Ancient Monuments Laboratory
Report 2/99

BEERWAY FARM, SHAPWICK,
SOMERSET,
REPORT ON GEOPHYSICAL SURVEY,
1996

P Linford & M Cole

AML reports are interim reports which make available the results of specialist investigations in advance of full publication. They are not subject to external refereeing and their conclusions may sometimes have to be modified in the light of archaeological information that was not available at the time of the investigation. Readers are therefore asked to consult the author before citing the report in any publication and to consult the final excavation report when available.

Opinions expressed in AML reports are those of the author and are not necessarily those of the Historic Buildings & Monuments Commission for England.

Ancient Monuments Laboratory Report 2/99

BEERWAY FARM, SHAPWICK,
SOMERSET,
REPORT ON GEOPHYSICAL SURVEY,
1996

P Linford & M Cole

Summary

A geophysical survey was carried out over Church Field at Beerway Farm, Shapwick in Somerset at the request of the Shapwick Project to extend previous geophysical survey work carried out there. The magnetometer results detected a complex palimpsest of archaeological anomalies from both Roman and medieval periods including the remains of the original church at Shapwick, which existed between the 8th- and 14th-centuries AD. Complementary archaeological evidence gathered by the Project helped to interpret the anomalies visible in the survey results.

Author's address :-

ARCHAEOMETRY BRANCH,
Ancient Monuments Laboratory,
ENGLISH HERITAGE,
23 Savile Row,
LONDON W1X 1AB.

Historic Buildings & Monuments Commission for England.

Report on Geophysical Survey, 1996.

Introduction

Geophysical survey was undertaken at Beerway Farm, Shapwick, Somerset, in response to a request from the Shapwick Project. The aim of the survey was to aid the investigation of an important and complex multi-period site centred around the remains of a former church. The latter, possibly founded as early as the 8th-century AD, was abandoned in the 14th-century when the present church was built (approximately 1km to the west).

The layout of the church was revealed by parch marks in aerial photographs (APs) taken during 1989 which also indicate that it is located within a large rectangular enclosure. The latter appeared to contain numerous other linear features of unknown origin. An earthwork survey (Aston, Bond & Ingle 1994) mapped the bank and ditch of this enclosure as well as other mounds and the location of a spring. Excavations by Dr Chris Gerrard in 1993 (Gerrard 1995) demonstrated the complex structural development of the church and its boundary earthworks. These excavations also produced material evidence of prehistoric and Roman activity in the vicinity. Subsequent fieldwalking surveys by the Shapwick Project yielded a wealth of finds relating to the church including roof tiles, slate and quantities of human bone (Gerrard and Gutiervez, 1997). Additionally, a sufficient quantity and diversity of Roman material was collected (including glass, Samian ware, tesserae, roof tiles and hypocaust flue tiles) to suggest the presence of a well- appointed suite of buildings, possibly a villa, as well as a variety of worked flint dating from the Mesolithic through to the Bronze Age.

Geophysical survey (undertaken by Geophysical Surveys of Bradford in 1993) over a limited area of the site demonstrated its suitability for magnetometer survey. This survey made it clear that more extensive survey coverage had the potential to produce highly informative results and led to a request for assistance from the Ancient Monuments Laboratory.

The site at Beerway Farm is centred on ST 42 38. The underlying solid geology consists of Lower Lias clay with some limestone (Institute of Geological Sciences 1973). The soil association, Evesham 1 (Soil Survey of England and Wales, 1983) is a well drained calcareous clayey soil which might be expected to exhibit a high magnetic susceptibility, especially when its naturally occurring magnetic mineralogy is modified by anthropogenic activity.

Method

Field Procedure

As the effectiveness of magnetometry had already been demonstrated, this technique was chosen as the most suitable to employ. Unfortunately insufficient time was available to undertake a resistivity survey, although this deficiency was mitigated by the excellent response of the magnetometer to subsurface building foundations.

