Eynsford Castle and its Excavation

EYNSFORD CASTLE AND ITS EXCAVATION* By S. E. RIGOLD, M.A., F.S.A., F.R.HIST.S. THE small but massive castle of Eynsford, lying low beside the Darent at N.G.R. TQ 542658, was the Stammburg of the greatest of the archbishop of Canterbury's knights. Throughout this report, unless otherwise indicated, the 'Castle' must be understood to mean the flint-walled enclosure within a broad but shallow moat, although this was, or became, the inner bailey of a larger complex, not easy to define. Apart from a survey and partial excavation in 1835, ably recorded for the time by Edward Cresy,1 it was httle studied2 until the Society for the Protection of Ancient Buildings, which had acquired it in 1937, at the instance of the tenant, Lady Fountain of Little Mote, placed it, in September 1948, in the guardianship of the then Ministry ofWorks. The conservation proceeded slowly enough for the writer to observe every stage and to conduct simultaneous excavations with one labourer, or at rare intervals with more. Most of the work was done between 1953 and 1961 and the results, here modified in a few points, were summarized in notes for Medieval Archaeology3 and in the official guidebook. Subsequently, particularly in 1966-1967, further deep sections were made to verify earher ones, and the bridge was totally excavated: at this stage, the writer was greatly assisted by Mr. D. C. Mynard, who also studied the pottery and drew most of it. The long sections were completed in 1971, with the help of Mr. J. Haslam, who worked further on the finds. The plans and sections were finished by the drawing-office staff of the Ministry, now Department of the Environment. WhUe this report was in preparation, a smaU excavation, started by an accidental discovery in a garden outside the guardianship area, has revealed a subsidiary medieval buUding within a presumed outer bailey. Thanks to the discoverers and to Mr. S. R. Harker, a summary of the findings, to date, is appended. I t is the first positive contribution to our knowledge of the Castle's outworks. * This paper has been printed with the aid of a grant from the Department of the Environment. 1 Archceologia, xxvii (1838), 391-7. 2 Hi Sands, in Some Kentish Castles (1907, from Memorials of Old Kent), adds little to Cresy. 3 Med. Arch., i (1967), 156-7; vi-vii (1962-1963), 322; ix (1965), 190. 109 S. E. RIGOLD The writer has been engaged in many hnes of research connected with Eynsford: (i) Documentary, including the wider activities of the Eynsford famUy; (ii) Topographical—the Castle in relation to the village and valley; (hi) Interpretation of the upstanding ruins; (iv) Excavation of the walled area; (v) The place of Eynsford in the history of fortification, for, though small, it is unquestionably a stronghold, not just a moated site; (vi) The excavation and comparative study of the unexpectedly weU-preserved remains of a timber bridge, several times reconstructed. The bridge is still under examination and will only be mentioned here as far as is necessary for the interpretation of the waUed enclosure.* The rest will be discussed in the aforenamed order, save that questions of comparison and reconstruction will go at the end, but before the description of finds (which does not include those from the bridge). The ponderous quantity of pottery has produced a series of local types from the late eleventh to the early fourteenth centuries, unmatched by any site in the district and dated by its relation to the structures and by a pace of evolution appropriate to a single consistent series, rather than by external references. The series here speaks for itself and the few analogues, even from the Darent and Cray, are not cited in detail. The only exception is the pottery from the medieval site overlying Lullingstone Roman viUa, 1 -3 km. away, which is much richer than Eynsford in the earhest phases. This has been studied and drawn by D. C. Mynard and, with the approval of Lt.-Col. G. W. Meates, F.S.A., a selection is published here to supplement that from Eynsford. HISTORY [Note. For brevity and convenience the several valuable recent historical works on the tenures of the Archbishopric and of Christ Church wUl sometimes be cited rather than primary sources. Even in the more controversial parts, the sources wiU be referred to as simply as possible and close argument avoided.] Direct documentation of the Castle is shght, but there is much circumstantial evidence, and enough about the principal tenants to describe their careers in some detaU (which is not attempted here). How Christ Church first acquired Eynsford has been often recited, 4 The bridge will be treated, with comparative material, in Med. Arch.; the historical material, also of wider than Kentish interest, has the makings of a long thesis. 110 EYNSFORD CASTLE most recently by Professor Du Boulay. & This was in Archbishop Dunstan's time, when the possessions of the chapter were not fully distinguished from those of the archbishop and his familia. Subsequently, it was stolen, and then recovered by Archbishop Lanfranc,6 who kept the territorial lordship for himself and left the advowson to the monks. It is more germane to the Castle to ask when and from whom it was recovered. It was not by the famous plea at Penenden in 1075,7 nor by a simUar action three or four years later.8 Nevertheless, the culprit was probably Odo, whose tenures are still very thick in this part of Kent in the time of Domesday, with his tenants and those of the see closely interlocked in what looks hke a compromise order. Lanfranc persisted and his latest opportunity to make Odo disgorge, and the most hkely one for the recovery of Eynsford, is his arrest and temporary echpse in 1082. In Domesday, Eynsford seems aheady to have the extent that marked it out as the caput of a barony that was to last until the Reformation held not in chief but of the archbishop. Nevertheless, the knighthood of the see had not then been fully organized, as it had by 1093-1096,9 and not improbably before Lanfranc died in 1089. The barony then appears in final form, the largest lay tenure, assessed at 7 | knight's fees. There is no mention of a castle in Domesday. When the material for this was collected, c. 1085, the archbishop's tenant was Ralf, son of Unspac, or Hospac, ancestor of the line which took its name from Eynsford and held it in unbroken male succession, through five generations, aU caUed Wilham, untU 1231. Brief accounts of the family have been given by the late Professor Douglas,10 by Mr. Colvin,11 by Professor Du Boulay12 and by Dr. Urry.13 AU these contain, or repeat, minor errors, and Professor Du Boulay introduces a new confusion by a too hteral reading of a deposition made in 1261,14 which ascribes, wrongly but understandably after more than a century, the surname 'Goram' or 'Gurham', properly of WiUiam II, to William I: all other evidence indicates that by 'WiUiam, the first lord', called 0 F. R. H. Du Boulay, The Lordship of Canterbury, London, 1966, 33-5. "Ibid., 42. ' Ibid., 37. 8 Ibid., 38. 0 Domesday Monachorum, ed. D. C. Douglas (1944), 106. 10 Ibid., 44-7. 11 H. M. Colvin, 'The Archbishop of Canterbury's Tenants by Knight-service', Medieval Kentish Society, Kent Records, xviii (1964), 16, etc., does not enlarge on the genealogy but discusses the extent of the barony. There is no trouble about its member in Topsfleld, Essex, which was attached to an archiepiscopal holding in Hadleigh, and not the part of Topsfleld which was of the Honor of Boulogne. 18 Op. cit. in note 5, 108-10. 13 W. Urry, Canterbury under the Angevin Kings, London, 1967, esp. 54r-5. The main concern is with urban tenures in Canterbury, not connected with the barony of Eynsford. " Lambeth MS. 1212, 417; see note 12. I l l S. E. RIGOLD Goram, they mean WiUiam de Eynsford I. Note that the first lord is not Ralf. Little is known of Ralf. He was not an Anglo-Dane,15 but presumably a lay retainer from the abbatial lands of Bee or Caen. He held nothing else, except perhaps Croham, Surrey,16 which was certainly held by his descendants, also of the archbishop. He was perhaps dead before the transcription of Domesday, certainly by 1087, when Lanfranc states that he had enfeoffed his son, Wilham I17 in language implying that this was not automatic but was a position of special trust, a trust that was later to outweigh the strict law of primogeniture in the succession of WiUiam i n . It seems that the years 1087-1089 saw the organization of the knights, the creation of a lordship for Wilham I on wider terms than his father's, and, in view of the historical circumstances, the entrusting to him of the building of the Castle on Lanfranc's authority, not his own. A fortification of this quahty, a smaller version of the castle at Rochester that, in these very years, Bishop Gundulf with difficulty persuaded Wilham Rufus was economic for him, Gundulf, to build on Rufus's behalf, would not normaUy have been tolerated in private hands. After Lanfranc's death, the long vacancy, with the lordship in royal hands, and then Ansehn's exile, would hardly have been opportune for the founding of an archiepiscopal castle or the permission for a private one. But the years 1087-1089, when Lanfranc and Gundulf stiU stood in the confidence of Rufus, but on the defensive, with Odo scotched, not killed, provide the most arguable occasion for the consolidation of the militia episcopatus, with the barony and castle of Eynsford as its spearhead. This is 'Phase W—see below. A long and active hfe confirmed Lanfranc's view of the abihties of William de Eynsford I. As chief tenant, we see him managing ecclesiastical affairs, with the general approval of Gundulf and the exiled Anselm, and with the disapproval of the ambitious Prior Ernulf.18 Later, he moves into a more independent position as sheriff of Kent for a long period19 and, briefly, of London, Essex and Hertfordshire.20 At the end of his career, he may have overreached his powers: soon 16 Unspac occurs as a name, e.g. at Lincoln, before the Conquest, but both Ralf and his son, of age by 1087, had Norman names. 10 Near South Croydon station, about TQ 335640. The Ralf that held it in Domesday may be fitz-Unspao, or his brother-in-law, to whom William I succeeded. 17 Cartulary of the Priory of St. Gregory, Canterbury, ed. A. M. Woodcock (Camden Soc. 3rd ser., Ixxxviii, 1956), 2. 18 Patrologia Latino,, ed. Migne, CLIX, col. 233, lviii, col. 235, lxi (letters of Anselm). Ernulf is surely aheady prior of Canterbury, not Rochester, on this occasion. 1 » This is the natural interpretation of, e.g. Regesta Regum Anglo-Normannorum, I I (ed. Johnson and Cronne), 1093, 1189, 1191, 1497, 1867. „ 20 Cartul. Monasterii de Rameseia (Rolls Ser. no. 79, 1884), I, 139—sheriff of London, surely not far short of 1130; Pipe Roll, 31 Hen. I, 63 (cf. H. Round, Geoffrey de Mandeville, 298)—allowance for what remained of his cancelled fiveyears farm of Essex from 1128. 112 EYNSFORD CASTLE after 1130, having laid down, or been forced out of his office, he retired as a monk to Christ Church,21 which, with his wife Hadwisa, he had endowed with Ruckinge without derogation of his obhgation of knightservice. 22 I t appears that it was he that acquired most of the considerable possessions of the family outside the barony, in particular large holdings of the bishop of Lincoln in Buckinghamshire and Huntingdonshire. 23 A man of patient business, a bulwark of ecclesiastical power, and, if we can judge from his acceptance as a quire-monk, an early instance of a hterate knight. His successor was a less considerable figure. His pubhc activities were slight, and to him is ascribed the beginning of the ahenation from the see that was to flare up briefly under WiUiam III.23a His tenure coincided, approximately, with the disturbed years of Stephen. This is, historically as weU as archseologicaUy, the most likely occasion for the building of the HaU and gate and the heightening of the curtain (Phase X), in every sense a re-fortification and one for private benefit, totaUy and permanently changing the Castle's character. Such would have been frowned upon under Henry I and probably forbidden under Henry II. WiUiam II, called Gurham, was dead or retired by the late 1140s; it is his son, WUliam III, who about this time attests frequently, as attendant knight, usuaUy in association with Ralf Picot, later sheriff, to deeds of Archbishop Theobald.2* That is to say that he served in that archbishop's famous and educative household. He named one of his sons Theobald, and was personaUy obhged to the archbishop for his succession to lordship, in preference to the children of his deceased elder half-brother John. From John descended the line of Lese, or Peckham, who on many occasions, one the instance of the deposition of 1261,25 tried to reclaim their interest or part of it against the main, 21 Op. cit. in note 9, 46; 'c. 1135', but it may have been a little earlier. 22 Ibid., 109, etc.; A. Saltman, Theobald, Archbishop of Canterbury (1956), 269-70. Hadewisa may have been an heiress (? of William of Adesham, the Domesday tenant of Ruckinge). 23 In Stoke Mandeville and Great Staughton—Red Booh of the Exchequer, I, 376; the Pipe Roll—Mag. Rot. Scacc., 31 Hen. I, 48—suggests that at least Staughton goes back to William I. Other holdings included Foots Cray (Red Booh, 191), the Canterbury tenures (op. cit. in note 13, passim), and, possibly connected with these, tenements assigned to the Templars, as of, but not necessarily in, Strood (B. A. Lees, Records of the Templars in England in the XIII cent.— Br. Acad. Rec. Soo. Econ. Hist., IX (1935), xcvii, 20; see op. cit., in notes 27 and note 5, 66 for small enclosures in the Weald. These do not exhaust the widespread interests of the family. m Op. cit. in note 9, 109; pater meus et ego ipse (William HI) mouimus calumpniam . . . 