Tag Archives: time slices

Batford Mill

I have a number of small surveys which remain unreported that I need to catch-up on, and so here is the first of them.

Earlier in the year, Mike Smith and I assisted Alex Thomas (University of Bristol) in undertaking a Ground Penetrating Radar survey of land lying to the north of the B653 at Batford, Harpenden, Hertfordshire (TL 148150, Fig. 1). The survey was undertaken over the weekend of 2nd/3rd April 2016. Earth resistance and magnetometry surveys had been undertaken in the area previously.

Fig. 1: Location of the site at Batford.

Fig. 1: Location of the site at Batford.

The underlying geology of the site is Lewes Nodular Chalk formation overlain in places by the Kesgrave Catchment Subgroup sand and gravel.

A Mala GPR with a 450mhz antenna was used, identical to the one CAGG borrow from SEAHA. The survey transects were at a 0.5m spacing collected in a zig-zag fashion. The survey started in the NE corner and proceeded east-west. Radar pulses were set for 0.05m intervals with a time-window of 73ns. The newer Mala systems do not allow manual selection of sample numbers which are determined by the machine, in this case 376 samples per trace.

For the amplitude slices presented here, as usual, the software system developed by Jeff Lucius and Larry Conyers was used (http://www.gpr-archaeology.com/software/). This necessitates the conversion of the Mala rd3 files into GSSI dzt files using the companion conversion program.

For this posting, the slices were 3ns in thickness starting at 3.5ns  From these, it appears that the second slice, 6.5–9.5ns represents the immediate ground surface. This agrees with the estimate of the first reflection from the individual radargrams at about 8ns as examined using RadExplorer. Beyond slice 7 (>24.5ns) the signal has completely attenuated. This means that all the usable returns lie in the band between c.6.5 and 24.5ns. This is not unusual for Hertfordshire where the clay soils do not allow for the GPR surveys to penetrate particularly deeply.

As with most of the GPR surveys reported in this blog, the numerical output from that software was turned into images using Surfer v.8. Kriging was used to interpolate the values into a 0.1 x 0.1m grid. The resulting images where then imported into Google Earth.

Six amplitude maps or `time slice’ maps were produced and are shown in Figures 2–7. The topmost map (Fig. 2) shows two strong reflections to the north-east and the south. The second map (Fig. 3), which represents the 9.5–12.5ns range, has the clearest set of features. A number of long linear features are visible, two of which I have labelled A and B. There is a odd-looking curved linear feature with two parallel lines, labelled C, into which a pair of parallel lines cuts, labelled D. Further fainter linear features can be seen, such as those at E.

slice-2

Fig. 2: Time slice 2 (6.5–9.5ns).

Time slice 3 (9.5–12.5ns).

Fig. 3: Time slice 3 (9.5–12.5ns).

The third map (Fig. 4) has fewer clear features, most of which are probably `echoes’ of the features seen in the previous map. The next three maps (Figs. 5–7) have successively fewer features in them, none of which are especially clear. By the last map, the GPR signal has largely attenuated and little can be seen. At best, we are getting a depth penetration of about a meter, probably somewhat less.

Fig. 4: Time slice 4 (12.5–15.5ns).

Fig. 4: Time slice 4 (12.5–15.5ns).

Fig. 5: Time slice 5 (15.5–18.5ns).

Fig. 5: Time slice 5 (15.5–18.5ns).

Fig. 6: Time slice 6 (18.5–21.5ns).

Fig. 6: Time slice 6 (18.5–21.5ns).

Fig. 7: Time slice 7 (21.5–24.5ns).

Fig. 7: Time slice 7 (21.5–24.5ns).

The question arises, therefore, as to what the long linear features may be. If Fig. 3A is a wall, it would be at least 35m long, and Fig. 3B would be at least 55m long. One possibility is that they represent old field boundaries. Looking at the 1898 OS map (Fig. 8), there is nothing to suggest an origin for those features. The 1799, map now in the Westminster Abbey Muniments Room does, however, show a field boundary behind some buildings to the north of the road. A crude overlay of an extract of this map (Fig. 9) on the Google Earth image with the GPR data, shows a remarkably good correlation between the field boundary and the one of the linear features (Fig. 3A).

Fig 8: The survey overlain on the 1898 OS map.

Fig 8: The survey overlain on the 1898 OS map.

Fig. 9: The 1799 Westminster Abbey map overlain on the Google Earth image.  Westminster Abbey map used with permission.

Fig. 9: The 1799 Westminster Abbey map overlain on the Google Earth image.

