Specialization: stoneware pottery production in northcentral Texas, 1850-1910

Abstract

Ceramic specialization, a key concept in both scientific and cultural evolutionary archaeology, continues to be poorly defined more than ten years after Rice published her model of ceramic specialization (Rice 1981). Rice and others continue to struggle toward an accepted definition of what specialization is and how it can be identified in the archaeological record. Two critical models underpin Rice's model of specialization: an attribute model specifying a correlation between specialization and kinds of artifact attributes, and a distribution model specifying a correlation among specialization, the distribution of production loci, and the finished products.I evaluate Rice's model of specialization using historical and archaeological data available for nineteenth- and twentieth-century stoneware pottery-kiln sites and domestic farmstead sites in Northcentral Texas. Deed/title records, population census schedules, tax rolls, and business advertisements provide historical information on the temporal and spatial distributions of production methods, and resource, production, and domestic loci. I conduct macro- and micro-analysis of archaeological stoneware ceramics from seven pottery kilns and five farmsteads and of clay specimens from seven clay outcrops to verify the historical information and to evaluate the ceramic attributes that Rice identifies as appropriate for examining ceramic specialization.Although the historical and archaeological data indicate that stoneware production in Northcentral Texas was specialized, no correlation was found between Rice's attribute model and specialization, e.g., an association between paste composition and specialization. Rice's attribute model fails because specialization is not how pottery is made but rather who makes the pottery, i.e., a small number of individuals. Her attribute model specifies correlations between production types, e.g., specialized or nonspecialized, and specific attribute expressions rather than specific attribute relations.Contrastively, Rice's distribution model is not rejected and can be strengthened by developing appropriate empirical criteria for identifying low-volume specialized production sites. Historical and archaeological data reveal that contemporaneous resource, production, and household loci are not coterminous as assumed by Rice's distribution model. Stoneware pottery production was carried out by a small number of potters, with each nonceramic-producing household acquiring stonewares from several stoneware pottery-kiln sites.