A grid of 30m squares was established over the site by the Shapwick Project team (see Figure 1). Each of these squares was then surveyed using Geoscan FM36 fluxgate gradiometers, according to the standard technique outlined in Annex 1, note 2. The magnetometer survey was conducted during two separate visits to the site, the first visit having been curtailed by the severity of winter conditions. Owing to the clarity of the response, a limited area, focused over the remains of the former church, was subsequently resurveyed at a higher resolution (traverses spaced 0.5m apart). The location of this area, which covers parts of grid squares 13 and 14 is indicated by the darker shading in Figure 1.

Data Processing and Presentation

The data thus recorded is presented in this report in the form of greyscale and graphical trace plots. In these, it has been treated with a localised adaptive thresholding median filter to reduce the intense response to modern ferrous litter (Scollar et al 1990). Figure 2 shows a greyscale plot of the magnetometer data at 1:2500 scale superimposed on the Ordnance Survey map of the area around Church Field. Figure 3 shows the same plot of the data at 1:1250 scale and Figure 4 shows the data in trace plot form, also at 1:1250. Figure 5 shows the greyscale plot from Figure 3 superimposed on the earthwork survey of the field again at 1:1250 scale. Figure 6 shows the higher resolution data taken over the remains of the church in both greyscale and trace plot format at 1:500 scale.

Results

It is immediately clear from the above plots that soil conditions were favourable for magnetic survey and that a remarkable number of potential archaeological anomalies have been recorded. An interpretation diagram, superimposing these anomalies on a plan of the field boundaries, is presented at 1:1250 scale as Figure 7 and the numbers in bold in square brackets in this text (eg: [x] ) refer to numbered features in this figure. This diagram presents a bewildering palimpsest of anomalies, all potentially archaeological in origin, reflecting the multi-period occupancy at the site. Hence, in Figure 8 an attempt has been made to separate the overlying anomalies into plans representing different phases of activity at the site and these are plotted at 1:2500 scale. This interpretation has benefitted from information supplied by Dr. Chris Gerrard of King Alfred%s College, Winchester, who has excavated at the site in 1993 and 1998; a report on the 1993 excavations may be found in Gerrard (1995).

One non-archaeological feature has been marked with a yellow dashed line running east-west across the field in Figure 7. This is the line of a water pipe noted during excavation in 1993. The strong magnetic signal it generates has, unfortunately, saturated the response of the fluxgates within 3m either side of its position, limiting archaeological interpretation in this region. Nevertheless a large number of potentially archaeological features have been identified, particularly in the area towards the north-east of the field.

Negative magnetic anomalies (those where the causative feature is less magnetised than the surrounding soil) show up in black in the greyscale plots of the magnetometer data and have been marked in red in Figure 7. Such anomalies are unusual but can occur when masonry wall footings composed of non-magnetic stone are present beneath a cover of well magnetised (often clayey) soil. In the present survey, the remains of the church are a particularly clear example of this phenomenon and their location is indicated in the interpretation plan [14] . More conjectural examples, where the response is similar in magnitude to the general level of magnetic noise recorded in the data, are indicated with dashed lines. Also, possibly significant areas of lowered magnetic response, perhaps due to buried stone rubble or floors, are indicated with a red stipple (eg [2] ).

Positive magnetic anomalies, which show up in white in the greyscale plots, are indicated in blue in Figure 7. These anomalies are caused by features with a stronger magnetisation than the surrounding soil and are often caused by infilled ditches, the fill of the ditch having been magnetically enhanced by anthropogenic action on the site. Again, more conjectural possibilities are signified with a dashed line. Non-linear positive anomalies are shown in green in Figure 7 and these might be due to pits or simply areas of soil exhibiting enhanced magnetic susceptibility owing to human activity in the vicinity.

One class of faint linear positive anomaly has been distinguished using brown rather than blue in Figure 7. Invariably these describe the arcs of circles, all of similar diameter and a particularly clear example is seen [1] . It is possible that these represent ring-ditches and are the remains a prehistoric phase of the site. For this reason they have been isolated and depicted separately in Figure 8a. The example at [1] may contain evidence of internal features, and these have also been indicated in brown. However, this interpretation is uncertain owing to other linear features of a probable later date being superimposed on the same area.

