Cherry, Glebe and Sands

Regular readers of this blog will know that over the years we have surveyed both within and outside the walls of the Roman “small town” of Durobrivae near Peterborough. Before Christmas we completed the Boat Field just outside the SE gate of the town. Historic England asked if we could survey some transects across three fields south of the A1, just to the west of the town. Figure 1 shows the areas we have surveyed with magnetometry survey.

Figure 1: Areas surveyed by CAGG at Durobrivae.

The sample survey transects are across three fields: Cherry Holt, The Glebe and Glebe Sands (Fig. 2).

Figure 2: The three new survey transects.

The weather was not very pleasant. There was a cold wind blowing in from the east with squalls of rain to add to the delight (Figure 3). We did, however, manage to complete the three transects with time to spare. We completed 4.28ha of mag survey in the two days and had time to take magnetic susceptibility readings over the three transects in the afternoon of the second day. We swapped who was pushing the mag every 30m or so just so that the people holding the poles could warm up!

Figure 3: Ruth pushing the Sensys magnetometry cart towards Jim holding one of our target poles.

The first field we surveyed was Cherry Holt to the west. We completed a 340m x 50m long transect across the field by mid-afternoon. We chose the line of the transect in the hopes of detecting the double-ditched feature that can been seen in the Google Earth image, and has been seen in various Historic England air photos (Fig. 4, norrthern light blue arrow).

Figure 4: Magnetometry survey in Cherry Holt. Arrows explained in main text.

The first thing to notice in the magnetometry results are the east-west stripes and the north-south striations. The east-west stripes (the dark blue arrows indicate just one) are land drains. They show very clearly in the Glebe. The north-south striations are cultivation marks probably from when the field was last ploughed. The most obvious archaeological features are the double ditches indicated by the red arrow. I’ll discuss these in relation to the next transect.

The line of rather ferrous looking responses indicated by the yellow arrows look like an old fence line. To check this I downloaded the 19th century OS maps from Edina. After various failed attempts to get Google Earth into QGIS or the map data into Google Earth, I resorted to quickly digitizing the key features in QGIS and then importing those into Google Earth. Figure 5 shows the result. The red lines are the field boundaries and as you can see, one lines up perfectly with the feature in the mag data (Fig. 6). Hurrah! I have also plotted the “Roman house”, “Roman villa” and “iron works” from the OS map (Fig. 5).

Figure 5: field boundaries (in red), the footpath and archaeological sites from the 19th century OS maps overlain on Google Earth.

Figure 6: Detail of the mag survey in Cherry Holt. Arrows explained in the text.

The rather strange shaped feature in the southern half of the transect (Fig. 4, pink arrow, Fig. 6 red arrow) is hard to interpret. Due to its odd shape I’m guessing it is likely to be natural, but that is just a guess. The blue arrows in Figure 6 show the line of two linear features, maybe early field boundaries? They do not occur on the OS map so are earlier than the first surveys in the 19th century.

One entertaining observation regards the footpath shown in Figure 5. On the northern edge of Cherry Holt is a kissing gate with a bridge across the drainage ditch which runs alongside the A1. Judging by the brambles on the little footbridge I doubt anyone has used it in a while. Besides, one would have to be mad to try and cross the A1 on foot at this point (Fig. 7)!

Figure 7: the kissing gate at the north edge of Cherry Holt.

In Figure 6 I have indicated three linear features with yellow arrows. We had originally thought these might be roadside ditches, but they do not show up at all in the mag data. The survey in the Glebe to the west suggests that at least some of them have a more prosaic origin.

The aerial images of The Glebe (Fig. 4) show lots of features including the corner of the double-ditched feature and an enclosure towards the south. We placed the transect (Fig. 4, red line) to catch these features. Figure 8 shows the results.

Figure 8: mag survey in The Glebe. Coloured arrows explained in the main text.

The red arrows show the corner of the double-ditched enclosure. This feature was partially excavated when the A1 was widened. Stephen Upex is currently working on this legacy excavation and states that it contains Flavian pottery. The enclosure looks rather military even though it doesn’t have the classic “playing card” corner shape.

The field is covered in field drains. The yellow arrows show just one of them. The pink arrows indicate the southerly drain at which point the others run at right angles to it. This land drain lines-up with one of the linears shown with a yellow arrow in Figure 6. This suggests that our “road” might be connected to the drainage instead.

The blue and orange arrows indicate two further complexes of ditches. We have only just clipped the one indicated by the blue arrow but the crop marks suggest there is more to the west. The southern one matches the clear crop marks seen in the Google Earth image (southern light blue arrow in Fig. 4). Although this is near the “Roman villa” marked on the early OS maps, my gut feeling is that this is earlier, possibly late Iron Age. The straight ditch marked by the green arrow might be a boundary ditch for the villa if it lies slightly to the east of our transect.

From our survey and the aerial images, this field appears to be very busy archaeologically!

The final field to the east was Glebe Sands (Fig. 9). The transect went up a slope and then at the north end was a plateau. This can be seen clearly in the lidar image (Figure 10).

Figure 9: Magnetometry survey in Glebe Sands. Red line: old field boundary.

Figure 10: Lidar data for the area.

The flatter plateau marks the edge of the sands and gravels with alluvium downslope to the south. There are quarry pits along the edge of the plateau which is possibly the origin of the fieldname: Glebe Sands. The “iron working sites” marked on the 19th century OS map also lie on the plateau edge. The rather amorphous magnetic features are probably the remains of these quarry pits (Fig. 9). At the southern edge of our transect are some faint traces of field drains running NNW to SSE, one of which is indicated with the white arrow.

