(Institute of History, Tallinn University / Department of Archaeology and History, University of Tartu, Estland/Finland)

Keywords: town archaeology, retrospective analysis, Eastern Baltic, find analysis

The boom of rescue excavations arrived in Estonia in the mid-1980s. As a result, there are numerous archaeological projects, the documentation of which has remained unfinished. Some of the sites are of key importance for understanding the development of a certain place, such as the medieval town of Viljandi, where the only large-scale excavations next to the town market place (1993–95)are the subject of this presentation.
The main aims were to assess the possibilities to reconstruct stratification, plot structure and the overall importance of the site with changes in time. To achieve this, comparative analysis of the main remaining sources: 1) written notes of the excavation leader; 2) skitches and drawings; 3) photos; and 4) specialist analyses has been conducted. In the presentation I will discuss the problems met during the retrospective analysis, and the main expectations as well as problems that were encountered. As a result, it was established that the reconstruction of stratigraphy was the most difficult task, but absolutely vital for more specific use of find analysis and for the interpretation of other specialist data. An assessment of the level of existing documentation in regard to the methods and resources available will give a more realistic estimation of the feasibility of the planned approach. In our case, the role of the site in the development of the town was given additional details, but for a more detailed development scheme, the excavations with the methods and tools available today would give several additional impulses. To conclude, it was certainly worth going through the existing material, especially as no more excavations of that scale are likely to take place in the forseeable future, yet even small-scale studies at the same property might increase the potential for interpretation of existing data.