The Eastern Korinthia Archaeological Survey: integrated methods for a dynamic landscape: experiments to evaluate and calibrate results.
All regional-scale surface surveys would benefit by having an experimental component operating in tandem with the main archaeological reconnaissance. EKAS conducted a series of experiments in 1999 and 2001 with two specific aims: (1) to measure the effects of field conditions upon archaeological procedures for purposes of calibration, standardization of practice, and development; and (2) to reflexively test the efficacy of our methods. With regard to field performance, one might ask: How well do we recognize cultural material when half the surface is covered with vegetation, or when the color of red-fabric potsherds and terra rossa-stained limestone gravel is essentially the same? How does the pace of fieldwalking affect identification? Will a pace twice as fast decrease recognition by half? These questions are important ones as they influence the archaeological document that a survey produces.
EKAS assembled a team to address these and other questions. The most extensive experiments (Table 8) investigated the effects of field conditions upon artifact recovery through a series of seeding experiments in which a team of researchers planted potsherds, which had been previously photographed and described, along a 50-m tract and plotted their positions. (91) Two kinds of potential survey conditions were tested. In the first experiment, equivalent sets of artifacts were placed in a tract with optimal visibility (100%, i.e., with the ground surface not obscured), in another with poor visibility (20%), and in two others with average visibility (50%-70%). This experiment focused on the extent to which ground cover such as weeds and grain stubble hinders artifact recovery. In a second experiment, a tract with high background disturbance (92) was tested against a tract in which there were few visual distractions, in order to measure differences in artifact recognition under these two conditions. The analysis took into account vegetation cover, artifact type and appearance, and background confusion as they affected the rate of artifact recovery. The determination of recovery rates over a broad range of conditions was expected to provide a better sense of the total range and quantity of material, not only in the inspected portions of survey units, but also in the uninspected portions.
By feeding the results of the seeding experiments back into everyday practice, we contributed to the acquisition of more robust data for analysis and to a more refined interpretation of artifact patterning in our standard survey units. In 1999 and 2001, the experimental team used the seeding results to create a standard for surface visibility rankings with the aid of detailed written descriptions, photographic images, and in-field collaboration with team leaders and fieldwalkers. Team leaders were charged with rating effective visibility as a percentage of potential visibility (in increments of 10 from 0% to 100%, where 0% represents total impediment to surface visibility, and 100% represents no surface or ambient impediments to visibility). The continuity of key personnel in the team leader positions ensured consistency in these observations. To serve as a further, independent check, two "visibility" photographs were taken for each DU: one overview of the unit, and one detail of a representative surface. These photographs may be used retroactively to assess the visibility rankings, which are crucial since the relationship between surface visibility and artifact recovery forms one basis for GIS analyses (described above) and ultimately for our interpretations of the survey data.
The seeding experiments generated a large dataset of records of artifact recovery by individual walkers in a variety of field conditions (Table 9). The full dataset shows that artifact recovery and relative visibility have a linear relationship, although not an equivalent one (Fig. 15). In other words, while recovery increases consistently as visibility improves from 40% to 80%, that increase is not twofold. (93) A series of recovery curves, correlated with specific discovery conditions, might be used to apply quantitative corrections to normalize visibility for all units. When visibility and artifact collection data are combined, formulas to extrapolate to adjusted artifact counts, and derived statistics such as total artifact densities, are within reach. (94)
[FIGURE 15 OMITTED]
Without such experimental data, there is no basis for explaining or attempting to control for the sources of variability in artifact recovery, although we acknowledge that myriad other factors must be taken into consideration as well, (95) some of which were also addressed by the experimental teams and are currently undergoing analysis (Table 8). Although these experimental data may be informative principally for the eastern Corinthia, they offer an excellent comparative dataset for surveys in the Mediterranean and beyond. (96) Similar experiments, adapted to local field conditions and methodological interests, would be relatively simple to devise and implement.