Period 0-1 | Period 2 | Period 3 | Periods 4-5
Period 0: The pre-bank surface | Pre-bank cultivation
Period 1: The bank | Front revetment | Rear revetment | Supposed palisade trench | Intra-mural walkway | The ditches | Corner turret
The pre-bank surface
In nearly all of Wainwright's trenches the pre-bank surface was examined and indicated in section, a situation which has resulted in the exercise of some interpretive licence by Radford. Nowhere is it clear from the drawn sections of any differences between 'turf' and 'subsoil'. The layer on which the bank was built was of varying thickness, being of rather greater depth in sections in the north-west corner than elsewhere. In one section, at least, W.II, the depth of soil shown on Wainwright's original drawing is by no means as definite as on Radford's published section. The existence of a greater thickness of soil in one of the trenches of this area, which could be taken as implying that turf was not stripped from the bank area, could explain the lack of humus in the make-up of the bank, though the chances of such direct correlation in practice in a single section are slight. In a number of places Radford has described the soil under the bank as representing an 'old turf line'. This is not warranted by the evidence - and is anyway inconsistent with Radford's invariable conclusion that the pre-bank land surface was under cultivation. All the evidence recorded by Wainwright is consistent with the inference that turf and presumably some of the subsoil was stripped for use in the construction of the bank from the areas of the bank, berms and ditches as a matter of course. This interpretation appears to have been favoured at least on one occasion by Wainwright himself who, in referring to trench W.XVII (centre of west side), records in his site notebook now in Wiltshire Heritage Museum (notebook p.49): 'Old topsoil had been scraped away before bank raised in position - sometimes the scrape went deeper than at other times'. This situation duplicated exactly that observed in the 1975 excavations in the south-west corner.
Radford has suggested in several places that variations in thickness of the soil sealed under the bank have preserved evidence of pre-bank cultivation, and that the lack of such variation indicates the absence of cultivation. Thus in trench W.V (north-west corner) the greater depth of surface soil under the bank is seen as being 'probably connected with the post-Roman ploughing, the section being cut along the line of a ridge' (1972, 67). The same conclusions are drawn in trench W.IX (south-east corner, W-E section, east defences), where Radford states explicitly (1972, 76) that 'the differences in thickness [in the soil layers noted by Wainwright] must be the result of pre-bank activities, probably ... ploughing; the ridges of a ridge and furrow pattern would be protected under the bank, but elsewhere degraded by later cultivation'. In trench W.1 Radford merely states (1972, 67) that the soil under the bank 'had been cultivated' without adducing evidence, and has drawn a similar conclusion in relation to the soil under the bank in W.XI (1972, 70). In trench W.XIII (north-east corner), Radford postulates (1972, 72) the existence of a 'plough ridge cut obliquely' which 'would conform to the pattern of ridge and furrow visible east of the borough'. In trench W.VI Radford states (1972, 68) that the observation that a layer of 'black soil' underneath the bank in Wainwright's section gave way horizontally to an 'old turf line' which 'indicates that cultivation had stoppped short at the head of the scarp above the river flats'.
In short, Radford has tended to assume almost as a matter of course that the soil was cultivated before the bank was laid over it, and that every variation in its character or thickness noted by Wainwright can be adduced as evidence that this was the case.
If such conclusions could be substantiated they would clearly be of some considerable interest, in view both of the general paucity of early evidence of ploughing in general and of ridge-and-furrow in particular, and of its potential importance for the history of agriculture (Fowler 1976, 28-9, 45 and passim). However, Radford's conclusions about the positive evidence of ridge-and-furrow, or about patterns or the existence (or indeed the non-existence) of any sort of pre-bank agricultural activity, are not sustainable from the evidence. Wainwright's original records were nowhere detailed enough to allow these sort of distinctions to be made, or any inferences about pre-bank activity to be drawn from them. His reliance on vertical sections, the fact that no variations in soil thickness in the horizontal plane were recorded, and the absence of any differentiation in his drawings between soil horizons, means that such variations in the thickness of 'soil', or distinctions between 'soil' and 'turf', have little evidential value. They are as likely to have been the result of variations in the depth, degree or method of turf-stripping prior to the construction of the bank as any other cause.
