Eynsford Castle: The Moat and Bridge

EYNSFORD CASTLE: THE MOAT AND BRIDGE* By S. E. RIGOLD, M.A., F.S.A., F.R.Hist.S. and A. J. FLEMING, M.A. THE first report on Eynsford Castle1 did not cover excavations in the moat, which has not yet been completely cleared, nor any aspect of the bridge, which is in fact a succession of timber bridges, of exceptional interest and wide relevance. A brief structural and comparative study of these bridges, presented at the Collogue du Chdteau Oaillard in 1972, is in the press and a fuUer version is in preparation for Medieval Archozology. This aspect is therefore treated summarily in what foUows, which concentrates on the excavations in the moat and the associated finds, in continuity with the previous report. THE QUESTION OF THE ENTRANCE From the end of Phase X, that is from the construction of the HaU and Gate-tower, the entrance to the enclosure was at the southeast and, on aU evidence, the curtain waU below the gateway was rebuUt from the ground at the same time. Since there was no trace of any bridge in this quarter assignable to Phase W, it was suggested that the original entrance had been elsewhere. The most likely position seemed to be the north-west, where the curtain was thickened inside and out and the moat appeared to be narrowest;2 a possible alternative was at the north-east, where there is a narrow, but apparently primary, breach in the waU, below which it is disturbed by a brick-lined weU of the kennel period (Phase K). In 1972 a section was dug by A. J. Fleming to test the 'north-western' hypothesis, from a point where, for reasons which wfll appear, the waU had been underpinned and buttressed in the time of E. D. TiU (1897 onwards) and needed further stabilization when the buttress was removed. The results gave no positive support to the hypothesis and showed that the apparent narrowing of the moat was deceptive. The 'north-eastern' position has not yet been tested but a section previously cut across the moat a Httle to the south of it, at v, showed nothing in favour of this position save that the moat was relatively shaUow in this sector and the same width in its bed, about 9 m., as in section t, less than in the bridge-area. No cut ditch is now discernible on the west side, where the moat runs into the flood-plain. * The Department of the Environment has contributed to the cost of the publication of this paper. 1 Arch. Cant., Ixxxv (1971), 109-71. a Ibid., 122. 87 S. E. BIGOLD AND A. J. FLEMING This report covers (Fig. 1): 1, the narrow north-eastern cutting at v; 2, the whole environment of the south-eastern approach and bridge, with co-ordinate sections at IT (continuing ij) and a; 3, the north-western section, i, described by Mr. Fleming, with corrections to. the description of the curtain at this point. THE NORTH-EAST CUTTING Here the footings of the curtain had been exposed, presumably by flooding, and later masked by topsoil. The chalk raft on which the curtain is everywhere based (except where rebuilt at the south-east entrance, and on section t) is here 30 cm. thick and below it is a slightly spread footing of two layers of flint bedded in gravel and Hme. This lies on natural clay-with-flints ('brick-earth'), which is eroded to give the appearance of a flat berm about 3 m. wide but which would have sloped from the top of the footings at a gradient of about 1 in 9. Some 10 m. from the curtain what appears to be the rising profile of the original ditch, cut in the natural, crosses the horizon of the base of the footings and rises quite steeply, about 1 in 4, to a counterscarp about 1 -5 m. above them. The bottom of the ditch was not sounded throughout but these measurements hardly aUow it to have been much deeper than 36-3 m. (119 ft.) above O.D., i.e. rather higher than in the bridge area or in section i, indicate a channel about 9 m. wide. The filling towards the bottom was loose and carbonaceous, but the dark sUt seen in the other sections did not appear and there were no early finds. The channel, such as it was, had been filled up with flinty soil capped by a layer of flints and over this a lens of redeposited clay, making the bed practicaUy level. There were no late finds either, but this may be the work of Phase K, as also, perhaps, a layer of flint 30 cm. deep which covered the counterscarp for at least 3 m. This limited test showed that the north-east approach had no advantage of defensive strength. THE AREA OE THE BRIDGE (Figs. 2-6) When the castle came into Guardianship access was by an earthen bank, rising gently, about 1 m. in 20 m. This was as 'improved' by E. D. TiU. The upper soil (Fig. 6, 1) was removed to reveal a bank, sagging down to a point about 12 m. from the curtain and rising thence about 1-5 m. to the entrance-gap, which provided the approach in Phase K. It had a rough metalling, bedded on a mass of flint rubble of which the upper part was purposely thrown up rather than faUen (2), generaUy over 1 m. in depth but graded up towards the curtain and very shaUow about 6 m. in advance of it, where a winged waU (Y) 88 EYNSFOBD CASTLE: THE MOAT AND BBIDGE % X '/A 1)11 / ' y < / A # © k / f ^ ^ • $ $ \ $ $ < ^ •/, S> ^ \ \ < J I \ \ \ \ \ \ \ / I t It \ / •r»j 10 O 30 /0 & A> re. i • ' i ' T • • i • * X.L-=_J £=£ £—a ar FIG. 1. soon emerged, of smaU, weU-coursed flintwork, simUar to, but not identical with, that of the haU chimney-support (u). This was obviously a bridge-pier or the retaining waU of an earthen abutment. MisceUaneous loose soUs were seen sloping up towards the curtain, but they did not form a compact promontory right up to waU Y, and none of it had been removed in Phase K, since not only the flint rubble (2) but two layers of earher debris, 3a, yeUow and mortary but with tile and flint, 89 S. E. BIGOLD AND A. J. FLEMING and 3b, more compacted, containing much more tUe, a Httle pottery and strictly comparable with the Phase D layers within,2 dropped sharply down towards waU Y and piled in depth behind it. The abutment, then, was essentiaUy hoUow, and the upper parts of the sloping deposit were loose, mainly buff denatured mortar (4), looking more like a 'dump' deposited not long before the dismantling represented by 3a and 3b. Transversely, they were not evenly spread and a pUe of crumbled Greensand (4a), breaking the surface in the northern half is D1 C1 PHASE I FIG. 2. B1 U4 1 1 1 1 1 1 4 6 6» perhaps better associated with the compactor layer below (5). The rear of waU Y was weU preserved and protected by the dismantling debris (3a, 3b) piled against it. The front was badly eroded or robbed, except at its lowest courses, though simflarly protected by the debris (3d, 3e—on this side the upper layer contained more tile), which extended down its eroded face. Even if much of the debris derives from the final dismantling of Phase E, rather than of Phase D (ascribed to 1312), the condition suggests that waU Y, though certainly not older than Phase B, had been exposed for a long period and had been buUt much nearer to Phase B than Phase D. It appears, however, to have lost httle in height and still riseB to -within about 50 cm. of the surfacelevel in the entrance-passage. In the first stage of excavation only the unstable upper deposits Ibid., 122. 