24 Saltman, op. cit. in note 22, charters nos. 69, 60, 61, 151, 153, etc. 26 See note 12. The Lese estate was on the southern edge of Eynsford, perhaps extending into Shoreham; the names of Lese and Peckham are preserved in Leize and Pecken Woods, about N.G.R. TQ 557615, just in Shoreham. Pace Douglas, op. cit. in note 9, 45, it was not John, who was already dead, but his unnamed successor, who was excommunicated with W illiam III. 113 S. E. RIGOLD but technicaUy cadet, line. Between 1148 and 1151, Wilham III (tercius qui nunc est) formaUy confirmed the donation of Ruckinge.26 He died not earher than 1193, when, stUl calhng himself tercius, he made, among other rehgious bequests, one to St. Radegund's, founded about that year.27 He names his wife as Beatrice, who has been generaUy, making nonsense of the chronology of the records, confused with Eleanor, her daughter-in-law.27a Wilham Ill's contributions to the Castle were minor (Phase Y), but he is best known of the line for his part in precipitating the fatal dispute with Archbishop Thomas. It was a passive part, and what he held in chief, though enough for the king to make a test case of him, must have been smaU—small enough for Thomas to regard him, as 'his' man. He had stood surety to Thomas for a large sum, and, having made his peace with see and priory under Archbishop Richard, recouped the forfeit from offerings at his martyred predecessor's shrine.28 Apart from this one incident, he was as good a churchman as his grandfather. WUham IV's tenure was extremely short. Only the deposition of 1261 assures us that he did not die vita patris. WiUiam V, surnamed Rufus, who came of age in 1200,29 reverted to the type of secularminded baronage. We hear of him on John's remarkably successful expedition to Ireland in 1210,30 and among the hard-core of the baronial party captured when the keep of Rochester was breached in 1215.80a After a period of forfeiture and imprisonment, he was back in royal favour in the 1220s, constable of Hertford castle and steward of the Household. His tenure corresponds roughly with Phase A, but he did little to the Castle unless the burning and reconstruction of the Hall took place before bis death in 1231. This seems possible: the finds are consistent with a gap between the rebuUding and the brief, penultimate occupation of Phase B. He left no son and his death was followed by at least six, probably twelve,31 years of minority and wardship. This would imply that his daughter, and probably her husband too, Henry 20 Saltman, op. cit. in note 22, 269-70 (charter no. 42), with commentary. 27 Bodleian Lib. Gough MS., 18, p. 70, cart. 622—this is a careful extract of a lost cartulary. For the foundation of St. Radegund's in 1192-1193, in preference to Monasticon's 1191, see V.O.H. Kent, ii, 172. William's contribution is a parcel in Penshurst. 2' a As in the pedigree, op. cit. in note 9, 47, with one or two of the numerous references to the two ladies. See especially Placitorum Abbrev. (1811), 1 and 64; the former leaves no doubt that Eleanor was the mother of William V, widowed and remarried by Michaelmas 1194. 28 Op. cit. in note 13, 55; op. cit. in note 9,110. 20 Rot. de Oblatis et Finibus (1836), 165. 30 Rot. de Libert, ac Misis et Praestitis, ed. Hardy (1844), 182. 30a Gervase of Canterbury, opera (Rolls Ser., 73,1880), I I , 110 (continuation of Gesta Regum). 31 Many references to the minority between 1231 and 1236 (Close Rolls, I, 564, II, 112, 114. I l l , 158; Extr. Fine Rolls, 217, 229, 313, etc.). For continuation (or renewal?) of wardship to 1242, The Book of Fees (1920-1923), 669, 678. 114 EYNSFORD CASTLE de Ruxley, had died in her father's lifetime. We know from the deposition of 1261 that the WUham de Eynsford VI, recorded as tenant in 1254,81a was their son, not a male-hne de Eynsford, and that he died young, leaving an infant WiUiam VII, who too was dead in 1261. With this pathetic chUd the united lordship ends, but Phase 'B' may represent the short-lived household of his parents. The outcome of the enquiry of 1261 was to ignore the claims of the Lese line and divide the whole inheritance—not only the barony of Eynsford, but every other tenure, great and small—such was the pedantic impractioahty of later feudahsm, compared with the pohcy of Lanfranc and Theobald—between the representatives of two sisters and ultimate co-heiresses of Wilham V, Joan, for whom he had bought the wardship of a potential husband, Hugh de Aubeville, in 1212-1213,32 and Beatrice, wife of Stephen Herengod. The result, for the Castle, was effective desertion. Joan's grand-daughter brought her share to the Criols (Keriels), a powerful family in east Kent (Westenhanger), and elsewhere. Beatrice's brought hers to the neighbouring and poorer Kirkebys of Horton Kirby. In 1265, neither hne was enjoying their share, since the heads of both were on the losing side in the recent baronial troubles, and it was the turn of rival royal officials to quarrel over the custody.33 One can imagine the aggrieved feeling of Alan de la Lese, who hved in the neighbourhood and would have run the business better as a whole. The Kirkebys were ready to dispose of their share, first by 1292 as a life-lease to Ralf de Sandwich,34 who may have had some hereditary interest too shadowy to argue here, and then, by 1307, by outright sale to William Inge,35 a judge who was buying up unwanted tenures in all directions and treating them as brass-bound parvenus do. In three years or so, he made himself well enough hated to bring old rivals together. In doing so, he seems to have put his own caretaker into Eynsford Castle, or even thought of occupying it himself, when in the district—at least he made sure of his hunting rights. Certainly, in Phase D, there are clear signs of a short but intensive re-occupation, after a long interval, about, or soon after, 1300, and then a final cataclysm, of which we have both the archseological and the legal record. The roll only preserves Inge's side of the case; excavation shows that he did not exaggerate the physical facts of it. In June 1312, a commission of oyer and terminer was issued on Inge's complaint that certain persons had broken down the doors and windows of 'his' (there is no mention of partnership) manors of Eynsford, Ightham 3ia Arch. Oant., xii (1878), 234. 32 Pipe Roll, 14 John (P.R. Soc, N.S., 30), 15. 33 Cal. Inq. Misc., I (1219-1307), 220, no. 719 (Ralph of Famingham versus Geoffrey de Marisco). 34 Placita de Quo Warranto, Ed. I-III (1818), 188, 363. 36 Cf. Cal. Ch. Rolls, III, 107. (Free warren to Inge in 1307.) 115 S. E. RIGOLD (part of the Eynsford barony) and Stansted, committed damage and let loose his stock.36 The leaders of these 'angry young men' were none other than the Alan de la Lese of that generation, Nicholas de Criol III himself, and two of his brothers. Even the 'whimper' that follows this 'bang' has its archaeological echo in the sterUe patching-up of Phase E, and its documentary one in a dry legal decision two years later that the manor of Eynsford was stiU notionaUy indivisible.37 I t had long ceased to be a living unit, and the only subsequent inhabitants of its capital messuage were not people but dogs. The later manorial history is not relevant, save that the lordship that passed to the Harts and Dykes of Lullingstone, was the capital lordship, formerly of the archbishop,,that the subtenure was lost sight of or absorbed, whUe some of the prestige, but not the rights, of the subtenants, nor the occupation of the actual Castle site, passed to the holders of Little Mote, the SybiUs and BosvUles. When, inthe eighteenth century, the Dykes estabhshed their hunting-kennels in the castle, they were on demesne land. The removal of the kennels was foUowed by the excavation and survey of 1835, and then by further neglect, until much of the riverward waU feU in December 1872. The first steps towards the preservation of the Castle were taken by an incomer and benefactor to the viUage, but on his own terms rather than those of the natives, E. D. TiU (d. 1917). His great enthusiasm was Arbor Day, and tree-planting to commemorate every possible event and sentiment, but he bought and repaired several old houses in the viUage, including what remained of Little Mote, though he did not live there. In 1897, he took a 50 years' lease of the Castle from the Hart- Dykes, and buttressed its weakest points. What remained of the lease passed to his successor at Little Mote, Agnes Lady Fountain (d. 1953), who thought out a positive plan of preservation, that the Society for the Preservation of Ancient Buildings should buy the freehold on the understanding that H.M. Office of Works would take guardianship. In the event, the lease of 1897 had to lapse before the whole could be accomphshed, but Lady Fountain hved to see the conservation begun. THE CASTLE IN ITS SURROUNDINGS (Fig. 1) The existing ford at Eynsford (Fig. 1, F) and the stone bridge beside it, more or less Gothic, but apparently later than 1596,38 he at a shght gorge in the valley, where the 60 m. contours, about 25 m. above the summer water-level, draw within 600 m. of each other. At this point, the Darent, after flowing eastward, swings sharply north and the present main stream and backwater, whose roles have been reversed 30 Cal. Pat. Rolls., Ed. II, I (1307-1313), 478. 37 Eyre of Kent, 6 and 7, Ed. II (Selden Soc. 29, 1913), 168. 38 I t is not shown on Symonson's map of that year. 116 EYNSFORD CASTLE II," // s* * // \X/ s L* ?sx X ii N ^ u \t D u \i u ;WV V1 B '/ o ^ // =} w \ \\ \\ W Fio. 1. Eynsford: Site Plan. for the needs of mUls and the railway, cross over each other. On the west bank, tracks converge on the ford; on the east stands the church (Ch.), which seems at first sight to guard the ford better than the castle, and the convergence hes a httle downstream. Two subsidiary passages over the river, shown on Fig. 1, may be relevant to the medieval domain: I, upstream, leads from the site of the present Home Farm to an area known as the Park; II, just downstream of the Castle, is on the track (X) to Little Mote, which Cresy caUed an old causeway, but also on an earher path (Y) which can be traced in drought. This passage is nearer the Castle and main settlement, nearer stiU to the presumed site of the original haU, and also near to the recently discovered (see below, p. 137), probably Christian Saxon, cemetery. 117 12 S. E. RIGOLD Nevertheless, ford F, the shortest and soundest under wet conditions, seems always to have been the principal (and eponymous ?) one, and the Castle (E) is in a position to control all three. From its exposed west and south-west flank, it has an extensive view upstream, even at the level of its earthen platform. The high curtain, or before that, the central tower (see below, p. 120 ff.) would have mastered any obstructions to a comparable view downstream. From its low site, the Castle dominates not merely the crossings but the whole valley throughout the original parish. Eynsford, including its former chapelries of Farningham and Stanes39 was a large parish, containing much woodland, isolated homesteads, of which the most notable were Orkesden, Austin,40 Charton and Pedham, and what became a considerable secondary settlement, besides Farningham and Eynsford, at Crockenhill. Stanes, if rightly identified with the medieval site overlying the Roman at Lulhngstone, seems to have been a village shrunken by the thirteenth century. Eynsford itself still gives the appearance of a settlement in three parts; this is more noticeable on the Tithe Award map of 1842,41 which has been freely used in the making of Fig. 1, on which the areas built-up in 1842 are hatched, with existing medieval and sub-medieval structures shown individually. It is uncertain how far this preserves the early medieval distribution: area A, around and south of the church has one or two seventeenth-century buildings but did not necessarily amount to much before that; area B, across the ford, contains at least two sixteenth-century houses, antedating the bridge, but the shape of the plots suggests no great antiquity; C alone, south and south-east of the Castle, looks hke an integrated settlement. Occupying a distinct terrace, it has toft-like plots of regular depth, several containing latemedieval houses, and within it the road sweUs to the breadth of a green. Off the widest point of this, near the present entrance to the Castle, is an inlet from the Darent, where drainage works have revealed only riverine deposits, though it is now dry and built over. It slopes down to the Castle moat between two clay terraces, that of the village, on the south, and that east of the Castle (Fig. 1, D), which, from the second phase (X), formed an outer bailey. Where the inlet widens and the Village Hall now stands, the 1842 map shows a quadrUateral ditched enclosure (Z) abutting the present stream, which in the Middle Ages, was probably a backwater feeding the moat. This enclosure seems too low for habitation, and certainly no occupation-material came to hght 30 Op. cit. in note 9, 108; in Domesday the secular tenures of Farningham are quite distinct from those of Eynsford, the several tenants including Odo's man, Wadard. 40 Now Austin (Lodge). 