The Sydney Cyprus Survey Project
A Bernard Knapp, Michael Given, 2003. https://doi.org/10.5284/1000208. How to cite using this DOI
Data copyright © Prof A Bernard Knapp, Dr Michael Given unless otherwise stated
This work is licensed under the ADS Terms of Use and Access.
Primary contact
Dr
Michael
Given
Department of Archaeology
University of Glasgow
The Gregory Building
Lilybank Gardens
Glasgow
G12 8QQ
Scotland
Tel: 0141 3305690
Resource identifiers
- ADS Collection: 323
- DOI:https://doi.org/10.5284/1000208
- How to cite using this DOI
Overview
- Survey Area and Sampling
- 'Sites' and 'Siteless Survey'
- Mapping
- Transect Survey
- GIS Analysis
- Conclusion
SCSP's main goals were to identify and locate industrial sites and agricultural villages; to reconsider and redefine the existence of a proposed site hierarchy (mining/resource sites, agricultural villages, urban centres); and to reconstruct early industrial and agricultural landscapes. SCSP was interested not only in locating sites, but in determining human-land relations throughout the landscape. In fact we were interested in the total physical and social landscape. Our framework for interpretation was based on the belief that the study of people cannot be separated from the study of their environment and, conversely, that the environment will alter as humans alter their society. SCSP dealt with the human transformation of a landscape over a period of 5000 years. In evaluating this landscape we integrated diverse fields and techniques such as archaeology, ethnohistory, geomorphology, ecology, GIS and satellite imagery into landscape reconstructions, and into social and economic reconstructions of a distinctive region. We considered not only how humans transformed the landscape they inhabited, but also how natural history and resources impacted on socio-cultural development and change.
In writing about landscapes, archaeologists often tend to overlook the human experience of place. Such places can be imbued with deep personal, ideological and economic significance. SCSP's methodology and philosophy represent our attempt to engage field survey within a social archaeology. The long-term, socio-historical perspective of this project formed an important research focus of its several principal collaborators. The interactive study by diverse specialists using innovative technologies to reconstruct past landscapes has enabled SCSP to make a different kind of contribution to the archaeological history of Cyprus. Only through interdisciplinary survey, excavation, exploration and social analysis can one gain a better awareness of the early agricultural and industrial landscapes that typify the island of Cyprus.
Survey Area and Sampling
Defining the survey universe is perhaps the most basic decision in regional survey methodology. In order to select such a study region, it is necessary to consider the nature of the boundaries chosen, and whether the area chosen should be stratified into smaller analytical units. As it may seem quite straightforward, determining the parameters of a cultural unit is problematic for several reasons:
- by definition, regional surveys involve diachronic research, usually covering millennia of human history and the concomitant shifting of cultural boundaries;
- even where the problem orientation of a survey is focused on a particular stage in the past, the spatial extent of the targeted cultural expression is seldom clear;
- permits to conduct survey work are handled by modern administrative and political units, not past cultural groupings.
The concept of a survey universe wields a major influence in the design of regional survey projects; it may be regarded as a methodological backdrop against which field research is carried out. Given a finite amount of resources and a expanse of territory defined as the survey universe, the entire area can be covered at a certain level of intensity, or else a sample of the whole may be carried out at a greater intensity. SCSP's survey area was a naturally defined area, the interface between the igneous (metal-bearing) ore bodies and the sedimentary (arable) soils. This contact zone is transverse to the drainage network, and includes drainage basins that have a wide variety of discharge and arable land surface. Because of the great variability in terrain and archaeological material across the survey universe, we had decided by 1995 to stratify our survey universe according to purposive principles, and then to sample within those strata to test our assumptions.
Our initial stages of fieldwork were designed to provide a systematic coverage of the survey universe and at the same time to generate information that could be applied constructively to the succeeding stages of research. The goal was to develop a broad understanding of the cultural remains and recoverability potential across the survey region before any stratification was undertaken. Accordingly, we adopted a multi-stage research design and tried to make it as flexible as possible; during the early stages we chose to sacrifice spatial extent in favour of intensity. As we moved on to the succeeding stages of the fieldwork and research, we had very explicit reasons for adopting a different strategy, but the overall goal remained to examine as much of the survey universe as possible with a level of intensity appropriate to the particular recovery goals of that stage of research.