The poor survival of the structure of the bank, of which in most sections only the lowest half metre was preserved, has meant that little evidence remains for its method of construction. Some conclusions can be drawn, however, from a comparison of the 1975 excavations with earlier records. In the first place, Radford's statement that 'there were traces of three lines of timbering - inner, medial and outer - related to a 30ft wide bank' (1970, 101) is not supported by the evidence. No timber strapping is recorded in any of the trenches across the defences; the bank consisted of a dump of clay which in most of the trenches was interleaved with irregularly disposed layers of darker material. These must, as Radford has concluded, represent layers or lenses of turf. The bank on the north side of the defences (shown in Wainwright's record of trench W.V - north-west corner), however, comprised numerous layers of gravel with inclusions of different coloured clays, suggesting that the material was derived from the lower flood plain immediately to its north. The greater concentration of turf at the corners of the bank as opposed to the sides, and the general use of turves as front and rear revetments (below), demonstrate that the turves were used as structural material to contain the clay. In the south-east corner, however, the presence of turves near the centre of the bank (also observed in the south-west corner in 1975 - see above) suggested to Radford 'the existence of a wooden structure, possibly a turret'. This inference is not warranted by the evidence.
Table 1: widths of the bank from excavated trenches
Measurements are between the front of the wall to the inner edge of the intra-mural walkway (nearest the bank)
|Position in trench (feet)
|Bank widths (feet)
|Bank widths (metres)
The use of turves to revet both the front and rear of the bank was observed in the 1975 excavations. These features appear in nearly every section across the defences, although they have in many places been misinterpreted by Radford. In his own trench R.V (south-west corner) he interprets the 'darker ... and more loosely packed' material fronting the bank as material filled in behind the added wall (1972, 80), though it is clear from the re-excavation of the same trench (section 4) that this material was in fact part of the original turf revetment, which at this point was quite substantial.
The same reservations exist over Radford's interpretation of trench W.XI (east defences). He states that the foundation trench for the wall was cut into the front of the bank with a 30-degree slope (i.e. 30 degrees to the horizontal, implying incidentally the removal of most of the bank itself) with the space behind the wall later filled with loose material. Wainwright's section however shows a near-vertical face on the line of the back edge of the stone wall, which he indicates as the 'presumed line of bank'. The 'reddish clay and soil' marked on his section behind this line can therefore more easily be interpreted as the remains of the front turf revetment. This conclusion is supported by several of Wainwright's photographs in which these relationships are clearly shown (especially no. IV/4 and Q.7).
Elsewhere the dark 'interleavings' within the clay of the front of the bank observed by Wainwright are correctly interpreted by Radford as part of the front turf revetment of the bank (e.g. trenches IX [E-W diagonal and N-S sections], W.XIII [N-S section] and W.XX ). In several of these places the turves formed roughly horizontal bands interleaved with the clay. This contrasted with the discrete mass of turves observed in other places, which implies that the mass at the very front of the bank had been cut away. This is one reason for accepting the inference favoured by Radford (against Wainwright - see below) of the later insertion of the stone wall into the front of the bank. Furthermore, in a number of places the clean clay inserted as packing behind the lowest courses of the wall (see below) butted up against a near-vertical face of this same dark gritty material, which was itself interleaved with the clay of the original bank. In these instances the dark gritty material must represent the remains of a front revetment of turves cut back with a vertical or near-vertical face to receive the wall and the clay packing behind it.
The so-called 'rear revetment' at the back of the bank has also been consistently misinterpreted by Radford. In most of the trenches where Wainwright excavated in this area he recorded a flat layer of stones some 1-1.5 metres in width lying over a layer of turf, or the hollow representing the robber trench of this same feature. This feature has in every case been interpreted by Radford as a rear revetment wall (e.g. trenches W.I, W.II, W.IV, W.V, W.XI and others) - probably following Wainwright himself who referred to it as a wall. The same feature has, however, been shown in the 1975 excavations to have been an intra-mural walkway.
