A weekend at Willington

Firstly, apologies to everyone for the delays in posting. The day job got in the way, but I am hoping to catch-up with several posts over the next week or so.

In late April, we spent a pleasant weekend surveying at Willington, just south of Bedford, for the local group that have been investigating the medieval complex, of which the dovecote and stable block survive.  Both listed monuments and are looked-after by the National Trust. We had to obtain a Section 42 licence from Historic England to undertake the work.  The site has been surveyed once before: an earth resistance (“resistivity”) survey was undertaken in 1997.

Peter Alley using the mag cart at Willington with the dovecot in the background.

Fig. 1: Peter Alley using the mag cart at Willington with the dovecot in the background.

The area available for survey was not very large, a mere seven of our 40x40m grids covered the area we needed to investigate and we completed the mag survey in a day.  Historic sites are, however, often poor candidates for magnetometry due to the large quantities of ferrous material, such as nails, that get left behind.  We decided, therefore, to undertake a GPR survey as well.  Many thanks, once more, to SEAHA for lending us their Mala system.  The GPR survey took us two days to complete a slightly smaller area.  We also undertook some survey inside the buildings.  That will be the subject of another post.

Results of the magnetometry survey at Willington.

Fig.2: Results of the magnetometry survey at Willington.

The first impression of the survey, shown above, is how “noisy” it is.  The large quantities of ferrous metal fragments, and possibly even relatively magnetic bricks, along with two roads and some services, make for a very messy result.

Interpretation of some of the main features showing in the survey.

Fig. 3: Interpretation of some of the main features showing in the survey.

The northern road marked on the image above is the modern drive to the buildings to the SW. The older road which curves around the stables dates back at least as far as the Ordnance Survey maps in the 1880s, and was still in use in the 1970s as it is marked on the maps of that date.  The one feature our survey shows, which does not appear on the early OS maps from the 1880s and 1900s, is that the older road forks to the SW.  This split, however, shows clearly on the Google Earth image, which dates from 2009, shown below.

Google Earth image showing the fork in the old road as a parch mark.

Fig. 4: Google Earth image showing the fork in the old road as a parch mark.

This extra road does, however, show on the maps of the 1920s and must have been built sometime in the early 20th century.

As the survey progressed I only surveyed in one of the manhole covers, which is marked in Fig. 3.  One can see how clearly these show in magnetic data.  The large and very noisy area represents a pond which is shown on many historic maps including the OS maps from the 1920s, but has been filled-in by the 1970s map.  A more thorough investigation of the map evidence that I was able to do online might tie down these dates more closely.

In general, it is hard to see any details of the buildings we know existed from the 1779 estate map.  I have re-drawn it for comparison here (Fig. 5).

Fig 5: The 1779 Estate map redrawn.

Fig 5: The 1779 Estate map redrawn.

The GPR survey took two days and we covered two 74m long strips.

Fig. 6: the GPR starting the survey near the stable block.

Fig. 6: the GPR starting the survey near the stable block.

The GPR results are affected by changes in ground and surface water.  It can be quite difficult to get surveys split over several days to match, and this is the case here.  I have created time-slices 4 ns thick.  It would be possible to spend quite some time adjusting the slices and the contrast to make the edges match more cleanly than is shown here.

GPR time slice 1: 8 to 12 ns.

Fig 7: GPR time slice 1: 8 to 12 ns.

The topmost time slice (Fig. 7), merely shows the two road surfaces, the modern one to the north and the older one to the south.

GPR time slice 2: 12 to 16 ns.

Fig. 8: GPR time slice 2: 12 to 16 ns.

The second slice (Fig. 8) shows the southernmost road more clearly.  The lighter band to the SE is partly where we stopped for lunch:  as the sun dries the upper layers we can see this in the GPR results.  But this doesn’t explain why there is a change back.  I doubt this is archaeological as it follows the line of the GPR transects.  We are, however, probably still quite close to the surface at this point.

GPR time slice 3: 16 to 20 ns.

Fig. 9: GPR time slice 3: 16 to 20 ns.

The third time slice (Fig. 9) is starting to get down a bit as we are beginning to see through the road. The area to the east of the stable block is still remarkably noisy.  If we compare this image with the 1779 map (Fig. 5) we can see that “noisy” parts of the time slices are roughly where there are buildings, and the quiet parts of the slices are where we have empty spaces on the map. Sadly, however, we cannot really see any buildings clearly in the GPR data.  There is a broad stronger response (shown in dark tones) heading towards the entrance of the stables (may be a track?) and a linear feature running away from that at roughly a right angle which is much narrower.  Maybe this is an earlier wall?

GPR time slice 4: 20 to 24 ns.

Fig. 10: GPR time slice 4: 20 to 24 ns.

The fourth timeslice (Fig. 10) shows a little bit more towards the northern corner of the survey.  The thin line is almost certainly a modern service.  There is, however, a broader linear feature running SW towards the stables.  Perhaps this is another track?

GPR time slice 5: 24 to 28 ns.

Fig. 11: GPR time slice 5: 24 to 28 ns.

The last time slice (Fig. 11) shows a scatter of strong reflections but generally very little.  The signal has “attenuated”, and we are getting little sensible information.  Some parts of the road is still showing as an “echo” of the upper features.  One useful thing, however, is the strong linear feature heading towards the southern corner of the stables.  This is probably a modern service but it does show clearly on the radargram (Fig. 12).

Screen grab of one radargram in RadExplorer.  the strong reflection in Fig. 11 is indicated by the red arrow.

Fig. 12: Screen grab of one radargram in RadExplorer. the strong reflection in Fig. 11 is indicated by the red arrow.

Why do we care?  This strong reflection can be used for hyperbola matching, i.e., drawing a curve on-screen which matches that in the data indicated by the red arrow in Fig. 12.  This process provides a speed for the radar waves through the soil at this site.  I tried several spots and radargrams, and a speed of about 8.5 to 9cm per ns seems about right.  Using RadExplorer I can then see the strong reflector (?pipe) is about 60cm deep.  Assuming the top of our top slice is about ground level, each of our time slices is about 12cm deep.

A maximum depth of only about 60cm is not all that much for GPR, and we must presume that the soils are not that conducive to this sort of survey.  Having said that, I would be surprised if the top levels of the archaeology were any deeper than 60cm.

Unfortunately, the surveys have not provided the impressive and exciting results that we have obtained at some sites which is a shame.  Both the magnetometry and the GPR surveys show noisy areas which match well with the areas of buildings shown in the 1779 Estate Map, but we do not have any clear indications of surviving building remains, although there are a couple of possible track ways.

 

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s