a geographi- cally restricted temper, for example, may serve as a "tracer" analagous to the radioactive tracers used in biological research. Petrographic and chemical analyses have exploited this observation to great effect (e.g. Bishop and Rands 1982; Shepard 1936; Blinman and Wilson, this volume). Even when ceramic clays and tempers are both widely distributed, however, differing recipes for paste preparation may produce consistent differences in ceramic paste compositions among sites or regions (¢.g. Feinman et al., this volume), Variability in the paste composition of ceramics and raw materials is usually compared through the application of one of several multivariate numerical methods. Some of the more commonly used methods are multiple discriminant analysis, cluster analysis, principal components analysis and factor analysis. Useful discussions of the relative merits of various techniques and data transformations in the analysis of geological materials are provided by Davis (1985) and LeMaitre (1982). The specific application of multivariate numerical methods to ceramic compositional analysis has been addressed by Bishop and his colleagues at Brookhaven National Laboratory, as well as several other authors (e.g. Bishop, Rands, and Holley 1982:300-313; Bishop 1980; Bishop, Harbottle, and Sayre 1982; Bieber et al. 1976; Wilson 1978:226-233). 298 Christopher A. Pool Manufacturing Technology. The transformation of a plastic lump of raw paste into a rigid, fired vessel is accomplished through a sequence whose constant steps are forming, drying and firing. Depending upon the methods used, vessels may be decorated when they are still moist, after they have dried, after they have been fired, or between bisque and glaze firings. The forming process can also be divided into sequential steps. Rye (1981:66- 95), for example, defines primary forming, secondary forming, and decorative forming stages of vessel manufacture, and identifies characteristic surface marks and grain orientations by which they can be recognized. Potters, however, sometimes employ different sequences for different portions of a pot, and often use the same techniques in primary and secondary stages of forming. These practices make Rye’s scheme difficult to apply to some collections (Rice 1987b:124; Pool 1990:43-57). This difficulty may be overcome by recognizing that different sequences usually apply to different portions of the vessel, instead of being repeated on the same part of the vessel (see, ¢.g., Nicholson and Patterson, this volume). For these reasons Ihave elsewhere suggested a somewhat different classification of techniques by their use in the building of a vessel portion where it did not exist before, the shaping and thinning of that portion into an approximation of the final shape, and the finishing of the vessel portion to refine the vessel shape, obliterate earlier marks, and apply decoration (Pool 1990:43-57). Building techniques include modeling, coiling, molding, and throwing. Shaping techniques include beating and scraping, and finishing techniques include smoothing, burnishing, polishing, texturing, plastic decoration, slipping and painting, These techniques and others are thoroughly described by Rice (1987b:115-152) Rye (1981:66-95) and Shepard (1980:53-69). The ability of potters to replicate the images of finished pots they and their customers hold in their minds depends upon which of these techniques they employ and the skill with which they apply them. Some techniques are inherently capable of producing vessels with a narrower range of variation in particular attributes than are others. Thus greater standardization in forms is possible with molds or with the wheel than through hand modeling, stamps and seals can reproduce designs with greater accuracy than free-hand incision or carving, and the range of paste, slip and paint colors may be narrowed by controlling the firing atmosphere in kilns rather than firing the vessels in the open. The need potters feel to reduce variation in their products, however, depends partly upon their own culturally bounded evaluation of the benefits of greater efficiency or predictability in production, and partly on their appreciation of consumers’ tolerance for variation. Form and Decoration. The choices potters make among the raw materials and manufacturing techniques available to them find their most

You might also like