41 Now lodged by Rochester diocese in the Kent Record Office. I am grateful to Dr. F. Hull for permission to consult it. 118 OURTYAR c 10SS /'Z tf{. Century / J f/x Century [XX] £aHr V\>OT-£, Mi 60 10 If FIG. 2. Eynsford Castle: General Plan. [face p. 118 EYNSFORD CASTLE when a bungalow was built on the site, at a level raised with spoil from the Castle excavations. Of the two groups of houses on the east bank, B by the church and main ford, might be thought more likely to represent the pre-Conquest settlement, but it has httle room for a village, and the terrace-site of C, separated from B by a piece of former glebe-land, vacant until quite recently, is much more convenient. There is no proof of early settlement in C: the cemetery may well have been away from habitation, and no trace of Romano-British occupation is reported. AU the Roman tiles re-used in the Castle are presumed, on present knowledge, to have come from Lullingstone or Farningham. The terrace of C and the terrace of the outer baUey, D, appear to be complementary and suggest that the settlement, with its regular plots, may be a deliberate creation, attached to an unfortified manorial site, whether Norman or earher in origin, rather than to the Castle. The boundary of D continued to the main road until the Baptist chapel was built. North-east of this boundary is an area, E, defined only by the track X, but this contains Little Mote, which has been a capital dwelling since the early sixteenth century, stands within the remains of its own flint-walled garth, but may perpetuate the approximate site of the Norman manor-house, in an oval enclosure beside and eroded by the stream. South-west of the boundary he bailey D and the Castle in the narrow sense. I t must be emphasized that, at first, the Castle did not contain a proper hall, and that it is separated by a made ditch from the terrace of D and from the area containing Little Mote (M). As far as the excavations show, it did not even have a shght natural knoll to recommend it, except possibly at the extreme south-west, and is best regarded as an artificial island, whose moat cuts into the manorial site and leaves no berm beneath its walls. As such it is not the nucleus of the manorial site or the village, but additional to both. The large fields surrounding the village today might suggest open-field cultivation, but on the Tithe Map they are much subdivided and give no support to the idea that, Norman creation or not, settlement C ever foUowed this un-Kentish practice. Superficial deposits cover the entire settled part of the valley and the chalk is nowhere exposed. Since this reach of it includes little gravel, both chalk and gravel are used sparingly in the mounding of the Castle. The chief materials are those of the surrounding surface—either black riverine silt or 'brick-earth', to stretch this phrase to cover a range of red-brown to yellow-brown clayey soils, containing eroded flints and usually some gravel. In the moat, this heavy, but not particularly sticky, earth proved surprisingly watertight and sterile and preserved the timber plates of the bridge only a few centimetres beneath the loose and rapidly draining build-up of the causeway and actually above the water-table in dry summers. A much purer red clay was used 119 S. E. RIGOLD for clean surfaces and may, in cases, represent the dissolved daub of buUdings. [Note. In aU the descriptive sections that foUow the hall-block wUl be treated as though its long axis were due east-west. Everything behind it, within diagonal axes, is the 'North sector' (N), and so with the East and West sectors (E and W), between the short ends and the curtain. These are productive sectors: the relatively barren area in front (south) of the hall and forebuilding, where the level was not much increased from an early date, and in parts, perhaps lowered, is called the 'Courtyard'. The gate-tower and the 'old', or 'great' kitchen are specified: the areas under the fairly straight sections of curtain between these and the acute end are 'south-east' and 'south-west' respectively. The acute end itself is 'South' (S).] THE VISIBLE STRUCTURES Before the Ministry began its works, the Castle appeared as a ring of flint waUs rising nearly 9 m. directly from the flood-plain, covered with ivy and broken by a gap forming the entrance at the south-east and by a coUapse towards the north-west. From the platform within, more than 3 m. above the plain, only the highest parts of the ruined hall protruded. To the entrance led an earthen causeway, hned with trees whose initials spelled out a piece of E. D. Till's Arbor Day wisdom. The Curtain The enclosure, as it appeared on clearance and was already known from Cresy's survey, is oval, not elliptical, in that it has one acute end and one obtuse. In detail, it forms an irregular polygon with twenty facets of different lengths, the most exposed being the longest. The full height, c. 8-8 m., was achieved in two stages, the lower accounting for about two-thirds of it (c. 5 -8 m.), with the division very clear internally but better disguised on the facade. Both parts are of coursed flintwork, but the lower is better built, as the differential weathering on the faUen section shows. The flints here are larger, both in the core and the face, where they are often canted to give a shghtly herring-bone appearance. There is a little ironstone, particularly for spanning channels. Half-way up the lower stage, there is an offset phnth, roughly rounded in flint, giving a basal thickness of about 1 -8 m., but often more on the short facets. A hne of small putlogs or weep-holes (?) appears at varying levels, a httle below the plinth. Though the two stages follow the identical plan, they wiU be shown to be half a century or so apart in date. There are no dressed quoins at the angles in either stage. The few dressings in the earher stage are of tufa, a oharaoteristio of early Norman work in Kent. In the upper stage, they are of Roman tile. 120 PLATE I Crown Copyright reserved Acute Angle of Curtain with Garderobes. face p. 120 PLATE II • -. • V—*• • W**m _': •*i>«f»-^ Crown Copyright reserved Solar Undercroft and North Curtain. PLATE 111 Crown Copyright r West Angle of Old Tower (OT), showing seating for Plate. PLATE IV ' 4 * 4 * & mm, ^1S 'i^p *a-'-X:»«~:--*' ' ' ' , i • X v - \ 'Jm - > if Crown Copyright reserved Section fi looking towards exterior of E. Wall of HaU. EYNSEORD CASTLE Two of the facets are thicker than the others in that the surface of the offset plinth is carried right to the top of the wall, mcluding the second stage. One of these (a), the northernmost facet of all, is also thickened on the inside, but, at least as far as it is preserved, only at the lower stage. The other facet (b), which contains the present entrance, has been rebuUt from the ground when the upper stage was added (see p. 123 ff.); it is uncertain whether it, too, was originally thickened internaUy, but the outer edges of the external thickening look primary and contain channels that suggest timber-lacings rather than draw-bar holes. The internal thickening of the northern facet carries the seatings (c) for the treads of a fairly gentle stairway, which terminated at the top of the lower stage of the wall, but apparently began less than 0 -5 m. below the final medieval occupation-level. Beside the thickening is the matrix (d) of a timber post, 28 cm. square and, probably, connected with the same staging as the stairway. The upper part of the thickening now carries a garderobe, the survivor of a pair (el, e2) with Roman tile dressings and covered by segmental arches, which, in turn, could have served as the floor of a turret a httle higher than the general wall-top level of the second stage. There are now four embrasures whose intradoses are near the toplevel of the first stage: three of the these (fl, f2, /3), close together at the acute end of the enclosure, were garderobes. /3 was altered and widened in the second phase, but aU three have been broken through. / I had its outer face repaired; later (in the Kennel period 1) it was blocked, and has recently been unblocked. It retains its tufa chute, part of its tufa jambs, and its internal rendering, with two waU-recesses, as in the others. It is probable that aU three were identically equipped in the primary buUd, though their floor-level is higher than that of the final medieval surface of the courtyard. The fourth embrasure, towards the north-east (g), may also be primary, but its linings have been renewed, and it contains a weU, at least in its present form, of the Kennel period. A line of rough recesses (h), cut into the top of the interior of the first stage of walling east of the HaU, is almost certainly also from the Kennel period and receptacles for fodder or harness.418. In the upper stage there are eroded suggestions of embrasures at the acute end and along the west side, including the faUen section, but none can be taken as positive evidence. On the other hand, the wellpreserved eastern and northern stretches are absolutely unpierced at either stage, save at the entrance and the north-eastern embrasure, and it seems that this was the intended aspect of the whole curtain at both stages, any piercings being conveniences of a later, presumably thirteenth-century, date. 41a They are not bee boles: this is the conclusion of consultation with the late Mrs. V. M. P. Desborough and Dr. Eva Crane. 121 S. E. RIGOLD Neither stage shows any trace of a continued stone parapet or wallwalk. This was looked for with particular care whUe consolidating the wall-tops, since Cresy's section shows a parapet. It is probable that he was generalizing from the short parapeted section (j), turned in Roman brick, beside the entrance and approached from the gate-tower of which he knew nothing. This and a similar look-out above the northern thickening could have covered most of the eastern and north-eastern circuit, and there is positive evidence that another form of waU-walk, a jettied timber hourde, existed in the second phase on the exposed south-east and south-west flanks. From the entrance southward and round the waU as far as it is preserved, but definitely not north of the entrance, there is a level series of channels, suitable for carrying the transverse plates of such a hourde. A row of holes, at much lower level, on the inner face of the south-west wall would have accommodated narrow joists, appropriate to the Kennel period rather than anything medieval. The position of the original entrance is not obvious. From the second phase of the curtain onwards the entrance was certainly in its present position. At this phase, the wall here was rebuUt from the ground, and an internal gate-tower constructed and, no doubt, bonded into it. The few courses of the gate-tower that remain show that the passage, several times re-surfaced on a bed of chalk, was flanked by two shallow recesses on the south and deeper ones on the north, the division between them suggesting that there may have been an intermediate arch across the centre of the passage. The inner arch, of which the debris was found, was turned in Roman tile. Only the thickening of the wall and a low-set channel, as though for a massive draw-bar, and apparently penetrating into the primary wall, suggest that there was indeed an original breach in this position. Yet the remains of the bridge include arrangements that go back to the rebuilding in the second phase, but show no trace of an earher structure. The primary breach, then, would seem to have been a postern or saUy-port, above the level of the foot of the waU but below the primary waU-top. There is no other primary embrasure sufficient for an entrance, those at the acture end being garderobes and the north-eastern one, if original, much too narrow. The hkeliest place for the original entrance is that suggested by Cresy (even though he did not reahze the wall was constructed in two stages), namely over the top of the wall at the northern reinforcement (a), and approached internally from the steps seated on the inner thickening. This would have led from the enclosure now containing Little Mote and the ditch here is narrow enough to be spanned by a single drawbridge. Such an entrance would have been of a purely defensive type, consistent with the original nature of the enclosure, not in constant use and not containing any domestic buildings. 122 EYNSEORD CASTLE The easy but protected entrance of the second phase befits the enclosure's new status as an inhabited baUey instead of a quasi-motte. The flintwork, the detaU, in Roman tile with a httle roughly squared greensand, and the whole logic of the structures indicate that the hall-block, the heightening of the curtain and the entrance and gatetower are of one construction. The haU, which Cresy envisaged as a three-storey keep, was not a great tower but a normal first-floor hall and the defence remained in the curtain, which now mastered the whole lower floor and some 2 m. of the upper. The floor level of the upper storey, of which nothing remains, can be deduced by completing the extrados of the arches that supported it across the centre (Fig. 7) and from the garderobes that were buUt on top of the northern reinforcement (a) and which must have served the solar of the hallblock, via a bridge. The Hall The hall-block, though not a 'keep', is a buUding of some strength, lighted only by narrow loops on the ground-floor, even in the part intended for habitation. The best preserved is turned in Roman tile; all have a steep internal downward splay. The most 'keep-like' detail, the porch-turret, or forebuilding in the narrower sense (I), covering the entrance to the first-floor haU, is an early addition, and so, surprisingly, is the stone base of the external staircase (m), parallel with, but not attached to, the south front of the block, suggesting a gentle staircase, perhaps largely of timber, unprotected at its foot, but apparently with a pair of entrance-arches, separated by a trumeau. When the porchturret was built, it was hnked to the staircase by a wall containing a door (n), which was subsequently blocked and the whole forebuUdingcomplex enlarged by a projection along the front. This and other, probably thirteenth-century, additions will be discussed together below. The hall-block, with its early extensions, which hardly differ in masonry from the original, will be treated as a unity. Apart from the stairway (m) and the gap between it and the porchturret (n) which could at first have been spanned by a drawbridge, the undercroft-level comprises three cells, which were almost certainly repeated on the upper floor—that in the porch-turret and two in the haU-block proper. The longer and eastern of these was under the haU itself, since the stair and porch-turret indicate an entrance at the east end of the south wall. The western ceU, which has no communication with the other at underoroft-level, is assumed to underhe the solar, or