This presents the author's doctoral research, originally submitted in 2011. Previously, Brindle was a Finds Liaison Officer and so had practical experience of the Portable Antiquities Scheme (PAS).

Chs 1 and 2 discuss the PAS, overall quantities of Roman period finds on the database and their distribution, and the potential to contribute to knowledge of Romano-British landscapes and settlements, as opposed to the more frequent research use of PAS to study types of artefacts. Ch. 3 describes the methodologies adopted for examining the PAS Roman period data in each of the study areas, and assesses factors affecting the deposition of metal artefacts, post-depositional processes, and the additional biases introduced by the specific recovery methods of the metal-detector user and by the recording process. This useful summary draws on a fairly small (sample size 65) survey of detectorists, included as an appendix. As much of the analysis of the data was based upon mapping, questions of reliability and bias in location are examined. The key methodological decisions were about how to define a ‘findspot’ in terms of spatial associations: a distance of up to 200 m between each record was defined as an association. What quantity of artefacts per findspot would suggest a potential settlement? Such questions are familiar to a Historic Environment Record (HER) officer making decisions about what defines a ‘monument’ (or ‘heritage asset’ in current planning language). B. settles for a classification of ‘medium’ (5–10 metal artefacts) and ‘large’ (11+ metal artefacts) findspots as potential ‘sites’.

The meat of the book (chs 4–8) is the five case studies, using the PAS data for Wiltshire, Worcestershire with Warwickshire, Northamptonshire, North with North-East Lincolnshire, and Cumbria. Each is approached in the same way, plotting the PAS Roman finds against the HER sites, identifying high-activity areas using kernel density analysis, observing similarities and dissimilarities in the distributions and densities of the data. Then some local areas, usually a relatively productive parish, are selected for detailed study, again examining potential sites in the PAS alongside the HER. The PAS sites are also analysed by artefact type to identify distinctive chronological or functional patterns in the assemblages. Only in one case, Great Bedwyn in Wiltshire, was there any related fieldwork: geophysical survey identified a square enclosure, potentially a shrine. Some of the maps, essential for understanding the discussion, were not easy to read, particularly those with greyscale shading to illustrate topography (and this shading was inexplicably reversed in the Northamptonshire section).

A primary aim of the research was simply to validate the PAS data as a useful addition to our knowledge of Roman Britain. It shows that the PAS is identifying numerous sites that were not previously on the HERs and confirms the density of activity per sq km previously suggested by other types of intensive surveys. This is an unsurprising conclusion to those of us working in Norfolk and Suffolk where such data have been recorded and incorporated into the HERs since the 1970s, but it was important to make the case for a range of English regions. Inevitably it contributes less where there is minimal arable land, for example in Worcestershire as opposed to Warwickshire (53) and most starkly in Cumbria where geographical constraints combine with a paucity of all types of Roman artefact on rural sites.

I have one reservation about the methodology, the comment (29) that ‘Regions represented by particularly high quantities of PAS, such as Norfolk and Suffolk, have not been included because of the practical difficulties of analysing such large quantities of data’, raises the question as to what will be feasible in the future when other counties build up to similar quantities of recorded objects. A test on the 19,694 Roman PAS records for Suffolk at the end of 2014 showed 1,490 Brindle findspots, with between 1 and 1,024 records in each. By comparison the largest group in his study areas was Worcestershire and Warwickshire with 7,499 records from 596 findspots. The Suffolk material would give 244 ‘large’ and 209 ‘medium’ assemblages as compared to 63 and 64. The approach provides a method for selecting Roman sites to add to the HER — essential for ensuring that it is used in the development control process — and a foundation for future work that more fully integrates the PAS data for a region with other sources. The volume is well produced and contains a useful collection of data for future comparisons. It complements the volume on Roman coinage using PAS data (P. Walton, Rethinking Roman Britian: Coinage and Archaeology (2012)) and includes similar ideas to the more detailed integrated study of early Anglo-Saxon Norfolk (M. Chester-Kadwell, Early Anglo-Saxon Communities in the Landscape of Norfolk (2009)).