90 PLATE I Crown Copyright reserved A. Bridge-area, from Curtain, showing inner Abutment. Crown Copyright reserved B. Wall Z, Plate A3 and Foot of Brace. [ face p. 90 PLATE II ; ^ ~ Hfe, * I /X :* • •' DX'.M. Crown Copyright reserved Base of V2, showing Matrices for Timbers and Post (W). PLATE III ft ">*5as B I ; M S ^ Crown Copyright reserved Plates B2, B3, A3 and Wall Y. PLATES IV AND V X^ r PLATE IV. Planking on felled Trestle. Crown Copyright reserved Crown Copyright reserved PLATE V. Felled Trestle on DI. between Y and V1-V2. EYNSFOBD CASTLE: THE MOAT AND BBEDGE were sectioned and stripped—the topsoU (1), the flint-debris (2), the dismantling debris (the '3' layers) and the friable upper deposits against the curtain (4, 4a). These and corresponding layers were removed right across the moat, revealing (see Fig. 4), besides waU Y: —(i) another retaining waU (Z) on the outer Hp, not winged but identical in masomy with Y, similarly eroded before it was buried (in soU, 3c, rather than d6bris) and obviously part of the responding abutment; (fi) a pair of more or less trapezoidal piers (VI, V2), flanking the D1 D2 C1C2 PHASE l ' I I I ' FIG. 3. entrance-gap and helping to retain the dump-like earthern abutment, but different in masomy from Y and Z, being of large broken flints, poorly bonded. Subsequently, low and feeble wings (yl, y2), sloping towards the curtain from the foot of Y, were traced, though poorly preserved—perhaps kerbs to the abutment. Examination in depth foUowed in three stages. In 1961, soundings at the angles of VI, V2, Y and Z showed the unexpected potential of the area. VI and V2 were in two overlapping builds which together descended for nearly 2 m. from the level of the entrance-passage, but entirely through unstable deposits; they were not, as had been thought, reHcs of an early drawbridge-pit. Their lower part, particularly in V2, had been formed around a series of round or roughly trimmed poles, some at least standing on a plank, as sole-plate. Taken together the piers constitute a clumsy, asymmetrical pair, and it is not clear whether 91 ii S. E. BIGOLD AND A. J. FLEMING the timbers were intended as a lacing or just a shuttering. The topmost edges of the flintwork foUow the expected ahgnment of a bridge carried on waU Y and the footings do not penetrate much below the weU-drained layers of the 'dump' against the waU. Both builds may be accepted as late. The masonry, as found, was dry, leached and crumbly, and the matrices of the timbers quite empty (Plate III). On the other hand, the lowest facing-courses of waUs Y and Z were sohd and weU-preserved, with the impressions of upright timbers PHASE ml B3 A3 6M. FIG. 4. in their rendering or overspflt mortar, and just below the footings of each a timber plate (C3, A3) was found in moderate condition, with mortices corresponding with the impressions of the uprights. Preservation of wood was quite unexpected: the level was not far below the empty slot at the base of V2, dry in summer and certainly not permanently waterlogged. The sterile conditions that prevailed up to a distinct horizon, some 35 cm. above the footing of waU Y and just below the base of VI and V2, are stiU not entirely explained but may be due to a combination of the filtering effect of the flint and Hme above, a solution of iron in the seepage and genuinely waterlogged conditions below. It was decided to leave further investigation until it oould be done on an extensive scale. In the event this was in two campaigns: in 1963-4 and, under the direction of D. 0. Mynard, in the summer of 1966, when the whole stratification between waU Y and the curtain 92 n LZDC1 o + o f—\ m Fios. 5A (upper right), B (upper centre) and 7 (Same scale). [face p. 92 XII c. MASONRY XIII c. | TIMBER S/V/\T\\e debris natural Clay/Flint redeposit >XyChally6 ravel X X X X * x x x*xjGreensand o fea| Hint HMF, lint/gravel PV7: I X i, Topsoil Old ^jhumo us Dark gravel soil humus ;J',|'J.'|;;|;,i clayey Gravel Dark silt TTl TTH curtain i WWWMIW7T777I 20 Ft vx/> I 1 1 I! D1 D2 C1 C2C3 Fio. 6. B1? f B3 T B2 / / / A3 A2 A1 EYNSFORD CASTLE: THE MOAT AND BRIDGE was removed, aU the timbers that were sufficiently rigid were lifted and waU Y, which had been, in effect, buUt upon a timber raft, was underpinned with concrete. The width of investigation was 10 m. right across the moat. As in the previous report, the layers down to D or its equivalents, 3a-e, are described in descending order. This has aHeady been done, including, for convenience, the 'dump' deposite 4 and 5b. Likewise, the lower layers are described in ascending order from 'natural', the remains of the bridge and aU masonry treated in context. THE STRATIGRAPHY OF THE BRIDGE (Fig. 6) The base of the sohd curtains, flat and almost level, at about 37 m. above O.D., is separated from the shghtly less even natural soil (almost everywhere clay-with-flints, or 'brick-earth', of varying density) only by the chalk raft and, in some places, by a spread flint footing beneath it, as in sections j8 and v. Towards the south-east entrance, however (section -q; see first report) the natural was falling to nearly 1 m. below the (rebuilt) curtain and outside it it was nearly twice as deep. Prom this point, right across the moat, though no doubt affected by scouring, it was remarkably even, rising from just above 35 m. to about 36 m. at the outer Hp (i.e. under the artificial abutment retained by waU Z) varying by only 0 • 7 m. in 17 m. Here at least, the impression is not that of a dug ditch but an essentiaUy natural flood-plain, with the enclosed area as an island whose steeply falling margin coincides with the curtain, hence the coUapse and rebuilding. The surface of the clay-with-flints seemed more indurated than elsewhere, described by the diggers as 'ballast'. In places, especiaUy where protected by the abutments associated with waUs Y and Z, it was covered by an artificial layer, perhaps an attempt at lining, of mixed baUast and clay (11). The strata are built up either on the natural or on 11 and none contains pottery of phase W, even against the curtain, where what is almost certainly the spread footing of the older curtain (Fig. 6, F) remained beneath the rebuilt one. Below the base of the rebuilt curtain was a spfll of washed flints (7a) at about water-level under ordinary high-water conditions, which may represent an attempt to stanch erosion. VI and V2 were based on this spUl. Below it was a pfle of gravel (9c) against the rough flint footing, F, 50 cm. deep, which seems to correspond with the footing, only two courses deep, in section v. 9o is not primary: below it and driven into the basal baUast, through the ubiquitous dark silty layer, 10 (see below), was a row of riven spiles, G, apparently forming a secondary 'corset' to prevent the spread of the footing, rather than a primary marking-out. The strata rising from the base to the destruction-layer, 3, may 93 S. E. RIGOLD AND A. J". FLEMING be divided into the weU-compacted layers from 10 or 11 to 7, that is to say those up to the construction of waUs