For the past couple of weeks, I've been working on preparing the data from the Ohio Boeotia Expedition (1979-1982) for re-analysis. This is all working toward presenting an expanded interpretation of the results from this field work at a panel put together by Kostis Kourelis and Sharon Gerstel at the 2010 Archaeological Institute of American Annual Meeting. The panel is called "First Out: Late Levels at Early Sites". The site of Thisve is primarily known for its relatively well preserved Hellenistic fortification wall and substantial corpus of published Greek inscriptions. My paper will focus on the surface remains from across the broader region with particular attention to Late Roman and later material (although the surface assemblage no matter how you excavate, is the first out.)
As of this weekend, I finished keying all the density data from the survey transects that can be reasonably mapped. There were three or four transects which I've not been able to map in accurately in the GIS. Despite this missing data, I think that the mapped transects reveal something about the distribution of ceramics across the plain south of the ancient city of Thisve (and the modern villages of Thisvi and Doubrovna).
One of the interesting things is that this was not the first time that this data was mapped. Three maps of the archaeological "topography" of Thisve appeared in T. Gregory, "Archaeological Explorations in the Thisbe Basin" in the Boeotia Antiqua II: Papers on Recent Work in Boiotian Archaeology and Epigraphy (Amsterdam 1992), pp. 17-34. The OBE team plotted these maps using Surface II software against a digitized map of the Thisve plain. Unfortunately, from what I can gather, this original map of the basin no longer exists. So I've had to reconstruct the location of the transects from the notebooks in which the data was originally recorded.
The comparison of these two images and the processes that created them is a nice, small case study both for archaeological visualization and for the curation of digital media. I think that my more recent map of artifact densities and transect presents a more accurate picture of the distribution of ceramics across the landscape. That being said, even my plan has generalized. The samples from most of the individual survey units, mapped as squares in my plan, were taken from a 1 m sq area. I've extrapolated them across the entire unit (i.e. the width of the transect x the sampling interval). The lower images, generated by Surface II plotting, have simply extrapolated the density of artifacts across the entire Thisvi plain. I suspect that the linear arrangement of survey units (an early form of the Sydney Cyprus Survey Project's "Souvlaki method" of surveying the landscape) exposed the distribution plots to various kinds of "edge effects" as the software was asked to extrapolate artifact densities farther and farther from known or established data points. Moreover, the jagged abstraction of these figures makes it difficult to assign the surface densities to real space on the map (note the lack of scale or even a north arrow on figure 3.6!).
The disappearance of the Surface II data is another important issue. While it is easy to criticize a project for failing to responsibly curate their data, in fact, the field notes books and survey sheets from the project are well-maintained and organized. The maintenance of data produced over the course of secondary analysis is a challenge for a small project like the OBE which worked in the area for only three years and published their analysis and then dispersed. Survey projects, in particular, suffer from rather ephemeral constitutions (as opposed to the usually more permanent relationship between excavators and a particular site). If the relatively low impact of survey archaeology on the landscape tends to attenuate the link between the archaeologist and a particular place, then the combination of paper and high tech applications ranging from relational databases to GIS mapping applications adds a layer of complexity to curating the digital data that these projects produced. In most cases, data was (and I'd argue still is) collected from the field in paper form and then keyed and plotted into digital databases of various descriptions. So the digital data represents the first phase of analysis rather than a primary data collection. Perhaps this is part of the reason for failing to maintain the digital data as carefully as the paper forms and notebooks. In recent years, a more serious approach to the practices involved in curating digital data (and survey data more generally) will undoubtedly change future practices. Hopefully our work with the data from Thisvi will represent an important case study for the curation of digital data in the context of re-analysis.