However, a layer of dark material in some sections, between this feature and the clay of the bank, indicates the presence of a turf revetment which is generally rather slighter than the equivalent revetment at the front of the bank. It can be inferred that this was the main way that the back of the bank was stabilised. This can be seen in the following trenches:
W.I - width 2ft (0.6m). Shown in section, not mentioned in Radford's report.
W.II - width 1ft (0.3m), thickening to 2ft (0.6m) at a higher level. Wainwright's section shows this material to have been darker than the layer overlying the walkway (the latter representing degradation products of the bank), which is confirmed by his photo no.29, spool 5, 1954.
W.X - width 3.5ft (1.05m); interpreted as a rear revetment (possibly in the absence of the stones of the walkway, which appear to have been removed at this point).
W.IX (W-E section) - width 1-2ft (0.3-0.6m), probably represented by the soil between the back edge of the walkway and the back of the bank.
W.XIII (N-S section) - width 2ft (0.6m); interpreted by Radford (1972, 71) as a rear revetment.
Supposed palisade trench
During the 1975 excavations, the sections across the south defences showed the presence of a gully or depression along the line of the front of the inserted wall of period 2, into which was usually placed the lowest course of stones or the external plinth of the front of the wall. This same situation has been observed in the following trenches:
W.II This feature is interpreted by Radford (in my view correctly) as the setting of the external plinth of the wall;
W.IV (N defences)
W.IX (N-S section), where a marked gully, containing what must be the bottom stone of the front course of the wall, is shown on the section at the front edge of the wall, but not mentioned in the report. Radford mistakenly says that 'No trace of the wall remained in position' (1972, 76). A similar gully appears in several trenches in the same corner of the defences, but is not mentioned in Radford's report.
Trench 3, square XV: a rounded gully on the line of the front of the wall, overlain by the mortar and stone spill from the robbing of the wall (period 3), is shown in one of Wainwright's photos (1954/8, no. 24, site notebook p.13, Aug. 1954).
Similar features are drawn on Wainwright's sections (unpublished) of squares XX and XXIII, S side.
W.X (S defences)
W.XX: Radford describes in square E10 - also shown on the section - a 'slot' containing the stones forming the front of the wall.
This feature is shown in unpublished drawings of all four trenches on this side.
Conclusion - Radford concluded that this feature was a trench dug for a palisade of vertical timbers, set on either a horizontal baseplate or in the soil, to support the front turf revetment. It has been found in trenches on all sides of the defences. However, no post-holes or marks of a beam emplacement within it have so far been found on any occasion. Even allowing for the fact that Wainwright was not looking for post-holes - since he considered the wall to be a primary feature - there exists some considerable doubt in my mind whether this feature represents a palisade trench. In the cutting back of the bank and the preparation of the foundation of the wall the original line of any former palisade is likely to have become obliterated. The fact that the lowest course of the front of the wall is invariably placed within this feature therefore implies that it must be associated with the construction of the wall rather than with an earlier palisade. It seems therefore more probable that it is merely a marking-out trench for the front of the wall, or, perhaps more probably, a foundation trench for the front course of the wall, which formed a plinth on which the upper course (or sometimes two courses) were offset, to prevent its lowest courses from slipping forward under the weight of the completed structure. This is discussed further in relation to the wall, under period 2.
A feature which has been recorded in trenches on all four sides of the defences was recognised in the 1975 excavations as an intra-mural walkway, placed at the original ground level around the rear edge of the bank. This provides a considerably better explanation for these observations than its interpretation by Radford as a rear revetment wall (see above). In excavations on the south-west corner in 1975 this feature comprised two principal components: a) one or two layers of flat stones laid horizontally over b), a layer of dark gritty soil which was invariably thicker than that under the bank, and which is interpreted as undisturbed turf. To demonstrate the validity of the hypothesis that this feature was an intra-mural walkway or 'wall street' it would be necessary to show that it ran around the whole of the inner perimeter of the defensive circuit. There is in fact abundant evidence in Wainwright's original excavation records, and in some of the published sections, to show that it originally did so.