To the north a confusing pattern of intersecting linear anomalies is apparent. However, the circular area of low magnetisation, corresponding to a circular rise in the earthwork survey is clear [2] . The lowered magnetisation suggests the possibility of a spread of buried masonry rubble and, indeed, excavations in 1993 (trench 73/93/4016/F) revealed this to be a stone walled building chocked full of stone blocks, the only finds dating from the Roman period.

To the north, a negative linear anomaly is marked [3] apparently forming a rectangular enclosure with rounded corners, measuring about 35m east-west by 30m north-south. The low magnetisation once again suggests that this is caused by a buried wall footing rather than a ditch. Inside this enclosure, towards its north-eastern corner, areas of significant magnetic enhancement occur, perhaps indicating a focus of activity here. The stone footings of a building might also be tentatively be suggested, as may those of a second larger building further south. Excavations in 1998 have gone some way to confirming this interpretation, showing the southern building to be the base of a 12th-century AD Chamber block (trench 73/98/4016/Y) with stone walls and stone flagging on the floor. To the north, in the centre of the enclosure, finds of significant quantities of animal bones suggest a nearby kitchen, indicating a possible function for the wall footings detected in the north-eastern corner and explaining the significant magnetic enhancement detected here (heating activities leading to magnetic enhancement of associated soils and structures).

A linear ditch anomaly [4] , thought (C. Gerrard pers. comm.) to represent the remains of a Roman boundary and excavated in trench (73/98/4016/Y), cuts through the centre of this enclosure running north-south. Further north the same ditch anomaly also appears to underly the remains of the old church near [14] , supporting the hypothesis that it predates the medieval occupation of the site. If this is so then it suggests that the enclosure discussed above [3] , as well as the buildings in it, post-date the Roman period. This would accord with the fieldwalking evidence which found the greatest concentrations of medieval pottery in the area to the south of the old church. Hence, it is possible that the enclosure is contemporary with the old church.

West of these anomalies at [5] a particularly strong, roughly circular positive anomaly is apparent. This has a peak magnitude greater than 200nT and may well be due to modern ferrous litter. However, it is also possible that it could be caused by remains from an ancient industrial process. Whilst in the proximity of the excavated Roman building mentioned above, its association with it is unclear and an interpretation as a bell pit related to the church might also be considered.

To the east of the enclosure at [3] several more sets of possible wall footings have been identified [6, 7, 8 & 9] . Interpretation in this area is hampered by the profusion of linear boundary features running across it, possibly dating from a later period (see below). The existence of masonry remains of buildings at these locations must therefore remain highly conjectural. However, if this interpretation is entertained then the strong positive anomaly at [7] , at the eastern end of two tentatively identified parallel east-west footings, may be significant. This has a peak anomaly strength in excess of 30nT, consistent with the field strengths that can be caused by thermoremanent magnetisation, perhaps suggesting some fired structure associated with the inferred building.

With the rather fragmentary remains visible in the magnetometer plot it is not possible to estimate the date of these structures. However, the greatest concentration of Roman finds discovered during fieldwalking came from the area of [8 & 9] , suggesting damaged Roman deposits %leaking" material into the ploughsoil. Excavations in trench 73/98/4016/Z over [8] showed an intersecting grid of Roman ditches, some stone filled and cut through the bedrock. It is also notable that [6 & 7] are contained within ditched enclosures and that all similar ditch anomalies are clustered between [6] and [10] in close proximity to the Roman activity at [8] . Hence, it might be conjectured that enclosure ditches (as opposed to the enclosing wall around the medieval buildings described above) are characteristic of Roman buildings on the site, and thus [6 & 7] may also date from this period.

It is also possible that Roman structures extend into the field to the east and some anomalies possibly due to masonry have been tentatively identified at [10] . However, the confused readings, caused in part by modern ferrous material, including buried machinery, make detailed interpretation in this vicinity extremely difficult.

A clear boundary, possibly consisting of separate positive and negative anomalies runs south past [5] before turning east to [12] where it seems to bifurcate and run either side of the anomalies at [7] . The initial section in the vicinity of [5] corresponds with a linear feature identified on the earthwork survey and the existence of both positive and negative anomalies might suggest it is formed of wall footings running inside a parallel ditch. Whilst it is also possible that a ditch feature with a strongly magnetised fill could cause both a positive anomaly and a negative %shadow", excavations in 1993 across this boundary have confirmed the former interpretation (trench 73/93/4016/A).

