Attempting to model Symon's Castle inevitably confronted a number of these issues head-on. Indeed, these issues assumed even greater significance the high degree of structural interpretation that was both possible and necessary as a result of the nature of the site meant that a flexible approach that enabled alternative reconstructions to be easily generated was required.

Two separate approaches to flexible modelling can be identified:

  1. The production of a model in which selected parameters can be modified by the end user. A rare example of this is the Canterbury Roman theatre model <>
    ( ), in which the model, implemented in Javascript and VRML, incorporates a heads-up display allowing control of the number and dimensions of seats and the overall height of the building, for example.
  2. The production of a number of 'traditional' or static models (including interactive VRML and QTVR movies) in which the developer and archaeologist can modify the parameters during the creation of the models, and hence easily generate alternative models, but the end user can only view them. Here, the flexibility exists at the production end rather than the client end of the process.

Although the second approach is more restrictive, it is also attractive for several reasons. Allowing the end user to manipulate the virtual environment and change the model provides a powerful visualisation tool, but at the same time does not allow the archaeologist to limit the interpretation except within broadly defined parameters. This is not to suggest that the end user approach is in some way 'bad'; rather, that a greater degree of control may be desirable. Roberts and Ryan recognise this in the form of their alternative 'require new' model, in which the archaeologist generates multiple instances of the model containing alternative possibilities and arrangements and the client downloads each model as required. This means that the user can be guided round, the order of presentation is controlled and the models explained, and hence the user needs only basic interaction skills and little specialist knowledge. However, Roberts and Ryan argue that with this approach:

'A limited or biased viewpoint can be easily, even unconsciously, presented. The user is not so immersed in the three-dimensional world, nor do they have much control over what is being viewed; with the publisher providing the choices for the user. Appreciation of scale and making comparisons between observations and viewpoints are difficult unless the publisher chooses to provide images of the same scale and viewpoint on a single page.' ( ).

This seems to presuppose that providing the end user with a configurable model avoids problems of bias, whereas it perhaps only disguises it within constraints imposed by the publisher (for instance, the selection of parameters available for change and the extent of allowable change).

However, one aspect of this end-user approach that could usefully be adopted and applied to the more traditional approach is the ability to define parameters interactively for elements of the model. As things stand, most modelling packages employ a range of primitives which by definition are a long way from a typical archaeological object or structure. Automating the production of these, generating structural types in specified locations and to specified scales, would greatly ease the process of model production and experimentation. Two methods can be suggested:

Examples of each method can be illustrated using the Symon's Castle model.