More on mapping pottery

Further to my previous post on mapping broad-brush pottery distributions, I was reasonably content with the maps for prehistory and the Roman period (albeit that they had significant shortcomings in terms of temporal currency), but I was not really satisfied with the amount of data I could find for the early medieval period. One particular shortcoming was the lack of data for the earlier half of the period, for which I had been able to discover very little.

After my blogpost, Helena Hamerow found a map in a publication by Catherine Hills which included a map of earlier Anglo-Saxon pottery. A little bit more investigation showed that this was adapted from Myres 1969 (Map 1). As such, it is clearly a very old source, with no evidence included from the massive post-1990 explosion in developer-funded archaeology, but it seems to remain the most complete national map for the period. I therefore digitised the dots in this map and the Blinkhorn (2012) map referenced previously and turned them into density surfaces. With this simple task complete, it felt like my picture was becoming more useful:

em_pot_v2
Blinkhorn, Myres (labelled Hills) and Vince overlaid.

However, the fact that the Myres data is quite so out-of-date suggested we ought to find some more modern proxy for the ceramic evidence. Letty suggested we ask Toby Martin, a British Academy postdoc here at the Institute, if we could use his corpus (2011) of Anglo-Saxon furnished graves (C5th to C6th) as just such a proxy, insofar as she felt that the people buried in such a fashion should also be people who use pottery. Toby was happy to oblige and so I created a model using a density plot of his data in addition to the previously mentioned three sources. Because Toby’s material is not actually ceramics but just being used as a proxy, I gave his data a lower weighting in the model. I did the same for Vince (1993) as his zones are rather too vague in extent for my purposes here. So, essentially, all of the four sources were normalised by their maximum value (so that they varied between 0 and 1), and then combined as follows (in two steps, so I could separate out earlier and later):

(Myres + 0.5 Martin) + (Blinkhorn + 0.5 Vince)

em_pot_v3_all
Combined model inc. Martin in addition.

However, there is one very clear problem with this model and that is that all of the sources used are explicitly “Anglo-Saxon”. In other words, where are the “Britons”? Toby and I did a bit of investigating and found a very interesting PhD thesis by Imogen Wood (2011) which included three maps of Cornish pottery of the early medieval period. This was exactly what I needed to help colour at least one of the none “Anglo-Saxon” parts of the map. Rather nicely, Wood’s first map was largely temporally coincident with Myres and Martin and her second map largely temporally coincident with Blinkhorn and Vince. So, I simply expanded the model as follows (again split into two stages):

(Myres + 0.5 Martin + Wood_early) + (Blinkhorn + 0.5 Vince + Wood_late)

em_pot_v4_all
Final model also including Wood.

The final model is shown above. I also have a couple of maps which split this out into the earlier (C5th-6/7th) and later (C7/8th-9th) parts of the early medieval period, but I feel that the combined model is probably the most robust. Although some of the input data is not perfect due to its age (Myres) or its spatial vagueness (Vince), I feel that is probably the best model we can currently construct for broad brush early medieval pottery presence / absence, at least without putting in substantially more work.

Any thoughts or disagreements are more than welcome, however!

Chris Green

References

Blinkhorn, P. 2012. The Ipswich Ware Project: Ceramics, Trade and Society in Middle Saxon England. Medieval Pottery Research Group Occasional papers.

Martin, Toby F. 2011. Identity and the cruciform brooch in early Anglo-Saxon England: an investigation of style, mortuary context, and use. Unpublished PhD thesis, University of Sheffield.

Myres, J.N.L. 1969. Anglo-Saxon pottery and the settlement of England. Oxford: Clarendon Press.

Vince, A. 1993. “Forms, Functions and Manufacturing Techniques of Late Ninth- and Tenth- Century Wheelthrown Pottery in England and their Origins.” In D. Piton (ed.), Travaux du Groupe de Recherches et D’Etudes sur la Céramique dans le Nord – Pas-de-Calais; Actes du Collque D’Outreau (10 -12 Avril 1992). Numéro hors-série de Nord-Ouest Archéologie, pp.151-64.