The northern end of our transect is very busy with a series of linear features, almost certainly ditches, and two circular features, one 10m in diameter and one 22m in diameter. These may be barrows, but it is curious they are cutting one another. Perhaps one is prehistoric and the other Roman or Saxon? Whatever the precise dating of these features it is clear that the gravel terrace attracted substantial multiperiod settlement.

Now that we have our own magnetic susceptibility meter, I am starting to collect some data at any site where we undertake a magnetometry survey. Over time I’m looking to build-up a database of readings compared to geology and our results. Ruth, Jim and I spent the last part of the afternoon collecting readings at roughly 25m intervals. In Figure 11 I have represented the results as shaded circles as the spacing seems too big to interpolate a continuous surface.

Figure 11: the magnetic susceptibility survey results.

The results show some interesting contrasts. Cherry Holt has a more even pattern than the two other fields, and generally low readings. Geologically, according to the British Geological Society’s viewer, the field is split between river gravel terraces 1 and 2. The eagle-eyed of you would have noticed that the plot in Figure 4 is cropped to +/- 3nT whereas the Glebe and Glebe Sands images are cropped to +/- 4nT. Apart from the Roman double ditches at the north end, the features are relatively faint. In The Glebe, the readings at the north end of the field where the corner of the Roman ditches can be seen is high as would be expected where there is dense human occupation. The readings drop, however, as one gets closer to the south and the alluvium. The enclosure to the south of the transect does not show as a particularly high area of readings. In Glebe Sands, the southern area on the alluvium is very low, with the highest readings of all on the river terrace where all the activity is and the “iron workings”.

We have, therefore, two processes at work. The soils which develop on the alluvium have a lower magnetic susceptibility to those which develop on the river terraces, especially terrace 1. The magnetic susceptibility readings are then enhanced by anthropogenic factors where there is evidence of occupation.

Just to finish off, Figure 12 shows Jim with our new Sensys which conveniently has slots for carrying the target poles across the field.

Figure 12: Jim and the Sensys.

Messing about in boats

Well, not actually in boats, but in the boat field at Durobrivae. Those of you with long memories may recall that we did some work at Durobrivae in 2016 and 2017 (if you use the drop-down box on the right of this screen you can filter the posts for Durobrivae and see the previous posts.) The aim of the 2016 survey was mainly to test which of the three main survey techniques we had available to us would work on this site. The answer was: all of them! A short article about the results was published in ISAP News (Issue 52, November 2017, pp. 5-9).

We then went back to extend the surveys in December 2017, particularly in the area of the mysterious ‘mound’. Although we got some good data, the mag died (yet again) and not only did we have to stop surveying early, but we lost the grids we had collected that day. Due to the excellent results of what we did manage to survey, Stephen Upex was able to obtain a grant to pay for a towed magnetometry survey of the entire town. In 2019 Stephen teamed-up with Peter Guest (then of Cardiff University) to undertake some trial excavations, including across the temple we had surveyed in 2016.

The field to the SE of the Roman town, just outside the town walls and next to the lay-by off the A1, is Boat Field. Along its northern edge runs the River Nene (with lots of boats moored-up). Ermine Street, the main Roman road from London to York, runs through one corner of the field. Until recently, nothing much showed-up in aerial photographs of this field apart from Ermine Street. The more recent photographs, however, showed that there were indeed features in this field. That isn’t really surprising considering the location just outside the town walls, and the fact that there are several areas of “suburbs” known in and around Durobrivae. Ruth, stalwart member of CAGG, has a long-standing interest in this site and was keen to try our new magnetometer in this field. Given the new machine, the distance from our usual haunts, and the time of year we decided to keep to a small team of just three, myself, Ruth and Jim. The first day of survey was in late September, and we managed four visits in total, finishing the field yesterday before lunch. Completing 6.73ha of mag survey in 3½ relatively short days shows how efficient the new system can be. As we finished Boat Field before lunch yesterday, we did one more grid over the mound so that we could compare the two surveys. Figure 1 shows the 2016, 2017 and 2022 magnetometry surveys.

Figure 1: the 2016, 2017 and 2022 magnetometry surveys.

For the earlier surveys, please see the earlier posts. In this post I am going to concentrate on the new survey in Boat Field, and comparing the old and new magnetometry surveys.

Figure 2 shows the survey of Boat Field.

Figure 2: The Boat Field survey.

As can be seen, there is a great deal going on. The faint stripy-ness in the data is most likely to be old plough scars. The field is covered in a network of ditches with a large number of other features. Figure 3 zooms in on the SE end of the field.

Figure 3: the SE end of Boat Field.

In Figure 3 I have indicated various features with coloured arrows. First of all, the yellow arrow indicates two concentric circles. These are very likely prehistoric, maybe part of a ploughed-out burial mound. The field on the other side of the A1 has quite a few circular features seen in aerial photographs. I have posted Fig. 4 previously, but repost it here.

Figure 4: Oblique aerial photograph of the field to the south of the town showing the Roman suburbs and earlier prehistoric circular features. Photograph courtesy of Stephen Upex.

There are a great many linear magnetic features in Boat Field, most of which are probably ditches. There are two pairs of parallel ditches, running roughly at right angles to each other which might be trackways. These are indicated with blue arrows. As well as the linear features, there are large numbers of “blobby” magnetic features (technical term that…), many of which are quite large. I have indicated one with the green arrow which is over 4m across. Most of these are probably pits although some are quite strongly magnetic. Would they have needed wells so close to the river? A couple of the blobby features are very magnetic and might be pottery kilns. I have indicated these with the red arrows.

