Rochester Castle, 1976

ROCHESTER CASTLE, 1976 COLIN FLIGHT, M.A. AND A. C. HARRISON, B.A., F.S.A. The structural history of Rochester Castle (Fig. 1) was first worked out by Canon G. M. Livett in 1895, 1 and his account has been followed in general by subsequent writers. For the most part he seems to have described the evidence accurately and interpreted it correctly, in the light of the information available from documentary evidence;2 but there are a few points which need to be reconsidered. One problem which Livett failed to solve concerns the site of the earliest castle at Rochester. A solution for this problem occurred to us as a result of observations made during the 1960s and has now been confirmed by excavation. The main purpose of this article, therefore, is to present the evidence in favour of our solution. We also correct a number of other mistakes into which Livett was led by his failure to identify the site of the earliest castle. The earliest reference to a castle in Rochester occurs in 1086, when it was recorded in Doomsday that the Bishop of Rochester had been given land in Aylesford 'in exchange for the land on which the castle stands'. 3 A little later, during the rebellion of 1088, a castle at Rochester was held against William II by the supporters of his brother, Duke Robert, but besieged and taken. At about the same time, though the sequence of events is not entirely clear, Bishop Gundulf agreed to fortify the castle 'for the king in stone at his own expense', in return for 1 Arch. Cant. xxi (1895), 21-38. The keep had already been described in detail by G. T. Clark, Arch. Journ .• xiotii (187S), 210-23. Some further information on both the keep and the outer defences was supplied by G. Payne, Arch. Cant., xxvii (1905), 177-92. 2 For the documentary evidence Livett relied on a paper by C. H. Hartshorne, Arch. Journ .• xx (1863), 205-23. This is now superseded by the contribution from Professor R. A. Brown, in R. A. Brown, H. M. Colvin and A. J. Taylor (eds.), The History of the King's Works, ii (London, 1963), 806-14. Also by Professor Brown is the official guidebook, Rochester Castle, (London, 1969). 3 V.C.H., Kent, iii (London; 1932), 209. 27 C. FLIGHT AND A. C. HARRISON Boley Hill + Spot heights in feet OD. cf Fig. I. Rochester Castle and Boley Hill, based on the Plan by Livett (Arch. Cant., xxi (1895). (For key to hatchings, see Fig. 2.) 28 ROCHESTER CASTLE, 1976 the confirmation of a grant of land in Buckinghamshire. 4 The wording here clearly implies the existence of an earlier castle not constructed in masonry. Livett's problem was that he could not find archaeological evidence for any fortification of the castle site earlier than the parts of the curtain-wall which he identified-no doubt correctly-as the work of Bishop Gundulf. In an attempt to resolve this apparent contradiction between the archaeological and documentary evidence, he suggested that the 'castellum' mentioned in accounts of the rebellion of I 088 was not a castle in the medieval sense at all, but rather the city itself, surrounded by its Roman wall. 5 Understandably dissatisfied with this solution but still faced with the same difficulty, Canon S. W. Wheatley put forward the idea that the earliest castle at Rochester was on a different site from GunduJrs-on Boley Hill, to the south of the existing castle and outside the line of the Roman wall, where a mutilated earthwork still survives. Formerly thought to be a Danish siege-castle of the ninth century,6 Boley Hill was now identified as an eleventh-century 'motte and bailey'; and this identification,7 in spite of its inherent improbability, has been accepted faute de mieux by subsequent writers. No attempt has been made to confinn the hypothesis, except for an inconclusive excavation in 1960.8 It can be fairly said that the only argument in favour of an eleventhcentury date for Boley Hill has been the alleged absence of preGundulfian defences on the castle site itself: but even this argument can now be shown to be wrong. In fact there is ample evidence to prove that the castle site was already surrounded by large-scale earthworks before the Gundulfian wall was built. Some of this evidence was not accessible in Livett's time, some of it he missed and some of it he misinterpreted-though he came close to the right answer when he noted that the north curtain-wall was built upon the gravel upcast from the ditch.9 4 'Textus Roffensis (ed. Hearne, 1720), 145-8. The exact date of the agreement is unknown, except that it must fall between the accession of William Il in November I 087 and the death of Archbishop Lanfranc in May 1089. The siege of 1088 took place in May, according to Ordericus Vitalis: King's Works, i, 28-9. 5 Arch. Cant., xxi (1895), 22. 6This was the view taken by local antiquaries in the eighteenth century and also by Livett, Arch. Cant., xxi (1895), 30. 7 S. W. Wheatley, Arch. Cant., xxxix (1927), 159-64; xii (1929), 127-41. 8 A. J. F. Dulley, Arch. Cant., lxxiv (1960), 197-8. 9 Arch. Cant., xxi (1895), 28. 29 70 30 ½so l20 60 C. FLIGHT AND A. C. HARRISON 1 ashlar r-----., ____ ..J 20 metres &,,ashlar 􀀄►-. chalk ? ... ....