Wood, Imogen. 2011. Changing the fabric of life in post-Roman and early medieval Cornwall: an investigation into social change through petrographic analysis. Unpublished PhD thesis: University of Exeter.

Archaeology and the Map: Critique and Practice

This is just a short advertisement for the forthcoming day conference Archaeology and the Map: Critique and Practice. The conference will take place on 23rd May 2015 at the University of Leicester. EngLaID’s Chris Green will be there, talking about his slightly crazy theories on using Heisenberg’s Uncertainty Principle as a guide to good cartographic practice. The other papers all look even better! The conference will be £6 for students and £10 for others.

arch&map

Modern or archaeological? Understanding archaeological data

Understanding how data are generated is important for any archaeological research project: is what we see a ‘real’ pattern reflecting past reality, or is the picture distorted by more ‘modern’ processes? These are important questions with far-reaching consequences for the conclusions that we, as archaeologists, can draw. The EngLaId team, in particular our GIS expert Chris Green, has explored the implications of this issue on a national level on previous occasions, both in the context of PAS finds (https://englaid.wordpress.com/2014/03/17/pas-affordances/) and sites and monuments (https://englaid.wordpress.com/2014/08/19/affordances2/). This blog explores some additional considerations in understanding regional differences in the archaeological record between case study regions.

EngLaId case study regions
EngLaId case study regions

Although significant effort is made in by archaeological professionals in England to attain certain levels of nationwide standardisation, regional differences nevertheless exist. In his previous blog post on affordances relating to the survival and recognition of archaeological sites and monuments on a nationwide scale, Chris Green identified the following affordances: first, the opportunity (or not) to undertake archaeological excavation, and second, modern land use and obscurance (https://englaid.wordpress.com/2014/08/19/affordances2/). A further level of refinement one might add to this, and one determined to a large degree by the same factors that Chris flagged up, is the question what type of archaeological investigations take place in different regions.

To give a simple example: prehistoric field systems survive better in upland areas that have seen little arable exploitation. But what is more, prehistoric field systems are also more easily recognised through large-scale landscape or aerial/remote sensing surveys than through keyhole investigations such as watching briefs – more typical of built-up areas – which are more likely to identify individual ditch sections, requiring a further level of interpretation and extrapolation before these can be ‘glued’ together into more extensive landscapes.

The EngLaId project database currently contains more than 900,000 records, and it is both beyond our scope and outside of our remit to recreate such past landscapes; instead, we aim to base our research entirely on already existing digital data. (A very good discussion of this issue, written by my colleagues Anwen Cooper and Chris Green, has just been published in the Journal of Archaeological Method and Theoryhttp://link.springer.com/article/10.1007/s10816-015-9240-4). Nonetheless, a broad awareness of differences in investigation types between the different case study regions is important. For that reason, we carried out a comparison of archaeological investigation types across the country as a whole and in the various case study areas.

Methodology

Two relevant datasets that record archaeological ‘events’ (rather than ‘monuments’) on a nationwide scale were identified, the AIP and the EI (http://archaeologydataservice.ac.uk/archives/view/304/). As the AIP was closed down not long after the EngLaId project started, and the EI is actively in the process of incorporating all AIP records, it was decided the EI was the most complete source (also see Evans 2013). Only records from 1990 onwards were incorporated, to reflect the situation since the implementation of PPG16.

As the EI incorporated a large variety of events, a simplified classification was devised into broad ‘types’ of events that were thought most likely to affect the nature of the archaeological record. These were: 1) intrusive (open area), incorporating both open area excavation and strip, map and sample investigations; 2) intrusive (keyhole), incorporating trial trenching, test pitting and watching briefs; 3) survey (geophysical/aerial/earthwork); 4) field walking/metal detecting; 5) other; whereby 1) and 3) were considered most likely to result in the identification of large-scale archaeological landscapes. Records not fitting any of these categories were deleted.

