Plog et al. define an archaeological site as the ‘‘discrete and potentially interpretable locus of cultural materials’’. This definition of site - and leaving aside issues arising from nonsite and distributional approaches - still leaves much to the imagination. Archaeologists working within different survey traditions or with different research questions in mind inevitably make varied decisions concerning the material that need be present, and the means of delimiting sites. Surveys are best designed when decisions of this kind are explicit and applied consistently in the field so that survey results are replicable and comparable with other projects. One guideline is to demarcate sites restrictively, since it is far easier to combine sites than to split them apart during later stages of analysis. Another important guideline is that surveyors should be able to illustrate the complete data flow from initial site visit to final settlement pattern map (see Sites: Mapping Methods).
Apart from defining sites, the growth of survey methods brings further diversity in technique with differing approaches to sampling, intensity of coverage, collections, site recording, and standards of publication. The use of full-coverage survey techniques has become the primary means to study regional variation and change over time for much of problem-oriented settlement archaeology. The rationale for full-coverage survey is the unpredictable nature of prehistoric settlement, such that no subregion or sample adequately describes the whole. The results of full-coverage surveys, moreover, allow for accurate characterization of site hierarchies, the spacing between sites on the large scale, discovery of rare or unique sites, and analyses of boundary conditions within and among regional settlement systems.
The dispute over whether full coverage is better than probabilistic sample surveys has died down, having for a time resembled the old Miller Lite beer commercials in which ex-survey stars could be imagined to say, ‘‘Tastes great! Less filling!’’. Today, wherever possible, and most especially in arid and semiarid regions with complex, sedentary populations, full-coverage survey is almost always the chosen method (notably in Andean South America, Mesoamerica, Mesopotamia, and the Southwestern United States). Sampling nonetheless remains entrenched in European archaeology and much of the United States, owing to difficult surface visibility in temperate forests and the needs of cultural resource management. Similarly, survey in tropical lowlands has emphasized transects and probabilistic sampling rather than full coverage (e. g., surveys in the Maya area). The real question, however, is not which method is better but how best to integrate full coverage with probabilistic samples in long-term, region-oriented research programs. For most areas, probabilistic sampling makes more sense if it follows a full-coverage survey, or once there was knowledge of the actual universe of sites. The global reach of the highland Mesoamerican survey technique illustrates the adaptability of the full-coverage method, as well as the more general issues arising from surveys of all kinds.
World History
