I believe that the Portable Antiquities Scheme (PAS) is a hugely important archaeological resource and this is something that I wanted to demonstrate with my PhD research. I have recently completed my PhD for which my thesis was entitled ‘Unstructured Data and Archaeology: The Use of Large
Datasets in Archaeological Research’. When I started my research, at The University of Birmingham with my supervisor Dr Roger White, I was looking to continue the research I did for my undergraduate dissertation. My dissertation studied the continuation of La Téne art into the Romano British period. I began my postgraduate research by considering a similar study of finds across the Romano British to Anglo-Saxon transition period. However, upon downloading the PAS data, it became apparent that this was a research project within itself. The vast quantities of data and remaining debates about its usefulness made it an important topic of research.
My thesis demonstrates the use of unstructured data more broadly and PAS specifically in archaeological research. The PAS is still occasionally ignored or disparaged by archaeologists and relationships with metal detectorists are not always positive across the board. There are still research projects which omit PAS data or use it in very specific ways. I wanted to test the
effectiveness of research using the PAS database on a broad scale and determine whether it could be used to go further than the studies of land use and settlement patterns often seen.
I approached the research methodology in two parts. First, an entire dataset approach to determine long-term and nationwide trends. This also produced a control dataset to determine which areas and time periods deviated from the broad trends. Whereas broad PAS trends may more accurately reflect modern recovery and reporting, deviation from these is more likely to represent genuine historical data. For example, on the whole the finds in the PAS database are concentrated in lowland areas to the south and east. Statistical analysis, in the form of a Chi-squared test, was carried out on the number of finds in each of these ‘zones’ to determine that the difference was statistically significant.
The second element to the methodology sought to go beyond distribution maps and assess the types of finds in a particular case study period and how they differ or are similar across England and Wales. The Romano British to Anglo-Saxon transition period made an excellent case study period as it was a time of great change in the archaeological record and there is still much debate as to the nature of the transition. I reduced the PAS data to a case study period of 300 – 600 AD. The fact that the PAS database is so vast means that it has greater statistical power. This was used to determine whether there is a statistically significant difference in the makeup of finds assemblages across various zones and periods. The overall pattern of finds suggested that, whereas the Roman finds assemblages differed across the boundary of highland and lowland, the Anglo-Saxon boundary was further east. The east is characterised by decorative items such as brooches and the west by more domestic type finds. Further, there appeared to be some difference in finds between the east and
west of the Roman province of Britannia Prima. One possible outlier identified was Staffordshire, which is located in the western zone but had a high proportion of Anglo-Saxon finds. Finds such as the Staffordshire Hoard possibly suggests that high status Anglo-Saxon metalwork was, in fact, used further west than currently thought.
Fragment of an Early Medieval gold and garnet cross pendant (ID: KENT-9D33EB) and gold bracteate (ID: KENT-0163F3) from Kent demonstrating the new, decorative style of material culture that dominates the assemblage in the east.
Some examples of Anglo-Saxon finds from Staffordshire: a copper alloy lozenge-shaped strip brooch (ID: WMID-054B67), a fragment of copper-alloy gilded square headed brooch (ID: WMID-922C17).
The conclusion of my research was that the PAS is clearly an incredibly useful tool in analysis of the archaeology of broad areas and time periods. Other research into the use of the PAS such as that by Katherine Robbins and Tom Brindle has focussed on the PAS being more useful for small-scale
research. For example, Katherine Robbins discusses the use of techniques such as analysis of field rewalking and specific land use. I wanted to focus on the use of the PAS on a broad, nationwide scale as the sheer volume of finds in the database appeared to lend itself uniquely to being able to assess
nationwide trends with statistical analysis.
For the case study period, the data could go some way towards answering specific cultural questions. For example, it identified several zones of material culture assemblages based on artefact type and tests of statistical significance. The data mainly provided a coarse grained overview of
regionality during the Roman to Anglo-Saxon transition. Several limitations on using the data at this scale were identified. The lack of secure dating, artefact type identification and association with other archaeological material affected the ability to conduct a more detailed analysis using the data. The more the PAS data was reduced, the greater the effect of these limitations. In effect, the greatest strengths of the PAS database appeared to be those of all large datasets. That is, its variety, velocity (the rate of new finds being added) and veracity (large datasets average out errors more). In
order to produce a more fine-grained analysis of the case study period, it was deemed that further, more structured, data would be required. One of the areas I identified for further study was the possibility of constructing a brand new methodology for using the PAS in conjunction with excavated