Find below my first effort at an AIA paper that I will be co-writing with Tim Gregory. It's rough around the edges, but I think on the right track.
For more on this research:
Reclaiming Thisve Data
Thisve Basin, Archaeological Visualization, and Curating Digital Data
First Out: A First Draft of An Intro for New Views on Old Data
Survey Archaeology Finds as Data
More on Thisvi in Boeotia
Fine ware and Function at Boeotia Thisvi
New Views on Old Data: Reinterpreting Intensive Survey Results After 30 Years
William R. Caraher, University of North Dakota
It seems natural to include a paper on survey archaeology on a panel entitled “First Out”. After all, the surface assemblage is, by necessity, the first out for any excavation. At the same time, the study of surface assemblages has fit into the definition of “First out” intended by the organizers of this panel by contributing significantly to our understanding of post-Classical periods in Greece over the past four decades. In fact, the ground breaking work of many of the participants on this panel has made clear that the rigorous documentation and analysis of surface finds has expanded our notion of what constitutes an archaeological site to well beyond the built up centers of ancient polis and across every century from the end of antiquity to the modern era. Intensive surveys in Boeotia, Laconia, Messenia, and the Corinthia are rewriting both the ancient and post-Classical landscapes of these well-studied regions.
If I can continue to play with the idea of “first out”, it is also clear that this phrase could apply to the first generation of intensive, pedestrian “siteless” surveys in another way. Like the first phase of excavation at major sites across the Mediterranean, the first efforts at intensive survey often relied upon assumptions and methods that were unrefined or unsophisticated in comparison with more recent work. In fact, the constant refinement of survey techniques and the ever more robust datasets that they produce often include explicit and implicit critiques of earlier survey methods. This continuous critique has not only weakened the status of survey among a sometimes skeptical archaeological establishment, but also served as a tacit justification for neglecting the results of earlier surveys. Technological barriers, irregular recording practices, and the incomplete publication of data sets have further impaired archaeologists’ ability to redeploy data collected from the first wave of surveys for newly formed hypothesis.
While the methodological concerns associated with revisiting early “second wave” survey data prose problems, this data nevertheless preserves evidence for the ephemeral surface record in Greece. Both ever-expanding development of the Greek countryside and the irregular patterns of surface visibility, agricultural practices, and erosion patterns obscure and threaten the surface record. As Albert Ammerman famously observed based on the results of several seasons of the systematic resurvey in Italy, sites tend to blink on and off in the landscape like traffic lights. What a project documents one season may not be there the next. Consequently, intensive survey data often captures a single distinct and unique view of the landscape which is not susceptible to reproduction even using similar methods.
This paper will use the data collected from the Ohio Boeotia Expedition between 1979 and 1982 from the (modern village and) Boeotian polis of Thisvi. The results of this survey were published in a series of short articles between 1980 and 1992. While these articles provided for a broad discussion of method and a basic report on the project’s finding, they did not publish finds or quantitative data extensively. Our goal with this paper is to take the first step in re-introducing data from the OBE into the broader conversation about settlement and survey data in both in Boeotia and across Greece more broadly. To do this, we would like to first discussion briefly the process of curating the survey data produced by the OBE and then go on to analyze this data in the context of some recently published survey work from Greece.
The first step in preparing the OBE data for analysis was the keying of records preserved in a series of notebooks and binder pages. These records included counts of artifacts from survey units, which were generally 1 meter square total collection circles as well as from more robust collection procedures conducted at a number of sites across the survey area. We also keyed the finds data from both survey units and the sites into an access database. The finds tables were in turn normalized. Here it is interesting and perhaps valuable to recognize that the quality of data recorded over the course of original fieldwork was quite high, but it was hardly normalized and consequently unsuitable for systematic, quantitative analysis. The lack of normalization was perhaps, in part, the result of the novel character of desktop-level tools for quantitative analysis (e.g. SPSS-X and IBM’s iconic SQL-powered DB2 debuted the year after OBE competed its fieldwork – 1983; the Macintosh personal computer was introduced in 1984.). This is not to suggest that quantitative analysis of archaeological data did not occur prior to the early 1980s, but rather to point out that the creation of normalized practices of data-recording and well-defined hierarchies of object identification became a higher priority after desktop database and statistical tools became more common. By normalizing the robust data sets produced by intensive survey, the database became as important as the traditional artifact catalog for analyzing the chronology and function of sites across the landscape.
At the same time as we keyed data from notebooks and binder pages, we also sought to remap the location of the transects using GIS software. At some point in the 1980s [some additional historical clarity here would be helpful], the artifact counts and location of transects was entered into the Surface II software program which produced a contour map of the artifact densities across the Thivi basin. While versions of these maps were published, the data behind these maps appears to be lost. Unfortunately, at present the disappearance of this spatial data has made it difficult at this point to place the western-most transects on the ground. The written description of the locations of the western transects relies upon points of reference that are not visible on the Greek Army Mapping Service 1:5000 maps and have been destroyed on the ground as a result of the construction of a massive pipemaking factory. There is hope that we can find the location of these transects from older aerial photographs of the area.