By the end of our fieldwork, SCSP had built up an extensive database of quantitative cultural, environmental and social information about the survey area; its natural resources have been mapped and catalogued; initial patterning in the relationship between human activity and the landscape has been observed and discussed; and our field and laboratory methodologies had been refined and tested.
'Sites' and 'Siteless Survey'
We now need to clarify our position in the debate over the meaning of the term 'site,' which has a direct bearing on regional research programs and strategies. Attempts to operationalise the concept of a site in archaeological field survey at times have been quite arbitrary: for example the Southwest Anthropological Research Group defined a site as an analytical unit having a minimum of five artefacts per square metre. Most archaeological sites are difficult to define in the field, specifically because their limits are not immediately apparent. Within the SCSP survey area, this was particularly apparent in the alluvial river terraces. Even where surveys seek only to define distributional densities in the field, in their final analyses they often attempt to examine clusters as possible indicators of a site. In our view, a site is a locale that can be interpreted not only with respect to its constituent parts, but also within the regional context of other sites.
Early proponents of 'siteless' survey pointed out that non-sedentary populations leave minimal material evidence in the landscape. We would add that, at times, even sedentary peoples carry out activities in the landscape that leave few material indicators. Whilst we agree that it is important to problematise the notion and definition of sites, the lack of chronological control does not necessarily mean that material cannot be dated. Nor does the boundedness of concentrated cultural material mean that spatial analytics cannot extend beyond its limits. Archaeological survey methodology, including sampling theory, is often misunderstood because of a failure to be explicit about the nature of the units discovered. In order to define the makeup and stratification of SCSP's survey universe, it is necessary to discuss the terminology associated with the analytical unit of discovery.
SCSP sought to examine the total landscape and its use, exploitation, development and change in the face of long-term human activity. In such an approach, a site is more of an interpretative construct than something that can be strictly observed and defined. Defining a site also requires a detailed knowledge of localised patterns of artefact distribution and landscape use across the full range of spatial and chronological scales, as well as an awareness of taphonomic and geomorphological processes. In other words, there is a great deal of variation in what exists during different time periods, under differing conditions of visibility, and within different parts of the survey area. Another significant issue in defining sites as artefact clusters or even on the basis of habitational elements is that a site's boundaries are often indeterminable. Thus it is not so much a question of where the site is as where it isn't. If a 'site' is an artificial construct, then site distribution is meaningless without adequate information on overall artefact-density distributions, meaningful criteria to construct sites from these density-distributions and the capacity to assess environmental data. Another relevant point in considering the SCSP region is that we have large, very broad areas of high artefact density as well as other evidence indicating human exploitation of almost the entire stable landscape; only areas stripped of all surface soils (usually by erosion) appear to be devoid of artefacts. Such factors actually confound the concept of 'site,' which might be defined more readily and clearly in regions with less dense land-use.
Our decision to begin with a systematic sample and to employ the methodology of a siteless survey reflected a commitment to gain as much information as possible about the area of our investigations before making any attempt to characterise it, and ultimately to stratify it. In this mode, our analytical unit was the survey unit itself. Wherever we found agricultural plots clearly defined in the field and on aerial photographs, they formed our basic recording unit. Such units are characterised by roughly uniform land-use, topography, vegetation cover, often even slope and aspect; they are distinguished from other units by a change in any of these factors, or by a field boundary, fence, hedge, road, stream etc. Where agricultural fields, vineyards, orchards and similar aspects of the landscape were not clearly defined, we used transect recording (see above) and defined units according to topographic, geological or even artificial boundaries (e.g., roads). We also attempted to limit these arbitrary units to 100 m or less in length. The data from any analytical unit consist of raw numbers of artefacts and quantifiable observations on geomorphological factors, ecofacts and other spatial components (e.g., slope, aspect and the like). Yet we assumed from the outset that our higher level research goals, which impel us to understand the social aspects of any human community as expressed by material in the landscape, would require us eventually to shift our analytical unit from landscape features to something that is itself a construct of our research process.