At [12] where this boundary splits, the southern arm continues straight, almost due east to the present eastern field boundary. As this appears to mark the southernmost extent of the conjectured Roman wall footings, it is tempting to suggest that it is an original Roman boundary. The northern arm describes a dog-leg to run between [6] and [7] before apparently turning north to pass just west of [11] as a broad positive linear anomaly. It is possible that this represents a later, perhaps medieval modification to the boundary no longer respecting the Roman layout of the site.

The structure at [11] just north of the water pipe may represent the remains of a further masonry structure, surrounded by substantial areas of positive magnetic enhancement indicative of intensive activity. Excavation near the spring to the north-west of this area (trench 73/93/4016/E) discovered tight, burnt cobbling, the probable cause of this enhancement. The lack of material above the stones suggested a medieval date.

Immediately to the northeast, a further complex of possible wall footings has been identified around [13] , associated with several amorphous areas of enhanced magnetisation suggestive of human activity. Within this complex, the short positive east-west ditch to the north of [13] was identified as Roman during excavations in the neighbouring field in 1997. It does not appear to respect the alignment of the putative wall footings, implying that the buildings are later, perhaps dating from the medieval period. Within one of these conjectured buildings, at the place marked by [13] is a circular positive anomaly exhibiting strong magnetisation (shown in orange). Studying the trace plot of Figure 4, it can be discerned that this has a double peaked form with a maximum anomaly strength of about 25nT. It is possible that this is caused by the subsurface remains of a kiln or furnace and it is perhaps significant that a spring has been identified immediately adjacent to this complex of anomalies. However, if this is the case then the structure must be fairly deeply buried as a much higher peak anomaly strength would usually be expected. Hence, given that other burnt stonework has been discovered nearby, it is also possible that it is caused by the burnt masonry.

The positive linear anomaly running north-south through this group of features has been identified as a previous field boundary known to have existed between 1759 and 1839.

To the west, a clear imprint of the final, 14th-century plan of the medieval church is evident at [14] . However, no evidence of any earlier phases of construction between the 8th- and 14th-centuries can be discerned, even when studying the higher resolution survey shown in Figure 6. The church is surrounded by amorphous areas of strong magnetisation and it is possible that these mask the responses from earlier footings. The amorphous anomalies are strongest along the north wall of the church, where they are linear in character, and also either side of the entrance in the south wall, where two prominent individual anomalies are visible. It is possible that all the positive anomalies around the perimeter of the church are caused by the interment of human remains; certainly graves were discovered during excavations to the south (trench 73/93/4016/B), one of which was covered with a grave slab.

North-east of the church, further possible wall footing anomalies have been indicated, as has a rectangular area of lowered magnetisation [15] . This is surrounded by more amorphous areas of enhanced magnetisation and it is thus possible that a building associated with the church stood here, possibly even an earlier incarnation of the church itself. Furthermore, immediately south of [14] and partially masked by the anomaly caused by the water pipe, a tentative boundary ditch has been indicated as have several more possible wall footing anomalies. Excavations in this area in 1993 (trench 73/93/4016/B) revealed that these wall footings were part of a building of medieval date. The linear anomaly running north from [4] through the nave of the church was also cut by this trench. Surprisingly, whilst the positive magnetisation of the anomaly suggested a buried ditch, it was found to be a deeply buried Roman stone wall. This might possibly be explained by magnetically enhanced soil building up against the wall during its life due to human activity in the vicinity. Thus, it is this enhanced soil that is being detected rather than the negative anomaly that would be caused by the wall itself. Close to the position of this excavation trench is another very strongly magnetised circular anomaly, indicated in orange. As with the similar, larger anomaly at [5] the possibility that this represents a bell pit should be considered.

In the far south of the survey area, some faint linear anomalies running approximately north-south are visible [16] . Any identification of such indistinct features must remain tentative; nevertheless, it is possible that they reflect medieval cultivation in this area. Certainly they are close to where faint ridge and furrow was detected in the earthwork survey, lending weight to this interpretation.

