Geophysical surveys of approximately 2.4ha were conducted over the site of the Roman amphitheatre at Richborough, Kent (SAM KE25). This lies approximately 600m SW of the Roman fort built in the later third century to dominate and protect the Wantsum Channel that separated the Isle of Thanet from the mainland (Johnson 1999). Both the fort and the amphitheatre are located on a former island, with the latter at the highest and most visible point. The relationship of the amphitheatre to the fort is unclear, although it has been postulated that it was a later Roman construction built on the fringes of the settlement surrounding the military base at the main entry point into Britain (Johnson 1999, 10).

Current knowledge of the amphitheatre is inadequate and stems from excavations by a Mr Rolfe in 1849 (Roach-Smith 1850). It was described as an ellipse measuring 200 feet x 166 feet (~ 61m x 50.6m), with the surviving perimeter (~1.0m thick) interpreted as the exterior wall. The latter was found at a depth of only 0.3m from the ground surface (ibid 161-2). Three entrances ('north, south and west') were identified, although only one of these (the 'northern') included a gap (~2.7m in width). This 'northern' entrance was believed to have been arched (ibid 163). However, the contemporary account of this, and the other 'entrances', is difficult to understand; and in particular the orientation and character of the entrances cannot easily be matched with the geophysical evidence to be described below. There are no accurate site plans and despite the obvious topography of the field (a raised mound with circular depression in the centre), the location of the walls and supposed entrances remained unknown until the present survey.

The aim of this survey, in light of this limited knowledge, was to undertake an evaluation of the geophysical potential of the site, and to attempt to establish the nature of the monument and define its extent. It was also intended that the survey might provide some initial information towards the planning of a wider project researching the development and context of settlement at Richborough (Kendall and Wilmott 2001).

The amphitheatre field (TR 320 598), in the guardianship of English Heritage, lies on deep and mainly calcareous clayey soils of the Newchurch 2 association (Soil Survey of England and Wales 1983) developed over Brickearth and Thanet Beds (Institute of Geological Sciences 1966). At the time of the survey the field was in pasture.

Magnetometer survey has been shown to be very effective on other Roman settlement sites such as Wroxeter (Gaffney et al 2000), Silchester (Martin 2000) and Verulamium (Linford 2000) and was therefore the method of first choice at Richborough. The survey was conducted over all the grid squares (Figures 1 and 2) using the standard method outlined in note 2 of Annex 1. Plots of the data-set are presented as both an X-Y traceplot and a linear greyscale, at a scale of 1:1000 on Plan A and superimposed over the base OS map (1:2500) on Figure 3. The only corrections made to the measured values displayed in the plots were to zero-mean each instrument traverse to remove heading errors and to 'despike' the data through the application of a 2m by 2m thresholding median filter (Scollar et al 1990; 492) to reduce the detrimental effects produced by surface iron objects. In addition the lower and upper values of the data have been trimmed for presentation as a traceplot on Plan A.

A more limited earth resistance survey was conducted in the centre of the field, where the topography is most suggestive of structural remains, in an attempt to clarify subtle anomalies noted in the magnetometer data. Measurements were collected with a Geoscan RM15 resistance meter, MPX15 multiplexer and PA1 mobile probe array in the Twin-Electrode configuration. Readings were collected using the standard method outlined in note 1 of Annex 1. Plots of the data-set are presented as both an X-Y traceplot and a linear greyscale, at a scale of 1:1000 in Plan B1 and B2. Plan B3 and B4 show the same plots of the data after a high-pass filter has been applied. A linear greyscale of the filtered data has been superimposed over the base OS map (1:2500) in Figure 4.

A trial electromagnetic survey was carried out over six grid-squares (Figure 2) using a Geonics EM38B with on-board digital acquisition. This instrument allows a compatible data logger to record simultaneously both the in-phase and quadrature response of the received signal, operating at a frequency of 14.2kHz. Due to the limited time available for the survey the data was collected at 1m x 1m in a vertical coil orientation only, where the in-phase and quadrature response are proportional to the subsurface magnetic susceptibility and conductivity respectively (eg Linford 1998). Plots of the data-sets are presented as both X-Y traceplots and linear greyscales on Plan C1-2 and 5-6. In addition the quadrature phase (conductivity) data has had a high-pass filter applied, the results of which are presented on Plan C3-4. Note that for consistency with the earth resistance data, values of high conductivity (low resistance) are plotted as black and low conductivity (high resistance) as white.

A comparison of the data-sets recorded by all the techniques, in the area surveyed by the EM38 are presented as linear greyscales in Plan D.

A graphical summary of the anomalies discussed in the following text is provided on Plan E1.

The magnetic response is subdued (~ ±1nT) over most of the site, although the area has been disturbed in several areas by recent intrusions. Examples of the latter include [1], a response from the fence surrounding the field, and [2-4] - possible sites of military air defence activity during WWII (Mr Daw pers comm). Four isolated responses [5-8], probably indicate ferrous debris, perhaps from excavations. The disturbed linear anomaly [9] is likely to be a buried cable or ferrous pipe (see below). A diffuse area of slightly raised magnetic response [10] may be connected with [9].