The origins of the curved and parallel linear features can be seen if one takes into account the location of the machine-dug test trench marked in Figure 10.  These parallel lines, only some 1.8m apart, represent areas of soil compression from the machine used in the excavation of the test trench.  Examination of one of the radargrams (Fig. 11) would seem to confirm this.  The origin of the reflections, marked with blue arrows, occurs at the very surface and is highly suggestive of compression rather than construction.

Fig. 10. Slice 2 showing the location of the machine dug test hole.

Fig. 10. Slice 2 showing the location of the machine dug test hole.

Fig. 11: Radargram showing areas of surface compression.

Fig. 11: Radargram showing areas of surface compression.

The survey results appear to be largely connected to (a) earlier agricultural use of the land in the form of hedgerows and so forth or (b) the recent impact of the excavation of the test trench.  It appears highly unlikely the GPR results indicate any sort of structure although the golden rule of ‘absence of evidence is not evidence of absence’ must be applied.  The lack of pottery or ceramic building materials on the surface makes it unlikely that a building is indicated.

Perhaps I should have saved a more exciting post for #100!

Insane?

The sun was shining, we have a GPR and a Park to play in, so why not? Mike, Peter and I headed out to Verulamium Park to do a little more GPR, although with a difference. We used a 25cm spacing between lines instead of 50cm. We managed a 40m x 60m block.  Without taking into account turning around, or getting to the spot and back again, that is 9.6km of walking, or just under six miles in old money. The results were great, however.  Later, I will do a posting comparing the 0.5m spacing with 0.25m to see if the extra 4.8km was worth it.

First things first.  Where in the Park were we?  I am especially interested in Insula XXIV simply because the Wheeler’s did not dig anything there and we have very good mag results.

Location of the mag grids in the northern half of Insula XXIV.

Location of the mag grids in the northern half of Insula XXIV.

Here is a more detailed view of the mag results.

Detail of the magnetometry results from the northern half of Insula XXIV.

Detail of the magnetometry results from the northern half of Insula XXIV.

The stark black-and-white feature in the NE corner is a practice cricket wicket made of concrete and green carpeting. It overlies the Roman road.  The diagonal feature in the NW corner is the other Roman road, and at least two buildings show clearly as white lines against a dark background.  The western 40m square was surveyed by Ralph Potter at 0.5m intervals.  The middle square and the western 20m of the eastern square were surveyed yesterday.

I used Larry Conyer’s program to slice the data we collected.  The slices start at 8ns and go down in 4ns thick pieces.  Ground surface in the Mala radargrams is usually at about 7ns and so the first slice is very much the top surface of the field.

Slice 1: 8 to 12ns.

Slice 1: 8 to 12ns.

The cricket wicket can be very clearly seen in the top-right corner.  The change across the plot at about 27m is where we went for lunch.  But what on Earth is the big circle?  Here is a clue…

GPR in the Park.

GPR in the Park.

Yup, it is the centre circle of the soccer pitch.  The second slice reveals the striping we have often seen in the Park.

Slice 2: 12 to 16ns.

Slice 2: 12 to 16ns.

I am guessing these are cultivation marks before the land was turned into the Park in the 1930s.  How old they are I am unsure, but could they be the residue of cultivation ‘strips’ as seen at Ashwell?

In the third slice, we begin to see the Roman archaeology.

Slice 3: 16 to 20ns.

Slice 3: 16 to 20ns.

Notice how there is a white line running through the lower building cutting the black lines representing internal walls.  This is a wall which has been partly robbed.  We also appear to have some surviving floors as shown by the larger black areas in the building.

Slice 4: 20 to 24ns.

Slice 4: 20 to 24ns.

In the fourth slice down we can now see the road running across the top-right hand corner in a similar, but not identical position to the cricket wicket.  I wonder if we might be seeing evidence of a hypocaust in the middle of the lower building?

Slice 5: 24 to 28ns.

Slice 5: 24 to 28ns.

In slice five we can see the outer walls of the building are much fainter suggesting shallower foundations, but the main wall which was only a light line previous is now showing more clearly.  This is presumably what is left at the bottom of the robbed wall.  There is also a building along the road which was not visible in the higher time slices.

Slice 6: 28 to 32ns.

Slice 6: 28 to 32ns.

Slice six is very black-and-white.  This is because the radar signal is now very weak and the slices are either showing something or nothing.

All in all, a very nice set of results.  As always, these are my rather rough-and-ready time slices.  With some experimentation and practice they could be better.

Here are Ralph’s grid square (processed by Mike Langton of Mala) and our new one roughly plotted together in Google Earth.

Ralph's and our blocks of GPR data plotted together using Google Earth.

Ralph’s and our blocks of GPR data plotted together using Google Earth.

I think it would be worth surveying the whole of Insula XXIV if we can!