Figure 5: the NW end of Boat Field.

Figure 5 shows the NW end of the field. The possible trackways continue as shown by the blue arrows. Ermine Street, the main road from London to York, is indicated with the red arrows. It actually shows less well in the mag data than on the ground where it is a substantial bank. The yellow arrows indicate the line of the ditch outside the town walls. The survey by Durham showed that their was a double ditch to the defenses and we seem to have the outer one here.

There are a number of negative magnetic features which show as lighter lines. Things that are magnetic in the soil have a positive and a negative, a north and south pole if you like. That is why magnetic features have both a dark (positive) and light (negative) elements. The average background level of magnetism is shown in mid grey. Where there is something in the soil which is not magnetic — like a wall foundation made of non-magnetic materials — it will show as a lighter line as the foundations have displaced the slightly magnetic topsoil. In this survey we have some linear “light line” features which might be walls but they cut across the track. They might be land drains leading into the Nene? I have indicated one with a green arrow. There is also a curious circular “light line” feature also marked with a green arrow. Ideally, we should run the GPR or the Earth Resistance meter over those features as those will show foundations more clearly.

Having completed Boat Field before lunch, we decided to re-do the survey over the mound (Figure 6) so that we could compare the Foerster survey with the Sensys. Two things to keep in mind: (a) the Foerster’s odometer was playing up and (b) part of the site has been excavated in 2019. Figure 7 shows the earlier Earth Resistance survey overlain with the contours from a dGPS topographic survey (re-posted from an earlier blog entry), Figure 8 shows the old magnetic survey and Figure 9 the new survey.

Figure 6: The mist shows the location of the “tumulus” beautifully.

Figure 7: contours overlain on the Earth Resistance data.

Figure 8: The 2017 magnetometer survey.

Figure 9: the new 2022 survey.

Thankfully, the old and new surveys look pretty similar! There are, however, a few differences. The most noticeable one is that the building on the southern edge of the mound is much clearer in the new survey than the old one. This may be because of two factors. Firstly, one of the 2019 excavation trenches cut across this building so some of the magnetic overburden will have been removed. Secondly, the odometer problem with the Foerster in 2017 caused a certain amount of stagger error. Look at the strong magnetic linear feature running away from the mound in Fig. 8 and you can see the distinctive “saw tooth” effect caused by this. The second problem is that there is a slight north-south shift in the results. I do not know (yet) if this is my conversion of the GPS coordinates from the original survey (using a website) or some other issue with the new data, but I will investigate!

The new machine (Figure 10), however, is clearly a success. Ruth’s S42 licence lasts for another few months so we are planning to return and undertake some more Earth Resistance and/or GPR surveys. Watch this space!

Figure 10: the Sensys in action in Boat Field.

GPR at Gorhambury 2022

Regular readers of the blog might have been wondering what is happening this year. I can only apologise that I haven’t been able to write my daily update during the 2022 season. I also have a backlog of other sites to write-up which will get posted when I have the chance. In this post I am going to concentrate on the GPR survey. We have some exciting news on the magnetometry front too which I will save for another day.

Last season (see previous posts for August 2021) we surveyed roughly half of the field to the north of the drive at Gorhambury. This field is called “Blackgrounds” although the team usually call it “the macellum field” after the building excavated there in 1938 by Miss K. M. Richardson. We had completed the magnetometry survey of this field in 2016 (Fig. 1).

Figure 1: Magnetometry survey in Blackgrounds.

The magnetometry survey revealed the course of Watling Street very clearly. Some buildings show extremely clearly, and there are many indications of walls seen in the results as white lines. Remembering that something magnetic has a positive and negative pole (north and south if you like), why do walls show as negative? Mid-grey in these figures represents “neutral”, or the average background value for magnetism. Topsoil and archaeological sediments are often more magnetic than subsoil which is why things like the aqueduct show so clearly. A flint wall is not magnetic, but it occupies a space in the slightly more magnetic surrounding deposits, and therefore shows as a negative reading. If you zoom in to the image you can start picking out many walls, and some make clear rooms and buildings.

Last summer we completed about half of this field using Ground Penetrating Radar (GPR). We use the UCL Institute of Archaeology’s Mala GX with a 450mhz antenna. Many thanks to the IoA for allowing the use of this equipment! We mainly collect data in 40x40m grid squares at 0.5m transect intervals. The radar sends a pulse into the ground roughly every 3cm (Figure 2).

Figure 2: Nigel (NHAS / NCAG / WAS) using the GPR.

The GPR results last summer were really nice, but left us hanging as some really fascinating looking features were starting to show (see “A happy ending”). Well, this year’s survey has not disappointed! Up until now, the survey results have been great but mostly we have found the sorts of things one would expect in the town: roads, small buildings, big buildings, pits, kilns and ditches. This season’s results have had us asking “what is THAT building?” Why? Because they are huge and of unexpected forms. Figure 3 shows all the results up to the end of August 24th.

Figure 3: the GPR results up to the end of 24th August 2022.

Figure 3 shows the Insula XXXVII Building 1 very clearly. This building has been long known and can be seen very clearly as a parch mark on Google Earth. The line of the 1955 ditch — the first century boundary of the town which went out of use, according to Frere, in c.125 — doesn’t show very clearly in the GPR data but is important. Why? Well, the buildings to the NW of the ditch (i.e., outside the early town) all look like the sort of domestic / small business properties one would expect and have seen elsewhere in the town. The buildings to the SE of the ditch, and north of Watling Street, have a very different feel to them.

Figure 4: Block 1.

