2.3 Comparative survey and metadata

So far, discussion has focused on the analysis of legacy data from individual surveys. However, there is growing interest in comparing between different surveys and regions (Alcock and Cherry 2004). The potential value of comparison around and beyond the Mediterranean has long been recognised (e.g. Cherry 1983, 406), though until the last five years, progress has been slow. The most obvious stumbling block has been the diversity of survey methodologies. Simple questions such as 'what is a site?' have generated competing conceptual models and diverse field methods. The result has been much discussion and relatively little progress towards integrating and comparing surveys. Over the past decade, however, the achievement of this goal has been advanced by:

  1. growing awareness of the variability of regional settlement through the ongoing publication of projects, including final reports by some of the larger and most ambitious surveys of the 1970s, 1980s and early 1990s (e.g. Libyan Valleys survey, Barker 1996);
  2. increasing emphasis on the connectivity and interdependence of Mediterranean societies (e.g. Horden and Purcell 2000), itself partially based on recognition of the results of field survey as evidence for high levels of exchange and communication;
  3. expanding availability of integrative spatial technologies (GIS, GPS, etc.) and digital data.

Comparison of regional settlement patterns around the Mediterranean during the Archaic, Hellenistic, Roman and Medieval periods should offer great insights into variable political and economic connectivity and large-scale demographic processes. Legacy data assume an important role in any such programme. No individual project can encompass the breadth of an individual country, let alone the Mediterranean and beyond, thus comparative surveyors will always need to work with legacy data.

As already discussed, these legacy data are important because the surface archaeological record is deteriorating. But they also have another particular significance for comparative survey. One of the key methodological trends over the past 25 years has been a significant increase in the intensity of coverage and a dramatic decline in the size of regions covered. The merits of this development are hotly debated (e.g. Fentress 2000; Terrenato 2004). Specifically, it is argued that this reduction in the scale, and the incompatibility of off-site methodologies, has derailed the comparative survey agenda (Blanton 2001). Some have therefore turned to the older surveys as the basis for inter-regional comparison (Blanton 2004; Fentress forthcoming); others have argued for more variety, using the older surveys as a wider framework for the greater detail provided by smaller surveys (Witcher 2006, 61-2). Notably, few have dared to reject older regional surveys outright, as this is tantamount to rejecting significant progress with comparative survey altogether (but see Ammerman 2004, 182). However this debate is resolved, it is clear that legacy data lie at the heart of comparative survey. Legacy data have to be combined with new datasets to explore large-scale questions of settlement and economy over time and significant issues of data compatibility must therefore be confronted.

How then are these methodological differences to be addressed in order that surveys can be compared? The answer is metadata; in other words, not the results of surveys, but data about the results (Wise and Miller 1997). A simple example illustrates the point: a survey using walker spacing of 10m will recover all sites more than 10m in diameter, while surveys using wider spacing will recover only a percentage; if these small scatters represented for example, burial sites, spacing of more than 10m will systematically underestimate the scale of funerary activity. Metadata therefore allow the significance of individual survey results to be understood and, in turn, allow meaningful comparison with other surveys.

The range of metadata which can be documented is varied. Some aspects of a survey can be directly measured (e.g. walker spacing, artefact sampling). These metadata can be used to analyse and compare results formally. 'Proxy' metadata are also common (e.g. person days per square kilometre). Although these can be measured, their relationship with survey results is not direct but can help to explain survey results (e.g. a survey of one person day per square kilometre will probably find fewer sites than a survey of twenty person days per square kilometre). In addition, to these 'formal' metadata, there are other forms of 'data about data' which defy measurement - even categorisation - but which can assist understanding. For example, survey aims have a strong influence on results through methodology and interpretation, but this influence is difficult to measure directly. Similarly, knowledge of local ceramic sequences is incredibly variable and profoundly affects the ability to date material and, hence, sites. The scale and implications of such influences are difficult to measure, but can be highly significant. In what follows, I will maintain a broad distinction between formal metadata and what I term contextual metadata.


© Internet Archaeology/Author(s) URL:
Last updated: Mon Jun 30 2008