2. Methodology and Models

A mainly phenomenological approach was used, implemented via site visits, a Geographical Information System (GIS) and a Virtual Reality (VR) system, to understand the Neolithic Avebury region from social perspectives. GIS was the best way to address the landscape as a whole and also allows for detailed and intelligible output of the results. But, despite the cognitive insights offered by GIS viewshed analysis, it does not place one in the landscape, but rather looks down onto it - a problem known as God Trick (Wheatley and Gillings 2000, 14). The existence of relationships between features within the landscape is made clear but not necessarily their meanings. Thus a VR system was used alongside the GIS, allowing one to experience the landscape in narrative terms, as the view is from within the landscape and one must move between points (specific views) via other points (specific views) rather than skipping directly to places of interest. While travelling through the landscape in this way, views might suddenly and surprisingly appear or one's movement might be directed to specific sites.

Because both GIS and VR were used it was decided to design a bespoke GIS suite that could output data as a traditional GIS map view or as a VR world by auto-generating VRML code. The GIS application is implemented in JAVA with functionality specific to the project in hand. Topological data was acquired from the OS's Digimap resource as Land-Form Panorama DXF files (© Crown Copyright/database right 2005) and point data was acquired from the Wiltshire SMR and site reports. The VR worlds were run in a web browser using the Cortona plug-in. It is important to stress that the VR worlds use schematic rather than photo-realistic imagery as an understanding of the relationships between landscape features is sought rather than detailed life-like recreations. Despite its age and the heightened photo-realism of newer systems, VRML was selected as its simplicity and the ease with which its code could be manipulated made it an ideal fast, fuss-free, method of obtaining the required results. Indeed Eiteljorg (2000) describes the increase in realism of VR modelling as a 'double-edged sword', that is to say that representing the landscape in too detailed a way can give the model a type of false digital authority; it suggests that the model is so detailed that it must be correct. In comparing wire-frame and detailed VR models of the Athenian Acropolis, Eiteljorg finds that both models are as worthy as each other in terms of examining the structural relations within the building. In fact, the more detailed model has the greater potential to include erroneous information. A 3-D modelling approach has previously been applied to the Avebury area (Earl and Wheatley 2002; Gillings et al. 2008, 131) but was not developed as focus was shifted to excavation.

The GIS models cover an area of around 15km². The VR model is slightly smaller than this, as computer memory and speed constraints are encountered when attempting to reproduce such a large landscape. The VR model is not included in this article as many of the features were modelled such that they appeared correctly from the required paths of movement but were either hollow or incorrect when viewed from alternative areas. Creating fully rounded features would have required more memory and thus reduced the size of landscape modelled. .WMVs of the journey along the avenues are included. Also included are screen dumps of the VR world, which give better quality images than the .WMVs.

The River Kennet and Winterbourne stream are not visible from the majority of the landscape, being relatively narrow with small but steep banks. It was therefore decided that they would not be modelled by draping a water-coloured band across the landscape. However, it seems that although much of the lower areas of the landscape were cleared of trees at this time (Evans et al. 1985, 307), the rivers were still bordered by trees or scrub (Gillings et al. 2008, 193), thus the watercourses are represented by lines of trees. The chalk banks of the earlier Neolithic monuments, the long barrows and the Windmill Hill causewayed enclosure, may have been grassed over by the later Neolithic. However, there is evidence to suggest that they were still visited and that activity took place there at the time represented by the model (Pollard and Reynolds 2002, 77; Gillings et al. 2008, 57). They are represented as having cleaned chalk banks in the model because this makes them stand out. If represented in the model with grassed banks they do not stand out as well as they do in reality. Some doubt has recently been cast on whether the West Kennet Avenue was a continuous row of paired stones in its central section (Gillings et al. 2008, 139). It may be that the avenue was discontinuous, was only made up of a single line of stones, had longer intervals between stones than other parts, or was subject to a combination of all three (Gillings et al. 2008, 139-40) at this point. As both the layout here and the length of the deviation from the pattern seen elsewhere are uncertain, the avenue is modelled with only the stones around stone 63 missing but it should be borne in mind that the deviation from the paired stone format may have covered more of the avenue.


© Internet Archaeology/Author(s) URL:
Last updated: Tues Oct 27 2009