-,(gravel) 􀂋 SO feet 2 • Roman l!!l!!lll Gundulfian D 12th-century 􀂌 13th-century 􀂍 14th-century C modern ?ldipitocfh 4 wroidaedn ed 20 feet 0.0. Fig. 2. Rochester Castle: Profiles through the Defences. THE PRIMARY EARTHWORKS The primary earthworks can be identified on all sides of the castle except on the west, where the Gundulfi.an wall is built partly up against and partly on top of the Roman town wall. Profile 1 (Fig. 2), based on 30 ROCHESTER CASTLE, 1976 a section drawn by Livett, 10 shows both these walls, and also the wall of a thirteenth-century building abutting them. Of the Roman wall, best viewed from the Esplanade, three courses of the ashlar facing survive. Livett was told by the clerk of the works in charge of the underpinning carried out in 1870-2 that these three courses extended inwards beneath the wall, forming its foundation ; 1 1 but this is doubtful, particularly as his suggestion of a similar foundation at the south-eastern angle of the Roman town 12 has been disproved by a recent excavation. 13 Further south the Roman and medieval walls part company. Profile 2 shows the Gundulfian wall built in a shallow trench on top of a bank of chalk rubble. This bank has been cut back externally into a nearly vertical face about 15 ft. high, perhaps when the ditch was filled in and the area landscaped during the eighteenth century. Livett mistook the chalk for 'solid rock'. 14 Until 1970, when the whole slope was faced with stone, it was in fact quite evident that the wall stood on made ground composed of chalk rubble, with a few broken flints embedded in it. In any case, the chalk cannot possibly be bedrock, because the Roman wall runs inside the Gundulfian wall, at a much lower level. George Payne found that the stones forming the edge of a modern path here had been 'bonded into' the top of the Roman core. 15 There is thus no doubt that a bank of chalk rubble was piled up against the outer face of the Roman wall, burying it completely, and that later the Gundulfian wall was built along the crest of this bank. These facts suggest very strongly that the chalk bank does not belong to the same structural phase as the Gundulfian curtain-wall. Clearly whoever threw up the bank was not intending to build a wall as well; and clearly the bank had had time to consolidate before the Gundulfian wall was built upon it. What looks like a relieving arch can be seen on the inner face of the Gundulfian wall at about the point where this breaks away from the outer face of the Roman wall and climbs up onto the chalk bank. 16 This seems to illustrate the problems posed for Gundulf by the existence of the primary earthworks. 10 Arch. Cant., xxi (I 895), Pl. I, No. 7. Livett's figure of 38 ft. for the difference in level between the Castle Gardens and the Esplanade is incorrect. 11 Arch. Cant., xxi (1895), 27. 12 Arch. Cant., xxi (1895), 27, note. 13 A. C. Harrison and C. Flight, Arch. Cant., lxxxiii (1968), 59, Fig. 4. 14 Arch. Cant., xxi (1895), 25. 15 Arch. Cant., xxvii (1905), 189. 16 This feature was not noticed by Livett. It is mentioned by Brown, Rochester Castle, 27, but not identified as a relieving arch. 31 C. FLIGHT AND A. C. HARRISON On the opposite northern side of the castle the curtain-wall can again be shown to follow the line of an earlier earthwork. The bank here is of gravel, not chalk, a fact easily explained by reference to the topography of the site. Like Boley Hill, the south side of the castle occupies a low ridge of chalk, but the :northern part, like most of the rest of the ancient city, was built on brick earth and gravel. 17 Of the north curtain-wall all that survives is a stretch of the Gundulfian wall, much dilapidated and pierced through at intervals by doorways giving access to the gardens behind High Street. In the garden of 34-36 High Street, where Profile 3 was drawn, the side of the ditch has been dug away and the ditch itself filled in. Subsidence affecting an outhouse behind the shop may possibly mark the position of its outer lip, about 82 ft. from the wall. Gravel underlying the Gundulfian wall, though not exposed in this garden, was observed by Livett further to the west. 18 The bank appears to have been about 15 ft. high. The east curtain is known from documentary evidence to have been rebuilt almost entirely during the fourteenth century, between 1367 and 1370. The side of the ditch has been dug away along much of its length, so that the foundations are exposed to a depth of about 10 ft. in the little graveyard opposite the cathedral, between towers 2 and 3, and of as much as 15½ ft. north of tower 3. They consist of massive wedgeshaped piers of masonry linked together by crude pointed arches 19 - exactly the same construction that was used in the east wall of the city, where this was carried across a filled-in ditch. 20 Until they were blocked with stone some years ago the four arches south of tower 3 could be seen to be filled with roughly horizontal layers of sand and gravel. The four arches north of tower 3 have been emptied out, but patches of gravel can be observed adhering to the exposed surfaces. It is clear, therefore, that the fourteenth-century foundations were sunk into a bank of gravel, as shown in Profile 4. A recent excavation has provided valuable evidence confirming this interpretation (see below). TOWER 2 Unlike tower 3, which is all of one build, tower 2 is set upon an earlier foundation. This feature needs to be described in some detail, partly 17 The deepest and most informative section geologically is that recorded in the Deanery Garden: Arch. Cant., lxxxiii (1968), 66, Fig. 9. 