EMF - Previous Meetings

Chemical Fingerprinting of Pottery Meets Geochemical Mapping

Simon Chenery, Emrys Phillips and Phillip Green, British Geological Survey, Keyworth, Nottingham, NG12 5GG in conjunction with various members of Scottish archaeological community (MARG).

Researchers, whether archaeologists or geologists, are increasingly asked to think on regional rather than site specific scales. To do this, suitable tools must be developed and validated. The British Geological Survey (BGS), in collaboration with members of the Medieval Archaeological Research Group in Scotland and funded by Historic Scotland, undertook to test the geochemical and mineralogical fingerprinting of Scottish medieval redware pottery. Only the geochemical data will be discussed here. Two sets of pottery samples from 11 sites in 5 geographic regions were chemically analysed by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS) for a large number of elements (approximately 45). The first series were supplied as known sites, from 5 river valley groups i.e. G1: West Pans, Throsk, Stenhouse and Fife; G2: Elgin and Spynie; G3:Berwick; G4: Dundee and Perth; G5: Glasgow Govan and Glasgow Cathedral. These were from the ‘training data-set’. The second series were supplied to the analyst as unknowns, but from sites known to the archaeologists. These were the ‘blind data-set’.

Graphical and statistical techniques were then applied to fingerprint and assign the blind samples to known pottery sites. These included: element plots; element ratio plots; cluster analysis; principal components and factor analysis; and canonical discriminate analysis (CDA). A final site assignment was made on the basis of more than one technique but CDA proved to be the most powerful. Twenty out of 23 blind samples were correctly assigned to their sites. Two of the incorrectly assigned samples were from a site where only two training samples were available, indicating the limitations of small data sets. After disclosure, the whole data set was accurately re-modelled successfully using CDA.

At the majority of the sites investigated it was assumed that the pottery was made from local source material. It was therefore decided to compare the element concentrations in the pottery with available geochemical data. The BGS is conducting an on-going Geochemical Baseline Survey of the Environment in the UK (G-BASE), based on the collection and analysis of stream sediment samples collected on an approximately 2km2 grid. Conveniently, for this study, samples are sieved to enhance the ‘fines’/clay fraction. Key elemental concentrations in the pottery were compared to the geochemical maps and although, absolute values did not coincide, broad trends were in agreement suggesting that it may be possible to identify potential sources of material.

Until recently, the data were available only from the database or as coloured gridded maps. However, it is now possible to access the G-BASE data in a user-friendly Geographical Information Systems (GIS) environment such as ‘MAPINFO’. It is then possible to overlay G-BASE data such as sample sites and elemental concentrations onto geographical features (rivers, roads, etc), geological features (clay formations) and archaeological sites. In this way a useful synergy between geochemical mapping and archaeology should be able to be fully realised.

The authors are grateful to Historic Scotland for funding this project as Scottish Medieval Pottery: Geochemical and Mineralogical Fingerprinting. Published with the permission of the Director, British Geological Survey