inner chamber, which in turn would have had access to the garderobes on the northern thickening of the curtain. All three cells had very high ceUings for undercrofts (Fig. 7): this is stUl apparent, even though the floor levels now displayed represent a raising in the thirteenth 123 S. E. RIGOLD century. The original height was nearly 6 m. The floor-joists of the hall, which must have run transversely to the block, were supported across their centres by an arcade of three arches on square piers with Roman tile- quoins, the arches themselves, as far as can be deduced from the shght remains of springing, included flue-tiles and roof-tiles. Though these bear on the relatively thin partition-waU, the arcade was not carried across the solar-undercroft, where the joists must have run axially to the block. Both the porch-undercroft and the hall-undercroft were entered separately from the space beyond the stair-base, each door having a draw-bar: that of the porch is convenient to its present level, that of the hall-undercroft low-set and convenient only to original internal level, though the threshold of the doorway was not raised with the floor and must have required internal wooden steps. At the north-west corner is a spiral staircase (o) linking the undercroft to the Hall above, the arched doorway turned in Roman tile. In this corner, the floor-level does not seem to have been raised. Both the hallundercroft and porch-undercroft have no other facihties than for storage. The solar-undercroft, on the other hand, formed a complete and independent residential apartment, high enough to have contained an internal staging, though there is no trace ofthis, but not communicating with the solar above. The entrance, down steps, is in the west wall: at the foot ofthis is a weU (p), in a recess, which has been shown to be a relic of an earher building on the site (see p. 121); beside this is a rounded wall-fireplace (q), backed in Roman tile, and certainly original though altered in the thirteenth century. In the partition-wall is a rectangular aumbry, and in the north-west corner another doorway turned in Roman the (r), leading, via a bent stairway in the thickness of the waU, to a small privy, the outlet of which was extended by a cess-chamber (s), of which the outlet, when found by Cresy, retained its segmental-arched head. Apart from the privy-outlet there are two projections from the north facade of the haU: the wall is thickened where it contains the spiral staircase and again towards the east end, and it is possible that the waU of the haU above was carried on an arch at the projected face. The hall block contains a number of alterations in less regular, flush-pointed flint rubble and weU-tooled, if friable, greensand ashlar. They all seem to belong to a rehabilitation consequent on the thirteenthcentury fire, which excavation shows to have devoured aU three cells of the building. They comprise: the raising of the floor of the hallundercroft by nearly a metre and the solar-undercroft by a much less amount; the enclosure of the western pier of the undercroft arcade by a mass of flintwork (t) which almost certainly carried a central 'pedestal hearth' and suggests that the fire may have been caused by the failure 124 EYNSEORD CASTLE of a hearth resting on the arcade alone; the improvement of the fireplace in the solar-undercroft by a surround with bar-stopped chamfers and probably a hood, and the addition of a block of masonry (u) to the end wall, almost certainly to carry a chimney for the solar above—the fragment of a conical chimney-top of stone, with trefoiled openings and no sign of soot on it, must come from one or other of these; the making of a chamfered ashlar door-case, standing shoulderhigh in Cresy's day and approached by steps, at the end of the passage between the haU and the external staircase, thus finaUy making the space before the haU- and porch-undercrofts into a proper room, and the enlargement of this space by a projection on the south facade; and probably, though it may be earher, the retention of the privy outlet. In this rebuUding the hall was completely roofed in tiles, which had been used only on a hmited scale before, and at least the east window of the first-floor haU was glazed. It was now a comfortable house and had lost any resemblance it may have had to a donjon: on all the evidence, it was only occupied for two short periods, but it is the remains of the building in this state that the visitor sees. Other Structures Low walls remain of two kitchens. One, the 'Great', or 'Old' Kitchen (OK), fits into an obtuse angle of the curtain, near the well in the courtyard. Pottery associated with its substantial footings, probably not the first on the site, and the re-used look of its tufa dressings indicate a date in the twelfth century but later than the Hall. It was several times re-floored as the external level and threshold were raised, and the final floor, in use until the dismantling of the Castle, included a broad hearth of tile-on-edge (v) with a short spere beside it. The other, 'New', Kitchen (NK) was between the Hall and curtain at the north-east, and so hardly more convenient for the main staircase than the Old Kitchen. It probably had its own external stair at the other end of what would have thereby become a proper screenspassage, and it seems, for this among other reasons, to be part of the post-fire reconstruction. It was a timber-framed buUding on narrow ground-waUs with seating for the posts, whose feet thus 'passed' the interrupted sole-plates. The fire-back, against the curtain, is of greensand and tile-on-edge. The only certainly medieval lean-to range (L) against the curtain ran from the gate-tower to the acute angle. This had a broad, though mutilated ground-waU, possibly twelfth-century, which contained the seating for at least one post. Beside it were fragments of hghter structures (yy, zz). AU the other flimsy scraps of walling, now mostly removed but shown on Fig. 3, date from a patching-up after the dismantling, or from the kennels. 125 S. E. RIGOLD N sin e w ) ww e? NK N hh I II a % id t D ci* $ •/ i ® <« ^ OK n'x O m OT ^ E / / >• i i '» ; X1 x& / |vvj£j f'luisc K ^ // F--3| /*«»• E ^ ['•'•] I imp Mortar Floor!. «I (' n [1 Metrfs 5 (i n 20 in 41) « t Eio. 3. Eynsford Castle: Exoavation Plan. The position of an early central tower (OT), antecedent to the hall and contemporaneous with the lower stage of the curtain, is marked on the surface. It wiU be discussed in the excavation section. The foregoing description is without prejudice to absolute dating. The architectural detaU is shght; the plans aUow some latitude of date; the associated pottery provides a comparative dating but has few 126 EYNSFORD CASTLE external references except at the end. I t wiU be convenient, therefore, to summarize the dates argued on historical grounds. The lower two-thirds of the curtain and the defensive structures associated with it (see below, p. 132) are ascribed to the early career of William de Eynsford I, in the late 1080s. The hall, gate-tower and heightened curtain to William I I in the late 1130s. The Great Kitchen and perhaps the forebuildings seem to be the work of Wilham III, some time in the later twelfth century. The reconstructed haU and the new kitchen are from towards the end of the undivided tenure, a decade or so before or after 1240. The dismantling was in, or just before, 1312. The subsequent repairs were ephemeral. The kennels lasted from before 1783 to 1835. EXOAVATION The excavation of 1835, reported by Cresy42 but led by two local clergymen,43 though exceUently surveyed, was, of course, just a wallfollowing. They missed the kitchens, the gate-tower and even the internal piers of the hall. Their trenches, clearly seen on all the sections (Figs. 4, 5), were seldom deep enough to damage significant stratigraphy. The value of the report hes in the precise description of parts since fallen, including the west curtain and certain details of the Hall. The excavations between 1953 and 1961 were to some extent determined by the consolidation of the ruins and their exhibition at a suitable ground-level, which, according to Ancient Monuments practice, is that of the final occupation of the Castle as such. This suited conditions of heavy overburden in a confined area: deep cuttings were generally made subsequently. I t did not affect the work on the bridge and moat, which is here treated only incidentally. The traces of the eighteenth-century kennels were reckoned as expendable but worth recording. The first soundings, in and just north of the Hall, showed this occupation lying on a considerable depth of flint rubble, leached white and free of soil, and distinct from a layer of debris beneath it, in which red roof-tiles predominated and rubbish, especiaUy pottery, abounded. This pottery immediately suggested, and no evidence to the contrary has since appeared, that the debris was to be associated with the damage complained of in 1312, and that this involved, beside breaking-down of doors and windows, stripping of much of the roof but not systematic demohtion. Subsequently, it appeared that, in a few areas, particularly north-east and north-west of the HaU, there was flint rubble under the tUes or thick clay between two layers of tUes, and in other places, notably in the porch-turret, small areas of lime floor immediately above the tUe-spread and under the deep flint rubble, while two pieces of rough walling (w and ww) were 12 See note 1. 43 B. Wenston and A. W. Burnside, an early member of this Sooiety. 127 S. E. RIGOLD built upon the tiles debris. These (see below, Phase E) must represent an ephemeral attempt to patch the HaU up after the unroofing, and a little of the upper tile debris may weU come from the patched-up roof. These remams had no associated pottery and were neghgible from the point of display. In general, the original supposition was sustained: the tile-debris represented the end of the last serious medieval occupation and immediately covered the floor levels to be exhibited. There was hardly a single find between the early fourteenth century and the later eighteenth, and, though some levelhng of wall-tops was in preparation for the Kennels, most of the accumulation of flint can be ascribed to nearly five centuries of neglect, without much interference from the villagers. Controlling sections were cut, where feasible, on the same intersecting lines and the overburden removed as on a 'face'. The designation of the section-lines and the stratigraphical terminology estabhshed as work progressed wiU be used in this report. Though the lines form an ad hoc grid, recording below the exhibited level was almost entirely by section-trenches and httle was examined in breadth except within the hmits of a trench. From short lengths dug at different times, it has been possible to piece together the equivalents of long continuous sections, in some cases right across the enclosure, as shown on Figs. 4 and 5. The disadvantage of piecemeal excavation has been that it has been impossible to compare the strata visually throughout a section, and for practical reasons, it has not always been possible to cut every part of a composite section absolutely on the same line. In one or two cases structural features near, but not on, the hne are shown in elevation. The sections were called by Greek letters. For short, the whole section wiU be referred to, e.g. as 'Section a'. Only the section-lines, not the trenches, are marked on Fig. 3. The terminal points, e.g. al, a l l , always lead from west to east or south to north. The main west-east sections, shown on Fig. 4, are: a, north of the hall, with 0, just behind the hall j8, through the haU and solar undercrofts y, through the forebuilding complex TJ, obliquely, through the gate-tower passage, towards the well. The main south-north sections, shown on Fig. 5, are: 8, through the garderobe and solar undercroft, with 8* showing rather different conditions 2 m. east e, through the haU undercroft , south of the haU, spanning the early tower (OT). Short sections, approximately west-east, shown on Fig. 4, are: A, on the edge of the Great Kitchen exposed by the collapse of the waU. 128 & Q SECTION 1PJ 5J METRES (yt SECTION '///////, '/&ZZ- aass5>?^be?K§sp«o Wmg^sm^. P&>5«9o©2 3 SECTION r^ssswss 1«^^ /x4o£'9/4&~7777~, X SECTION 5 10 METRES WAV////////; i-EET Fio. 4 Eynsford Castle: W-E Section. (Key overleaf) [face p. 128 £ SECTION 0 ot ^ ^ ^ ^l^^^x\\\x>^^^^ _ _ j 5 SECTION W R lumi cISS^pOo^ooo JOo nn ° ° T - . ^ ^ ^ ^ ^ ^ ^ ^ S ^ O ^ ^ m ^ e p ^ R W102 W102-+X K SECTION 1/ SECTION