We accept, therefore, that there is a certain tension between the type of siteless survey conducted in field plots and transects, on the one hand, and on the other the attempt to discuss sites, settlements and settlement hierarchies in more theoretical terms bearing on the research goals and questions in which SCSP was interested. Methodologically, our field strategy evolved to incorporate the concept of POSIs (Places of Special Interest) and SIAs (Special Interest Areas). POSIs, it may be noted, are limited in horizontal expanse and in their diversity of material: e.g., lithic scatters or dense pottery concentrations, a water mill, a sheepfold or the remains of a smelting furnace. SIAs, on the other hand, are more complex, broader in horizontal area and usually multi-functional, containing material from different time periods, and occasionally made up of several different POSIs. In dealing with SIAs, a variety of techniques are involved - from casual walk-overs, to sample transects, to 'block' survey, to gridded collections and measured plans and sections. In effect POSIs and SIAs represent a further stratification of our survey universe.
In order to address our research goals it was essential to develop definitions which would facilitate the analysis of settlement patterns through time. Given the diverse types and dating of material recovered by SCSP teams over five field seasons, we were finally in a position to assess exactly what constitutes specific uses of the landscape - e.g., lithic production areas, smelting sites, mines, slag heaps, rural sanctuaries or churches, farmsteads and agricultural settlements. Our ultimate goal was not to create static definitions of sites or settlements, but rather to make meaningful statements about their internal organisation as well as their role in the emergence and development of social landscapes and human communities in the northern Troodos foothills of Cyprus.
Mapping
One of the chief problems SCSP had to overcome was the lack of the standard Cypriot 1:5000 topographic maps in the survey area. In a refinement of earlier attempts to address this problem, we made use of enlarged aerial photographs to create a base map of the entire survey region. Using the GIS program MapInfo, these photographs were scanned and registered to the UTM grid, with grid lines of 100 m spacing superimposed on the base map. Mapping sheets were generated from the base map for use in the field. Each A4-size mapping sheet (or field sheet), aligned in a north/south direction and produced at a scale of 1:2500, showed a 500 x 200 m portion of the aerial photographs, with the UTM grid overlaid on it. his process was fairly efficient for transect work, but ultimately did not offer the degree of horizontal accuracy required for more detailed investigations. For example, on a 1:50,000 map, a pencil point dot is roughly the equivalent of a circle of 10 m in diameter. Add to that camera distortion and plane tilt, and the inaccuracy can rise to 100+ m.
This problem was overcome through the generous assistance of the Cypriot Geological Survey and the Cypriot Lands and Survey Department, from whom we acquired a set of 1993/4 aerial photographs and corresponding camera fiducials. These images, as compared to those from the 1960s, more closely resembled the built and natural environment in which we were working. From Lands and Survey, we were given GPS (geographical positioning system) readings for survey points in our study area. The GPS points we were given were purposely degraded in accuracy by +/- 2 m in three dimensions.
With accurate survey points from 1993, our task was then to gather further GPS readings for our 1993/4 aerials. We did this in two phases: first, tying together all the survey points of the Department of Lands and Survey with very accurate readings; and second, from these points taking at least four and often upwards of a dozen GPS readings for each aerial photograph. The Trimble 4600LS Survey Grade GPS that we used consists of two receivers, one of which acts as a mobile base station and the other of which acts as a rover. Accuracy for static survey (rover stationary) is sub-centimeter within 10 km.
The 1993/1994 aerial photographs were then rectified using the ERDAS IMAGINE 3.1.0 geometric correction, rectification and export utilities. Geometric correction of each photograph was required to remove image distortion caused by lens distortion, plane tilt and parallax. Focal length, lens distortion and fiducial coordinates used for geometric correction were obtained from the Cypriot Department of Lands and Survey. Once camera components were corrected, a minimum of four GPS ground control points (< 5 cm precision) were used to geo-reference accurately the image pixels. Photo rectification/warping was achieved through a maximum of ten iterations of bilinear correlation and least squares analysis. Once rectified, the images were cropped to remove the photo headers and fiducial marks using the ERDAS raster resampling tool. The images were then exported and restored from an ERDAS pyramid format (.img) to TIFF format for use in MapInfo. The TIFF images were then registered as image tables in MapInfo using rectified geo-coordinates measured in ERDAS IMAGINE. Once registered as a MapInfo table, the rectified images were easily integrated into the SCSP MapInfo database.