Simplification was automated, and therefore by definition broad-brush (for more information about the methodology, please contact Letty at letty.tenharkel@arch.ox.ac.uk). (As a control to test this broad-brush methodology, a more detailed characterisation of AIP investigations in selected case study areas was carried out, using the same methodology; comparison with the broad-brush EI investigation types characterisation revealed slightly different patterns, but the overall picture between case study regions remained broadly the same.) The results are plotted in the two charts below.

EngLaId simplified investigation types: 1) intrusive (open area) – brown; 2) intrusive (keyhole) – pinkish brown; 3) survey (geophysical/aerial/earthwork) – yellow; 4) field walking/metal detecting – blue.

Relative occurrence of EI investigation types per case study area in comparison to nationwide (1990-2010).
Relative occurrence of EI investigation types per case study area in comparison to nationwide (1990-2010).
Number of different EI investigation types per square km per case study area in comparison to nationwide (1990-2010).
Number of different EI investigation types per square km per case study area in comparison to nationwide (1990-2010).

The implications of the results have not been fully thought through, but the resulting patterns will be taken into account in our assessments of the archaeological patterns that emerge from our analyses of the archaeological sites and monuments in our database. Some emerging patterns are immediately obvious, such as the much higher density of fieldwork of any kind in our two southeastern case study regions – the Lea Valley and Kent – especially when compared to Cornwall, Cumbria, Devon and North Northumberland. Also of interest is the lower occurrence of recorded metaldetecting/fieldwalking events in western and northern case study areas, such as Cornwall, Cumbria, Devon, the Marches and North Northumberland, areas with large swathes of land use types that are not conducive to successful metal-detecting and which are traditionally associated with low numbers of archaeological ‘finds’. Finally, North Northumberland, the Mendips/Somerset Levels, the Marches and the Isle of Wight have the largest relative percentages of non-intrusive surveys, and therefore may produce ‘good’ data for prehistoric field systems or extensive settlement complexes.

Any feedback at this stage is very welcome – please contact us if you have any comments or questions!

References

Cooper, A. and Green, C. 2015, ‘Embracing the Complexities of ‘Big Data’
in Archaeology: the Case of the English Landscape and Identities Project’, Journal of Archaeological Method and Theory
 22 (1). DOI 10.1007/s10816-015-9240-4

Evans, T. 2013, ‘Holes in the archaeological record? A comparison of national event databases for the historic environment in England’, The Historic Environment: Policy & Practice 4: 19–34.

Sensing the Past

I have just returned from the Sensing the Past conference in Frankfurt on remote sensing in archaeology. It was the final conference for the ArchaeoLandscapes Europe project (ArcLand), which has been running for five years under the directorship of Axel Posluschny of the Roman-Germanic Commission. ArcLand’s main aim was to create a European network between users of remote sensing methods in archaeology and to encourage the wider take-up of and training in said methods. As the conference revealed, in this context remote sensing means a mixture of different non-invasive prospection methods, including LiDAR, geophysics, aerial photography, hyper-spectral imaging, and some satellite imaging.

The conference itself was very engaging intellectually. I was particularly taken with two elements: the work of Michael Doneus and his team on the latest prospection methods (particularly hyper-spectral imaging, which has the potential to be the next great breakthrough); and the work of Kevin Barton with community groups in Ireland, where low-cost aerial imaging has the potential to open up community engagement with archaeology in a country where you need a license to undertake almost any other type of archaeological work. Another fun paper was given by Gabor Bertok which included the successful usage of data collected by GPS-enabled combine harvesters for archaeological prospection (which would be of limited applicability for archaeology, but potentially immensely useful for reconstructing past river channels across wide landscapes)!

Amongst the most useful achievements of ArcLand partners has been the creation of two toolboxes for processing LiDAR data: Ralf Hesse’s LiDAR Visualization Toolbox (LiVT) and Žiga Kokalj’s Relief Visualization Toolbox (RVT). By their own description, RVT is the more user-friendly and LiVT the more customizable when undertaking analyses. I shall certainly be experimenting with them more myself.

In any event, overall the conference was very worthwhile and it is to be hoped that the brilliant work of the ArcLand network continues even as the funding comes to an end.

Chris Green