The applications of compositional analysis to provenance studies of archaeological pottery in Southern Africa: a geochemical perspective using XRF spectroscopy
Pottery arrived in the South African archaeological record some 2000 years ago. There are two major traditions of pottery making although they may have had the same origins. The first is that of the Khoisan peoples who inhabited the drier coastal and inland areas. They were mobile pastoralists who spent their year in independent groups in the search for water and grazing for their animals. Their pottery, although distinct, is not easy to classify into discrete spatial and temporal phases which is probably a reflection on their migratory lifestyle and lack of any centralised political authority. The second group were Bantu-speaking farmers, also referred to in archaeological terms as the Iron Age. They spread through the wetter areas suitable for farming and lived in more settled communities many of which reached urban status with thousands of inhabitants. Their pottery, in terms of decoration, is highly formalised and, in contrast to the Khoisan, is organised into well defined discrete phases with temporal and geographical coherence. Both groups were involved in trade and exchange networks which, for certain Iron Age communities, became continental in scale and resulted in major social transformations taking place in their society. This thesis looks at the problem of sourcing pottery in order to measure the direction and intensity of these networks. The main objective is to assess the viability of sourcing studies in different parts of the country. The first problem deals with the role of additives such as temper and whether this can mask the original clay signature. Experiments were conducted which show that temper can alter the original clay chemical profile but the scale will depend upon the difference in composition and the amount of temper used. Other problems deal with diagenesis: burial can alter the composition under certain circumstances but at present only phosphorus can be definitely seen to be a contaminant. A number of provenance case studies centred on various regions in southern Africa are presented. Each one presents its own specific methodological problems. The most successful results came from two localities. The first, Mutokolwe, in the Soutpansberg, and the second, a succession of sites including Schroda, K2 and Mapungubwe in the Limpopo Valley provided definite evidence to confirm on geochemical grounds that a number of pots were non-local. The other case studies were geochemically successful but indicated that certain regions, for geological reasons, might not be suitable for undertaking provenance studies. Recommendations are made for future research in this promising field.