and, later, showed a not very massive flint wall with yellowish mortar, like those in the undercroft, standing to full height, i.e. to the seating for a timber sole-plate (PlateIII), not far below the yellow-brown surfacing of the 'platform'. When the ends of this were located, it appeared that the lateral walls running off it had been broken down, so as to slope steeply towards the Hall, and that they bounded a quadrilateral, but not rectangular, building, which had been destroyed to make way for the HaU and was connected with the waUs under the solar undercroft. From its central position and the seating for the plate, they were evidently the ground-waUs (though 2 4 m, high internally) for a timber tower such as is known to have formed the central feature of many mottes, hereafter referred to as the Old Tower (OT). Further trenching was directed to examining the relations of the platform and its make-up with the curtain, the Old Tower and the Hall. Sections against the Curtain. These had aheady been cut at the east ends of fi and y, at K on the south, and at the north end of 8. They were not very conclusive and showed different soil-conditions in each cutting—darker soil under the yellow-brown capping in the first two, and hghter graveUy clay below that; undifferentiated dark soil in K and chalk and flint on a graveUy bed (W 102) in 8. But they all showed that the curtain was founded on an artificial bed of chalk (confirmed by an external cutting at v), that the surface-rendering of the waU was everywhere well preserved to a height of about 2 m. above this chalk, that aU internal soil had been pUed up against the curtain after it was buUt but that a broad and persistent channel, in places re-cut below the weU-preserved rendering, ran round the interior of the wall. Not untU the more accessible west ends of fi and A were cut in 1967 did the true sequence emerge: about 0-7 m. of varied strata, running 132 EYNSFORD CASTLE fairly level, and including a productive soil layer (W 101), had accumulated against the curtain, which, on falling, had carried some of its chalk-bed with it,40a and leaving the taU of it under the strata. Then, clearly after some interval, but not long enough to damage the rendering, the bulk of the platform was piled up as a low mound with sloping sides, thus creating the channel or guUey between it and the curtain. A thin clay-capping visible in K, A, and fi, west, was added shortly before the demolition of the Old Tower, and the sohd yellow-brown capping after the building of the HaU, marking the transition from Phase X to Phase Y. The Old Tower: external Conditions. The base of section^ showed that the inner face of the Tower was also founded on a bed of chalk, but that the outer face did not reach the same depth, and was apparently trenoh-buUt down to a possibly natural bed, above which was an accumulation topped by humus and almost certainly equivalent to the level basal strata in fi and A. Only above this is it well-rendered, having been buried, almost to its top in the 'second mounding', which comprises a variable accumulation of clay, gravel and brick-earth, topped by the clay capping noticed in K, A and fi, and, over that, another flint and gravel surface, aU within the lifetime of the Tower. It appears, then, that the Tower was founded after the fairly level strata had begun to accumulate against the curtain (probably a short enough time to treat them as consecutive), that a humus layer (W 101) accumulated against both, and, after a moderate interval, the second mounding buried the whole external face of the tower but left the channel round the curtain. AU wares associated with these phases of building and mounding are treated as 'W', and show no perceptible development. The first presumption, that the Old Tower, on a very low motte, preceded the curtain was disproved. Interior Conditions of the Old Tower and its Yard. That the Tower was not completely embedded in the second mounding was evident from the attached waUs under the solar undercroft (66, dd) and from the fact that all was demohshed except the south waU, protected by the mound. In fact, it had a yard which formed the emplacement of the HaU and provided its deep undercrofts, and the platform, though extensive on the south, was reduced on the other sides to a mere inner ring round this yard. Neither Tower nor yard can be fuUy recovered in plan, but some indications are given by the section of fi immediately west of the Hall, by the section of 8 at the north end of the solar undercroft, by the distribution of rubble under the footings of the haU and by a point of depression in section a just east of the 400 Fig. 4 here shows the saltire (greensand) symbol; it should be crosses (chalk), and W101 on seotion X should be taken as the horizontal layer included below, not level with, the marking. 133 13 S. E. RIGOLD garderobe, which may respond to a corner of the yard. The west part of section fi is most informative: the Hall wall was not built in a trench but in the free area of the yard and the extent and inclination of this is given by a timber-slot (ee), presumably the sole-plate of a revetement for the 'second mounding', which is here largely of flints, and shows at ff a mutilated fragment of its upper surface, at the normal level of about 2 m. above the yard, also bearing signs of timber-slots and carbonized wood. This was damaged by the final reddish clay-capping of the second mounding, which, in turn, antedates the hard, yellowbrown capping that seals Phase X. As the section shows, the reddish clay was laid down after the revetement of the second mound had collapsed and a layer of soU had gathered over the slope. This, as well as the variety of pottery forms, shows that Phase W was of long duration. On section 8 a cushion of clay may indicate the edge of the level yard, and north of it a line of flints and yellow mortar (hh) may suggest the alignment of yet another structure. More flint and yellow mortar appears on fi just east of the cross-waU, at this level, and again probably near the limit of the yard. The waUs in the undercroft (bb, dd) are cut into by those of the Hall; they seem to belong to a stone well-turret at the corner of the Old Tower, and higher and stronger than its ground-walls, having a level stone stage before it. Over bb and dd and in the Tower itself the conditions of destruction are well seen. The mortar floor of the Tower had broken up under damp and silted up before its walls, as well as 66 and dd, were buried in a mass of yellow rubble from the destruction of other parts of the complex. This was masked by an even heavier deposit of dense black-brown flinty clay, evidently introduced. Both layers contain 'X' pottery, uneroded. The black-brown clay spreads even over the top of the Tower waUs and extends (section fi) into the area of the Hall undercroft; it is all sealed under the hard yellow-brown capping of Phase X or under the floors of the HaU, totally obliterating the Old Tower except for its well. The well by the Great Kitchen has the same yeUow mortar and the 'second mounding' requires that it should be of similar date. X. The Construction of the Hall No stratigraphy attends the upper stage of the curtain, and the rebuilding of the entrance section is better treated in the context of the timber bridge. The gate-tower is contemporary but needs no further explanation. The original Hall-block stands out clearly on the drawn sections, built on deep footings of flint-in-clay, well spread under the cross-wall and south side (near the site of the Old Tower), more confined elsewhere. 'X' pottery, though found more in the destruction layers than in the foundation-trenches, occurs in the mortar of the HaU walls. The east wall was trench-built, the filling containing scraps 134 EYNSFORD CASTLE of the Koman tile used for dressings. The north waU, though not examined to its base externaUy, seems also to have been trench-built. It is probable that the eastern end and much of the northern part of the Hall lay beyond the earher yard. Soil conditions change near the hne of section e, and a charred layer, a temporary hearth (aa), containing 'W type pottery, sealed by a bed of clay underneath and just west of the east wall of the Hall, seems to have been preserved by the mounding cut away to build these parts of the Hall. The primary floors of the undercrofts, of thick hme-mortar, bedded on clay, are much sunken and reflect irregular conditions beneath them. The external finishinglayers, after the building, show the hard, yeUow-brown capping on the south and west and, outside the trench-built east and north walls, a red clay layer covering a depth of sterile black sUt. The first stages of the forebuilding-complex come somewhat later. The porch-turret has a building trench of different profile from that of the east wall and pottery from a pit under its floor resembles 'Y' rather than 'X'. Even the great stair-base, which should antedate it, is built on top of the yellow-brown capping, which itself produced a rim of 'Y' character, west of the hall. Y. The Building of the Great Kitchen From this phase occupation-rubbish becomes plentiful, as though the Hall, though built in the late 1130s, had been httle used until the tenure of Wilham III. As indicated above, the forebuilding seems to begin early in this phase and 'Y' sherds characterize the gravel surface east of and consequent on the building of the porch-turret. The finishing layers that followed the buUding of the Hall did not extend to the west curtain, where the preceding reddish clay-capping remained in a depressed belt, probably containing outbuildings. In Phase Y' (see sections fi and A) this was raised by a mass of chalk, part of which formed the emplacement of the Great Kitchen, and thus sealed a deposit of mature Y' pottery (Y 105) on the slopes of the intramural gully. A hardly distinguishable deposit (section v) was in the reddish soil (Y 106) that covered the chalk and the first, short-lived floor of the Kitchen itself. The first of a series of mortar floors, often patched and sealing small quantities of pot between them, around the western end of the HaU seems to date from this phase. The sherds from the temporary hearths in the channel against the east curtain (sections fi and y) are also of 'Y' rather than 'Z' types. Z. LeveUing of the margins No structures are associated with this phase, except on slender associations, a light structure with a fire-reddened hearth (xx), towards the south. But there are numerous deposits between the mortar floors (see above) and in the final fillings of the marginal channel, which now 135 S. E. RIGOLD effectively disappeared. The largest of these, a pit sunk below the offset level of the Great Kitchen, filled with loose red earth and sealed with clay (section p,) provided the sample of pottery here taken as representative. A. Further accumulation and Industrial activity Again no structure, but the rubbish which had hitherto been confined to discrete pockets is spread everywhere: 'A' wares predominate wherever the final humus hes directly under the 'D' debris. In particular, on the north side of the Hall, which had hitherto been generally clean, the old topsoil, lying directly over the hard yeUow-brown capping, has much charcoal at its base and contains almost exclusively 'A' type pottery. It also contains slag, not tap-slag, which, unless it is the result of some abnormal vitrification (nothing from BB suggested it was caused by the fire in the Hall), must be the result of iron-working in the north sector. The only sign of an instaUation connected with this was a shaUow U-shaped trench, kh, lined with reddened clay and based on chalk—not a bowl furnace, whatever else it was. A sealed midden with 'A' pottery overlapped the sealed 'Z' pit on section p.. BB. The Fire in the HaU. 'A' was clearly a long and busy phase; the succeeding B was a short phase before a long period of desertion. The Hall as reconstructed after the fire shows little sign of occupation. Nevertheless, it is probable that the fire-levels (BB) in the Hall are not, as originaUy thought, contemporaneous with B, which is also characterized by an ashy layer, but represent an incident towards the end of Phase A. The pottery from the fire-levels is not quite enough to prove this: the one complete bowl had a 'B' profile but 'A' fabric (BB1); the hard non-shelly wares are best paraUeled in A. The best indication is that B is later than the New Kitchen, which seems to have been built after the fire, but the question probably only involves a difference of a decade or two. The destruction was extensive, with quantities of heavy timber falhng, completely charred, into the undercroft, but very httle other material among them. Charred wheat was found in the solar undercroft. The reconstruction, quite luxurious for its day, has aheady been described. In the undercrofts of the hall and porch-tower, the floor was raised by a considerable depth of dry flint and gravel, sterile enough to preserve recognizable remains of small mammals. Saving the question of B, no effects of fire were traceable outside. B and C. The Ash-deposits and subsequent sealing B was the first sealed deposit to be recognized. It is confined to the right-angled triangle between the HaU and the New Kitchen and 136 EYNSFORD CASTLE consists of a layer of fresh, white wood-ash, different from the black deposits of BB and always distinct from the charcoaly 'industrial layer' of A. I t was packed with pottery, especially of the characteristic final, soapy-surfaced form of shell-filled wares, and also full of naUs, as though the wood had been charred roofing material. It is better to interpret it as exhausted fuel from the New Kitchen, rather than as the remains of a pot-store destroyed in the fire, and as representing a short and intensive use of the reconstructed Hall and Kitchen. It was immediately sealed by a thick blanket of red clay, containing a httle of the same types of pottery. The clay also sealed the exposed 'A' soil north of the HaU, and its effect must have been that of a complete tidying-up after half a century of messiness north of the Hall. IntentionaUy or not, this coincided with a phase that left no deposits at all until the pottery types had completely changed. I t is difficult not to associate this newly and finely refurbished house, so suddenly deserted, with the premature death of Wilham VI, while his infant son passed into the Archbishop's wardship, himself to die in 1261. III. THE BUILDING IN OUTER BAILEY, D I am grateful to Mr. S. R. Harker for a description of this, now being excavated by Mr. and Mrs. J . M. Allan and himself. I t overlies a cemetery of Christian orientation, with flint pillows under the skulls, and practically without grave goods, but was built in ignorance of it. The date of construction is uncertain; 'A' type pottery is associated with its demohtion, after which it was partly covered by the building rubble that makes up most of the apparent baUey bank (whatever its original age or form). The ahgnment of the building is not far off that of the Hall. I t is 5-65 m. wide (easily spanned by a tie without aisleposts) and has been traced for over 8 m. to date. The side-walls, 60 to 65 cm. thick, indicate a flint, not timber, construction, but the short end-wall is much hghter. There is no sign of other than a clay floor, nor of any partitions. AU this points to a