Figure 4 shows what I am going to call for the purposes of this blog post “block 1”. Here, we seem to have traces of some buildings around an open courtyard. The building in the north-east corner looks well preserved, the others are harder to make out (further data processing might help). The southern edge along Watling Street appears empty of buildings apart from one small on in the SE corner.

Figure 5: Block 2.

Figure 5 shows block 2. I have used an “overlay” analysis to try and improve the visibility of the building. On the NE side of the building we have a range 80m long with a 60m colonnade. It has about 20 columns. To give you a sense of scale, the nave of St Albans Abbey is 85m long. The colonnade looks out over the River Ver which runs to the north. The building seems to be subdivided into many smaller rooms. At either end is a projecting wing, the one on the east end having an apse.

Behind this main range there appears to be a corridor, then an open space, and then another very long building. This one has a central protruding room and two wings, but does not seem to have the multitude of internal rooms of the northern range. Behind this is what might be another courtyard with more rooms in the southern corner.

This building begs many questions. Is it all one phase? Are there connecting rooms or corridors (there are hints)? Do some rooms have floors surviving? Some of these questions can, hopefully, be answered with a programme of detailed data analysis: producing more time slices using different parameters and filters as well as looking at the all-important radargrams.

The biggest question is, however, what is this building? I’ve shown the plan to a few people and no-one has come back to me with an unequivocal answer. The word “palace” has been mentioned by several. If it is a “palace”, to whom does it belong? There are definitely more questions than answers at the moment.

Figure 6: Block 3.

Figure 6 shows block 3. Starting from the NE corner, we have faint hints of a road surface which appears to have gone out of use and been built over. Then there is a large structure, again with an apse. This might be an upmarket house? Behind that is the most curious area. There appears to be a courtyard building with rooms on the inside on the southern side, but rooms on the outside on the northern edge. Aligned SW-NE is an aisled building with quite large foundations. This building is roughly 45m long and 16m wide. It is at a slightly different alignment to the buildings in the next block and appears to cut by then suggesting it is earlier and went out of use. In the courtyard, perhaps attached to the aisled building are two large rooms. Further out into the courtyard, and quite deep in the radargrams, is another building only visible as fragments but with large buttresses.

Figure 7: block 4.

Figure 7 shows block 4 which is bounded on its SW side by Watling Street (Niblett and Thompson Street 14) and on its SE by Street 24. Its NW side is bounded by block 3 discussed above and the NE side faces the River Ver. Starting on the SW side, we can see two, probably, buildings facing onto Watling Street. Then behind them there appears to be a series of buildings within another courtyard. It is quite hard to make out what is inside and what is outside, but it certainly appears as a coherent, and almost symmetrical arrangement of rooms and buildings with that area. In the NE corner, just outside “courtyard” is a large single room building some 10m by 13.5m in size. Projecting from this at an odd angle is another feature, maybe a drain? Street 24 is an important road as it heads to the so-called “theatre gate” and there is some evidence of the road crossing the Ver at this point and going into the field opposite, but that is the topic for another blog post.

At the bottom of Figure 7 there is another building on the other side of Street 24. This building is the “macellum” excavated in 1938 which I will discuss in another post.

Surveying at Verulamium has always been very satisfying, and we have got some excellent results since the first season in 2013. The latest results will, however, generate a lot of discussion as to what these buildings are, and what they tell us about this fascinating Roman city.

Many thanks to the team for all their work this year especially Ruth, Pauline, Rhian, Jim, Mike, Nigel, Graeme and John. More updates soon!

Looking for the Holy Well

The Leighton Buzzard and District Archaeological and Historical Society (LBDAHS) having been trying to locate the “Holy Well” at Linslade since 2018. The well was very popular in the medieval period, especially in the 13th century. The waters reputedly had healing properties. The well is marked on old maps such as the first edition OS map, but the accuracy of the location is uncertain. The Society have managed to locate a small cottage of 18th or early 19th century date on the banks of the canal which they have partially excavated (Figure 1). In 1915 local historian J. G. Gurney mentioned an old cottage and garden that lay “on the exact site of the Holy Well”, and included a sketch map. This is certainly the cottage which is being excavated.

Figure 1: the post-medieval cottage excavation.

As well as a variety of post-medieval finds, some of the features have included some sherds of medieval pottery of the right date, and one trench a little to the north of the cottage contained a Romano-British pottery sherd.

From a geophysics point of view the site is quite difficult. The main area of interest is the strip along the edge of the canal which has thick riverside vegetation. The excavation trenches regularly fill with water. The site is, however, on the edge of a steep slope down the canal and so gets drier quite quickly as one moves upslope. What were we hoping to find? One suggestion was that because the well was so popular there might be some form of path or track to the well. If we could detect that, perhaps this would give a clue as to its location.

Pauline, stalwart member of both LBDAHS and CAGG, arranged for us to undertake an exploratory Earth Resistance survey at the site. Due to covid and other commitments this took a bit of time to organise, but finally we managed to earmark a couple of days on site (Fig. 2).

Figure 2: the Earth Resistance survey underway.

The survey was somewhat jinxed! On the first day, we couldn’t get a reading at all to begin with. I spent quite some time checking the cables and connections and generally making sure all seemed OK to no avail. In desperation I reset the machine to its “factory” defaults and hey presto, all started to work again. Then, as we were working, one of the welds on the frame broke. Luckily we could keep on working by holding the frame on the sides, and the next day a temporary repair was effected by the liberal use of duck tape. On the second morning I found that we had dropped a bracket from the GPS the day before when packing-up. Thankfully, using the GPS to locate where we packed-up and some careful searching around that point, we managed to find it again. Phew.