The final step in the production of this data is recording comprehensive metadata for the all of the data that we entered. Once the keying of the data and metadata is complete we plan to make this data available to the public via the internet. This step is especially important for small projects because it distributes of digital data expands the curation process from the purview of the creator of the data to the community of users who want to make use of the data. Disseminating the data to end users, with the proper metadata, we make it possible for others to use our material and make it far more likely to be kept compatible with changes in technology.
There have been significant changes in our understanding of post-Classical countryside since the Ohio Boeotia Expedition published their results in the 1980s. The work of both excavations and survey in Boeotia and elsewhere in Greece alone has produced a foundation for the reinterpretation of our survey data. Recent work by Archie Dunn and a team from the University of Birmingham has begun to document the post-Classical finds at Thisvi itself and Jonita Vroom’s study of the post-Classical ceramics from the Cambridge-Bradford Boeotia Project has shed important light on the relationship between post-ancient ceramics and settlement patterns across Boeotia.
The OBE team produced the current dataset through a number of different methods. The diversity of methods reflected the early stage in the development of field procedures and an avowedly experimental approach to documenting the landscape. The area closest to the city walls, Area A, were surveyed using a series of 11, randomly placed, 30 m radius circular survey areas from which samples were taken. The team surveyed the plain itself using a series of long transects (Areas, C, D, and E) from which they typically took 1 sq meter samples for density and diagnostic artifacts. Finally, the teams also collected samples for area of particularly high density which they designated sites. They surveyed these areas using a flexible methods best suited for documenting the extent, chronology, and function of the material on the ground. In addition to these survey areas, the OBE team also conducted intensive survey on two nearby islands in the Gulf of Corinth, Kouveli and Macronisos, which I have not included in the aggregated totals produced in the analysis below.
The survey of the mainland counted over 8700 artifacts and documented over 1700 batches of unique artifacts from the four areas of the survey.
The artifact density data from the OBE shows a decline in the number of artifacts from the units closest to the city across the central part of the Thisvi basin. This pattern, noted in the original publication of the survey, may be at least in part a product of the erosion patterns. In antiquity, an ancient barrage, described by Pausanias, controlled the flow of water and sediment into the basin. In more recent times, the lack of ability to control the flow of water may have either covered some of the sites or, at very least, discouraged habitation there. The density of artifacts, however, clearly increases on the gently sloping, stony ground the along the south side of the basin.
Against the backdrop of overall artifact density we can show the distribution of post-Classical material across the survey area. In general, the survey area is dominated by Classical to Hellenistic and Roman periods which accounted for 2/3 of the datable ceramics. There were, however, several concentrations of both Late Roman and Byzantine to Medieval pottery which represented about 10% of the overall assemblage of datable material collected from survey. Modern material and a thin scatter of pre-Classical material accounted for the other 20% or so of material from the survey.
Area A encompassed the highest density areas immediately south of the plateau upon which the ancient city and the modern village stand. The post-ancient material from this area were very focused with most of the material deriving from three units. Unit A2 contained an abundance of post-Classical material including Middle Byzantine material. It is situated immediately to the west of one of the Hellenistic fortification’s towers which appears to have undergone some modification in the post-Classical period. Units A5 and A8 produced significant quantities of Late Roman – Early Byzantine coarse wares including the ubiquitous combed ware.
The transects immediately to the south of the urban center of Thisvi, Area D, show diminishing quantities of post-Classical material with distance from the Hellenistic walls and the presumed center of post-Classical habitation. Overall only 12% of the material there was post-Classical as compared 77% of the datable material dating to the Classical-Hellenistic period. Area C, which extends south of the city walls to the west of area D, showed a similar distribution of Late Roman and post-Classical material. In fact, the only variation between Area D and C was the rich assemblage of Late Roman material collected from the habor at Vathy which fell within Area C. This collection of pottery pushed the total quantity of post-Classical material from Area C to close to 18%; without this material, the total percentage of Late Roman material was 13% or only slightly higher than found in the neighboring Area D. Unfortunately, we have not yet been able to place the some of the transects from area C. It is clear, however, that significant quantities of material came from the southern edge of the Thisvi plain where a large pipe factor stands today. The harbor area at Vathy has been completely destroyed by an industrial harbor serving that factory.