barn, rather than a stable— certainly not to an inhabited building, but to one proper to an enclosure in advance of the present entrance. THE SIGNIETOANOE OIF THE CASTLE Treating the curtain and the timber tower on a stone base as both integral to the original conception, Eynsford is hard to match among early Norman castles. The parts of multiple-enclosure fortifications are commonly distinguished as 'inner' (Hauptburg), the classic instance being the Motte, and 'outer' (Vorburg), either or both of which may contain habitations, but if both, the inner usually of subordinate status. The primary Eynsford Castle contained no permanent accommodation and its connection with an unfortified outer enclosure is not 137 S. E. RIGOLD proven, nor, in any case, intimate. It is perhaps best to treat it as an independent enclosure, but with the function of a motte, as the place of observation as well as of last resort. Inner defences have been divided, sometimes more by permutation-typology than by observation, into various classes of motte and small ringwork, with intermediate 'platform-ringwork' or 'ring-motte' among them. Eynsford could be regarded as the hmiting case of several of these. The low but massive curtain, the first thing on the site, was the equivalent of the earth-walls of ringworks;44 the low mound, sloped round its margins, when it had been raised to the top of the ground-wall of the tower, might be compared with a motte, in that all mottes examined in detail carried, or more often contained, towers. But the mound is not just the earthen glacis of a tower but, with its sunken yard, a combined platform and ring-work in its own right, and with its shallow layering, artificially reproduces just that sort of accumulation that has been observed in several cases to transform a vulnerable Flachsiedlung into an elementary castle, but for which the basic reason, as here, was to avoid flooding. Eynsford is thus exceptional because it is so generalized and because it lacked accommodation. The paramount strength was in the curtain. If treated as an isolated enclosure, rather than as the notional Hauptburg of something else, Eynsford takes its place among those early Norman instances of simple enceintes, castella almost in the Roman sense, envisaging, on occasion, perhaps on an occasion that never supervened, a relatively large body of men to man the walls. With its three or even four garderobes, but no permanent accommodation, it seems to fit this contingency, and the supposition that it was built, not to protect or elevate a private dominium, but, in a time and region of contention, as an instrument of the milicia totius archiepiscopatus, by Lanfranc's most trusty man, William fitz Ralf, and perhaps with the counsel of the faithful Gundulf. The tapering oval plan may be fortuitous: the acute angle was not normally exposed, hke a cut-water, to direct flooding, and the terrain does not dictate the shape. There are, however, analogies in the large and early curtained enclosure at Saltwood, later in the archbishop's tenure, and closer in place and function, in the probably contemporary (1087-1089) Gundulfian castle of Rochester, built as a royal castle, as yet without a differentiated donjon and probably to assemble large forces. Here the acute angle points to the bridge, as that of Eynsford does to the ford, and the advice of Gundulf at Eynsford is hkely in the historical circumstances. It is not too much to see in Eynsford the archbishop's smaller version of Rochester. The Eynsfords themselves seem subsequently to have imitated the plan at Great Staughton. ** See J. Cathcart King and L. Alcock, 'Ringworks of England and Wales', Chateau Qaillard, iii (1969), 90-127. 138 EYNSFORD CASTLE The tentative reconstruction (Fig. 6, a) of the primary period 'W' (after the full height had been achieved) incorporates as a demonstrable element, the tower in its sunken courtyard (boundary uncertain on the east), and in default of a better, the high-level entrance on the north. The httle bretasches on the short lengths of curtain are speculations, but something of the sort is required in default of a wall-walk. The suggested construction of the tower is based on the earhest timber bell-towers45 and such details as the stair-base turret (?) at Rayleigh, Essex.46 From these are derived the tapering form and the strong cross-bracing, lapjointed on the face, and shown exposed in the lower stage, though it would probably have been completely sheathed in vertical boarding. The jettied platform is consistent with the tapering outhne and the roof has the authority of some of the motte-towers shown on the Bayeux Tapestry. The sole-plate (ee) seems to imply a near-vertical revetement to the yard. From Phase X, the character of the Castle changes to that of the inhabited inner bailey of a private domain, approached from the outer bailey, D, through the strong gate that remained in England, from Saxon times, the mark of armigerous status. In the reconstruction (Fig. 6, b) of Phase B, of which the essentials derive from X and Y, every element has some evidence, if not conclusive. The proportions of the Hall block are based on the longitudinal section (Fig. 7) which, though any error is more probably on the positive side, shows the great height of the undercroft and the form of the piers, both of which were examined and found to have doubly offset bases, dressed, like the quoins, in Roman tUe. The hipped roof and at least one polygonal, spired chimney-top are demonstrable. The central hearth implies, in this case, probably a louver, and a single-span roof of great height, however low the lateral waUs of the first floor were. Gabled lateral windows are a possibihty, but the Lake House at Eastwell47 has been used to suggest the general proportions of the roof and disposition of lighting. The hourde is more generahzed: it may not have survived at fuU length in the late phase, and may have been elaborated towards the acute angle, like the basically late thirteenth-century tower-gallery at Stokesay, Salop, which suggested that the hourde was fairly completely boarded over externally. The relative cleanness and south-western aspect of the ground around the porch-tower may indicate a garden. How wet the moat was at any period is beyond conjecture, as the level and direction of the river is uncertain: it was not much lower, since the well only just breaches the present water-table and the moat could, 46 As Brookland, Kent; Navestook, Essex; Pembridge, Herefs. 40 Trans. Essex Arch. Soc, N.S., xii (1913), 159; cf. D. F. Renn, Norman Castles in Britain, 85, fig. 1. " Arch. Cant., Ixxxiii (1968), 155-61. 139 S. E. RIGOLD h ifrvSJ^^^G^S^SSS^ m PHASE W s? PHASE B Fia. 6. Eynsford Castle: Reconstructions by S.E.R. 140 I I I I & S O L A R STAGING? H A L L HEARTH H GT~IS-HTE3 "—T3—fcitil I h l t s l l i J U a f c d L d b i i i J ^7777 fq 10ft 3m FIG. 7. Approximate Section of Hall Block. S. R. RIGOLD and stiU can, hold water at the right level. These considerations and the lack of a berm suggest that water always played a part in the defences, but the moat was never deep and the surroundings are shown as marshy. THE FINDS Sources are in brackets. Dimensions not stated where there is a drawing. LMC = London Museum, Medieval Catalogue, London, 1954. I. Building materials and stone (Fig. 8) 1. Fragment of hexagonal chimney-cap, with gabled lucarnes containing trefoil-headed vents, around pinnacle, and hole for cramp to next tier; in fine white limestone, more granular than clunch, with close toohng. Some weathering, consistent with fifty years or so of exposure, but no smoke-stains (D, solar undercroft). 2, 3. Not much loose dressed stone: a rectangular shaft (11x12 cm., 28 cm. long), roughly dressed to a cylinder, has coarse diagonal tooling on the original faces. Nearly all the rest, including chamfered and rebated fragments, is in poor rag or soft greensand with coarse vertical toohng, as on the bar-stopped chamfered jambs of the fireplace in the solar-undercroft; all this must be post-fire, and includes a piece of attached shaft, dia. 13 cm., and a moulded section, with half-round and cavetto, as 2. The bull-nosed section, 3, is in chalk. 4. Five pieces of fired red clay containing a httle sheU, pierced with circular channels about 1-8 cm. dia., outer surface flat; too regular in shape and fabric to be mere burnt daub, it suggests a dehberate material for specialized use (Solar undercroft, X, destruction-layer of OT). Roman tile: mostly bonding-tUe, a httle flue- and roof-tile. Many fragments from construction-layers (e.g. east building-trench, section fi) of Hall; whole tiles (D or over) from collapsed inner arch of gate tower. 5, 6, 7. Medieval roofing tile (vast amounts from D; smaU quantities from sealed Z, A, near Great Kitchen, and B; hardly any from BB. This seems to indicate that before the fire most of the HaU complex was not tUed, but that the Kitchen was. The usual red Kentish pegtiles with two holes, usually without reduced core; at least two varieties: the earher, 5 (from A and B as well as D), shghtly smaUer, more distorted, buffer in colour, the holes usually closer together; 6, the majority of those from D, larger, redder, often with two to six regular ridges made by a flattening tool, peg-holes variable but often asymmetrical; a few tUes with orange glaze. Ridge-tiles, 7, uncrested, often with orange, occasionally with ohve-green, glaze. One or two hip-tUes. 8. Piece of whetstone, with grooves both sides, in fine pale-brown sandstone (D). 142 FIG. 8. Building Material and Stone ({). S. E. RIGOLD II. Ironwork (Figs. 9-10) An instructive and tolerably preserved coUection, all from D unless otherwise indicated. Discounting the probably intrusive Fe 4, 13 and 20a, much of the material is stiU advanced by commonly cited comparisons, since pottery and other finds are entirely consistent with a date for the rebuilding of the Hall not later than c. 1250, and for its dismantling (D), c. 1300 (probably 1312). Fe 1, 2. Strap-hinges of small lap-boarded door, with 'moline' terminals, badly bent (north of Hall and doubtless from its reconstruction). Fe 3. Terminal of another strap-hinge, with edge-beading (same context). Fe 4. Half of butt-hinge, or fixture of long hinge, in tinned iron of good workmanship. Reported context as Fe 1-3, at depth, c. 1 m., below K period floor; but it is so hke an H-hinge that an eighteenthcentury intrusion must be considered. Fe 5-10. Horeshoes, aU of same general shape, with 'pointed arch', moderate calkins and no trace of an indented edge, except Fe 10, which has as good a D provenance, from the floor of the solar undercroft, as the rest. This (for a 'great horse', rather than some sort of 'hack'?) is much heavier and fullered, for which LMC can cite no case before the late fourteenth century. Fe 11. Key (D, east sector), medium-size, finely wrought, with oval bow (not made by sphtting the shank, but welded?), extended point and stop half-way along wards. This type (LMC Type VII), with kidney-shaped bow, is common in the later fourteenth and fifteenth centuries, but would be advanced by 1312—it was not necessarily part of the rebuilt Hall. Fe 12. Key (D, floor of HaU undercroft, and probably its door-key at the reconstruction), D-shaped, welded bow, sohd stem, suggestion of a stop but no remains of a point; if the wards were symmetrical the bit was broad and shallow. Fe 13. Square padlock, presumably K period, but it is diflicult to see how it reached the deep masonry-joint beside the weU in the solar undercroft at that time. Fe 14. Rowel-spur (near Great Kitchen, A associations, but not sealed, so that D cannot be absolutely excluded). Rowel-spurs in the early thirteenth century are not totaUy unevidenced, and this one has features, such as the simple terminals, one triangular (?), reminiscent of twelfth- and thirteenth-century prick-spurs (LMC). The fabric is weak and cylindrical in section (it is much bent) and the reinforcement at the heel is clumsy. Fe 15. Tang of large knife (B, section 6). Fe 16. Tang of file or chisel (same context). 144 Fe20 FIG. 9. Ironwork (£). M>^*6fy?<6S»f< «» Prefix Fe- B V^ FIG. 10. Ironwork (J). EYNSFORD CASTLE Fe 17. Rectangular-sectioned rod with S-terminal and plate (?) attached (D, east of hall); uncertain purpose, too light for a windowfitting or branding iron. Fe 18. Tapering rod or ferrule, 14 cm. long, 1 -0 cm. dia. at hollow end (same context). Not illustrated. Fe 19. Tang of heavy chisel (same context). Not iUustrated. Fe 20. Uncertain—terret ? (same context). Fe 20a. Small compasses (same context, or intrusive?). Not illustrated. Fe 21. Small cranked spike, not a nail (X, solar undercroft). Nails generally fall into three classes (the large mushroom-shaped Fe 22, from D, is unique): i—clouts, with round heads, shghtly domed, in all contexts from Z to D (Fe 23 is a trifle larger than average); flat or squarish heads (Fe 24, 25) are in a minority, but common in the kitchen: U—ordinary cut nails (Fe 26), common from Y onwards: iii—nails with 'figure-of-eight' heads, in two sizes, those with slender heads (Fe 27), common from Z onwards, and those with larger, more bUobed heads (Fe 28), very numerous in B (ash-layer, north of Hall) andD. III. Objects of copper-alloy (Fig. 11) All from D, east sector, except Cu 6 and 7. Cu 1. Strap-chape with zig-zag rouletting, in red alloy with traces of gilt. Cu 2. Scabbard-chape, plain, two pin-holes and roughly soldered at joint. Cu 3. Strap-end buckle, neat workmanship, single rivet. Cu 4. Gilt button with separate dome and base. Cu 5. Perforated strip. Cu 6. Folded strip with six holes, in yellowish latten (A just north of Hall). Cu 7. Jetton (courtyard, topsoil); the only numismatic find and practically the only find of its period (except 6 1 ? ) . Late Nuremberg, dia. 25 mm.; normal types—Reichsapfel in trilobe/three crowns and lys; name of Hans Schultes (fl. 1550-74) both sides. IV. Glass (Fig. 11) Vessel-glass: G 1. Frilled base (uncertain context, north of Hall); looks sixteenth-century, but little else of this date. G 2. Base of vessel, 8 mm. thick at bottom, thinning to 1 -2 mm. (certainly medieval—sealed A context, north of HaU, east of garderobe). Such vessels, rare in the thirteenth century, are usuaUy classed as lamps or medical urinals: the weighted base would suit either. 