The creation of these photo-maps began as a solution to a problem (the lack of sufficient detailed topographic maps), but ended up as an innovation that we believe improves on standard field techniques. The photo-maps were extremely useful in the field. Not only were they used for recording archaeological units but they were also used by other specialists for detailing field observations. Having photo-maps with overlain 100-m grids allowed us to address specific grid sections. Finally, when the photo-maps serve as a base layer for the survey units, they provide for documentation of the survey area that is easy to read and rich in information.
In our daily fieldwalking, surveyed units were outlined and numbered sequentially on the field sheets, while the data associated with each unit in the field were recorded on a separate form. Survey unit outlines were digitised from the field sheets using MapInfo. Data collected in the field for each survey unit - spatial, geobotanical, artefactual - were entered into the project's database and were exported regularly to MapInfo. The resulting table could then be opened and combined with the digitised outlines, whilst the outcome was saved as a new table for analysis.
Transect Survey
Our basic field strategies were:
- to walk 50 m wide transects north-south across the entire survey area at 500 m intervals, in order to obtain a broad systematic sample of the survey area;
- to utilise spatial information entered daily into the GIS in order to determine which topographic, geological and land use factors may have conditioned the occurrence of exposed cultural materials;
- to conduct block survey of 'Special Interest Areas' which showed extensive evidence of early industrial, agricultural or settlement activities;
- to investigate 'Places of Special Interest', or sites, designated by obtrusive remains or densities of artefacts.
The basic element which made up survey transects (as well as block surveys) was the survey unit. In uncultivated ground, units were 25 m wide, corresponding to five fieldwalkers walking 5 m apart. Units were defined at their beginning and end by natural divisions such as a break in slope, a significant change in vegetation, or a gully line. If there was no such division, the unit would be ended after 100 m, even if there were no artefacts. The outline of the unit was drawn on the field sheet, and later digitised. Transects were always two units wide, giving a total width of 50 m in uncultivated ground.
In cultivated fields we used a slightly different system of defining units, known as the 'souvlaki method,' so called because these survey units generally projected beyond the normal 25 m wide strip. A ploughed field is clearly one discrete unit: artefacts will be moved around within it by the plough, but not regularly from one field to the next; and in most cases a field's geomorphology, vegetation and physical characteristics are homogeneous. In cases where a field extended beyond the normal 25 m wide unit, the extra portion was included in the survey unit, with some team members walking an extra strip to maintain the 5 m spacing. This was done only when 50% or more of the entire survey unit lay within the original survey transect. In cases where the transect crossed the edge of a very large field (i.e., one in which the survey unit was less than 50% of the field), the survey units were confined to 25 m.
Within each survey unit, we recorded a range of locational, environmental and archaeological information. This included basic data such as the unit number, location, status (e.g., 'surveyed', 'standing crop'), and method ('transect', 'souvlaki'). The percentage of the ground visible beneath the vegetation was decided after discussion among the team, as was 'background confusion,' the extent to which red or grey stones, for example, made it more difficult to identify sherds. Topographical, geomorphological and environmental data collected consisted of topography, geomorphology, sediment cover, surface character, erosion pattern, soil colour, modern land use, slope, aspect and terraces. The choices were listed as a series of codes on the unit sheet, and were further explained in a detailed field manual.
In each unit the teams collected a representative sample of cultural materials: pottery, chipped stone, ground stone, metals, slag, ores and fluxes, glass and tiles. All other, mainly non-diagnostic, material was simply counted and left in the unit. Slag, ore and flux were estimated when they were too numerous to count. The pottery collection strategy as it evolved by our third season (1996) was designed to correlate quantitatively what was counted versus what was collected. Each of the five fieldwalkers collected one sherd of each discernible category of ceramics they found in their 5 m strip. Thus a common ware or form would have multiple examples in the collected sample, but we did not have to collect excessive quantities of pottery from the field. In each collection, accordingly, there might be as many as five sherds that were essentially identical (same fabric, same type or sub-type of vessel, same portion of the vessel. All of these data were recorded on Unit Forms using a series of codes). As well as the forms, one team member kept a field book with a running commentary on the units, the finds, any problems in procedure or interpretation and general comments on the team's health and state of mind.
Each of our specialised components - geoarchaeology, geobotany and satellite imagery, archaeometallurgy, pottery, historical archaeology and ethnography - established specifically designed methodologies for fieldwork and analyses, all of which were closely integrated into our general field methodology (as outlined above).