Y and Z (with which the uppermost timbers, Al, 01—see p. 90—must be associated), and the looser layers, 6 to 4, accumulated after the construction of Y. AU preserved timbers and aU excavation by stripping were in the lower, compacted strata. Layer 10 was a dark, almost black, riverine sUt, with the consistency of cheese when damp. It appears also on section t and something simUar, presumably redeposited, was found in pockets in the interior, as against the HaU and in the guUy within the curtain. Its even texture, free of peat, suggests rapid deposition under catastrophic conditions, but it was plastic enough to absorb sherds and other fragments. That it had been cast up the slope under the rebuflt curtain, spilling a Httle into the interior, whUe some at least of the earliest timbers had been laid into it in recognizable cut channels and others driven through it, indicates that it was present before Phase X, and the absence of yet earHer timbers or pottery in a context so conducive to preservation seems to signify that the area did not then hold a bridge. With one exception, there was nothing older than sheUy wares of the 'late X' and 'early Y' types associated with the completion of the haU and forebuUding (not the destruction that preceded the haU).3 Such wares were plentiful in layer 10, mixed with shghtly later forms and fine wares not certainly matched in the interior before Phase Z. The new entrance and the first phase of the timber bridge must be associated with the completion of the haU and gate-tower, with layer 10 exposed to subsequent dropping of rubbish. A comparable layer (10a), under the outer abutment suggests similar exposure for at least another 3 m. and the original bridge may have been one bay longer. The timbers set in layer 10, or ultimately covered by its fluxion, were: (i) two squared piles (W) and the probable 'ghost' of a third, driven through the layer into the 'baUast', 2-6 m. from the curtain; (H) two transverse plates, DI and the heavier CI, with six mortices each; (Hi) two massive baulks with chases along the top, Tl, T2, butted, not halved, to Dl and 01, but clearly articulated with them; (iv) another transverse plate with six mortices, Al, under the outer abutment. Though aU these dormant members have a structural conformity their bedding was not uniform: Dl was in a packed channel cut in layer 10, Al let into the hard baUast at the bottom of a scooped trench, while the others were apparently sunk into layer 10 where it was plastic. The distance demands at least one intermediate plate: Bl is conjectured from another channel in the baUast but if there were 8 Ibid., 162. Y5 and the other example oited (properly just west of the hall) axe the earhest stratified Y-types, the introduction of whioh seems to coinoide with the completion of the hall. 94 EYNSFORD CASTLE: THE MOAT AND BRIDGE others they need have left no mark below layer 10. These constitute 'phase I' of the bridge (Fig. 2). The next layers, 9a, 9b, 8, 7, though variable in texture, were essentiaUy dark, graveUy soUs, getting darker as they descended and absorbed more organic sUt and less graveUy as they moved from the curtain, where 9c was almost neat gravel, yet turning from grey to khaki in the same direction. This gradation was more marked in the rapid fills, 7 and 8, than in the slow deposit of 9a. The pottery content of 9a is indistinguishable from 10, with a predominance of 'Z' types, while that of 7 and 8 is distinctive, with wares down to those of Phase B, perhaps even later. In the outer abutment the ascending strata are a raised causeway of flints in sUt, 9d (it seems to date from the long exposure of 9a), and two further elevations, 8a, a thick sandy layer, and 8b, earthy but capped by a flint metalling. WaU Z is cut into these layers. GeneraUy associated with layer 9, but not necessarUy contemporaneous and each in a distinct context, were the timbers grouped as 'phase I I ' of the bridge and certainly of intermediate date (Fig. 3). These were: (i) a re-used member, D2, evidently once a door-durn, placed in the middle of the articulated frame 01, Dl, Tl and T2, to carry reinforcing shores, but set in a channel not unlike that of Dl; (H) a plate at higher level, 02, with only two mortices, apparently reinforcing, but not replacing, 01; (in) a transverse plate, B2, short like 02, with five mortices, bedded on layer 10 (9a was absent here); (iv) another plate, A2, set in layer 9a and clearly replacing Al. The reinforcement represented by D2, and probably 02, is generaUy covered by the fluxion of 9a and aU associations suggest that it was required not long after the phase I construction and weU within the twelfth century. Clumsy though it looks, it was effective and gave the original structure something Hke a century of further Hfe. Layers 7 and 8 represent a process and are best described in terms of the timbers they contained, the cleaning and lifting of which was the chief consideration in excavating at this level. Within the area of the inner abutment the process is clear: the superstructure of plate Dl, probably five uprights and one shore (U1-U6), sheathed on the side towards the curtain by six strakes of horizontal planking, H, was pushed bodily over, away from the curtain, without displacing Dl more than a few degrees, to form a raft or 'frame-gate' on which to bufld waU Y (Fig. 5, A before, B after, lifting of planks). The rest of the primary structure had been largely removed, but the casting-down of one transverse frame ensured its preservation to a height that would have been impossible had it remained upright. Layer 8, a khaki clayey soil with diminishing gravel, was sandwiched between the feUed frame and layer 9a. Layer 7, a spread of re-deposited graveUy soil, extends 95 S. E. RIGOLD AND A. J. FLEMING from the obviously contemporary 'stanching' of flint at the foot of the curtain (7a), over the frame and some distance into the moat, to form the final bedding of waU Y. It sealed off the driven pUes, W, and a long morticed plate, E, was laid obhquely in it in such a position that it could not have functioned as an articulated member, but merely as a binder for the earthen abutment, and perhaps for VI and V2 (Fig. 4). Of the new functional plates, 03 was laid into layer 7, B3 and A3 laid on the older surface: a secondary sUting, 7b, covered aU three. Subsequently, a pile of flints and Roman tiles, 5a, had been cast on the far side of the moat and had already acquired a topsoil, 5b, before the ditch was cleared out and then refiUed to some depth with soU and flints but no occupation rubbish, 5c. AU this took place before the final dismantling. A static section of timber bridging must have joined waU Y to the curtain since the inner earthen abutment was not completed as the outer was. These two, together with plates 03, B3 and A3, represent phase III (Fig. 4) of the bridge, which was never replaced; hereafter a ramp sufficed. The trestles carried on 03 and B3 were cast down in the final dismantling, that is at the end of Phase D, or at least, Phase E, since they are buried in debris, beneath which were also what are probably to be identified