Of greater archaeological significance is an extensive area of increased magnetic noise [11] on the north-eastern side of the field. Within this there are several linear positive magnetic anomalies [12] that are likely to be ditches, foundation trenches or burnt wall foundations (cf Gaffney et al 2000). The responses (~2.5nT) clearly stand out from the background (see Plan A1) and their rectilinear arrangement hints at the lay-out of buildings, with intervening spaces suggestive of routeways. In the vicinity of [12] are several non-linear anomalies [13-15] (6-12nT in strength) that could indicate pits or hearths.

Similar anomalies to those just described can be seen on the southern side of the field [16-17], although these are less distinct and any pattern to which they belong is difficult to detect against background variations.

The positive magnetic anomaly at [18], and an adjacent area of raised magnetic response, may be the location of an excavation (although it is unclear whether or not excavation took place here: Roach-Smith 1850). The weaker positive magnetic responses at [19-22] appear to indicate parts of a sub-circular pattern that seems the most likely candidate for the amphitheatre. However, due to the indistinct and disconnected nature of these anomalies, and the intense disturbance recorded nearby it is not possible to clarify this interpretation further.

A graphical summary of the anomalies discussed in the following text is provided on Plan E2.

The earth resistance survey has significantly enhanced the results of the magnetometer survey. The data is dominated by the high resistance anomalies [R1-2], some 7-10m wide, which clearly identify the general configuration of the amphitheatre. There is evidence for at least two entrances, measuring about 10m in width, one to the south-east and one to the north-west. Although the interior dimensions provided by [R1-2] correspond with those recorded for the exterior of the ‘ellipse’, there is otherwise very little correlation between the resistivity anomalies and the simple walled structure illustrated by Roach-Smith (ibid 162). It would appear that the latter is a considerable over-simplification of a more complex and larger structure.

Quite unexpected are the two large and very pronounced areas of high resistance [R3-4] that occur approximately halfway between the two gaps. These are difficult to interpret but imply the presence of substantial structural remains and/or rubble. Their positions suggest that these may be additional entrance ways. Roach-Smith noted the evidence for an arched structure at Rolfe’s ‘northern’ entrance (ibid 1850;163), and it may therefore be a possibility that these two opposed anomalies relate to elaborate entrance features. The two anomalies take slightly different forms, with an area of lower resistance [R5] encroaching onto [R3]. Along the lengths of [R1-2] are also linear areas of slightly lower resistance [R6-7] that seem to indicate either inner gaps between the walls, or robbing; [R7] in particular is suggestive of some later intrusion. [R8] may relate to the later magnetic intrusion of [4], rather to any underlying Roman feature.

The arcs of slightly raised resistance at [R9-10] also probably represent structural features such as walling, although their relationship with [R1-2] is unclear. It seems unlikely that the total width of the amphitheatre perimeter was ~20m and it may be speculated that the relatively weaker definition of [R9-10] implies a different level of preservation, or of depth - thus the possibility of differing phases of construction.

Low resistance anomalies at [R11-12] are located in the eastern entrance and could correlate either with excavation intrusions or features of the entrance itself. The distinct rectangular area of low resistance [R13] cannot be explained but may be a post-Roman feature such as a quarry. Its proximity to the magnetic disturbance at [2] may even indicate some sort of related 20^th century feature.

Other identifiable anomalies seem to be of less archaeological significance: the circular area of low resistance [R14] corresponds with an area of the field that had been used for feeding cattle. The linear low resistance at [R15] has only been partially captured by the survey and so is not possible to interpret. The one at [R16] is barely discernible, and may be of relatively recent origin (see below).

A graphical summary of the anomalies discussed in the following text is provided on Plan E3.

Low conductivity anomalies [EM1-2] replicate the high resistance anomalies associated with the walls of the amphitheatre, including the areas of very high resistance (low conductivity) at [R3] and [R4]. However, that the linear high resistance anomaly [R7] is not clearly reproduced in the conductivity data suggests that this may well be due to a near-surface source. Previous studies indicate that the depth of penetration obtained with the EM38 in a vertical coil orientation is greater than a Twin Electrode array with a 0.5m mobile probe spacing (cf Cole et al 1995). This would also suggest that the linear high conductivity anomaly [EM3], that is barely discernible within the earth resistance data ([R16]), originates from a more deeply buried feature, though probably of recent origin.

The course of the ferrous disturbance identified in the magnetometer survey [9] is also evident as a linear anomaly [EM4] in both the quadrature and in-phase data sets. This latter anomaly is almost certainly a recent service provision to the centre of the amphitheatre depression where the in-phase data has recorded a high susceptibility anomaly [EM5], coincident with a scatter of magnetic noise [10]. Whilst the latter may not necessarily be related to the service, this must be considered a possibility. If not some sort of agricultural provision, and given the location of the site and the natural defences afforded by the amphitheatre topography it seems reasonable to suggest that these anomalies might be linked to some form of military air defence, active during the last war.