147 S. E. RIGOLD cm Cu2 Cu3 :,A»«yJfc0 ^'^'"^ \ j c n c 02 =3- a 03 Fia. 11. Various Finds (J) 148 EYNSFORD CASTLE Window-glass (Fig. WG): about 20 fragments from D, east of HaU, representing a minute part of a high quahty window or windows. Red and blue strips, without visible painting, and grisaiUe (G 3) far superior to that from Temple Manor, Strood, set up and removed at about the same dates. Designs include drapery of fluent brushmanship and foliage, with some lines 'cotised', quite free of naturahsm and consistent with a reconstruction of the Hall around A.D. 1240. The characteristic trefoils with three circlets at the tip appear, e.g. in a window at Stanton Harcourt, Oxon., assigned by C. Woodforde to the first half of the thirteenth century.48 V. Bone artifacts (Fig. 11) 0 1. Offcut of antler (toggle?), possibly showing friction by a cord (Z, west slope under Great Kitchen). 0 2. Turned bodkin or 'stylus' with remains of iron pin in tip; spherical head and ornamental grooves (D, north side of Hall). These implements are fairly frequent on medieval sites, usually in earlier contexts than this one; their purpose must be domestic (sewing?) rather than writing on tablets.480, 0 3. Bone disc ornamented with trios of concentric circles, the inner circles faint; central hole and eccentric depression, suggesting attachment—a button or pommel rather than a gaming-piece (same context as 0 2). VI. Lead Offcuts of thin and thick (2 mm.) sheet lead from D. VII. Pottery This is classified under the main phases, as defined in the excavation section, and under three general categories: i, fully sheU-gritted coarse wares, quite distinct from any others, dominant in aU phases down to C yet totaUy extinct in D; n, unglazed sand-tempered wares, with many fabrics and intermediate varieties and a continuous gradation in shellcontent from fairly high (but distinct from category i) to nil: these wares form a very small, but increasing, proportion from W onwards, are stUl very much in a minority in A, approaching equahty in B, and absolutely predominant in D; in, glazed wares, a very small proportion, even in D. In category i each phase has its own characteristic rimforms, while the fabric modifies shghtly but graduaUy. After Y there is an overlap in rim-forms, smaU but sufficient to prove the completeness of the series. In W and X the material from Eynsford is insufficient to make a comprehensive series; that from Lullingstone enlarges the range but is not precisely enough stratified to point the development. 48 English Stained and Painted Glass (1954), PI. 3, r. 48a cf. Proc. Suffolk Inst. Arch., xxviii (1959) 145. 149 14 S. E. RIGOLD In category ii the early material is not enough to serialize forms and the general impression is that each separate fabric is remarkably conservative and simply gets commoner, common enough to serialize the later forms. Category iii shows a similar conservatism of fabric and includes a smaU proportion of continental imports, but in all periods and categories there are very few wares other than strictly local. Situations encoded as in excavation section. i. SheU-Gritted Wares W and X wares were at first classed together, but the stratification allows them to be separated, not always beyond dispute. In both phases, the cooking-pots are generally small and there are only two or three base-angle sherds to some twenty rims and necks. Lulhngstone, where the phases cannot be separated, shows the same proportion. Clearly the majority were globular. No thumb-strips, except possibly on a hand-made sherd. Phase W (Fig. 12). Three w^ares: (a) Shell fairly fine but irregular; fabric hard, compact and heavy; surfaces relatively smooth; body and surfaces dark red to black. WI, W2, everted rim with shght convexity and bead (both from aa, hearth under Hall). W3, upturned rim (A, primary humus, W101). W4, more upright rim, clubbed (S, basal gravel, W102). W5, similar (X context, solar undercroft). W6, shorter clubbed rim (context as W3 but in fi). (b) Crude, possibly hand-made, fine shell and chaUc, dark grey to brown. W7, stubby rim (context as W6). (c) Lighter weight, coarse sheU, approaching ware X c—wallsherds only, especially near the base of the curtain in fi, east; perhaps only in the second mounding and intramural gully. Phase X (Figs 12, 17). Again three wares: (a) Finely ground shell, generally dark grey, harder and thinner than W a. XI, everted, beaded rim (solar undercroft, demohtion of OT, X 103). X2, everted cablebeaded rim (in wall-mortar of solar undercroft), (b) Very thick, fine shell and chalk, buff-brown to dark grey; a smoother version of W b, surface almost 'soapy'; minimal rims (Fig. 1). X7 (X103, solar undercroft), X8 (X103, hall undercroft); another from fi west, under the Y chalk was possibly to be associated with the final W clay capping, (c) Much the commonest ware, with many wall sherds in X103 and the gully within the east curtain; very coarse sheU-grit, hghter, more friable, red, oxidized surfaces, sometimes quite bright in colour with conspicuous shell-flakes, the development of W c, and probably the ancestor of all the later shell-gritted wares. X3, 4, 5, 6, aU with simple everted rims, straight or nearly so—X6, a large lamp or small bowl. All from destruction of OT (X103) or footing-trench of cross-waU of HaU, except X3 (fi, east, under temporary clay hearth in gully— a Y context?). 150 .. I1'.., • * N- 00 fi OJ X j I OJ fo co X X IO nV !' co 1. CM O) I \ CD X in K CD CD S. E. RIGOLD Early Wares from Lullingstone (Fig. 12). The lighter fabric, as Xc, predominates and may include the transitions between Wc and Xc and between Xc and Y. Rims are: (a) curved, clubbed (nine examples), e.g. LI, L2; (b) curved with squared top, sometimes producing an inner bead (fifteen examples, two thumb-pressed), e.g. L3, L4, L5; (c) straight, everted and squared as Xc (about twenty, five thumb-pressed) including L6; (d) hand-made, uneven but hard, pale surface, not found at Eynsford and possibly very early, cf. L7, L8, L9 (lamp?). Phase Y (Fig. 13). Fabric derived from Xc but harder; shell still generaUy coarse and surface rough; hght Indian red to orange-buff, with grey, reduced core. Cooking-pots only, in many sizes, some now large, at least two-thirds with sagging bases and one or two with thumbed strips, necks and beveUed rim-forms of two varieties: (a) shghtly everted, with continuous gentle curve and hardly any shoulder; (b) well-everted, with well-developed shoulder, approaching Z form. Variety a: Yl (gravel east of Hall, Y104); Y3 (K, bottom of gully); Y4, hard (as Yl); Y5 (pit under porch-tower floor); another, in form and size like Y3 but with fine shell (filling of yard immediately east of Hall, under yeUow-brown capping, almost an X context). Variety b: Y2, very hard, thin, with coarse sheU (passage in front of HaU, displaced from footing of stair-block?); Y6, Y7, finer shell (A gully slope under chalk beneath Great Kitchen (Y 105)—a late Y deposit?); Y8 (as Y3). Also, Y9, clumsy neck (displaced in A); Y10, small (as Y6, Y7). Y forms are common at Lullingstone, e.g. L10, L l l; L12, with inner bead but very coarse sheU. Phase Z (Fig. 14). The sealed pit on section p, is taken as a sample. I t contained a few bowls and about seventy shell-filled cooking-pots, weighing over 15 kg. Cooking-pots again vary in size but get steadily larger; all seem to have sagging bases (Z2, Z3), but only four or five have thumbed strips; shell generally finer, surfaces generally duller buff-brown, core always grey and about ten pots completely reduced. The rim-bevel becomes a distinct hp, the neck shorter, the shoulder always pronounced, exemplified a tendency, widespread in the later twelfth century, to reduce the neck to a simple curve. Zl, Z4-Z11, sample of cooking-pots from the sealed deposit, among which the more forward-looking, with smoother interiors and fine brush-marks, inside and out, are Z7, with wide, down-turned lip, and Z10, Z l l , with slight inner bead. Bowls from same context, Z12, Z13, show brush-marks, upper bead or heavy clubbed rim. Z forms scattered elsewhere, sometimes with Y and A forms. Phase A (Fig. 15). Unless another source is indicated the sealed A deposit on section p, is used to represent this phase, but it may be a relatively early A assemblage, weighing about 5 kg. The shell is more finely pounded, the buff-to-red surfaces less patchy, the grey core more 152 EYNSFORD CASTLE CD >- >- OO >• (K j1 0) >- X_y' O >- CM ^ K 153 " CO II N .3 02 EYNSFORD CASTLE S. E. RIGOLD sharply defined, and the tendency to a 'soapy' surface, observed very rarely in Z, is fairly common in the sealed A deposit and more so in the widespread, and on average later (?), unsealed A deposits, most productive north of the HaU and between the Hall and Great Kitchen, which total quite another 10 kg. Greater variety of forms appear: the cookingpots are even larger, with a higher proportion of thumb-strips, and bases often thick and always sagging. The characteristic sign is the weU-marked inner bead, almost universal on A cooking-pots north of the Hall, but the sealed deposit shows four varieties of cooking-pot rim, all with weU rounded shoulders: (a) stiU somewhat triangular and reminiscent of Y, as A13; (b) developed from Z type, as A14; (c) typical, with strong inner bead, as A4, A5, A6, A7, A8, A9, A9, A10, All; (d) approaching B, A12, A18. Bowls of various sizes also have the bead, as Al, A3, and A2 (from south-west of Hall); A15 is small and completely reduced. 'Fish-dishes', oblong, not thrown, with thumbing, first appear, as A16, fully reduced. Lamps, as A17, not reduced. Also fragments of carinate rim, from a jug (north of HaU). In all, these wares probably cover the greater part of the first half of the thirteenth century. Phases BB, B and C (Fig. 16). BB, the fire-level, probably comes within Phase A, but the only reconstructible shell-filled vessel, a large bowl, BB1, has a B-like rim, without inner bead, yet an A-like fabric. The sealed B deposit, immediately north of the Hall, weighs quite another 15 kg. and is very consistent, but there is httle B pottery elsewhere. The shell is yet finer, the surfaces smooth, red, even and generally 'soapy', the grey core absolutely distinct, the fabric heavier, but the walls and bases often thin compared with the massive rims. About half the bases are now flat, as B l l , the rest sagging. The inner bead is found on only one or two pots in the sealed deposit; otherwise, it has disappeared and is replaced by two forms of rim, quite smooth internaUy: (a) thick, squared lip, as B5, B6, BIO; (b) broad flange, sometimes tapering or wavy, as Bl, B2, B3, B4, B8, B9 (unsealed); both tend to be down-turned. Bowls are few; B7, unsealed (north of HaU, with A wares). C contained about 1 kg. of sheUy sherds, identical with those from B, save that the form b rim, as BC1, prevailed. This was presumably the ultimate form; aU sheUy wares in D are derived and eroded; only one small bowl, DIO (Fig. 20), with a typicaUy D-style flange, was in a 'sandy-shelly' ware, not only relatively full of sheU but with the red-buff surfaces and grey core of the otherwise extinct fabric. ii. Sand-tempered and lightly sheU-fiUed wares The division between sand-tempered wares with Uttle or no sheUor chalk-gritting and those with a clearly deliberate admixture of sheU 156 w.a o m" « fl S. E. RIGOLD is hard to define at any period, owing to smoothed surfaces or dissolution of the shell. In the later deposits, from A onwards, the same general sub-varieties occur in the 'pure' sandy wares and the lightly shellfilled. A working distinction will be made between 'non-shelly' and 'semi-shelly' sand-tempered wares, within the dominant local range of fabrics, which are nearly always grey and reduced throughout and change so little that their classification must be on form rather than fabric. Those just outside the normal range will be isolated, as well as more obviously non-local wares, never very numerous. There is, however, a small series of non-local coarse wares with sand and shell temper, called for convenience, 'sandy-shelly'. The late instance DIO (Fig. 20) has been mentioned, but comparable fabrics, almost always oxidized where the local wares are reduced, occur in Z, as Z14, with a peculiar rim (sealed deposit, section p), and in sealed A, on same section, and elsewhere. The 'Main Series' of Sand-tempered and partially Shell-gritted Wares. This forms the other great mass of pottery, beside the fully shellgritted, and grows from a small proportion to become the overwhelming majority in D. The fabric is nearly always fully reduced and in some shade of grey, and can be divided, though not absolutely, into 'ST a', thicker, coarser and lighter grey, but always of relatively fine texture, and 'ST 6' thinner, finer, tending to dark grey or black and sometimes almost burnished; with these correspond lightly shell-fiUed versions ('shelly-sandy') 'SS a' and 'SS 6'. Where possible all wares not fully shell-tempered will be referred to these classes and only the totally unamenable treated separately. The fabric is generally akin to the east Surrey fabrics, of which the typical kiln-site is Limpsfield, 20 km. west of Eynsford, but it is not identical with any contemporary find from Limpsfield, nor with those from the manorial site at Netherne, near Coulsdon, 25 km. away 'as a crow flies', of which the excavator, Miss L. Ketteringham, remarks that the wares are generally less sharply finished than those from Eynsford. The exact parallels are from nearer sites, Joyden's Wood, near Bexley,49 a site in Dartford,50 and even Temple Manor, Strood,51 20 km. to the east. The source is therefore probably not in Surrey but in north-west Kent. The exceptional wares will be treated first. Various early Sand-tempered Wares. Represented by a few sherds only and generally not reconstructible: (a) Very coarse sand or flint, grey or pink surfaces (j8 west, W101): a type generally called Iron-Age, but occurring on medieval sites (Netherne, Pachesham, Surrey) as residues (?).52 10 Arch. Oant., Ixxii (1958), 18-40. 