On the first day, Pauline, Rhian, Kate and I surveyed a line of grid squares parallel to the canal. Partly due to the vegetation we did not get too close to the canal. Also, the waterlogging would result in featureless results and so it simply was not worth the effort (Fig. 3).

Figure 3: Rhian using the Geoscan RM85 Earth Resistance meter.

On the second day we surveyed a block of grids closer to the current road to see if we could pick-up any buildings in that area. As we usually do, we used the 1+2 method. The RM85’s built-in multiplexer (basically a fancy switching box) means that we collect three readings with each movement of the frame. The first reading has a 1m probe separation and the second two have a 0.5m separation. The 1m separation looks about a meter or so into the ground and the 0.5m separation roughly 0.5–0.7m into the ground. The deeper survey, however, has half the resolution of the shallower survey (1m transect spacing and a 0.5, sample spacing compared to a 0.5m transect spacing and a 0.5m sample spacing).

Figure 4 shows the results of the shallower survey, and Figure 5 the deeper survey.

Figure 4: the 0.5m probe spacing Earth Resistance survey. Black is high resistance.

Figure 5: the 1m probe separation Earth Resistance survey.

Looking at Figure 5 first, I would argue that most of what we can see here is geology rather than archaeology. The ground slopes from the bottom of the image to the canal at the top. We can see in the strip of survey at the top of image a band of low resistance readings parallel with the canal. This is almost certainly the transition from the more solid bedrock to river/canal-side deposits and the presence of water. In the area near the road, the high resistance areas near the building are again related to topography rather than archaeology.

The 0.5m probe separation survey in Figure 4 has a few potential features. I have labelled these in Figure 6.

Figure 6: the 0.5m survey labelled.

1. There is a faint higher resistance line running towards the excavated area on the canalside. This could be a path?
2. A small rectangular high resistance feature might be the foundation for something.
3. The blue lines indicate two low-resistance linear features. These might be some sort of cut feature (pipeline, robbed wall lines?).
4. A high resistance vaguely linear feature.

I wish I could be much more positive with these results and I am the first to admit that none of them are 100% convincing. It might be worth “ground truthing” my tentative interpretations a little more.

Lastly, a thank you to Mike who has subsequently rewelded the frame, and to Jim for making-up some new jump leads ready for the res meter’s next outing.

A trip to Oxfordshire

I was asked by a long time friend and colleague, David Griffiths, if we would be interested in undertaking a GPR survey at a site near Kidlington. David, along with colleagues and students from the Oxford University Department for Continuing Education (Twitter: @OxfordContEd), had undertaken some survey work at this site over the last year. Aerial photographs had hinted at the existence of a ploughed-out round barrow cemetery at the site. The 1945 imagery available via Google Earth shows some of these barrows (Figure 1).

Figure 1: The 1945 RAF imagery of the barrow cemetery.

The team from Oxford have undertaken three sessions of magnetometry using Bartington Grad-601s and the results have been very good (Fig. 2).

Figure 2: the gradiometer survey results.

As can be seen, there are 10 ploughed-out probable round-barrows, as well as some linear features including a very clear one running SW-NE with a break in it. An unexpected and quite exciting find, however, is in the red-box in Figure 2 consisting of two curved magnetic features running roughly parallel to each-other. The most likely explanation is that these are the ‘quarry ditches’ either side of a long barrow. The aim of the GPR survey was to see if we could add any detail to this feature. Ruth and I spent two days collecting data ably helped by Stewart and Louise, two of the students (Figure 3).

Figure 3: Kris and Ruth with the GPR (photo David Griffiths).

Following advice from Jarrod Burks, we decided to collect data at a transect interval of 25cm, and to make sure we covered the whole of the long barrow we had transects 50m long. We managed to cover an area 50m by 60m, which is just over 12km of radargrams. There were plentiful cow pats (Figure 4), some of which had baked into concrete!

Figure 4: cow pat blues.

After the first day, I processed the data for a block 30x50m and was singularly unimpressed. We kept going however on the Sunday to cover the whole of the long barrow. I was very puzzled as to why the radargrams on the screen of the GPR looked so promising, and the time-slices were so poor. I processed the whole data set today and realised that something odd was going-on with the data processing. After a bit of trial and error I managed to work something out and, much to my (and everyone else’s) delight, found that the results were really rather good. Figure 5 shows 11 time-slices (aka amplitude maps to give them their proper name).

Figure 5: Time slices from the survey.

Ignore the depths in Figure 5, I have yet to determine the velocity so that I can calculate the depths. The ditches of the long barrow and the round barrows can be clearly seen, however, in slices 5 and 6. Let us look at some of them more closely.

Figure 6: Time slice 1.

In the first time slice (Figure 6) we can see the track which runs across the middle of the area clearly (red arrows), and it matches the Google Earth image perfectly. Having access to a high-accuracy GPS makes our work so much easier! The white line seen in the mag data also shows in the GPR (blue arrow) and are probably land drainage. The oddest thing is the difference between the area at the north of the plot (as shown by the yellow arrows) and below. As we are dealing with the very top surface here, my guess is maybe the farmer used an electric fence at some point and had cattle in the north half but not in the southern?

Figure 7: Time slice 2.

The second slice (Figure 7) shows a series of parallel lines. These are almost certainly land drainage.

Figure 8: Time slice 3.

In slice 3 we can see some ‘strong reflectors’ in the middle, and the hints of the two ditches of the long barrow. The strong reflectors might be the remains of the mound of the barrow. The red line shows something cutting across the feature, which is suspiciously in-line with more modern track marks. Looking at the Google Earth image without the geophysics, however, shows the track curving and joining the N-S one noted above. It seems likely, therefore, that this is a ditch cutting across the barrow.