as shght remams of their coUapsed uprights. The truss on A3 may have decayed in situ, as the foot of a brace remained upright in its mortice. The finds were concentrated at the junctions of the layers, mixed with oyster-sheU and other food-remains, particularly on the surface of layer 7 and where 7 lay unconformantly on 9. The subsequent layers 6 (within waU Y, loose gravel and flint) and 6 (within the moat and providing a clay bed 03, B3 and A3), which completed the final bridge, were barren. Roof-tile and pottery, apparently already of 'D' types, begins again, in smaU quantity, in layer 5. Within the moat, but not on the axis of the bridge, and on the top surface of 5b, a Hght, nailed lattice-work of riven slats, Httle more than laths, was found. It can hardly have been part of a parapet and, if not just a hurdle, may be a piece of the filling of some 'open' timber frame. THE NORTH-WEST SEOTIOK (Figs. 7, 8) As already mentioned (p. 87) the reason for the section at i was to test whether an early access to the Oastle existed there. The trenches, dug with contract labour in 1972, were initiaUy laid out as co-ordinates but the cross became a Lorraine cross when extended to examine the timber (Fig. 8). Only the principal transverse section, completed on an offset Hne, is pubhshed; the rest of the trenches merely repeat the sequence. 96 EYNSFORD CASTLE: THE MOAT AND BRIDGE Five major stages emerge (Fig. 7): Stage Layers Description/interpretation V 1 Phase K (eighteenth century) and modern. IV 2-3 Phase D, occupation c. 1300 to dismantling in 1312. I l l 4 SUting-up of moat, perhaps mainly during divided tenure from 1261, or even from the time of mounding. I I 5-8 Late eleventh-century gravel mounding in advance of buUding the curtain waU. I 9-10 Bed of pre-curtain 'proto-moat' containing early wares. Pleistocene YeUow current-bedded gravel, not 'clay-with-flints'. Stage I. Layer 10 was a thin (max. 5 cm.) band of clayey gravel on the natural, compacted as though it had been trampled. I t contained the hand-made bowl-sherd (Fig. 10, 3). Layer 9 fiUs the primary moatchannel, a sticky black-to-khaki silt, simUar to, but 'fouler' than, layer 10 at the bridge (p. 94), containing wood-ash, wood-chips, twigs and leaves (indicating stagnant water) and wares with coarse sheU of the 'Wc to Xc' fabric. The 'proto-moat', with its flat floor is clearly artificial, even if made out of a natural river-meander. The only substantial relic of occupation before the curtain was a straight length of cleft timber, weU preserved and lying with one end in the mud and the rest aslant the strike of the tipped gravels. I t had no evidence of carpentry or nails and, whUe it may have served some temporary purpose at the time of mounding or actual buUding (it would fit a putlog-hole neatly), the fact that it could have been retrieved suggests that it was just thrown away, stuck in the mud and keeled over. Stage II. Gravel layers, 5-8, represent a mounding, visible also within the curtain (section 8), where it was cast up over the more typical clay-with-flints. It contains a very few fragments of pottery of the 'coarse-sheUy' fabric exemplified in groups Wc and Xc and nothing later, even in the upper layers 5 and 6. The rising sequence was: 8E, yeUowish, finer and more compact than 8D which was very unstable, of water-worn cobbly flints covered in fine silt percolated from above; 80, a lens of yeUow gravel; 8B, a thin band of khaki sUt, Hke layer 9, which may represent a respite, a flood during mounding or just a re-deposition; layer 8 proper, barren, mixed brown soU and gravel, becoming increasingly heavy and cobbly as it fanned out over the primary sUt; layer 7, brown sandy soU mixed with smaU yeUow gravel and one or two sherds. The freshly cut section showed yeUowish-green sandy streaks. Layer 6, a yeUow gravel, similar in appearance to the bed-rock of the moat, may represent a cleaning-out or even spoil left over from the cutting of the new bed, for layer 5, a mixture of small 97 S. E. RIGOLD AND A. J. FLEMING gravel and sUt, seemed to dribble across the moat-floor in a thin trampled layer, like 10. The flat bottom of the new moat, already achieved, had a hard metaUed surface of packed flints, extending to the point where the 'natural' rises to the outer Hp and showing the original fuU width at this point, beneath the later constriction. Stage III. Layer 4 consists of a thick bed of soft, clean brown silt, deposited in a sluggish stream—not, however, stagnant and prone to the suffocating effects of a build-up of leaves, as smaU freshwater moUuscs and quite sizeable freshwater mussels Hved there (as they did in layer 10 at the bridge). In view of the swift current of the Darent, we may infer that the flow was regulated in some way, a condition also imphed in the bridge area by the many smaU fragments of timber that remained in situ and were not swept away. The lowest three-quarters of layer 4 is crazed with mud-cracks, sealed by a 5 cm. thick sheU-bed, succeeded by silt which was glutinous when the stage IV dumping took place. Nearer the curtain the sflt is compressed, where not actuaUy mixed with buUding debris, whfle the largest lumps of mortar, flint and tile embedded themselves in the sflt matrix. The silt is varved. It is hoped that analysis of the varves, not yet available, wfll show the duration of the present sUting and, hence, whether the moat was cleaned out or not. It would seem unlikely that it was not cleaned so near the high garderobes, but unless the sUt is thus undisturbed it is impossible to discount an early access to the Castle at this point. Stage IV. Chiefly represented by the layer-3 matrix of variable brown soil, flecked with chalk and mortar and containing pottery, animal bones, Roman tUe, including tegulae, and medieval roofing-tile in great profusion, but dominated by heavy, packed flints, rammed down hard, especiaUy approaching the curtain. This dense material and the yeUow mortar and chalk of layer 2 ran under the forward face of the curtain as far as a Hne of resistance, almost certainly the face of its original footing (probed, not excavated, for fear of weakening the waU). The forward face at this level is clearly that of an added thickening, not weU bonded to the main body, though the upper thickening, supporting the garderobe, may be integral and corbeUed out. The rammed material was intended to underpin the basal thickening, which may therefore be ascribed to this late period, though too shaUow to be more than a 'strip-buttress'. At this point the waU itself is based on gravel: the fissure that began the coUapse of the riverward seotion is nearby and the waU was underpinned and buttressed by E. D. TiU. The range of pottery from layers 2 and 3 is pecuhar—sheUy wares of 'Y' and 'Z' types and then a 'jump' to a very consistent group of good-quahty 'D' type wares. Joins between sherds in 2 and 3 showed them to have been laid down together, but generaUy the sherds of the 98 EYNSFORD CASTLE: THE MOAT AND BRIDGE same 'D-type' vessels were distributed throughout layer 3 from the curtain down to the sUt, while the sheUy sherds, which must be