Diagnosing the undiagnostic Using sherd databases as a source of interpretation

In this paper, I offer a plea for the use of full sherd databases, rather than a small selection of catalogued items, as a starting point for interpretation of survey finds. While most sherds are not as precisely dateable or otherwise diagnostic as most items in traditional catalogues, in my experience a very large majority of survey finds is usually dateable in period categories, and most can be assigned a shape or function. By looking at examples from projects in Thessaly and Boeotia I am involved with, I hope to show that these ‘medium diagnostic’ sherds provide a solid starting point for interpretations of assemblages, even if conditions in the field or selective processing have stuck us with less than ideal collections. These examples also showcase possible approaches to find collection and processing.
Of course, the fact that much is possible does not mean that results are always straightforward or that there are no pitfalls. In the second part of the paper I illustrate some problems and possible ways to tackle them, and also reflect on some methodological issues, particularly implications of the limitations of our classification systems – which of course simplify results and produce biases. Finally, I will present some unsolved riddles. None of these difficulties, however, are problematic enough to invalidate or even seriously hamper the use of all kept material as the main source of our interpretations of survey finds. Most of the apparently