Article excerpt

Whilst much as been written on the decoration and spatial significance of Roman houses (Wallace-Hadrill 1994; Clarke 1991), an in-depth analysis of visibility and visual effects has never been carried out. Visual aspects are particularly relevant to the explanation of the Roman atrium house which, unlike mosaics and hypocausts, did not spread to areas beyond peninsular Italy. This distribution is not simply one of centre versus periphery, since by the end of the 2nd century AD the atrium house had been entirely replaced by a new form--the peri-style house--even within Italy itself. The substitution must reflect a change in social values in the Roman Empire as well as the role of the house (domus) itself. Analysis of new aspects of the domus will explain the subtle changes in building emphasis that heralded this transition.

The analyses

The most immediately intuitive representation of visibility is the `viewshed', in which all possible views are traced from a single point (Hanson 1988). The logical extension of this idea for the analysis of architectural spaces is to create a viewshed for each space, allowing the `viewer' to walk freely within the space. This `maximum' viewshed presents the largest possible range of views from any one room (FIGURE 1). Unfortunately, this diagram is not terribly diagnostic because free motion means that the viewer can see the entirety of every adjacent room. Discussion of the limitations on or control of visibility is impossible.


In their most simple form, rooms are either open or closed. The `open' form, or exedra, is a room with one completely exposed side. The `closed' form has walls on all sides with an opening. The differences between these forms can be characterized by how they limit visibility. An exedra will block visibility into it from either side, at an angle relative to the depth of the room.

A closed-room limits the visibility more than an exedra, but a viewer walking past an opening will gain a fairly complete impression of the contents of the room despite its `closed' opening. This is because his viewpoint pans across the majority of the internal space as he walks by (FIGURE 2). Even if the viewer's motion is limited so that he cannot get close to the opening (increasing [b.sub.1] to [b.sub.2] in FIGURE 2), the overall coverage of his visibility is not greatly reduced. Rather, the distance he must travel in order to achieve complete room coverage is increased (FIGURE 2). In neither case does the opening significantly limit the range of visibility from the outside. Visibility range is only truly reduced by the relative depth of the opening (e) to its width (d). …