However, the interpretative picture is complicated by the fact that Radford also interprets a wide scatter of stones observed in a number of places behind the bank itself as an intra-mural street. Both the 'rear revetment wall', and this scatter of stones forming an intra-mural street, are so marked on most of his plans of the trenches. This scatter of stones was shown in 1975 to have overlain and extended beyond the laid stones of the intra-mural walkway, forming a feature which can best be interpreted as the destruction deposits (period 3) of a wall built in period 2 part way up the rear of the bank. This same feature, showing similar stratigraphical relationships, is also discernible in many if not most of Wainwright's trenches, and can most conveniently be discussed here.
The occurrences of both the reinterpreted stone intra-mural walkway, and the overlying scatter of stones thought by Radford to be an intra-mural street, are listed below.
W.XX W.I - Radford notes 'a pile of dark soil or turf ....with a slight setting of stones'. This is shown in the published section as a horizontal layer of flat stones extending beyond the edge of the trench.
W.II - Similar feature as in W.I. However, Wainwright's photographs of the trench (spool 5, nos 26 and 31, 1953) show a layer of closely packed flat stones (quite dissimilar to those on Wainwright's drawn plan - which is reflected in Radford's published plan) which was isolated during the course of the excavation on a pedestal, implying a greater thickness of soil underneath.
W.IV - Wainwright's section shows a space 3ft (0.9m) wide, marked 'rear wall: no mortar', which can be interpreted as marking the position of the walkway, subsequently robbed. The scatter of stones inside this were interpreted by Radford (1972, 67) as an intra-mural street. This interpretation does not, however, accord with the information on Wainwright's original section, which shows several large angular stones which are marked 'scatter of stones tumbled, presumably from rampart, after wall [i.e. stones from the walkway] was removed'. These are shown on Wainwright's plan as an irregular scatter of bulky angular stones (probably Coral Rag) extending 17.5ft (5.35m) to the end of the trench.
W.V - Features similar to those in W.IV were found in this trench.
W.VI - The published section shows a disturbed area at the rear of the bank. Wainwright's section, however, notes a layer of 'biggish stones' forming a feature 3ft 9in (1.15m) in width in the opposite face of the trench.
W.VII (incorrectly described as W.VIII in Radford's text; 1972, 68). A near-horizontal layer of flat stones one course in thickness, cut into the back of the corner of the rampart, is interpreted by Radford as belonging to the same phase as the inserted front wall, and as possibly being the remains of a building. This feature is similar in all respects to that noted in 1975 in the south-west corner, and can on this evidence best be interpreted as the floor of an internal corner turret contiguous with the stones of the walkway, and belonging to period 1.
W.XI - A layer of flat stones recorded on the west end of the published section (not mentioned in the text), which could be the walkway, is shown as being raised above the level of the pre-bank soil.
W.XIII (E-W section, not published) - A layer of flat stones at the rear of the bank 4ft (1.2m) in width and one to two courses thick, is described by Wainwright as a 'rear wall', and is described by Radford (1972, 72) as lying on the layer of soil at the back of the bank. It is interpreted by Radford - together with a scatter of stones reaching 10ft (3m) behind the bank - as forming part of 'the base of a ramp or stairs leading up to the walk along the bank'. This latter scatter of stones is, however, neither a street (as Radford suggests elsewhere) nor a structural feature, but the tumbled deposits formed during period 2B/3A. The narrower layer of flat stones must here - as elsewhere - be the stone walkway. Both of these deposits are indicated on Radford's plan (Radford 1972, 70, fig. 3).
W.XIII (N-S section) - A possible stone feature at the rear of the bank, which could be the walkway, is partially drawn in the section, but not mentioned in the text.