residual, were in clutches. The unambiguous D-period assemblage, mixed with d6bris, suggests that the underpinning actuaUy foUowed the dismantling of 1312: it has abeady been noticed that the flimsy waUs of Phase E were mainly in this corner. Stage V. No deposit records the lapse of four hundred years to Phase KL, the kennels, but the hounds are graphicaUy evidenced by the large quantity of butchered bones from layer IB, a blackish dead humus which contains much broken flint, especiaUy in the hoUow which inexplicably cuts down to the moat bed. None of the upcast spoU was backfiUed, no stratigraphy was discernible and none of the few scraps of delft and country earthenware merits iUustration. From the topsoU, IA, stoneware ginger-beer bottles and marmalade pots, glass soda bottles and wfllow-pattern may record nineteenth-century picnickers. Conclusions. The most important is that an early moat, apparently with some occupation, existed, for however short a time, before the mounding that carries the curtain and, a fortiori, before the earhest Phase 'W deposits in the interior, which are contained by the curtain. Is this Ralf fitz-Unspac, rather than the young WUham I? The new moat, dug to run with the curtain, produced no positive evidence of access in this sector, nor indeed of any constructional activity until the repairs that must be very close to the Phase D dismantling. INTERPRETATION OE THE BRIDGE Definitive information about the species of the timber and dendrochronology are not yet avaUable, but there was a sharp difference between timbers lying hi simUar conditions of good preservation, far from the 'threshold' where they faded into sponginess and stains. Some were blackish, hardened on exposure, and were successfully lifted, for study ex situ; others were reddish-brown and fresh-looking, but decomposed rapidly and were impossible to lift complete. This difference covered heavy plates manifestly of the same buUd and was at first thought to be merely that between oak and elm, but it now appears that oak and elm were both present in the first category, and the second, provisionaUy, may mclude wUlow. DetaUs of toohng have been observed in the lifted timbers, as in the mortices (whether drilled and chiseUed or merely chopped out), or the pegs and peg-holes (whether round or oval, to allow for some wedging effect). These points are reserved for a comprehensive comparative treatment. Date. The arguments, stratigraphic and structural (for neither is conclusive in itseh7) have been given in the excavation section (pp. 88-96) for the unity of the timbers classed in phase I and in phase III, for the intermediate date, but not necessarUy structural unity, of those 99 S. E. RIGOLD AND A. J. FLEMING classed in phase II, for the high probabihty that the timbers (U1-U6) which stood in the mortices of plate Dl, together with their sheathing of boards, were the originals, from phase I, lasting untU they were cast down as a bed for waU Y, and for the association of the phase III plates with waUs Y, Z and probably VI and V2. Of the superstructure of this phase only the foot of one brace remained upright in plate A3 and ill-preserved timbers that appeared to have faUen forward towards the centre of the moat were probably those that had stood in the mortices of A3 and B3. Part of one substantial and apparently original upright, U7, remained on plate CI (Fig. 5), AU other fragments, including the relatively large and weU-preserved SP1, are impossible to place with confidence. The finds, almost entirely from the area of the inner abutment, i.e. around the C and D timbers, make it clear that the phase I structure (Fig. 2) dates from the transformation of the Castle which closes Phase X and begins Phase Y and is probably to be assigned to WUham of Eynsford II in the 1130s, possibly to the early tenure of WilHam III. They also indicate that the phase II reinforcements (Fig. 3, 02, D2) were inserted after quite a short interval, within phase Y, i.e. they are also mid-twelfth century. This does not necessarily apply to B2 or even A2 (which has a few associated finds): the trestles they supported were closer in form to those of phase III than to the primary ones, but there was a considerable interval between them and phase III, so that they can hardly be after the early thirteenth century. None of these include the notched laps which are the most diagnostic joints of this age. The dating of the phase III bridge (Fig. 4) rehes more on the time required for the erosion of waUs Y and Z and the accumulation within the moat, before the phase D dismantling, than on the finds associated with its construction. These at first seemed anomalous: the fresh sheUy wares are of types associated with phase B or, more or less identical and designated BC, with the only, and immediately conformable, layer from Phase C present in the interior*—types, in any case, extinct by phase D; but the sand-tempered wares are closer, as a sample, with a high proportion of rims without upper bevel, to those of phase D than of phase B. The hypothesis of an interval, c. 1266-c. 1300, of royal custody foUowed by divided tenure, does not preclude good management. A new, non-defensive bridge would fit weU early in this context, say c. 1270 and would give a new dimension to phase 0 and to the latest group of sandy wares. Structure. The basal supports—usuaUy aU that remains in situ—of «Ibid., 156. 100 EYNSFORD CASTLE: THE MOAT AND BRIDGE timber bridges in moats have been classified5 as: (1) earth-fast posts; (2) isolated transverse sole-plates,6 sometimes crossed by short toepieces, 7 which carried separate trestles; (3) quadrilateral groundframes with sole-pieces in both directions,8 which carried various rigid, belfry-like structures. Most known bridges faU into one only of these categories, but the phase I bridge at Eynsford incorporates elements of aU three. The categories should be regarded as functional rather than chronological. At Hen Domen, Montgomeryshire,9 the earhest bridge, older than anything at Eynsford, has a basal plate and its successors have earth-fast posts. Phase I (Fig. 2). It is suggested that the earth-fast posts, W, were the intermediate shock-absorbers of the simplest kind of drawbridge, a mere pivoted 'Hd', without counterpoise, without legs' and without parapet, but long enough (3 -60 m.) for its trailing edge to have reached the rigid structure, or pier, based on Dl, CI, Tl and T2. The weakness of this pier, as reflected in the phase II reinforcements which cured it so effectively, was in withstanding compression towards the axis rather than towards the curtain. It was an odd, somewhat hybrid structure, differing from most of its kind, including the probably earHer one at West Derby, Lanes.,19 which were closer to timber 'belfries' of the kind that existed in the middle of the first castle at Eynsford, having plates of equal scantling on aU four sides, trenched, or at least halved, over each other at the angles and meant to take the evenly distributed weight or faU of the bridge. The crude and apparently late structure at Elmer's End, South Norwood,11 seems also to be anomalous in this respect. At Eynsford, it included two transverse plates, 01, Dl, of different scantling but otherwise identical with each other and with the free-standing Al and Bl, not only in plan but apparently in the general pattern of the superstructure: six tenoned members, of which the central four were upright, the upstream terminal one, and less certainly the downstream, set obhquely, bracing horizontal to horizontal, rather than acting as a shore to a vertical. These members 6 Besides the forthcoming discussion in Ch&teau Qaillard, vi, cf. D. F. Renn in Trans. London and Middx. Arch. Soc, xxv (1967), 224^6, and F. J. Huggins in Med. Arch., xiv (1970), 60-3. 