GIS Analysis
A major component of SCSP consisted of using GIS-derived thematic maps to illustrate the analytical outcome of our field counting, collecting and recording strategy. Our main objective in using these maps, of course, was to interpret changing human activities in, and human use of the landscape. Our goal was to develop a measurement of pottery (called the Pottery Index, or PI, for which see further below) that would indicate the strength (importance) of a specific time period within a unit. Additionally, we wanted to be able to assign functional meaning to our pottery. We did so by incorporating the pottery data (density and distribution) into GIS maps that could convey a qualitatively and quantitatively different kind of information than prose could provide. The study of pottery, then, was the key analytical aspect in assessing the meaning and significance of survey units, in the broadest sense.
As fully evolved, our field strategy for developing reliable quantitative pottery data involved four observations taken in the field:
- the absolute count of pottery
- an estimation of the percentage of ground obscured by vegetation
- an estimation of the severity of background confusion
- an ordered sequence of geomorphological observations designed to estimate the depth and state of sediment erosion.
Because of the multiple factors that affect the discovery of surface pottery by fieldwalkers, a raw count of the number of sherds belonging to a particular period in a survey unit would be meaningless. In an attempt to overcome this problem, SCSP's database analyst Nathan Meyer and pottery specialist Timothy Gregory worked together to develop the PI (Pottery Index), a relative scale of density instead of an absolute number of pottery sherds. Our GIS analytical maps of material density and distribution in the landscape, and the analyses from which they are derived, are based on the PI, which factors in influences both natural and anthropogenic, and takes account of our collection methods. To be more specific, the PI was based on the following factors:
- square area of each survey unit (thus giving sherds per square metre)
- ground visibility
- background confusion
- projection from collected sherds to counted sherds (we collected a representative sample - about 30 percent - of pottery in a unit).
Because the sample is representative, we cautiously projected from the number of sherds of a particular period collected and analysed, to the total number of sherds of that period which were in the survey unit. The relevant statistical formulae were carefully tested by seeding experiments in 1997, and adjusted so that the final figures reflected more accurately the presence of the actual pottery in the field. What we propose, based on GIS and statistical analyses of quantitative pottery data and presented as GIS-derived maps of the PI, are some of the diverse possible uses of the landscape (detailed explanation of the PI, with full statistical formulae and analytical discussion, is presented in chapter 3.7 of Given and Knapp 2002). In the most basic and general terms, we have suggested that a PI of 500-1000 indicates a light scatter of pottery derived from agricultural practices such as manuring. In turn, a PI of 5000 might suggest a low-density habitation like a farmstead, whereas a PI of 10,000 suggests the very high densities found on major settlements like Tamassos.
Conclusion
Between 1992 and 1998 SCSP achieved the following:
- walked 1550 survey units at approximately 5 m spacing between fieldwalkers, covering 6.54 sq km, or 9.94% of the survey area;
- detected the background artefact scatter across the survey area, and estimated its density and correlation with topographic and environmental factors;
- identified and investigated 11 Special Interest Areas and 142 Places of Special Interest within the landscape;
- counted in the field 87,600 sherds of pottery, pithos and terracotta; 8,111 tile fragments; and 3092 lithics;
- collected and analysed 29,235 diagnostic, unique or indeterminate sherds of pottery, tile and terracotta, and 951 lithics; all of these were entered into the project's database and exported to the GIS.
All of these data are available in this archive, in database and GIS form.
Ultimately, a social archaeology at the regional scale deals with many facets of knowledge, both empirical and theoretical. We believe that SCSP made some significant strides in advancing survey methods in Cyprus. Our chronotype cataloguing and information system, for example, was integrated tightly with our pottery analyses and GIS mapping to present a new perspective on the exploitation and development of, and changes within a regional landscape. SCSP's Pottery Index (PI), theoretically and methodologically, sought to bring new rigour to the mapping of regional pottery data: our GIS analytical maps portray in a vivid and dynamic medium the level and types of materials we encountered. The integration of spatial, social and historical approaches has provided us further insight into the development of social organisation and resource exploitation in Cyprus during the last 8,000 years, the relationship between people and their landscape, and into the meaning and memory of the past.