Towards the west of the survey area, anomalies tend to become less frequent, suggesting that this part of the site was away from the centre of occupation. A few linear features have been marked, most notably those describing a right angle at [17] . These appear to traverse only a short distance in either direction before petering out, and it is not possible to ascribe a function to them with any certainty. However it has been suggested that they might represent a double ditched trackway (C. Gerrard pers. comm.). There is a noticeable depletion in their magnetisation at the point were the narrow ditch-like anomaly running north-south crosses them, suggesting that this latter is a later feature, cutting through them.

A final cluster of amorphous anomalies indicative of enhanced magnetisation may be discerned in the north of the survey area at [18] and some tentative linear negative anomalies have also been indicated. If a further structure existed here, then it is interesting that it would have been on a different alignment from the other buildings detected by the survey. However, its alignment would match that of the present road that bounds Church Field to the north, hinting perhaps at the presence of remains which post date the church at the site.

Conclusions

The magnetometer survey of Church Field has revealed a palimpsest of intense archaeological activity, possibly spanning the prehistoric, Roman, medieval and perhaps even post-medieval periods. The complex pattern formed by these superimposed anomalies makes their separation and identification as individual archaeological features almost impossible. However, the survey interpretation has benefitted from the extensive archaeological research undertaken by the Shapwick Project and it has been possible to come to some provisional conclusions about the nature of past habitation in the area.

Perhaps most striking is the number of possible Roman buildings apparently present within the field (and these are most abundant in the interpretation of the results). Medieval remains have also been detected and the church has been clearly delineated. However, no evidence for earlier phases of church construction is evident, prior to the final 14th-century plan, despite a higher resolution survey being conducted over the remains of this building. To the south of the church, a rectangular enclosure has been detected measuring some 40m by 30m. Within this enclosure anomalies suggesting a building have been detected at [3] and further anomalies consistent with a smaller building have also been detected in the north-east of the enclosure. Whilst conjectural, it is possible that these represent the remains of a medieval manor house on the site. Nevertheless, the interpretation of many of the other anomalies evident in the survey remains very tentative and only investigation via further excavation is likely to confirm their identification.

M. Cole
P. Linford 4th-6th March 1996

Report by: P. Linford
M. Cole

References

Aston, M. A., Bond, C. J., and Ingle, C., 1994,
%The site of the Early Church" in Aston, M. A. and Costen, M. D., (eds), The Shapwick Project: An Archaeological, Historical and Topographical Study. The Fifth Report, pp. 57-61.

Gerrard, C. M., 1995
%Excavations in the Church field (4016), Shapwick, 1993. Preliminary report." in Aston M. A. and Gerrard C. M. (eds), The Shapwick Project. An Archaeological Historical and Topographical Study. The Sixth Report, pp. 90-109.

Gerrard, C. M. and Gutiervez, A., 1997
%Denying nothing, distrusting everything." Interpreting pottery distributions from fieldwalking 1989-1994.' in C. Gerrard and M. Aston (eds), The Shapwick Project. An Archaeological, Historical and Topographical Study, The Seventh Report, pp. 45-82.

Geological Survey of Great Britain, 1973,
1:50,000 Solid and Drift Edition, Sheet 296, Glastonbury.

Scollar, I., Tabbagh, A., Hesse, A. and Herzog, I., 1990,
Archaeological Prospecting and Remote Sensing. Cambridge University Press: Cambridge.

Soil Survey of England and Wales, 1983,
Legend for the 1:250,000 Soil Map of England and Wales.

Enclosed Figures and plans

Figure 1 Location of the geophysical survey, 1996 (1:2500).

Figure 2 Magnetometer survey superimposed on Ordnance Survey map (1:2500).

Figure 3 Greyscale plot of magnetometer survey results, 1996 (1:1250).

Figure 4 Trace plot of magnetometer survey results, 1996 (1:1250).

Figure 5 Greyscale plot of magnetometer survey overlain on earthwork survey (1:1250).

Figure 6 High resolution magnetometer survey over old church site (1:500).

Figure 7 Interpretation diagram of magnetometer results (1:1250).