Figure 9: Time slice 4.

Figure 9 shows the fourth time slice. We can now clearly see the quarries either side of the long barrow (blue arrows), and the ditches from the two round barrows (red arrows). The southern barrow appears to be much less round than usual.

Figure 10: Time slice 5.

Figure 10 shows the next time slice down. The barrows can be clearly seen. In addition, the ditch coming into the grid from the NE shows well (red arrow). There appears to be another feature (possibly a ditch?) that does not show on the mag survey (as indicated by the blue arrows). There are also some long, thin, sinuous features which I do not understand as shown by the yellow arrows.

Figure 11: Time slice 6.

Time slice 6 (Fig. 11) shows the quarries for the long barrow very clearly, along with the various other features discussed above. The red arrow, however, suggests that the southern quarry is not uniform but has split into two.

Figure 12: Time slice 7.

Time slice 7 shows the various features discussed above. We can see, however, the strange network of thin features in the data which I have indicated a few with the yellow arrows. Suggestions on a postcard please…

Figure 13: Time slice 11.

Skipping now to the last time slice, No. 11 (Fig. 13). The round barrows are not visible, but vague hints of the long barrow quarries survive (red arrows). I should, perhaps, have opened the time-window up a little. The linear feature at the top of the grid is still showing very clearly, and some of the mysterious thin features can just be seen.

There is more to do looking at the radargrams and consulting colleagues about what all this might mean, but the main story is fairly clear. As I dislike ending on 13, the next figure is a “view from the office.” Many thanks to David and his team from the Oxford University Department for Continuing Education for inviting us down and getting us involved in this lovely site.

Figure 14: the view from the ‘office.’

Rectory Meadow, Wheathampstead

Rectory Meadow is an open area just to the NW of St Helen’s Church in Wheathampstead. Along the eastern edge is one of two Victorian “crinkle-crankle” walls in the village. These were on either side of the garden of the Old Rectory which has now been partially built-on, and partially turned into a Community Garden and boules court. Parallel to the wall, to the west, is a ditch and bank largely covered in trees and bushes. Daniel Secker has recently written an article suggesting this feature is the boundary of the Saxon settlement (see Mike Smith’s note on this). We hope, later in the year, to be able to cut a section through this feature to try and get some dating evidence.

Fig. 1: Ellen and Patrick MacNeill at Rectory Meadow. The crinkle-crankle wall can be seen behind the trees and bushes, and the ditch and bank is in the line of bushes.

Meanwhile, it seemed worthwhile to survey the open area to see if there was any sign of buildings in this part of the village. It also gave us a chance to test-out the newly repaired magnetometer. Although the extremely-expensive repair to the electronics had been completed, the wear to the wheel axles and odometer system needed some DIY make-do repairs. Our first outing showed that the make-do repairs worked well enough, but we suffered from the dreaded “sensor freeze” (this is where one of the sensors, or in this case two (!) get stuck on a single value). Yesterday, Ruth, Mike and I went back to redo that grid and extend the survey a bit. Thankfully, all seems to be well this time.

Fig. 2: Results from the mag survey.

Although it was great that the mag was working OK, the results did not show anything very exciting. We have lots of bits of metal, and three utility lines. The line running parallel to the south edge of the survey is an artefact of the data processing (it is hard to get lines only 7m long to match those 40m long). In this case the negative evidence is useful. If the ditch and bank is the boundary of the Saxon settlement, we would not expect structures outside that boundary.

The group have quite a few surveys in the pipeline including Cholesbury hillfort, Maldon leper hospital and others. Watch this space!

CAGG on tour

Mike and I undertook a survey for a colleague of mine on a site in Sussex. Last year, a Roman stone sarcophagus had been found (Fig. 1) and the question was “are there any more?” As stone sarcophagi are not magnetic, it seemed that GPR would be the ideal method to use.

Fig. 1: the sarcophagus (photo: Mike Smith).

I had hoped to be able to run the GPR over the site of the burial before they cleared it for excavation, but the timings did not work. It would have been ideal to be able to see what this looked like in radar data.

When we arrived, the GPS decided it was going to play-up. It took over an hour on the phone to get it to work once again. As a result, on the first afternoon we did a small grid 42m x 8.5m at 0.5m intervals. The top eight slices are shown in Figure 2.

Fig. 2: the initial grid, top eight slices.

There wasn’t anything too exciting in the data but I did wonder if the “blob” (“high reflectance feature”) indicated with the red arrow might be something. With the transects running E–W, only one or two would actually hit a sarcophagus also lying E-W. In order to test this, we did a second small detailed grid the next day, only 5m x 8.5m but this time at 0.25m transect intervals and N-S. The top eight slices are shown in Figure 3.

Fig. 3: the detailed grid.

The feature starts showing in 3, is clear in 4 and 5 but has gone by slice 6. Sadly, I don’t think this is a sarcophagus. Firstly, it’s shape doesn’t seem all that convincing, and secondly, it is quite thin, which shows even more clearly in the radargrams (Fig. 4).

Fig. 4: radargram across feature (NB N is to the left.)

It also does not show in the one Earth Resistance grid we had completed earlier in the training dig (Fig. 5).

Fig. 5: Earth Resistance grid.

Oh well, we tried! We also completed a larger block on the second day, some 50m x 40m at 0.5m transect intervals (Fig. 6).

Fig 6: the GPR in action (photo: Mike Smith).

The top eight radargrams are shown in Fig. 7. Despite there being features in the magnetic data (not collected by us), nothing much shows in the data beyond plough marks and cultivation patterns.

