Roman Samian pottery is incredibly well-researched, with much known about its shape, style, areas of production, and trade. Because of this wealth of information, visual identification is usually sufficient to be able to attribute finds to particular workshops. There are still many instances, however, in which such characterisation remains ambiguous or cannot be discerned due to poor preservation. In these circumstances, scientific analysis may provide a solution, and in this month’s ‘Science Notes’ we explore recent research by Richard Jones and Louisa Campbell, both from the University of Glasgow. Their aim was to assess whether portable X-ray fluorescence (pXRF) analysis can offer a quick and affordable way to categorise previously unidentifiable examples of Samian ware.
In the past couple of years, pXRF analysis has swiftly become a key technique for archaeologists, as it is a quick and easy way to assess a sample’s geochemical profile in the field. Similar to an XRF machine, just in portable form, pXRF analysers work by emitting an X-ray beam on to a sample. The atoms that are hit by the beam then emit X-rays in response. The energies of these X-rays are uniquely specific to each element in the sample. The instrument detects which elements are present and gives a quantitative estimate of each element.
While this method is easy to use, easy to read, and does not require any destructive sampling, pXRF’s portability does come with a cost. As Richard and Louisa acknowledge, pXRF analysers do not always have great precision, and destructive techniques will always provide much more accurate results. They have chosen to assess the possibilities of pXRF on Samian, however, because frequently non-destructive methods are necessary – especially when it comes to pottery from museum collections. Additionally, pXRF is a much quicker form of analysis that is able to be performed at scale.
To test how effective pXRF is at identifying Samian, Richard and Louisa applied it on 50 samples from South Shields Fort on Hadrian’s Wall, which acted as a maritime supply fort for the Romans. The researchers specifically chose samples which could confidently be assigned based on decorative style or stamp, and were from regions they knew had a specific chemical signature. In this way, they could compare the pXRF results with what was already known.
The results showed that the pXRF results did reliably cluster the samples into coherent chemical groups, which accurately reflected particular workshops or regions, based on stamp (where present) and style. No major discrepancies were identified. This demonstrates that, at a basic level, pXRF can successfully be used in place of visual identification when necessary.
The next step, however, was to compare each of these clusters with the published chemical data for Samian ware from the identified workshops. The image shows the way different workshops within East Gaul can be distinguished chemically from each other (clusters 1, 3, and 4 comprising examples from Rheinzabern, La Madeleine, and Lavoye respectively) and from those in Central Gaul (cluster 2 – mainly Lezoux). A few examples from South Gaul appear unclustered.
By relating the two forms of analysis, any lack of precision on the part of pXRF analysis can be accounted for. As expected, Richard and Louisa found that the element concentrations provided through pXRF were underestimated compared with the published data, but for the most part these discrepancies were within 10% of each other and could be corrected using a simple regression. Overall, while more work needs to be done to establish these baseline differences, pXRF shows great promise in its ability to make identifications of origin.
In the next stage of their project, Richard and Louisa are using pXRF to study unidentified Samian pottery from excavations at the Iron Age hill fort of Traprain Law in East Lothian (see CA 203 and 283), which is close to the Roman port at Inveresk just east of the Antonine Wall. The Samian ware here is poorly defined, and it is hoped that pXRF analysis will succeed in identifying source regions for at least some of the material – clearly demonstrating its usefulness for future research.
The full results of this study were recently published in M Hegewisch, M Daskiewicz, and G Schneider (eds), Using pXRF for the Analysis of Ancient Pottery: an expert workshop in Berlin.