M - Maddison's section of 1948 (not published by Radford) shows a horizontal layer of several flat stones at the rear of the bank (not mentioned in Radford's text, but marked on his plan, (1972, fig. 5).
W.IX (W-E section) - Wainwright's section and plan shows a horizontal layer of stones 4ft 8in (1.4m) wide with a straight western (inner) edge in plan, raised above the level of the soil underneath the bank, with other stones scattered on the surface to the west. Radford's description is misleading, and the scatter of stones mistakenly indicated on his plan as a roadway.
W.IX (N-S section) - The published section and plan shows a layer of flat stones one to two courses thick about 6ft (1.8m) in width at the rear of the bank. Wainwright's photo no. 1954/6, 23, clearly shows this layer at a higher level than the base of the trench to the north, suggesting (as in other instances) its construction on a greater thickness of soil/turf. The section also indicates a further layer of rounder stones (?Coral Rag) tumbled onto the surface of this feature and spread over the rest of the section - which deposit is noted by Radford (1972, 77) as 'a considerable spread of loose stones, which extended to the end of the section opened'. The interpretation of this latter as a separate deposit to the stones of the walkway is confirmed by Wainwright's site notebook (vol. 2 (1954), p.7), where these two stone layers are clearly differentiated. Radford correctly observes that the 'wall' [i.e. the flat stone intra-mural walkway] is contemporary with the bank, but comments (1972, 77) that '...it is tempting to interpret the "wall" ...as the base of a stone stair leading up to the wall walk along the top of the bank and to a possible wooden corner tower'. Since this is a linear feature, this is entirely fanciful; the temptation should be resisted.
W.IX (square XIII - inside corner of bank) - Plan and section not published. Wainwright's plan (site notebook 2 (1954), p.17) is unclear, but, together with the site photos, indicate a horizontal spread of flat stones over the eastern half of the trench and in the south-east corner (i.e. against the inner edge of the bank), raised about 9in (0.25m) above the soil level, As in W.IX (N-S section) (above), this must represent the walkway overlain by a further scatter of stones over the whole trench.
W.X - The former position of the (robbed) walkway is possibly indicated by a shallow trench 5ft (1.5m) in width.
Conclusion - The presence of the walkway of flat stones laid around the inner edge of the bank is established on three sides of the defences, and in three of its corners (all except the north-east). On the fourth side (the northern) what was probably this feature was noted on one side of one of the trenches (W.VI) though was disturbed on the other, and in two other trenches on this side (W.V and W.XIII) its position was marked by disturbed areas which were probably robber trenches. It can be stated with some certainty, therefore, that this feature ran around the inside of the whole length of the defensive bank. It was furthermore distinguished quite clearly in the original excavation records as being separate from a layer of tumbled stones which lay over and beyond it, and which has in all places been incorrectly interpreted by Radford either as the intra-mural street itself or (even more improbably) as a ramp to a stairway leading to a wall-walk at the top of the bank.
The difficulties of recognising the clay filling of ditches dug into clay have already been remarked upon. It is not surprising, therefore, that in several places earlier excavators have not observed their presence where they were anyway not expected. It is a bonus, therefore, that ditches whose significance was not realised at the time, have in fact been noted in a number of the trenches. Their positions are in all cases appropriate to those of the inner, centre or outer ditches noted in 1975 on the south-west corner, and are given below in Table 2.
Table 2: Observed occurrence of ditches in earlier excavations
|positions of ditch
|inner edge at S.8 - 30ft (9.1m) N of inner edge of wall
|probably inner ditch
|48-54ft (14.6-16.4m) E of inner edge of wall (Radford erroneously states 40-50ft)
|S.135-144ft: 46-55ft (14-16.7m) S of inner face of wall (not shown on published section)
|centre ditch 
|Inner edge at E.116: 29.5ft (9m) from inner face of wall (not shown in Radford's published section, fig. 6).
|inner ditch 
|E.129-141ft: 43.5ft to 56ft (13.2-17m) E of inner face of wall
|S.156-160ft (end of trench: inner edge at 47ft (14.3m) S of inner edge of wall.