Fig. 7: the main grid.

It was quite a shame that we were unable to detect something more exciting. As always, being able to survey a bigger continuous area would be ideal. Odd grids here and there are hard to put in context and a bit hit or miss. It is always preferable if one has time and resources to survey large areas.

The magnetic susceptibility survey

“Mag sus” is not used as commonly in archaeology as some of the other survey techniques such as magnetometry. It can, however, be very useful in specific circumstances such as the project I did in Italy a couple of years ago. Magnetometry (or to give its full name, magnetic gradiometry) is a passive technique. It measures the strength of the local magnetic field whatever the source of the magnetism, be that pits full of soil which is slightly more magnetic than the surrounding subsoil, or the keys in your pocket. At the start of each day, we find a “quiet spot” and adjust the magnetometer so that it reads zero whatever direction we are facing, and so all the sensors are reading zero over the same point. In that way, our readings are all variations against that local zero point. Even at the data processing stage, we make the general background level of magnetism be zero. As a result, we can see the “anomalies” (ditches, pits, kilns, old horseshoes etc.) as local variations against the background.

On occasions, however, we might want to see the broad pattern of magnetism that we cannot see in the gradiometry data. In other words, is the soil on this part of the site potentially more magnetic than the soil on a different part of the site, perhaps as a result of burning, or simply where people dumped their rubbish? To do this, a magnetic susceptibility meter creates a magnetic field, and then measures how magnetic the sample within that field becomes. In this way we can plot variations in absolute magnetism rather than patterns in the local field. The size of the sample measured depends on the size of the loop the meter is using. The MS2D sensor we were using last week has a diameter of 20cm and measures the top 10cm of the soil. In Italy I only had the MS2K sensor which has a diameter of only 25mm and is measuring only the top few millimetres which is why we had to use a mattock to allow us to take a reading from bare earth.

So, why were we trying this at Gorhambury? Fig. 1 is a slide I use in a lecture which shows the Earth Resistance survey results for part of the Theatre Field. I have circled two buildings.

Fig. 1: The Earth Resistance survey with two buildings indicated.

In the Earth Resistance data the foundations are showing as slightly darker lines, i.e., higher resistance features. We know from excavations that buildings at Verulamium have foundations of flint resulting in the higher resistance values. Now let us look at the magnetometry data (Fig. 2).

Fig. 2: the magnetometry survey with the same two red circles as Fig. 1.

In Fig. 2 I have circled the same two locations as Fig. 1. In the eastern circle we can see the building foundations as white, i.e., negative magnetism, but in the western circle we cannot see the building here at all. What is going on? We have to go back to first principles. We are measuring magnetism, and flint foundations are not magnetic so we should not be able to see them. My idea is that the soil around the eastern circle is quite magnetic in absolute terms, so when one digs a foundation trench one is removing magnetic soil and replacing it with non-magnetic material, hence the walls showing as white lines. In the western circle, however, the soil may be much less magnetic so that digging some away and replacing it with flint might not be visible. The magnetic susceptibility survey is an attempt to test this idea. If I am right, the soils near the western circle should have a lower magnetic susceptibility than those near the eastern one.

But what about the area of strong magnetic responses circled in yellow, and a similar area in the SW corner? These look like buildings which have burnt down, and were never replaced. If we look at the Earth Resistance data again (Fig. 3), it doesn’t appear that the area was built-on again. This is just an interpretation, and one that could only be securely tested by excavation.

Fig 3: the Earth Resistance data showing the location of the possible burnt building (circled in yellow).

To test my idea about variation in the background magnetic susceptibility, I picked out an area which contained the above buildings and we took 1,301 mag sus readings over a few days in the last week (Fig. 4). Once one got going, it was quite quick although a team of three does seem a bit overkill.

Fig. 4: the location of the mag sus readings.

By drawing a rough line around the points we can see the area surveyed on the mag data (Fig. 5). It includes both burnt buildings as well as the building we can see and the one we cannot.

Fig. 5: the mag sus survey area shown by the red line.

Fig 6 shows the results of the survey. White/red are areas of high magnetic susceptibility and blue is low.

Fig 6: the magnetic susceptibility survey results.

The results are a bit blobby — well a lot blobby! — but there is a clear trend for the higher readings to be in the NE half of the area and the lower ones to the west and south with one exception. The strong blob in the SW corner matches very closely to the location of the smaller of the two hypothesized burnt buildings. That gives us some faith in the method. Figure 7 shows the same results with the locations of the two buildings mentioned earlier indicated with red boxes (Google Earth doesn’t do circles!).

Fig. 7: the two buildings shown on the mag sus map.

The results are a little inclusive. Although the larger building to the east lies within the generally higher area of mag sus, its immediate location is not especially strong. The smaller building to the west, however, does lie in an area of low readings apart from one hot spot. If one looks at the survey points (Fig. 8), you can see that “spike” is the result of one reading. The magnetometry survey (Fig. 5) shows some spikes in the same area, probably modern iron.

Fig. 8: the mag sus sample points overlain on the results.

I do have some reservations. Firstly, the MS2D is only measuring the top 10cm. The grass and turf is quite thick in the field as it has been pasture since 2000. In Verulamium Park, the wear on the grass from walkers and footballers probably means the turf is less dense and thick, and the grass is certainly cut shorter. Perhaps we might get clearer results if we could measure a larger or deeper sample? Secondly, the topography of the field means that there has been soil wash / creep downslope (Fig. 9). Might the trend be a result of that?

Fig 9: topography in the survey area from the dGPS data.

