A recurrent puzzle for scholars studying the Roman provinces has been the relationship between textual and archaeological evidence. One classic example of this is the relationship between the'villa'mentioned in texts and the Roman-period rural settlements which archaeologists habitually term'villas'. The latter are usually recognised by their hypocaust heating systems, baths, wall-decoration and mosaic floors, and have a restricted (often regionalized) series of building plans. Sites with these attributes are often situated in distinctive geographical locations: on sheltered hillsides with expansive views, or by river or lake sides. This is a well-established category of archaeological sites across the former Roman Empire, but it is uncertain which of these settle-ments are what Roman provincials would have considered villas. 1 Several archaeologists have responded to this dilemma by separating textual and archaeological evidence completely. They have either refused to combine written and material evidence for'the villa'at all, or they have adopted a terminology differentiating between the'villas' of archaeology and villae of Latin texts. However, this simply avoids the issue, and it would be useful if material and written data could be more effectively combined. This paper explores one approach that might more closely link the villas mentioned in texts and archaeological evidence for the villa in the northwest provinces of the Roman Empire. 2

In order to achieve this, one would need to define a set of archaeologically recognisable characteristics identifying what (if anything) Roman writers specifically meant by the term'villa'. This would allow us to construct what might be referred to as an'archaeological signature'-a set of identifying material or topographical attributes-for the villa, enabling us to relate texts and material evidence more directly. Most of the detailed descriptions of the villa relate to Italy and date from before the early second century AD. Several Late Republican and Early Imperial authors-Cato, Columella, Martial, Petronius, Pliny the Younger, Varro and Vitruvius-described contemporary villas, or offered definitions of the villa in general. Unfortunately, this is distant in time from most of the provincial complexes called'villas' by archaeologists, which date predominately to the second to fourth centuries AD. 3