|centre ditch 
|* pne = part of ditch not excavated
Notes to table
 The inner ditch was probably located where the trench was not excavated in the area S.109-121ft. This seems to be indicated by a deposit of gravelly material on the lowest level of the berm between S.107-109 which, like a similar deposit on the inner edge of the inner ditch on the south-west corner, can be interpreted as a weathering hollow on the inside of the ditch, filled in period 2B. This deposit also appears to have been covered by the mortar and stone destruction deposits of period 3, which it can be inferred lay on the inner edge of the inner ditch.
 The ditch was probably located where the destruction debris of period 3 ended on the inner berm. The ditch was partly filled with large stones, not shown (see comments under period 3)
 The trench was not excavated on the likely position of the inner ditch.
In Knocker's trench (published by Radford 1972 - trench A1, centre of west side), there is evidence to suggest the presence of all three ditches:
Inner ditch: destruction debris from the wall (period 3) extended from the surviving outer edge of the bank 23ft (7.0m) westwards over the berm. If this included the width of the destroyed wall the width of this deposit matches exactly the width of the inner berm plus the wall observed in the south-west corner, over which destruction debris was also spread. It can therefore be surmised that the inner ditch was between 23-30ft (7.0-9.13m) from the remaining front of the bank (W 46-53ft on section).
Centre ditch: this probably lay 42-52ft (12.8-15.8m) west of the remaining front of the bank to include the eastern part of the 'disturbed clay', shown on the published section as running west from 42ft west of the bank.
Outer ditch: the ditch filling of 'disturbed clay' was recorded from 70ft 6in (west edge of concrete path) to 97ft (21.5-29.5m) west from the rear of the wall. This must be the wide outer ditch, the east edge of which is uncertain and the west edge of which was not reached in the trench.
The practical difficulties mentioned above of recognising features cut into clay, which are alluded to in Radford's account of Knocker's trench, mean that the exact positions and the dimensions of both the centre and the outer ditch in this trench are less certain than the fact of their existence.
Conclusions - The centre ditch has been observed on every side of the defences except the north, and the inner ditch on every side except the east. Since their positive recognition was in most places anyway fortuitous, it seems reasonable to conclude that at least the inner ditch encircled the town, and that the centre ditch existed on all sides except the north, where the land lies in the flood plain of the river. This conclusion seems to be supported by the greater distance of the inner ditch from the bank on the north side (trench W.VI). The wide outer ditch has been observed in two places on the west side (in 1975, and Knocker's trench K.1), and probably on the east side (Maddison's trench M, south-east corner). Topographical arguments (discussed in Part 3) suggest that, like the centre ditch, it also existed on the south side and was probably absent on the north.
The inner two ditches appear in all places to have been small, shallow (not more than 5-6ft (1.5-1.8m) in depth), and with a rounded U-shaped profile. Their depth is the same as that of the outer ditch. In Part 1 it is suggested the V-shaped profiles of the inner two ditches on the west side at the south-west corner, observed in 1975, were caused by repeated cleaning and over-deepening. The comparatively large widths of the centre ditch in several places may also have been caused by both natural erosion and repeated scouring. The dating and function of these ditches is discussed in Part 3.
The existence of a turret in the internal corner of the defences was inferred in the south-west corner from the presence of a substantial post-hole set in the stones of the intra-mural walkway. No equivalent evidence was found in any of the other corners, though in both the north-west and south-east corners the flat stone walkway on the internal corner formed by the bank was wider than elsewhere. It has already been pointed out that Radford's suggestion of the existence of a turret in the south-east corner, though it might be expected, is not a legitimate inference from the available evidence. However, the existence of such a turret at the two corners where the enlarged walkway was observed, as well as probably at the north-east corner, seems on the evidence available to be a strong probability.
Periods 0-1 | Period 2 | Period 3 | Periods 4-5
© Internet Archaeology
Last updated: Mon Jul 7 2003