Just as with anisotropic cost surfaces, all these frictional effects modifications can be implemented without the need for large amounts of specialist knowledge on the part of the GIS user. It must therefore be asked: why have so few practical applications of these appeared within archaeology? Why do the majority of the models that we do see continue to be dominated by cost surfaces constructed solely on the basis of terrain slope and terrain slope alone? Even more importantly, with the growing literature on these more refined cost surfaces being available, for instance through the proceedings of conferences such as the Computer Applications in Archaeology (CAA), why do GIS analyses continue to gloss and make excuses for the use of slope-based cost surfaces? The point of this paper is not to criticise directly those who have produced such analyses in the past, but rather to argue that we are now in a position where it is increasingly difficult to justify a cost surface, where, for example, the land is treated as a uniformly compacted asphalt-like surface, despite the known presence of swamps, lakes and forests within that landscape. In the same way that viewshed analysis has been forced to engage with the 'tree effect', modelling of pathways and catchments must come to a similar realisation with respect to travel costs.
Similarly, regional analysis of site locations and travel routes, with a few exceptions, have been undertaken either on sites located within large land masses or within a single island setting (see for example Bellavia 2002; Chapman 2003; Christopherson et al. 1999; Daly et al. 2000; De Silva and Pizziolo 2001; but see Indruszewski and Barton 2005). By and large these tend to ignore the effects of rivers, lakes and the sea on travel costs and affordances, either focusing only on foot travel within the study area, or glossing over the possibility of travel between islands and between island and mainland. This has resulted, for example, in assertions of the following type being made about the Iron Age: 'evidence of boats for river movement and wagons are not available, so we assume the settlers used to walk between the chosen settlements within the landscape' (Podobnikar et al. 2004). This is by no means an isolated example, and in fact it is notable for actually admitting its omission of non-foot transportation. It is suggested that these types of omissions are due to two factors.
First, there seems to be very little information available either in the archaeological literature or elsewhere on travel costs or travel rates in pre-industrial societies, aside from extremely general statements. We see statements referring to the sea corridor stretching from Portugal to northern Scotland such as: 'there was a high degree of communication along the Atlantic Scottish sea routes throughout the period' (Armit 2003, 92), referring in this case to the Iron Age, and Cunliffe argues that these water routes were key axes for both exchange and the transmission of ideas, perhaps from the Mesolithic onwards (Cunliffe 2004). We are told that water is important, but there is no way to quantify such a statement adequately when what one really wants is a method to calculate how much more important water transport was.
Where figures are presented, they generally relate to the bulk transport of cargo in the Roman Empire, and are derived from the relative costs of goods at ports versus sites further inland. This has allowed Kevin Greene, following Duncan-Jones, to calculate that land transport was 28 times more costly than sea transport (Duncan-Jones 1974, 366-69; Greene 1986, 40). However, this is an economic, not an energetic, cost and emerges in the context of a system of industrial production far more extensive than those present throughout most of human prehistory (it is derived in part from the relative costs of shipping an olive press from near and distant sources). It also provides little or no information about the speed or distances that could be covered by these different transportation methods. In essence, all this figure really tells us is how many oxen and drovers were required, and how their operating costs (money, food) compared to the costs of a ship and crew. We are therefore once again left with a general statement to the effect that water transport was probably more important in the past than it is now, and that lakes, rivers and oceans were possibly seen more as highways than as the barriers our land- and air transport-based society currently considers them to be.
Secondly, there is little critical analysis out there, both in the archaeological and historical/ethnographic literature about what constitutes an 'acceptable' travel distance, whether to one's field and back, or over the course of a day in the case of long-distance travel. Obviously this sort of information will be very much culturally determined and could vary substantially, but the problem here is not the variability of the available information, but rather that there is little if any information available! Researchers have tended to fall back on rules of thumb, like Higgs and Vita-Finzi's one-hour walking distance for agriculturalists versus two hours for hunter-gatherers (Higgs and Vita-Finzi 1972, 33), which are explicitly based on Lee's work with the !Kung and Chisholm's work on rural settlement (Chisholm 1968; Lee 1969) and often are presented with no evaluation of their suitability to the group being studied or the data at hand. Roper, in reviewing catchment studies to 1979, remarks: 'It is this almost mechanical use of these figures for delineating the analytic territory that was noted earlier as one of the major problems with site catchment analysis' (Roper 1979, 133). In 2005, precious little has changed. Even more importantly, in terms of travel using a variety of transport options in a variety of media (land, ocean, river), these one- and two-hour radii or distances are distances on land and land alone, with no consideration of what an acceptable travel distance over the water would be.
There is, therefore, a very real need for more work on how humans relate to their environment as they move through it, and for us to take a strongly critical look at how we are attempting to model that movement within GIS. How people travel from place to place is not just a matter of maximising efficiency, but is bound up in all sorts of social factors, as any child who has crossed the street to avoid a bully or barking dog can confirm. On a purely functional level, we need more experimental effort put into figuring out what is a reasonable distance to travel in the course of a working day.
© Internet Archaeology/Author(s)
University of York legal statements | Terms and Conditions
| File last updated: Tue Dec 20 2005