Area E to the east of the ancient city tells a similar story to areas C and D. It is notable that the overall assemblage produced by these units was smaller than either D or C (as was the overall area surveyed), and that Late Roman material accounted for close to 25% of all the material collected from this area and post-Classical material represented 27%. Much of this material, however, derived from substantial site situated along the southern edge of the basin and designated E1. Like Vathy, this single concentration of material exerted a substantial influence on the overall character of the assemblage from Area E. Without the material from this site, the overall percentage of post-Classical pottery declines to under 10%.
Since the most significant quantity of post-ancient pottery from the Thisvi basin can be dated to the Late Antiquity, it is perhaps most useful in this short paper to explore how we can reinterpret this distribution of Late Antique material in the countryside in light of the significant new analyses of material from this period in Boeotia and across Greece and with the help of more pliant dataset. It is significant, on first glance, that the distribution of material around Thisvi is similar to that recently published around the city of Thespiai to the east. The team from the Cambridge-Bradford Boeotia Project argued that the overall population of the city of Thespiai declined during Late Antiquity and, as a result, the residents of the city progressively abandoned the immediate hinterland of the city to intensive cultivation. In particular, this meant that the residents of Thespiai stopped the practice of regular manuring the fields near the city which, Bintliff and others argued, deposited ceramic material in a tell-tale halo around the urban core. In place of manuring, Late Roman farmers adopted less intensive agricultural practices and, at the same time, large tracts of land previously dedicated to feeding the urban population became part extra-urban agglomerations ranging from agricultural villas to self-sufficient hamlets.
The decline in artifact density visible for the Late Roman period in the Thisvi basin would fit well with this hypothesis as Late Roman (and more generally post-Classical) densities declining away from the city itself not simple as evidence for contracted habitation, but as the relationship between contracting populations and changing land-use patterns.
The work of the CBBP also revealed large extramural concentration of Late Roman material like those at the southeastern corner of the of the survey area, E1, and at the harbor at Vathy. The former coincides particularly well with kinds of developments documented by the Cambridge Boeotia Survey around Thespiai as villas. The assemblage from the site contained storage vessels consistent with some kind of agricultural installation as well as beehive sherds so common at Late Roman agricultural sites around Thespiai. Moreover, the site was outside the densest areas of ceramics around Thisvi even at its Classical-Hellenistic peek, and this too paralleled the findings of the work at Thespiai.
The harbor at Vathy was a more complex phenomenon. The material at this site was more diverse than a simple agricultural installation and included some of the few example of Late Roman fineware from the survey area in addition to a significant complement of transport vessels which would be expected at a coastal site. Vathy resembles more closely the assemblages present on the islands of Kouveli and Macronisos than the material present inland in the Thisvi basin or even neighboring Thespiai which likewise lacked significant quantities of Late Roman finewares. In contrast, at the Eastern Korinthia Archaeological Survey in the immediate hinterland of the important Late Roman city of Corinth some xx km to the southeast, fineware made up almost 10% of the total assemblage of Late Roman artifacts, and this is despite a collection strategy that would tend to under represent the proportion of fine ware to coarse ware.
It is the comparative context that allows us to begin to make sense of assemblage presented around Thisvi. When first documented and published in the mid-1980s the presence of Late Roman and post-Classical material in the countryside of Thisvi was worthy of remark. Now as “the busy countryside” of Late Antique Greece comes into sharper focus, the dearth of pottery present in the Boeotia countryside and its decidedly functional and non-cosmopolitan character gives pause. There is no question that southeastern Boeotian countryside continued to see investment in post-antique period with Late Antique fortifications extant at Thisvi, Thespiai, Khostia and on Mavrovouni. On the other hand, the lack of imported fine ware for Late Antiquity suggests a particular kind of investment in the countryside. The countryside around both Thisvi and Thespiai during Late Antiquity would appear to be less substantially invested in the kind of prestige rural habitation that is often associated with the concomitant decline in the urban core of the ancient world.
The title of today’s panel was “First Out” and we hope that our paper today extended the potential meaning of that phrase to include the post-Classical material from the first generation intensive pedestrian survey. Our paper today represents a point of departure for further study of both the material produced by the OBE across the Thisvi basin and the growing body of “second wave” survey material from Greece. While much second wave survey material has seen initial publication and has contributed to the present body of knowledge regarding the post-Classical landscape, we have shown the potential in returning to this material. For the Late Roman period, in particular, with think that returning to this material will allow us to move beyond the juxtaposition of rural prosperity to abandonment (a version of the old continuity or change question) and tease out indications of regional difference present in across the Late Roman landscape of Greece. The potential present in returning to the first sherds collected from the Greek landscape in an intensive and systematic way demands that we make the results of these early intensive surveys available in flexible digital formats. It seems like that a return to these survey projects will put an end to any lingering skepticism regarding the long-term archaeological significance of survey data and, in the processes, confirm the continued value of “first out”.