0 Perhaps closest to Eynsford, Bushwood, Lapworth, Warwicks., Med. Arch., vi-vii (1962-3), 336-7. ' e.g. Bodiam, Lord Curzon, Bodiam Castle (1925), 89-90, re-excavated 1970 by D. Martin, Bodiam Castle Medieval Bridges, Hastings Area Arch. Papers, 1 (1973). 8 e.g. West Derby, Lanes., Univ. Liverpool Annals of Archceol. and Anthropol., xv (1928), 47-55; Aoton Burnell, Salop., Med. Arch., viii (1964), 272-3; Leckhampton, Glos., Trans. Bristol and Gloucester Arch. Soc., xv (1928), 47-66; Beckenham, whioh also has horizontal planking. ° Ch&teau Gailltxrd. iii (1964), 15-27. 10 v.s., note 8. 11 v.s., note 8. 101 S. E. RIGOLD AND A. J. FLEMING on Dl (Fig. 5) were mere studs; on Dl, a much heavier plate, at least the second and fourth were substantial square posts that could have framed a kind of barbican gate above the walkway.12 There is a resemblance to a transverse and a terminal frame of an aisled building, which is carried further by the totaUy different treatment of the sides of the structure, analogous to a stave-like side-waU in such a building. The lateral plates, Tl, T2, are more massive even than 01, but only butted on to the transverse plates and secured at one end only by smaU pegs. Along the top of each is a wide channel which seems intended to take a continuous row of stout vertical planks, aU missing. This is a work of a house-carpenter rather than an engineer. Phase II (Fig. 3). Like the independent trestle-plates, A2 and B2, the reinforcing plates, C2 and D2, have mortices drilled before cutting. 02 carried two vertical members only. D2, laid in the middle of the existing quadrilateral frame, was certainly a re-used timber, but the mortices operative in the bridge seem to be those for a pair of uprights as in 02 and an obhque notch with a peg, like the seating for a rafter in a waU-plate, was almost certainly to take a shore from the upstream lateral waU. The original function of D2, which had an almost 'clubbed' head and a long rebate with pegs in pairs, may weU have been as a door-durn, with half of an arched head set in the rebate, but it is hardly strong enough to have served m the main castle-gate of phase W. A2 and B2 have five mortices, of which the outer pair is clearly for braces, passing rather than entering the uprights: the arrangement seems to have been identical with that of phase III, but no member of either phase was complete enough to measure the angle of intersection. Phase III (Fig. 4). The timber bridge was now reduced to the space between the abutment-waUs Y and Z, but the longitudinal plates must have carried back to piers vl and v2. Though more weathered than the other, plates A3, B3 and 03 apparently carried trestles of the same elevation, the outer mortices holding braces, the other three uprights. The foot of the downstream brace which remained in A3 was square in section and scribed round the curvature of the plate A3 and 03 must have derived some support from the flint waUs, whUe the freestanding B3 was broad enough for the mortices to be 'staggered', aUowing a strong halving of the braces and uprights. WaU Y, Hnked by plates to vl and v2, formed an essentiaUy hoUow abutment, never completely filled, and in a sense, a larger successor to the primary timber caisson. WaU Z revetted a solid embankment. One other substantial timber, E, lay obhquely just below the 'threshold' of observation, as though thrown down to bind the contemplated inner embankment and support the timber lacings of vl and t>2, but not 18 As in the reconstruction, Arch. Cant., Ixxxv (1971), 140. 102 EYNSFORD CASTLE: THE MOAT AND BRIDGE articulated with them. I t was longer than the others and clearly re-used, but the regular mortices and lack of scarf-joints do not suggest that it was once a longitudinal plate. THE FINDS I. Building materials and objects of stone 1. Dressed masonry: very Httle—slivers of coarse-tooled rag, some surfaces rounded, in destruction-layers of bridge, no doubt from the gate-tower, where the debris included a larger, buU-nosed section with a hoUow channel along one flat face, with traces of burning (a stringcourse with a drip-runnel?). Norman-looking, but coarse rag dressings at Eynsford occur in the haU after the thirteenth-century fire. 2. Roman bonding-tile: Httle, considering that at least the inner arch of the gate-tower was turned in this material. Occurs in the basal and destruction layers of the bridge and, re-deposited with flints, towards the outer slope (layer 5a). One piece bears the imprint of a boot—five rows of hobnails on the sole, a row round the heel and a looped ornament within it. 3. Medieval roofing-tile: plentiful from the destruction-layers, as in the Phase D strata of the interior. Also a number of fragments from layer 8, sandwiched between the basal timbers and overthrown uprights of the phase I bridge, not aU of the earHer type (first report, Fig. 8, 5) and including pieces with ohve and orange glaze. This, with the associated pottery, of phase B and apparently later, confirms the late date of the phase I I I bridge. 4. Rhenish lava quern (Fig. 9, 3) from the otherwise barren claywith- flints build-up within the gate-tower, contemporary with the new entrance (end of phase X, before the middle of the twelfth century). A sector of a lower stone, diameter over 60 cm., i.e. enough for a watermill, but with a shghtly raised rim. Coarse radial dressing, rough pick-tooling on underside. Source probably Mayen (Andernach and Niedermendig were trading-points). 5. Globose mortar (Fig. 9, 1) in coarse, not dense, pale grey calcareous sandstone, with a Httle fossfl shell, from bridge, layer 5. The form is usual in the late thirteenth century and c. 1300 for mortars not of Purbeck marble,13 save in having an everted base, not a baseroU, and the lugs squared off at the bottom. Fairly smooth toohng inside and out and Httle sign of grinding: the material was poor and it soon broke. 6. Truncated conical mortar with strip-lugs (Fig. 9, 2) in Purbeck 13 Cf. mortars of Caen-stone from Dover, J.B.A.A., 3rd ser., xxxii (1969), 82-4; of Burr-stone from Northolt, Middx., Med. Arch., v (1961), 279-84; and of rag-stone (?) from Beckenham. 103 *£€X FIG. 9 (i; 6 and 8, £). EYNSFORD CASTLE: THE MOAT AND BRIDGE marble, from bridge, layer 7 (destruction of phase I bridge in preparation for phase III). Only a lateral fragment, but the form is usual for Purbeck mortars of the second half of the thirteenth century.14 Exterior roughly tooled, verticaUy on body, horizontaUy and obHquely on lug; ulterior ground smooth but more towards base than sides. 