Proceedings of the 38 th Annual Conference on Computer Applications and Quantitative Methods in Archaeology, CAA2010 F. Contreras, M. Farjas and F.J. Melero (eds.) On the Road to Nowhere? Least Cost Paths, Accessibility and the Predictive Modelling Perspective Verhagen, P. CLUE, Faculty of Arts, VU University, Amsterdam, the Netherlands jwhp.verhagen@let.vu.nl This paper reports the first results of a study into the utility of using least cost path models and other indicators of accessibility for predictive modelling purposes. While accessibility and movement potential are potentially important variables influencing settlement patterns, a number of difficulties are associated with the application of cost sur- faces, least cost paths and network analysis for this purpose. The paper discusses the relevant issues, and presents a new way of creating maps of regional movement potential through the use of cumulative cost paths. It is concluded that a better theoretical foundation of (pre-)historic movement is necessary to apply these techniques more success - fully for the analysis and prediction of settlement patterns. Keywords: least cost path modelling, accessibility, predictive modelling. 1. Least cost paths, accessibility and predictive modelling In archaeological predictive modelling, it has always been considered difficult to include socio-cultural factors in the models used (VERHAGEN et al., 2010). One of these factors is the movement of people, animals and material resources through the landscape. Every archaeologist is aware that transport routes may have had a major influence on the density of settlement and the accumulation of archaeological materials; and the routes themselves are archaeological sites as well. From a predictive modelling point of view, least cost path (LCP) modelling may therefore be an important tool that can contribute to the prediction of locations or zones where (pre-)historic movement patterns may have concentrated. Yet, LCP modelling and territorial modelling based on cost surfaces have not been used extensively for predictive modelling purposes (but see WHITLEY and BURNS, 2008; WHITLEY et al., 2009). Most published studies consider the calculation of possible paths between known archaeological sites (e.g. VAN LEUSEN, 2002; HOWEY, 2007; ZAKŠEK et al., 2008), or even try to reconstruct (partly) known routes by means of LCP modelling (e.g. BELL and LOCK, 2000; FIZ and ORENGO, 2008; POLLA, 2009). The cumulative cost surfaces that form the basis for LCP modelling are used regularly to model settlement territories (e.g. SOETENS et al., 2003; ROBB and VAN HOVE, 2003; DUCKE and KROEFGES, 2008). The debate in archaeological computing literature has mainly centred on questions concerning the correct ways of defining the friction surfaces used, especially where it comes to the impact of slope on movement, and on the algorithms used for calculating the LCPs (e.g. LLOBERA, 2000; EJSTRUD, 2005; ZAKŠEK et al., 2008; GIETL et al., 2008; HERZOG and POSLUSCHNY in press; HERZOG, 2010). LCP modelling however is only one of the techniques available to model and measure connections between geographic locations. In economic geography, network analysis is used extensively for the same purposes (see e.g. RODRIGUE et al., 2009). Space syntax is a rapidly growing field of research dealing with similar issues. It is usually applied more in urban and built-up contexts (see BAFNA, 2003 for an introduction, and e.g. CRAANE, 2009 for an historical case study). Both approaches provide powerful tools to extract measures of connectivity between network nodes (places of departure and arrival). Measures of ‘resistance’ (friction) along the edges of the graph (the paths) can be used as well to find the travel routes with the least costs, especially over longer distances. The number of (landscape) archaeological applications of network analysis is however rather limited (see e.g. ALDEN, 1979; HARE, 2004; NUNINGER et al., 2006), and it has never been considered for predictive modelling purposes. Cost surfaces, LCP modelling and network analysis can be useful tools to identify places that are inherently Issues in Least-Cost Analysis 383