Over the last week or so I’ve been working on writing a draft of the preliminary report from the Western Argolid Regional Project. Yesterday, I returned to one of the more interesting issues facing intensive pedestrian survey in the Mediterranean: the matter of intensity.
WARP was a siteless survey and we took this quite literally. We did not return to sites after our initial survey to conduct more intensive investigations or gridded collections. In fact, we tended not to talk about sites at all (outside of the commonplace naming convention associated with particular known ancient places in our landscape, like the walls of the polis of Orneai) and assumed that high density scatters in the landscape could as easily represent a single period as the overlap of a number of periods through time. (Our experience in the Eastern Korinthia had suggested that many areas of high artifact density in the landscape reflected the overlap of single period scatters that may or may not be contiguous.)
This being said, we did recognize that more intensive practices – such as total collection or intensive sampling across small grid squares or other forms of “hoovering” – produced more robust assemblages of material. We experimented a bit with this on the Pyla-Koutsopetria Archaeologcial Project and while our results are very, very boring to read, they demonstrate that careful collection with our noses to the ground produced artifact densities that we much higher than traditional field walking. At the same time, the uniformity of the assemblage at Pyla-Koutsopetria ensured that more intensive collection strategies did not produce a more diverse assemblage. In other words, on PKAP, where our assemblage was pretty uniform, doing more intensive artifact collection did not yield more nuanced results.
In the Western Argolid, the situation is a bit different. The survey area is larger and encompasses a generally wider range of environmental conditions. Moreover, our artifact scatters across the survey area tend to reflect more complex and varied historical processes than we found at our “large site survey” on Cyprus. As a result, it seemed like a good place to run another series of experiments to see whether more intensive collection strategies produced different kinds of assemblages across the survey area. In particular, we were interested in seeing whether more intensive collection strategies revealed the presence of “hidden landscapes” consisting of pottery that tends to be overlooked during traditional survey field walking.
To do this, we set down slightly over 100 2m radius total collection circles across our survey area. These resurvey units fell within our existing units and we intended to use them to compare the resurvey assemblages to the assemblages produced by traditional intensive survey.
This proved pretty challenging for a number of reasons. Some involved research design. Our resurvey units were total collection circles with a 2 m radius (i.e. 12.57 sq m) and we typically did two per unit for about a 1% sample of the survey units which averaged about 2500 sq m. This sample could then be compared to the must larger, but less intensive ~10%-20% sample collected using typical survey procedures of field walkers spaced at 10 m intervals looking 1 m to each side. This comparison, of course, isn’t all that great. First, we recognize variation in the surface assemblage across the unit so it makes sense that our resurvey circle may well capture an assemblage with a different character than the assemblage produced by field walking. Ideally, they two assemblages are similar in some way, but some variation is likely to reflect that, short of total collection of the entire survey area, surface sampling is not designed to discover an example of every kind of sherd present in every survey unit, but a representative sample of the area generally. More than that, there was a general tendency to locate the resurvey circles in areas with higher visibility than the average visibility across the survey unit in general shifting slightly the recovery conditions for material.
That being said, our preliminary analysis of the material has produced some other challenges as well. First, it’s very difficult compare complex assemblages. We have remarkable ceramicists who are capable of defining pottery in very granular ways both in terms of typology and chronology. This granularity makes it difficult to compare two assemblages because the variation inherent in how we analyze ceramic artifacts. For example when I compared the resurvey and the survey assemblages this past summer, the vast majority of assemblages showed little chronological overlap based on the specific periods assigned to each chronotype. On the one hand, this meant that our resurvey circles were producing different kinds of chronological information from the larger survey units in which they were situated, but, on the other hand, this chronological need not result in substantively more knowledge about the artifact scatter. An artifact datable to the “Late Hellenistic to Early Roman” period is different from a “Hellenistic to Early Roman” artifact, to be sure, but this kind of granularity is as likely to represent the irregular distribution of chronological knowledge across a typology than it is to represent a general pattern of artifact dates that would influence the identification of the function of the site, for example. (I.e. some kinds of pottery, say fine ware, can be dated more specifically than other kinds of pottery.)
It is worth noting, however, that in general, standard survey units produced more artifacts with narrower chronologies than the resurvey units did. While this, in and of itself, is not meaningful (as, for example, prehistoric pottery dated quite specifically still tends to have wide chronological ranges than historical period pottery), it suggests that our standard survey practices produced assemblages that were susceptible to chronological analysis that are at least comparable and perhaps more fine grained than our total collection circles.
To mitigate this, I started to generalize the chronological categories a bit, grouping the finely honed chronology offered by our ceramicists into “Prehistoric,” “Greek,” “Roman,” “Greek/Roman,” “Medieval,” “Modern,” and “General” for artifacts only dated to broad spans of time. This aggregation, predictably, made the comparison between our two assemblages easier. For example, it showed that over half of the assemblages produced pottery of broadly the same date. Moreover, it allowed us to observe that the smaller the assemblage the less likely overlap occurs. In other words, our smallest assemblages from either standard survey or resurvey were not just producing worn, undiagnostic pottery that we tend to aggregate into more general chronological categories, but that they also produced variation.
It also gave us a way to see if the resurvey units had any particularly telling trends. It appears, however, that the resurvey units did not yield, as a general pattern, assemblages that could be dated to narrower periods than their standard survey collections. This tells us that our more intensive collection circles are not producing more narrowly datable pottery in general, but not necessarily that they don’t produce pottery datable to particular narrow date ranges. As an example, the only the resurvey assemblages produced any examples of Final Neolithic and FN-Early Helladic I in those units that we resurveyed. Other units in the survey area, of course, produced material from those periods.