I am continuing to work on analyzing the survey data collected from 4 seasons of intensive survey at Pyla-Koutsopetria in Cyprus. On Monday, I did focused on the relationship between artifact density data and surface visibility. The survey area for PKAP was particularly uniform in ground cover and field conditions. Most of the survey units had been under cultivation with cereals and were covered with grain stubble or were recently fallow.
Average visibility across the entire survey area was around 60%.
Average density of ceramic objects per area walked across the entire site is 2183 ceramic artifacts per hectare.
One of the key discussions in the analysis of surface density data is how to compensate for variable visibility. Our survey area consisted of only 471 units -- hardly a robust sample. The highest density areas of the site coincided with units of middling visibility and (to make matters worse) there were a significant number of very high visibility units that produced almost no artifacts (predominantly from an area that was an infilled ancient embayment and a coastal marsh drained in relatively recent times). So, the small sample size and these topographical and archaeological patterns make doing any analysis of artifact density and surface visibility across the entire site difficult.
I did, however, run a small analysis on the artifact densities and visibilities from the area of the Koutsopetria plain. We have reason to think that the distribution of ceramics across this area could be relatively uniform as most of these fields overlay buildings from the Late Roman period (if not other periods as well). At the same time, the units had some variation in visibility with the lowest visibility units being 20% and highest visibility units being around 60%. Most units had 30%-50% visibility. While the sample size is disturbingly small, it is interesting that there is a nice linear relationship between artifact densities and the surface visibility in Koutsopetria with an R-squared of 0.9229.
This does not hold up across the entire survey area where for the same range of surface visibility there is no evidence of a linear relationship at all and the R-squared is a laughable 0.5475. (Note that since the following graph compares units of different sizes, I compared them using density per area walked (per ha). The graph above compares units of identical size, so I can compare raw ceramic counts.)
For more installments of Preliminary Analysis of Pyla-Koustopetria Archaeological Data or Thinking Out Loud:
More as I continue to work through the data.