50 From High St.; examined and to be published by D. C. Mynard. 61 Arch. Journ., cxxii (1966), 128-30. 52 Information from Miss Ketteringham. 158 EYNSFORD CASTLE (b) A true medieval sandy ware, rather soft, with coarsish silvery sand and some chalk, grey core, pink to crimson surfaces; includes a joined-on sagging-base-angle (fi west, WlOl, and in the undercrofts in W, rather than X, contexts). (c) Fine, very thin, brown to grey surfaces, with rouletted ornament: X9 (Fig. 17) (Hall undercroft, under primary floor). (d) Medium to dark grey with chestnut overtones, a httle shell: bowls (Fig. 17), Yll, Y12, cruder, thumbed, and a jug (A, all on slope beneath Great Kitchen-Y105). May be from Medway area—similar to later wares from Temple Manor, Strood. (e) Hard, even, light grey; probably East Anglian, in 'Thetford' tradition; waU-sherds from Y, and Z16 (Fig. 17), cooking-pot with sharply recurved rim (p,, sealed). Sand-tempered Wares from London Area (?). Perhaps related to 'London' glazed wares. Two fabrics, the earlier 'ST c', even sandy texture, dark, almost black and burnished exterior, pink-buff to red lining, occurs in Z and A; the later, ST d, with grey core and smooth pink surfaces inside and out, almost in a shp, in A and B. All are jugs. Z19, with rich red lining and strip ornament, and Z23 (both p, sealed) represent ST c (Fig. 19). B30, B31, B32, with various styles of handle (all sealed B, behind Hall), represent ST d, which might possibly originate in east Kent, though the colour is not hke any known Tyler Hill types (Fig. 23). East Kent Wares (?). Very httle: a few sherds in A, BB and a little more in D seem to come from oxidized Tyler Hill, or similar, jugs (STe). Wares intermediate between ST c and Main Series. Grey wares, a shade coarser than usual, with black sand showing up in a generally bufhsh lining—possibly true Surrey wares, but more likely the predecessors of ST a; called ST /. Z17, Z18, collared jugs (Fig. 17), Z24 neatly stabbed handle (all p., sealed) (Fig. 19). Dark Grey Wares with striated Surfaces. SS g, dark grey, slightly shell-gritted wares, resembling SS 6, but thin for their relatively rough surfaces, which are marked by very regular latitudinal striations or brushings. This is a well known London-Middlesex feature—it may be a very refined form of scratch-marking. Not numerous, but including two practically complete vessels: A20 (Fig. 17), smaU cooking-pot, and B12 (Fig. 18), large cooking-pot with thumb-strips and advanced rim profile, approaching those normal in D. Both well sealed. Very hard Grey Wares. Perhaps just over-fired examples of ST a or ST 6, but more probably intentional and non-local. Jug-sherds of a fused, ringing quahty, core and interior grey, exterior often mauvish. Not numerous, but found in A, B, BB (not the result of secondary firing). 159 C E I CM CU CQ 5 CD 5 i— CQ CO T - CQ CQ O CM CQ CM CQ CO CM GQ rk r\ $ ! I EYNSFORD CASTLE IV fl S. E. RIGOLD Main Series as far as D. Very httle in Z, not treated under other headings, but Z20, a jug of true ST a fabric and Z21, a thumbed handle of ST 6, were sealed on section p (both Fig. 19). The material comes from A, including the sealed A midden on section p,, BB and B. There is no clear evolution in these phases; it is simply that before B there is little but jugs, but B contains a good proportion of cooldng-pots. These are almost all in hghtly shell-filled sand-tempered ware: consequently, the proportion of SS a and SS 6 to ST as and ST 6 rises as the fully shelled wares diminish. The jug and bowl rims are of many shapes but in A-B characteristically with a bevel or rounding at the upper angle, while those in D are characteristically flat-flanged, with all recurvature on the underside. There are a few exceptions. The skillet with hoUowhandle, B14, with typically A-B rim-profile may possibly come from D as another hke it certainly did. Other differences between the A-B varieties of the Main Series and those from D are in the ornament on some of the jugs: neat scorings, trelissed, as on A21 (Fig. 18), a tubespouted pitcher, in ST 6 fabric with dark exterior and buff-grey core, or on A30 (Fig. 18), similar, or parallel, as Bl, typical of many fragmentary jugs in paler, ST a fabric (all sealed) are broader and generally deeper than the more incised ornament on D vessels. Cooking-pots in SS a or SS 6 fabric (B15, B16, B17, B18, B19, B20, B21 (Fig. 17) (all from sealed B) and BC2 (Fig. 18) (from the red clay over them) show a wide variety of rim. B23 (Fig. 17) has a more archaic form, with thumb-strips, and B29 (Fig. 18) is quite exceptional. B22 (Fig. 17) is smaU, like A20 and in fabric nearer SS g. Bowls are similar but few. Jugs in ST a fabric have collared necks and various combinations of thumbing; slashing and coarse stabbing on the strap-handles as A22, A23, A24, A25 (Fig. 19), BB3 and a series from B, B24, B26, B27, B28 (all Fig. 19), are similar in form but in ST 6 fabric—a point which confirms that the fire was nearer in date to the sealed B layer than to the sealed A deposit on p,. Rope-decorated handles, a motif of twelfth-century origin, as A26, occur in ST 6 fabric in A and B, but the ST a jug, B25, with fine corrugations on the rim, anticipates D forms. Fish-dishes in SS fabrics continue with little change, as B33. Main Series in D. Apart from glazed wares and a few, apparently, Tyler Hill sherds, the D pottery is almost entirely ST a and ST 6, or SS a and SS 6. The grey colour is less universal, but buffer vessels are simply less reduced examples of these. Attempts have been made to subdivide the fabrics further and match these with variations in form, but the whole corpus seems to be a continuum with much variation and there is httle point in noting exact resemblances with vessels from Joyden's Wood53 or Temple Manor54 among those selected for illustra- 53 See note 49. " See note 51. 162 s CQ CO S. E. RIGOLD tion from a vast mass. Handles of at least seventy jugs have been noted and a sample weighing about 3 kg. produced almost exactly 50 per cent each of pure sand-tempered and of lightly sheU-filled; the proportion of the latter would have been higher if jugs had not been unusuaUy numerous, relatively few of these containing sheU. The characteristic flat flange without any upper bevel, a fashion seen around 1300 in pottery over a very wide area, is most marked on the bowls, but is generally seen on cooking-pots and on the large storagejars that now appear, and may well have been taller than as reconstructed. Cooking-pots (Fig. 20) vary in size, but many (eating-vessels?) are now smaUer again. They are generaUy without pricking or ornament and the rims fall into three classes, without respect to size (Fig. 20): (a) triangular, as D4, D5, D6, D8, D9; (b) level and flat-flanged, D7, D l l , D13, D14 and the transitional shouldered bowl, D10; (c) flat and down-turned, as DI, D12, D15, D16, D17. D3 is of the old, rolledover form. Most contain shell but its absence seems insignificant. Storage-jars (Fig. 20), in SS fabric, are more ornate, with wide, stabbed rims, rilhng and thumb-strips, as D18, D19. Bowls (Fig. 21) are very numerous, generally, but not always contain shell, and vary much in size and in the profile and surfaceornament of their flanges. Many (as D25, 26, 31, 32, 33, 35) have inner beads. Few rims are stabbed, but the decoration includes combing (D26, D34), wavy lines (D21, D30), slashing (D25, D29) and thumbing (D22, D23). Jugs (Figs. 22, 23 and 24, in part) are more often of bulbous (as D37, D39) than of baluster (as D40) form and often surprisingly thin-waUed. The finest are in dark grey fabric, as D37, D38, in ST 6, or D40, in SS 6. Wall-ornament is usually by dehcate combing in lattice-patterns, or more often, wavy hnes, as D37. Rims, handles and ornaments of the base-angle have been classified, but show all possible combinations. Rims may be: (a) plain, with upper edge sloped inwards, as D41; (b) triangular, generally with inner bead, as D39, D43; (c) of degenerate 'collared' form, as D42; (d) carinate and combed, as D37; (e) flanged, as D36. Handles nearly all have fine stabbing and, at most one central slash as D47, but this is rare at Eynsford (surprisingly it occurs in BB-BB4), commoner in the group from Dartford referred to above (? shghtly earher, or shghtly later). Handles may be of rod, oval (perhaps the commonest), or broad strap section. Two large but unreconstructible jugs in ST 6 fabric are 'cisterns', with spigot-holes (D67). Base-angles may be plain, pressed, as D51, pinched, as D50, or slashed, as D52. Fish-dishes, D48, D49, continue much as before. SkUlets include hoUow-handled vessels as B14 (which may have a D origin) and flat-handled ones as D20. 164 r yzszn X ao 8 PH 6 rM S. E. RIGOLD iii. Glazed Wares Glazes first appear in Z, as Z25, which is not early by standards north of the Thames, and essentially the same fabrics continue right down to D. They are never numerous and never enough to show anything like the organic evolution of the coarse wares, if, indeed they had any, being too self-conscious and too sure of repeating a good selling-line until it became ddmodd. None are strictly local; they are rare, probably, because the local potteries had no glazes to offer. However, B and D contain a fair selection. They are here classified under their probable sources. (a) London Area (Figs. 23, 24). Four or more sub-fabrics have been noted, but they are hard to define apart from their decoration and the wares will be treated as a continuum of sandy fabric from Z to D, with reduced core, generally orange-brown surface and lining, gradually getting harder, smoother and more brick-red. They include: i, plain mottled green jugs, as A28 and many fragments from B; ii, moulded vessels with more even ohve-to-green glaze, but rather roughly made, as B34, with rib-decoration (sealed B); iii, imitations of Nottingham or Yorkshire (?) and other fancy forms, as A27, with rich green glaze and bridge-spout, B38, thin, mottled glaze, rougher fabric, with relief of snake, and B39, with deep green glaze, trellis-ribs and 'hotcross- buns', and, in D, fragments of an elaborate, northern-style piece with crude figures (not a knight-jug); iv, imitations of Rouen jugs, with ornament of triangles, strips and pellets, in cream and brown, with red lining, or, in one case, brown-purple and cream with grey lining, as B35, B36, B37, from sealed B, but also fragments from late A, north of Hall. The archetypes go back to the 1240s, but not provenly any further; v, typical London balusters, apart from the collared base of a small one from B, not earher than D, but the most numerous form there, some with lightly thumbed bases—D54 is composite of eighteen vessels; D56, 57, 58, 59, show various ways of decorating the whiteslipped surface, with combing, rouletting or scales; D55 shows the typical rod-handle, thumb-pressed; D60, D62, less normal balusters of smoother fabric. Several fragments from D have brown-purple glaze with cream 'icing'. (b) Mid-Surrey (Figs. 23, 24). The next commonest fabric; whitish, rather coarse sand, but often thin, with deep and bright green glaze. Nearly all from D, but something like it appears in A, such as the stabbed and grooved jug-handle, A29. Balusters include striated and scaled vessels (D65). One bowl, D53, of a type common at Netherne, has patches of glaze. A jug sherd has green and yellow polyohrome decoration. (c) West Surrey (Farnham Area?) (Fig. 24). Coarse, pink-buff sandy ware with thin olive glaze, the grains showing through it. Occurs in 168 n / s F I G . 23. Earlier Glazed Wares, etc. (£). D48 Vi I / V )} zzz-il:k^Mm FIG. 24. Phase D, Glazed Wares, etc. (£). EYNSFORD CASTLE B and D, as the rod-handled jug D66. Also a small Surrey bottle, of fine fabric, D61 (D), as at Netherne. (d) North France (Rouen, etc.). Imitations, but only a few abraded scraps from B that seem to be of this fabric, common on Enghsh port-sites. (e) West France (Saintonge) (Fig. 24). All from D. Polychrome: ten sherds from at least three jugs, more baluster-shaped than usual, as D64. Fine white ware with decoration in purple, green (including part of a bird?), and yellow (including part of a shield), and plain strap handle. This, apart from the sum-total of coarse ware rim-profiles, is the most diagnostic element in D.65 Monochrome: similar ware, two wall-sherds, with mottled green glaze. (f) Low Countries (Andenne Type). One sherd from A; white ware with grooves and thick yellow glaze.56 (g) Low Countries (Holland or Flanders). Top of small, brick-red baluster D63. iv. Phase K A fair quantity in all, but nothing reconstructible. It includes large, cyhndrical stoneware beer-mugs, usually pale, treacly-glazed coarsewares, perhaps Wealden, a httle white earthenware, but nothing approaching porcelain, and black Basaltes, fluted, which sometimes occurs in plebeian contexts. The description of pottery rehes heavUy on the careful analysis of D. C. Mynard and the advice of J. G. Hurst and Dr. G. C. Dunning. VIII. Animal Bones Meat-bones have not been examined, but bones of small mammals were found sealed and in perfect condition in the post-fire floor-raising in the porch-tower and examined by W. G. Teagle, F.Z.S. They included the lower jaw and part of skuU of a water-shrew (Neomys fodiens) and jaws of a long-tailed field-mouse (Apodemus sylvaticus). 66 Now well known in contexts of c. 1300; of. Arch. Journ., cxx (1963), 201-14 and, for Kentish examples, Arch. Oant., lxiv (1951), 147, fig. 2- lxix (1955^' fig. 3,1; andlxxxv(1970), 110,ng.7. v '' 50 R. Borrennes and W. Waigninaire, La Ce'ramiqued'Andenne (1966), passim. 171