7. Slate whetstone (Fig. 9, 6), from bridge, layer 9, under coUapsed upright. Rectangular section, 5-5 cm. long, perforated at top for suspension.15 II. Objects of metal, with some unreported late material 1. Barrel-padlock of copper-coated iron (Fig. 9, 7), from bridge, top of layer 10, i.e. with strongly tweUth-century associations. Barrel diam. 3 -0 cm. Quite weU preserved, thin metal, dehcate workmanship, with fluted barrel and terminal hoops. The barrel, hasp and spring are shown from all aspects: a, barrel and tube; b, hasp and spring. The remams of another barrel-padlock, from the fire-level in the solar undercroft, BB, early to mid-thirteenth century, have been identified though not included in previous report. Also of coppercoated iron, barrel-diameter 2 -2 cm.16 2. Short iron knife-blade with traces of leather sheath (Fig. 9, 9), from bridge, top of layer 10. Cutting-edge hammered (possibly welded on) and honed.17 3. Horseshoes: several from the upper or destruction (3a-3b), and lower (8) layers of bridge, not weU preserved and generaUy Hke those from phase D in the interior (first report, Fig. 9, 5-8); one (Fig. 9, 10; from IT, 10) has more prominent calkins and rather different holes. 4. Edge or cladding of wooden implement (?) from 77, 8 (Fig. 9,11). In same layer, a 'doughnut-like' ring, diam. 8 cm., section-diam., 3 cm. Both of iron, very corroded. 5. NaUs: few, especially from lower bridge-levels; large structural nafls and low clouts (Fig. 9, 12, 13) from dismantling-layers of bridge, but not enough to suggest that the great gates feU outwards. 6. SmaU ferrule of copper-aUoy from bridge, layer 10. 7. Horse-beU (Fig. 9, 8), cast in copper-aUoy with surface-enrichment of tin, from uncertain high level, presumably phase K. 8. Whip-top (Fig. 9, 5), omitted from previous report, from behind final blocking of garderobe fl, presumably phase K. Oblate pear- 14 Cf. mortars from Northolt and Winchester, Med. Arch., as note 13, and from Dover, J.B.A.A., as in note 13, 82, 102, and xxx (1967), 107, 117. 16 Cf. one from Dover, mid-thirteenth century, J.B.A.A., xxx (1967), as in note 14, but the type is found at least baok to Viking contexts. 10 Cf. London Museum, Medieval Catalogue (1954), 146, but there is now material for a fuller study, e.g. a very close parallel in a fifteenth-century context (!) from Boston, Lines. Hist, and Arch. Soc, i, 7 (1972), 40-41. 17 From X-ray examination. 105 S. E. RIGOLD AND A. J. FLEMING shape, turned in fine-grain (fruit-tree?) wood, with one groove above, several below and iron spindle (point broken), wedged with wood. 9. Fragments of large iron chain (?) in moat south of waU Y. This and the foUowing (III) not yet analysed. The hinge, Fel, described in the previous report, has been found to have had its ends more recurved. III. Leather Off cuts of shoe-leather, in a baU, from bridge, layer 10. IV. Pottery General. The pottery from the moat augments the corporate series in the previous report and wfll be set against the classification proposed there, in which the coarse wares were divided into major groups labeUed, in upward progression, W, X, Y, Z, A, B, C, D.18 It was argued that the predominance of each group coincided with phases Hi the history of the castle and the pottery from the moat is consistent with this pattern. Reference to related examples in the first report wfll be made thus: 'D4', 'Z3' (i.e. to group and individual number, not to page or figure). Those first debcribed Hi this report wUl be designated by figure-numbers, under group-headings. Contexts are given briefly in brackets, the relevant section, t or w, foUowed by the relevant layernumber. (i) SheU-gritted wares (Cooking-pots unless otherwise noted). Very early types. Clumsy hand-made vessels from re-deposited material (i, IB); heavUy leached, fabric, externaUy dark drab-grey, internaUy Hght greyish pink and buff; not related to anything else known locaUy but late-Saxon rather than Iron Age.19 Fig. 10, 1, may be a bowl or part of a smaU, flared-rim cooking pot. In fig. 10, 2, the angle of the rim is uncertain, owing to unevenness. 'W-X' types. With large sheU-fragments, the fabric continuous as 'Wc' and 'Xc' and dominant in the early pots from Lullingstone. Fig. 10, 3 (i, 10, trodden into clean yeUow gravel floor of moat), hand-made bowl, rather uneven; the fabric ofthis, the earhest stratified sherd from the castle, is like L7, L9 but softer; orange, oxidized surface, grey reduced core; upper inner surface of rim decorated with Hght finger-tipping. WaU-sherds in typical Xc fabric (finer, harder, very thin, dark grey, directly paraUeled from footings of cross-waU of haU, section /3) occurred in t, 10 and IT, 10. Fig. 11, 2, probably from a smaU, flared-neck cooking pot, is in similar fabric but with smaU, relatively sparse sheU, approaching 'Xa\ 18 Arch. Cant., as note 1, 149. 19 In the opinion of Mr. J. G. Hurst. 106 EYNSFORD CASTLE: THE MOAT AND BRIDGE V T '3 XX \ \ \ ^ X ^ ^ > ^ r / / \ \ \ FIG. 10 (i). T' types. These pervade the bridge layers 9 and 10, at least 12 vessels in aU. The reconstructible profiles (Figs. 11, 3; 12, 2) suggest that cooking-pots are stiU generaUy globular. The two rim-forms, Y, a and Y, b, noted Hi the previous report are foUowed. Y, a rims: Fig. 11, 3 and 4, Fig. 12, 1 and 2 (aU w, 9b). The typical thumb-strips, smudged onto the body, contrast with the neat ones on the contemporary sand-tempered wares. Y, b rims: Fig. 11, 5 and 6 (w, 9a), Fig. 13, 2 (t, 3) augment the series. 'Z' types. Hardly any Z-type profiles from the bridge, except the precocious-looking Fig. 11,1 (w, 10). Fig. 12, 3 (77-, 8, so not sealed until much later) typifies the Z-type curvature, with shght thickening on the shoulder and wide, shghtly hoUowed, turned-down Hp. The re-depositions of section 1 produced more. Fig. 12, 7, 8, 9 (i, 2), variations of the Z repertoire, emphasize the coherence of this group: 8 is a variation of Z10, exaggerated in 7, while 9 echoes Z8. Fig. 13, 1 (1, 3) is an enormous, thumb-stripped pot (42 cm. high, 50 cm. at widest point, 38 cm. across the mouth) which could stand 107 12 " \ 'fa «~Y. K 'i Fio. 11 (£). ; u b ^ T T ^SSsE» FIG. 12 (i). FIG. 13 (i). Fio. 14 (±). [face p. 110 EYNSFORD CASTLE: THE MOAT AND BRIDGE alongside the great pot found and exhibited in Yaverland church, Isle of Wight. The high waUs appear to have been achieved by coilconstruction, indicated by the thick undulations of the lower body, which has been hand-smoothed whfle the rim and shoulder were finished on a wheel as opposed to a turntable. 'A' types. In the lower bridge layers, and even in TT, 8, only Y and Z sheUy types occur, but Hi7r, 7, they are supplanted by A and B types. Fig. 12, 6 (TT, 7) and Fig. 13, 3 and 4 (t, 3) have rims of A, c form, the most persistent form, with the inner bead, and the only one observed here. Fig. 13, 5 (i, 3), one of the rare sheUy bowls, could be classified under A or Z, but the soapy feel links it to the A group. 