As Garrison Keillor used to say of Lake Wobegon. Between Easter and catching up with my marking, the only attention I have given the project this week has been a couple of short conversations with Olaf and Mike. As usual we were talking about survey technology. What new toys can we find a use for on the project and how do we get hold of them.
The fieldwork so far has shown that gradiometer survey works well at spotting prehistoric pits and ditches. Its quick to do and gives nice consistent results. The other main geophysical technique, resistivity, has been more of a problem. It is much more time-consuming to do and, as Simon discovered last year, on this site it is harder to interpret. We have a new gadget for the resistivity meter called a multi-plexer, which speeds things up a bit by allowing you to take more readings at once. We will give this a go over the summer but generally I am coming around to the daleks’ view on this – RESISTANCE IS FUTILE – gradiometer survey is the way forward.
Even with the gradiometer we can’t survey the whole of the study area – well I suppose we could but we’d all be drawing our pensions before we finished. We need to divide it up so that we can sample a percentage of what’s there. The same principle applies to how we dig test pits and chose locations for environmental sampling. If we need to look at, for example, 10% of the study area with a particular technique how do we ensure that the 10% we sample is evenly distributed across the area and representative of all the different kinds of landscape. We have spent quite a bit of time thinking about what the best way would be to approach this problem.
Of course you could treat the whole thing as if it were in the south-western US, draw lots of straight lines at right angles to one another until you had 100 equal sized boxes and then survey ten of them in a nice neat grid. However, as a quick glance at any photo from the project will show, the Whitewell landscape is not that regular. There is a real danger that the grid approach could lead to us either sampling the same kind of feature over and over again or missing some kinds of landscape out entirely.
There is a method which has been used in the past, especially using GIS software, called predictive modelling. What you do here is compare the locations of known archaeology to other kinds of landscape features. So, for example, all the known sites might be on south-facing slopes within 10 metres of a watercourse. You would then focus your search for new sites around becks on other south-facing slopes. I’m not a big fan of this approach for two reasons. First, I think it over-simplifies the way that people and their environments co-existed in the past. The second problem is that, using predictive modelling, you are only likely to find more examples of things you already knew about.
We decided on a slightly different technique, albeit still linked to the characteristics of the landscape. Our plan is to divide the study area up into 22 smaller units and sample evenly within each one. These units would be irregular in size and shape because they will each represent one type of landscape – everything in the unit would have roughly the same geology, elevation, slope, aspect and soil type. This plateau that we will be digging again this summer is a good example of one of these units. This method is based on (i.e. shamelessly pirated from) the one used by the countryside management agencies and national parks to characterise the landscapes they manage, we will just be applying it at a much smaller scale.