Seminary XXV: why your Anglo-Saxon settlement maps need some rethinking

I have once before here mentioned my, well, friend is fair I think, Mary Chester-Kadwell, of whose research I am something of a fan. She works on archaeological landscapes in Anglo-Saxon East Anglia, but her approach is very technology-intensive and gets us a bit further than Myres’s distribution maps, and more towards what the context of our archaeological material is and how that explains some of what we find. This is in many ways the basic groundwork of archaeological interpretation, much like the basic `consider the author’ level of textual analysis, but it’s much harder to do in archaeology because you need so much context. Mary’s work draws on vast piles of records in archaeological archives and also, importantly, the ever-increasing body of metal-detector finds. Now there are arguments about the regulation, or lack of it, in British law about metal-detecting, and it is unquestionable that much more is found than is reported, but all the same the extra evidence we have because of this loose policy is undeniable. And on May Day Mary was at the McDonald Institute in Cambridge speaking to the Graduate Archaeological Seminar to the title: “The landscape of early Anglo-Saxon Norfolk: cemeteries, settlements and metal-detected finds”.

Mary is one of a number of scholars looking at approaches like this that involve a fairly serious reevaluation of our evidence. She is currently working on various forms of publication of her work, so rather than tell you what’s in it I’ll just give a few examples of the sorts of concerns she raises and therefore why I think her stuff is important. Her mapping is very dense: whereas many earlier interpretations of Anglo-Saxon settlement patterns tended to correlate only a few factors, settlement location against river access, against Bronze Age or Roman sites, against soil types, and so on, Mary is bravely trying to get all these things and more into play at once, and it is educational. In particular she was showing that we could, with such techniques, try and test some of the common assumptions about Anglo-Saxon settlement in the area, such as (i) that it’s usually riverside, (ii) that cemeteries often overlook significant places or routes, (iii) that the Anglo-Saxons favoured light well-drained soils because of not having the heavy plough, and (iv) that cemeteries are often placed near previous funerary monuments like barrows. Rather than just mapping the two things against each other and going, “Ta-dah! match!”, however, Mary computes baselines for a average distribution of, say, distances from rivers that are possible in Norfolk, and then applies the Kolmogorov-Smirnoff test to see how significant the data’s deviation from that expectation is. And in fact, in that instance, she finds that there is a stronger-than-expected tendency for settlement sites to be within 100 m of a river, for inhumation cemeteries to be about 200 m of a river, and cremation ones about 300 m from one, but that these are only trends and are easily countered with examples that form the tail of the distribution curves. So in that instance our understanding needs to be more complex. And this is the next step we have to be taking with evidence like this to start understanding what was really happening on the ground.

This sort of caution also allows one to start really facing the biases of the evidence. Example one: in archaeological digs of cemeteries, the metalwork that comes up is about as much iron as copper alloy. Metal detectorists don’t search for iron, though, so almost all of what they come up with is copper or precious metal, which means that sites only found by them look very different and perhaps shouldn’t. Example two: a very large proportion of Anglo-Saxon sites in the area are associated, or at least noticeably near, a Roman-period site. But there are shedloads more Roman sites known than Anglo-Saxon ones so that probably isn’t significant; it would be odder if they were not so associated just on probability. Example three: it is certainly true that a great proportion of Anglo-Saxon sites excavated have been on light well-drained soil. It is also however true that a vast proportion of all sites dug have been on such soil too, and there are plenty, if fewer, sites known from clay areas too. So we have to ask if really, that correlation isn’t more to do with where is easy to dig than where the Anglo-Saxons actually liked to live. Example four: there is a strong correlation between Anglo-Saxon mortuary sites that have been excavated and older barrows or barrow-like formations. But this correlation doesn’t exist with metal-detected finds, which suggests that the archaeologists are digging especially where there are barrows (as you’d expect), that the detectorists are avoiding such sites (which, since they’re not flat, I could understand) or both (which is probably the truth of it).

It all sounds terribly revisionist and destructive when I put it like that, I suppose, but firstly there is the usual argument for revisionism in such contexts, that it stops us saying things that are basically just plain wrong, and secondly there is the much more powerful argument that by trying to understand the complexity of the societies we’re looking at in all its horribly messy glory, using the sort of dense mapping techniques and data collection that Mary has done, we are likely to get further than we ever could by over-simplifying out most of the information. This way, I reckon, lies progress of a sort we previously couldn’t have made.