Catálogo de publicaciones - libros

This paper has emphasized the highly reflexive approach necessary for the correct identification and interpretation of the processes behind settlement patterns. In our opinion, the key challenges are: (i) to define a sample/study area and its levels of search intensity appropriately (correcting for or exploring “edge effects” statistically where necessary); (ii) to assess and sub-divide site size, function and date range (analysing comparable features only and/or arbitrating uncertain cases statistically); (iii) to account for the resource structure of the landscape (either by only considering environmental homogenous sub-regions or by factoring resource preferences into the significance-testing stage of analysis), and (iv) to use techniques of analysis that are sensitive to detecting patterns at different spatial scales. The latter in particular is an area increasingly well-explored in other disciplines, but as yet with minimal impact on archaeological practice. There remains some value in Clark and Evan’s nearest neighbour function for identifying relationships between sites at one scale of analysis, but it may fail to detect larger-scale patterning. More critically, the dichotomy it encourages between “nucleated” and “dispersed” is at best an overly simplistic model and, at worst, bears little relationship to the reality of settlement organization, which at different scales can show both nucleated and dispersed components. In our Kytheran case study, there is obviously further work to be done, but even with the existing dataset, we have shown that using a combination of Monte Carlo testing, frequency distributions, local density mappings and Ripley’s K -function allows a more sensitive assessment of multiscalar patters and therefore a more critical evaluation of the processes underlying settlement distributions.