Figure 8 Interpretation diagrams of magnetometer results, divided into possible phases (1:2500).
Annex 1: Notes on standard procedures

1) Resistivity Survey: Each 30 metre square is surveyed by making repeated parallel traverses across it, all aligned parallel to one pair of the square's edges, and each separated by a distance of 1 metre from the last; the first and last traverses being 0.5 metres from the nearest parallel square edge. Readings are taken along each traverse at 1 metre intervals, the first and last readings being 0.5 metres from the nearest square edge.

Unless otherwise stated the measurements are made with a Geoscan RM15 earth resistance meter incorporating a built-in data logger, using the twin electrode configuration with a 0.5 metre mobile electrode separation. As it is usually only relative changes in resistivity that are of interest in archaeological prospecting, no attempt is made to correct these measurements for the geometry of the twin electrode array to produce an estimate of the true apparent resistivity. Thus, the readings presented in plots will be the actual values of earth resistance recorded by the meter, measured in Ohms (%). Where correction to apparent resistivity has been made, for comparison with other electrical prospecting techniques, the results are quoted in the units of apparent resistivity, Ohm-m (%m).

Measurements are recorded digitally by the RM15 meter and subsequently transferred to a portable laptop computer for permanent storage and preliminary processing. Additional processing is performed on return to the Ancient Monuments Laboratory using desktop workstations.

2) Magnetometer Survey: Each 30 metre square is surveyed by making repeated parallel traverses across it, all parallel to that pair of square edges most closely aligned with the direction of magnetic North. Each traverse is separated by a distance of 1 metre from the last; the first and last traverses being 0.5 metre from the nearest parallel square edge. Readings are taken along each traverse at 0.25 metre intervals, the first and last readings being 0.125 metre from the nearest square edge.

These traverses are walked in so called 'zig-zag' fashion, in which the direction of travel alternates between adjacent traverses to maximise survey speed. However, the magnetometer is always kept facing in the same direction, regardless of the direction of travel, to minimise heading error.

Unless otherwise stated the measurements are made with a Geoscan FM36 fluxgate gradiometer which incorporates two vertically aligned fluxgates, one situated 0.5 metres above the other; the bottom fluxgate is carried at a height of approximately 0.2 metres above the ground surface. The FM36 incorporates a built-in data logger that records measurements digitally; these are subsequently transferred to a portable laptop computer for permanent storage and preliminary processing. Additional processing is performed on return to the Ancient Monuments Laboratory using desktop workstations.

It is the opinion of the manufacturer of the Geoscan instrument that two sensors placed 0.5 metres apart cannot produce a true estimate of vertical magnetic gradient unless the bottom sensor is far removed from the ground surface. Hence, when results are presented, the difference between the field intensity measured by the top and bottom sensors is quoted in units of nano-Tesla (nT) rather than in the units of magnetic gradient, nano-Tesla per metre (nT/m).

3) Resistivity Profiling: This technique measures the electrical resistivity of the subsurface in a similar manner to the standard resistivity mapping method outlined in note 1. However, instead of mapping changes in the near surface resistivity over an area, it produces a vertical section, illustrating how resistivity varies with increasing depth. This is possible because the resistivity meter becomes sensitive to more deeply buried anomalies as the separation between the measurement electrodes is increased. Hence, instead of using a single, fixed electrode separation as in resistivity mapping, readings are repeated over the same point with increasing separations to investigate the resistivity at greater depths. It should be noted that the relationship between electrode separation and depth sensitivity is complex so the vertical scale quoted for the section is only approximate. Furthermore, as depth of investigation increases the size of the smallest anomaly that can be resolved also increases.

Typically a line of 25 electrodes is laid out separated by 1 or 0.5 metre intervals. The resistivity of a vertical section is measured by selecting successive four electrode subsets at increasing separations and making a resistivity measurement with each. Several different schemes may be employed to determine which electrode subsets to use, of which the Wenner and Dipole-Dipole are typical examples. Geopulse earth resistance meter, with built in multiplexer, is used to make the measurements and the Campus Imager software is used to automate reading collection and construct a resistivity section from the results.