A high susceptibility response [EM6] and two discrete high conductivity responses [EM7] and [EM8] are found outside the NE entrance and correspond with the positive magnetic response [18] and the less readily discernible low resistance anomalies [R11-12]. However, the archaeological significance of these anomalies is difficult to ascertain given the history of probable military intervention and antiquarian investigations at the site. A similar high conductivity anomaly [EM9] correlates with both a low resistance response [R8] and an area of magnetic disturbance [4] that again suggests the underlying cause is of more recent origin.

To the N of the amphitheatre the in-phase data demonstrates an area of increased magnetic susceptibility [EM10] that correlates with the magnetic activity [11] revealed by the magnetometer survey. In addition, the location of the former cattle feeder has produced an extremely high conductivity response [EM11] (cf [R14]) replicated as a low susceptibility in-phase anomaly.

All the techniques applied were successful in producing a series of complementary and informative data-sets. The walls of the amphitheatre were readily detectable as earth resistance and conductivity anomalies although the structure was only poorly defined by magnetometry. However, the magnetic surveys provided considerable evidence for related occupation activity to the N of the site. More recent ferrous disturbance was also revealed by the latter surveys and it seems likely that much of this is related to military activity during the last war. Although of limited extent, the electromagnetic survey provided a deeper penetrating data-set than the resistance survey which was conducted only at single shallow mobile probe spacing of 0.5m (owing to the inclement weather at the time).

The results of these surveys indicate that the remains of the amphitheatre are much larger and more complex than those recorded by Rolfe (Roach-Smith 1850; 161-5). The only measurements that roughly correlate are those for the main dimensions of the ‘ellipse’ though this was attributed to the exterior rather than interior of the amphitheatre. The ellipse now appears more likely to have been the wall separating the arena from the cavea, the latter being untouched by Rolfe’s investigations (T. Wilmott pers comm). The resistivity data is suggestive, furthermore, of a complex site history - perhaps of more than one phase.

In addition to locating the amphitheatre structure the surveys have helped identify the probable presence of buildings alongside, especially in the NE part of the field. The relationship of these to the amphitheatre – whether they abut or underlie it - cannot be determined from the geophysical data alone. Expansion of the survey and/or excavation should clarify this relationship.

This initial work has very successfully highlighted the suitability of geophysical survey in the area. There is thus good potential for the deployment of these techniques more widely, for instance over the settlement area around the fort, to help record the full spread of remains at and around Richborough and thereby to help indicate potential areas for selective intrusive work. Furthermore, despite the saturated soil conditions at the time of the survey the conductivity measurements are not so high as to preclude the use of ground penetrating radar over the site of the amphitheatre at a later date.

British Geological Survey, 1966, Dover, England and Wales Solid and Drift, Sheet 290, 1:50,000.

Cole, M. A., Linford, N. T., Payne, A. W. and Linford, P. K., 1995, Soil magnetic susceptibility measurements and their application to archaeological site investigation, Science and Site: Archaeological Sciences Conference 1993 (eds. J. Beavis and K. Barker), pp144-62, Bournemouth University, Bournemouth.

Johnson, J. S., 1999, Richborough and Reculver. English Heritage, London.

Gaffney, C F, Gater, J A, Linford, P, Gaffney, V L, and White, R 2000 ‘Large-scale Systematic Fluxgate Gradiometry at the Roman City of Wroxeter’. Archaeological Prospection 7, 81-99.

Kendall, P and Wilmott, T., 2001 Richborough Castle Project. Paper to English Heritage Historic Settlement and Landscapes Advisory Committee, May 2001. Unpublished.

Linford, N. T., 1998, Geophysical survey at Boden Vean, Cornwall, including an assessment of the microgravity technique for the location of suspected archaeological void features, Archaeometry, 40, pp187-216.

Linford, N 2000 Verulamium, St Albans, Herts, Report on geophysical survey, January 2000 Ancient Monuments Laboratory Report Series 64/2000.

Martin, L., 2000, Silchester Roman Town, Hampshire. Report on Geophysical Survey, March 2000. Ancient Monuments Laboratory Report Series 65/2000.

Roach-Smith, C., 1850, The Antiquities of Richborough, Reculver and Lymne in Kent. London.

Scollar, I. Tabbagh, A. Hesse, A. and Herzog, I. (eds.), 1990, Archaeological Prospecting and Remote Sensing. Cambridge.

Soil Survey of England and Wales, 1983, Soils of England and Wales, Sheet 6, South East England.

Figure 1 Location plan of survey showing the relationship of the amphitheatre field to the Roman fort (1:5000).

Figure 2 Location plan of survey grid squares over base OS map (1:2500).

Figure 3 Linear greyscale of magnetometer data superimposed over base OS map (1:2500).

Figure 4 Linear greyscale of resistivity data superimposed over base OS map (1:2500).

Plan A Traceplot and linear greyscale of magnetometer data (1:1000).

Plan B Traceplot and linear greyscale of resistivity data (1:1000).

Plan C Traceplot and linear greyscale of electro-magnetic data (1:1000).

Plan D Linear greyscales of all datasets for the area covered by the electro-magnetic survey (1:1000).

Plan E Graphical summary of significant geophysical anomalies (1:1000).