One of the benefits of preparing the data from the Pyla-Koutsopetria Archaeological Project for publication in Open Context is that Eric Kansa has forced the us to remember all these little irregularities in our dataset. One that Eric noted this past week was that our crack survey team re-surveyed a group of units in 2005 that we had originally surveyed in 2004.
The reason for this had to do with plotting units in an open plowed field about a half a kilometer from our benchmark. In 2005, we accidentally plotted in units that overlapped entirely with a set of units surveyed in 2004 and caught the mistake when we produced our grid in our GIS. The result of our mapping glitch is that we resurveyed 6 units from 2004 in 2005.
This mistake produced some interesting results because it allowed us to compare the same general area under two very different field conditions. In 2004 the units were covered in fresh grain stubble and surveyed very late in the afternoon under unfavorable raking light conditions. The survey sheets noted that the glare from the grain stubble and that one unit (maybe 180) was the LAST UNIT of the season. The surface visibility of the units was around 40 (43.3 on average).
A year later, the units were almost completely clear of stubble with just a slight scatter of weeds. The visibility was 100% and we walked the units about an hour (and a month) earlier. The walkers did not mention glare or any other difficulties identifying the ceramics on the surface.
The results were almost too good to be predictable. The 2004 units produced 1292 artifacts per hectare and the 2005 units produced 2484 artifacts per hectare. Remarkably these units produced slightly over 50% more material in 2005 reflecting the improvement in visibility in something close to a linear fashion.
The units produced assemblages that are reassuringly similar. The 2004 survey produced 8 chronological periods and the 2005 units produced 14. Some key periods were represented in 2005 and absent in 2004: Archaic (1), Hellenistic-Early Roman (2), Early Roman (1), Late Medieval (1), and Ottoman sherds (1), but in all these cases it was no more than two sherds. The other periods that appeared in 2005 were all broad units (with time spans over 1000 years) like post-prehistoric and Medieval-Modern. From a chronological perspective, all of the narrow periods that appeared only in 2005 were accounted for in broader chronological periods in 2004 (e.g. Medieval and Roman) except the single Archaic sherd.
Despite these new periods, the overall chronological distribution of material from both sets of units was similar. The 2004 units produced 35% Late Roman pottery and the 2005 units produced 39%. The Roman material was about 5% for both surveys.
Interestingly, the survey in 2004 produced twice as much diagnostic fine ware than the 2005 survey, but otherwise the proportions were almost identical with utility wares (coarse, medium coarse, and amphora sherds) accounting for 83% of the assemblage in 2004 and 86% in 2005. Kitchen/Cooking wares accounted for 4% and 5% respectively.
The take away from this little analysis is that despite the significant difference in the number of sherds counted and collected, the two assemblages were nearly identical with only a handful of sherds appearing in 2005 that made any chronological or functional difference in the area.
Pretty cool, huh?