'B' types. Fig. 12, 4 (IT, 3, derived), with club-like rim-form, B, a. Fig. 12, 5, has the broad flange, turned down, the 'B, b' rim-form, dominant Hi IT, 7 and also among those grouped as 'BC in the earHer report, confirming that it is generaUy the later form, marking the demise of the fuU-sheUy fabric, save (?) for the foUowing: Fig. 12, 10 (i, 3). In this the fabric conforms with that in A and B sheUy wares, but the rim matches that of sand-tempered D wares, which can now (see below) be traced back to the age of TT, 7. (Ha) Sand-tempered wares with variable admixture of sheU. The 'sheUy-sandy' fabrics of the 'mam series' in the previous report pass in a continuum from sheUy with a Httle sand to fuU sandy with the sparsest addition of sheU and may be subdivided as foUows: SSa. Relatively thick, coarse, even crumbly when leached, body and core; generaUy, a fafrly uniform Ught grey. SSb. Relatively thinner, finer, tending to dark grey or black. SSg. Dark grey, shghtly sheU-gritted, resembling SSb but thin for their relatively rough surface with markedly regular latitudinal striations, a London-Middlesex feature. Where definition between SSa and SSb is ambiguous the term SSa/b will occasionaUy be appHed. In the moat excavations cooking-pots outnumber jugs, which exceed the number of bowls. Shelly-sandy wares stratified with T shelly in bridge, layer 10. In the Ulterior the 'main series' of sand-tempered wares could be traced, with little change of fabric, at least back to phase Z, but only Hi smaU numbers, IT, 10, produced a larger sample, firmly associated with Y wares as weU as Z. SSb. Fig. 15, 11, Jug-base, shghtly sagging, thin waUs, fine sand, sparse sheU, body grey throughout with smooth black surfaces. SSg. Fig. 15, 6 and 9. Thin, hard, very regular jug body-sherds, sparse, rather leached sheU, charcoal-grey surfaces grading to Hghter core. The distinctive striated surfaces are decorated on 6 with wavy comb-lines and on 9 (cf. the pitcher, Z19, in oxidized sandy fabric) with carefuUy appHed thumb-strips. m r d2 r-l EYNSFORD CASTLE: THE MOAT AND BRIDGE Shelly-sandy wares of B-types, i.e. with the shght upper bevel that was almost universal in the typical B layer behind the haU. SSa. Fig. 15, 14. Cooking pot from TT, 7, with the form of B17 but a hoUow upper surface to the flange, as B20. SSg. Fig. 14, 6 (t, 3) and Fig. 16, 11 (TT, Z or near) are both in a harder form of SSg than those with D-type rims (Fig. 14, 2, 3, 5, 8), preserving the sharp outlines to the flange as weU as the hard fabric of B12. The straight-sided waUs are shared by the comparable D-rim pots mentioned above. Shelly-sandy wares of D-types. Only the most typical profiles (D, b, c), with level or down-turned flat flanges, without upper bevel, are found. These are absent from the sealed B level behind the haU, but it cannot be asserted that absolutely none of the comparable vessels classed as D Hi the first report came from the A or B soUs covered by the debris, but otherwise unsealed. Since it is abundant Hi TT, 7, the unbeveUed form must have evolved quickly soon after Phase B. SSa. Fig. 15, 13 (TT, 7). Fig. 15, 7 and 11 (TT, disturbed layers); the former placed here for its ashen colour and coarseness, but striated as typical in SSg fabric. SSb. Fig. 15, 15 (TT, 8), thin-waUed unruled cooking-pot. SSa/b. with turned-down rim (D, c). Fig. 15, 10 {TT, 8). Fig. 15, 12 (TT, 7). Fig. 14, 7 and 9 (both i, 3), the last buff-coloured throughout. SSg. Fig. 14, 2, 3, 5 and 8, which, with Fig. 14, 7 and 9, form a consistent group from t, 3, are true phase D vessels of the last occupation. Notice the relatively thin walls of 2, 7 and 8 and the knife-trimming of the lower body of 8. Except for 3, the neatest cooking-pot found, which has a buff body, the surfaces are generaUy dark grey to black, sometimes with brownish oxidation just below the surface. Shelly-sandy bowls. SSa/b. Note the Hmer bead, generaUy characteristic of D-type bowls, with flat topped rim, occurs in the stratigraphicaUy early Fig. 15, 2 (TT, 8), blackish and smooth-surfaced, as weU as in the presumed late Fig. 14, 1 and 4 (t, 3); the latter is buff throughout, Hke Fig. 14, 3. Sandy-shelly jugs. Few, but typicaUy D-types, with stabbed handles. SSa. Fig. 15, 5 (TT, 8) and Fig. 16, 12 and 14 (TT, upper layers), stabbed strap-handles, unreconstructible. SSg. Fig. 15, 7, in the tradition of the much larger striated jugs, Fig. 15, 6 (with wavy combing) and 9, aU from i, 3. Fig. 15, 7 has a short, corrugated cyhndrical neck and flared, cusped, flat-topped rim, shghtly puUed out to pour. 113 J \s J EYNSFORD CASTLE: THE MOAT AND BRIDGE Other Sandy-shelly vessels, in SSa/b. Fig. 10, 4. Curfew, black surface, smaU smoke-piercing, radicaUy slashed top, handle reconstructed (TT, 7). Also, a skfllet (not Ulustrated); simple stabbed rod-handle (t, 3). (fib) 'Pure' sand-tempered wares. The characteristic grey, unglazed fabric can be seen in the bridge area, from which aU iUustrated examples come (though a few jugsherds were in t, 3) leading up to the early D types of layers 8 and 7 from something not very different except in the handles. Jugs. Fig. 15, 4 (TT, 9b), slashed handle, hard, Hght grey fabric. SimUar to Fig. 15, 3 (TT, 7), stUl with zig-zag slashing. Fig. 15, 5 (TT, 8), stabbed ovoid handle, simUar to Fig. 16, 3 (TT, 7) and 14 (TT, 3, or near). Fig. 15, 8, medium-coarse fabrio, surface dufl pink-buff (reminiscent of glazed West Surrey fabric, as D66) and akin to Fig. 15, 7 (»r, 8). f- ^3? J © \ / ® ^ k \ \ / ! s \ \ \ \ \ \ / / Vk / I / / \f \ \ I , 7 / FIG. 17 (i). 115 S. E. RIGOLD AND A. J. FLEMING Fig. 16, 8 (TT, probably 3), jug-body, flattish thumbed base, oxidized. Fig. 16,13, rim with shaUowly hoUowed neck recurved into body. Bowl. An isolated flat-flanged D type from7r, 3. (Hi) Glazed wares. The few glazed sherds, mostly from the bridge, faU within the known range of 'London-area' jugs, apart from the fohowing two sherds: Fig. 15, 1, in a remarkably fine, hard grey ware with smooth Hmer surface but badly burned external olive glaze; perhaps distantly related to Stamford ware, it comes from the lowest bridge layer, TT, 11, and is stratigraphicaUy the oldest sherd from the bridge and the oldest glazed sherd from the site. Fig. 16,2 (TT, 3), in whitish, hard Surrey fabric, quite to be expected. The 'London-area' wares show Httle variety and are akeady in layer 9b. Fig. 16, 1 (TT, 9b), a baluster, resembles in fabric and light, speckly glaze Fig. 16, 6 (TT, 8). Fig. 16, 9 (TT, 7), has brown, Hon-glazed vertical ribs; Fig. 16,4 and 5, are green-glazed; 4 with elongated scale-decoration, has a very rough exterior, poor, mottled glaze, duU red-brown lining and grey core, almost Hke Tyler HUl ware; 5, in simUar fabric, if not quite so coarse, has combed Hnes separating tiers of concentric cfrcles. Fig. 17, the London-area imitation of a north-country anthropomorphic jug, aHeady noticed in the previous report, is here reconstructed on the Hnes of the London Museum, Medieval Catalogue, PI. LXII, no. 2. ACKNOWLEDGEMENT In addition to his considerable part in the excavation and drawing, the classification of the finds from the bridge is largely the work of Mr. D. C. Mynard; Mr. J. E. Haslam and Mr. C. MiscampbeU also assisted. 116