One of my first opportunities to shape the research directions of a project came when David Pettegrew and I were allowed to help design the survey methods and goals of Australian Paleochora Kythera Archaeological Survey. I am pretty sure we didn’t do anything marvelous there, but I did meet my wife on that project. So, any work done on Kythera or in its general vicinity has continued to pique my interest over the years. The survey conducted by Andrew Bevan and his team on the island of Antikythera – a mere speck in the Mediterranean on the main sailing route between Kythera and Crete – has attracted my attention of late not only because he conducted it on a Mediterranean island, but also because Bevan (and co.) are among the most sophisticated survey archaeologist in the business right now.
In an article slated to appear in Archaeometry, Bevan and a group of collaborators proposed some new ways of measuring chronological uncertainty in intensive survey (here’s a preprint (pdf)). This is a long standing and vexing issue for survey archaeologists where artifacts datable only to broad or multiple periods are common. The absence of stratigraphy makes it impossible to propose narrower dates for these objects, so a number of strategies have developed to document the uncertainly associated with these objects. The Chronotype system, that we have employed at both the Eastern Korinthia Archaeological Survey (EKAS) and, in Cyprus at the Pyla-Koutsopetria Archaeological Project (PKAP) has provided one method for documenting artifacts of uncertain date. The Chronotype systems has a wide range of roughly hierarchical chronological categories into which we can group objects. For example, an artifact could be “Late Roman” or “Roman” or “Roman-Medieval” or even “Post-Prehistoric”. When we assign an artifact to one of these categories we are assuming that the artifact could appear with equal probability in any year (decade or century) in that chronological span. Of course, in practice, we know that artifact are unlikely to appear with equal probability in each decade or century, but this does provide a way to smooth chronological data and, when mapped across the landscape, it can help identify areas where handfuls of specifically dated artifacts appear alongside larger quantities of artifacts only datable to broad periods. Additional problems with this system, however, arise when artifacts can appear, for example, in one of two non-continuous periods (Hellenistic OR Late Roman).
Bevan and his colleagues have suggested a system where the ceramicist assigns a probability to each period in which an artifact might appear. An artifact datable in the Chronotype system to a broad period like Roman might appear in Bevan’s system as: 10% Early Roman, 30% Middle Roman, 60% Late Roman. This more subtle way of documenting the probability of an artifact appearing in any given period not only more accurately represents the way ceramicists analyze pottery, but also allows for artifacts appearing in nonconsecutive periods. For example, they noted that certain kinds of chunky prehistoric pottery could date with a fairly good probability to the Bronze Age, but might also date to the Medieval period. Moreover, this method allows for particular classes of artifacts to be identified by their distinct statistical relationships between periods and artifacts identified through these statistical measures could then by plotted spatially. The resulting maps would indicate where similar kinds of localized (un)certainties would appear.
Bevan notes that this system also allows for multiple readings of the same group of artifacts by different ceramicists who could assign different levels of chronological uncertainty to each batch of artifacts. This is particularly useful for types of artifacts that could date to discontinuous periods like our prehistoric or Medieval coarse wares.
(As an aside, its funny to note that Tim Gregory long had a category of “certainty” on his recording sheets. I think I made fun of it and claimed that the category was redundant within the “rules” of the Chronotype system. Now I wonder whether Prof. Gregory continued to keep that category … )
This past year the Antikythera team published the Roman period material from their survey in the Annual of the British School at Athens (106 (2011), 47-98): here’s a preprint (pdf) and here it is published form (pdf). While there is little evidence for the sophisticated system of probabilistic dating the assemblage of Late Roman material on the island is interesting to compare to our Late Roman material from Cyprus. The significant quantities of Phocaean fine ware from Antikythera find clear parallels with our assemblage at Pyla-Koutsopetria. It may reflect, as they Antikythera team has noted, the relatively late date for our Late Roman assemblage which was formed after the supply of the ubiquitous African Red Slip became attenuated.
It is also interesting to note the ratios of Late Roman 2 to Late Roman 1 amphora on Antikythera are almost reversed from ours on Cyprus. This is unsurprising, of course, since LR2 production sites are most likely in Greece or the Aegean and LR1 sites are in southern Asia Minor or Cyprus. The folks at Antikythera noted that LR1 amphora are commonly thought to transport wine, but they – like the LR2 amphoras – might have also served alternate household purposes like storing water or grain that led to their wide distribution across the island.
Finally, they suggest that the absence of material from the post-Late Antique period could indicate that the island was abandoned for a time prior to a Byzantine re-occupation. This fits well within the prevailing ancient (and modern narratives) for the chronology of settlement in the Aegean.