The last image is just to show how the aqueduct follows the contours (bar the odd dog-leg around something).

Fig. 10: topography with the line of the aqueduct indicated in blue.

Many thanks to all those I cajoled into helping me complete this little experiment. I shall keep pondering…

I have results from other surveys and projects we have undertaken, and so will post those as and when I can. My desire to write blog posts rather died during covid, but now I am going again I’ll try and keep posting the occasional piece beyond the results of the Gorhambury survey.

A happy ending

Most years two things have happened. Firstly, we have lost two days to bad weather and secondly, we’ve had some little job to finish off on the Bank Holiday. This year, although the dead Foerster mag resulted in us spending the first three days on another site using the Bartington magnetometer instead, and the late hay harvest delayed our start for the Earth Resistance survey in the Theatre Field, we have not lost any days for bad weather and we have met all my targets for the year so will not be back tomorrow. Go team!

Before I get stuck into the results and the usual end-of-season stats, I’d like to say a big thank you to all the team. Twenty people were involved in the work this season, some just for the odd day, some for almost every day. I’d especially like to thank those who helped ferry stuff and look after it over night, who helped fix stuff, and for ferrying me. You all know who you are. All your efforts are really appreciated and the project would not be the same, or even possible, without you. I’d also like to thank Lord Verulam for facilitating access to his land. There are some days when I am high on the hill overlooking the Roman town, with the cathedral on the horizon, that I cannot believe I can be so lucky as to be able to work here each summer.

The Earth Resistance team managed an excellent seven grids today to complete the southern transect to the hedge line (Fig. 1).

Fig. 1: the res team on the last day.

Over the last 17 days at Gorhambury the resistance team have collected data from 75 grids. Since moving into the Theatre Field they have averaged six grids a day. Those 75 grids cover 30,000m² or 3 ha. The team have stabbed the ground with the machine some 60,000 times, collecting 180,000 readings, 120,000 0.5m spacing readings and 60,000 1m spacing readings. The area covered is a bit odd looking because we have been adding to the edges of the area previously surveyed (Fig. 2).

Fig. 2: the 2021 Earth Resistance survey.

The GPR team had a few problems with the new system at the start of the season, but they soon got into the swing of things and have completed 4.12ha of Blacklands Field (Fig. 3). The field is 7.97ha, so they have completed 51% of the field, some 82,400m of radargrams which occupy some 3.14gb of my hard-disk!

Fig. 3: Blacklands Field GPR at the end of the 2021 season.

Despite Fergus’ indifference (Fig. 4), the res team did a great job today. The results are quite subtle (Fig. 5).

Fig. 4: Unimpressed.

Fig 5: the results from the Earth Resistance survey.

Although the results from the survey aren’t hugely obvious, we may have found a couple of buildings. I had indicated them with the res arrows in the image. I’ll have to cross-reference them with the mag and GPR images to be sure.

Figure 6 shows nine time slices from the GPR survey.

Fig. 6: nine time slices for the GPR survey at the end of the 2021 survey.

Today’s survey has revealed a series of buildings on the eastern edge of the survey area. Figs. 7-9 show three time slices in more detail.

Fig. 7: slice 5.

Fig. 8: Slice 9.

Fig. 9: slice 11.

We clearly have a series of buildings around a open area with Watling Street on the SW side. To the east, there appears to be a walled enclosure. Some of the buildings appear to have columns in front, especially the small one in the SE corner. The buildings do not look like the big town houses we have found in much of the rest of the town. I’m going to have to do some research into the plans of other types of building.

We also finished the mag sus survey today, We collected some 1,300 readings which I’ll have to type in. Unlike our other survey methods, the mag sus meter uses a Mark 1 pencil data logger…

I’ll post updates on the data processing, and on some other surveys I haven’t had a chance to write-up over the following weeks. Meanwhile, we’ve pulled-up our last flag at Gorhambury for another year (Fig. 10).

Fig. 10: the last flag.

Just one day more

If you are new to geophysical survey in archaeology, please visit the introductory page “Geophysical survey in archaeology”.

I always have mixed feelings at this time of year. I so enjoy spending the days out in the fields at Gorhambury with the gang, and the end of the season marks, for me, the end of summer and the start of getting ready for the new term. On the other hand, I’ve had, essentially, one day off in a month so I am looking forward to putting my feet up, at least for a day!

We ran all three machines today: the Earth Resistance Meter, the GPR and the magnetic susceptibility meter. All three managed a good area and hopefully we’ll all meet our targets by the end of tomorrow so we can avoid the traditional extra part-day on the Bank Holiday. The mag sus has now taken well over 1100 readings, but I’ll make a separate posting about that in due course.

Fig. 1: Gill, Pauline and the resistance meter, watched by Fergus.

The Earth Resistance team are a very long way from the rest of us, at the top of the hill overlooking the town. They are working eastwards towards Bluehouse Hill.

Fig. 2: the Earth Resistance survey after day 16.

The survey (Fig. 2) shows the 1955 ditch and the buildings I mentioned yesterday. There is quite a contrast between some areas on the edge of the dry valley and the buildings making them harder to see. With a bit of careful data processing I can, hopefully, bring them out more clearly.

The GPR survey’s two 40x40m blocks had one or two buildings in them. Fig. 3 shows time slice 9.

Fig. 3: GPR slice 9.

We seem to be in an area of public buildings and open spaces, rather than private houses like the “motorway services” (which I think is actually a very nice house and not services at all!). We are going to be left with very tantalising half-surveyed buildings until next summer (we hope).

Short posting tonight as it is getting late and tomorrow is our last day.