18 Arch. Cant., xxi (1895), 27-8, Pl. II. 19 Arch. Cant., xxi (1895), 35-6, Pl. V. Livett's interpretation is utterly unconvincing. See also A. A. Arnold, Arch. Cant., xviii (1889), 197-8, Pls. I-II. io Arch. Cant., lxxxiij (1968), 67-8, Pl. IIA. 32 ROCHESTER CASTLE, 1976 wall -==of inner --bailey- ... ---􀀐-- 􀀑 .... .... Fig. 3. Rochester Castle: Conjectural Diagram showing the south-east Corner of the Castle before the fourteenth-century Reconstruction. because Livett's account is grossly inaccurate,21 and partly because it has now been obscured by underpinning. The north face was exposed to a depth of about 7 ft., when the side of the ditch was dug away; some years ago the east face was also cleared to a depth of 2 to 2½ ft. The foundation does not consist of gravel, as stated by Livett, but of mortared rubble. In the east face is a crack, which starts near the northeast corner and passes obliquely downwards to the south; that part of the foundation lying below the crack has shifted outwards by up to 6 in., so as to produce a jagged, slanting offset. Tower 2 itself is not affected, and therefore has to be later. On the north face is an overhanging footing, thought by Livett to be secondary, but in fact of one build with the rest of the foundation. The north face returns into a short projection on the line of the curtain-wall, proving that the original tower was inse􀁥ted into a still earlier wall. This projection is separated from the fourteenth-century foundations by a narrow vertical gap, filled with gravel. For reasons which are explained most easily by means of a diagram (Fig. 3), it seems likely that this 21 Arch Cant., xxi (1895), 29. 33 C. FLIGHT AND A. C. HARRISON foundation is of twelfth-century date, 22 and that the curtain-wall to the south of it was re-aligned in the thirteenth century before being rebuilt in the fourteenth. Before the foundation was faced with modern masonry, patches of gravel could be seen adhering to the exposed surfaces. It was even possible to distinguish tip lines sloping downwards to the north and west. In a streak of dark earth sandwiched between layers of gravel was found a single sherd of medieval pottery, thin and black, shelltempered, and with a simply beaded rim. THE EXCAVATION In the autumn of 1976 permission was given by Medway Borough Council for the excavation, in advance of landscaping the area, of a narrow strip of ground lying to the east of the curtain-wall and north of tower 3. This strip was only 17 ft. in width, owing to a modern disturbance to the east and effectively about 40 ft. long as further to the north and ancient levels had been increasingly disturbed by the foundations of the recently demolished 'Castle Hall', destruction being complete after 55 ft. (Fig. 4). In section A-B, Fig. 4, it was possible to extend the trench for a distance of 7 ft. into the arch beneath the fourteenth-century wall. It should perhaps be noted that this vital area was only partly excavated, more than half of it being left for any future investigation that may be thought necessary. Stratification Section A-B (Fig. 5, Plates I and II). Diirectly under the modern floor of concrete a layer of the reddish gravel, which, as mentioned above, originally filled the whole arch, survived undisturbed for a depth of 2 ft. 3 in. with a rather more stony layer at its eastern end, which perhaps represents the remains of a marking-out bank. From the gravel were recovered scraps of Romano-British pottery and a coin of King Alfred (d. 899), which, though of great interest on itself (p. 55), must be regarded as a 'stray'. Beneath the gravel was a layer of fine black soil varying in depth from I ft. to 6 in. and containing pottery not necessarily later than the eleventh century (p. 41 ). Below this was 22 Square mural towers are typical of the twelfth century, for example on Henry II's inner curtain-wall at Dover built c. 1180. Wltile tower 2 might be as early as William de Corbeil's work of c. 1127 it is equally possible tltat it is connected with the repairs and strengthening of the defences carried out in 1166-1170, 1173-4 or 1191-1202 (see D. Renn, Norman Castles in Britain, 303, quoting the Pipe Rolls for those periods). After 1215 round towers had become fashionable, as in the south-eastern bastion and the rebuilt turret of the Keep c. 1221. 34 A S T L E 1p........, .., ,....... .., ..,_?1 ______ _,1p H 0 ,,.. Fig. 4. Rochester Castle: Site Plan. C. FLIGHT AND A. C. HARRISON A 29·5 FT. AOD 0 JOFT L-..􀀝--1-...L--.J...JL......L---'--'---'---' ,------􀀌-------- 0 3M C ? I l'j> fT 0 M D -!:f􀀆t;I · .;􀀚 • bosc of pier •:' projected B Pig. 5. (Upper) Section A-B. 1. Modern Concrete; 2. Modern Disturbance; 3. Reddish Gravel; 4. Soil with Flints; 5. Black Earth with Flints; 6. Brown Earth with Flints; 7. Black Soil; 8. Soil containing numerous late-Roman Coins; 9. Dirty Brickearth; 10. Pit; 11. Gully; 12. Fallen Plaster; 13. Burnt Clay; 14. Debris of Oven; 15. Wall Trench; 16. Orange Gravel (Natural Subsoil). (Lower) Section C-D. I. Modern Disturbance; 2. Dirty Gravel; 3. Filling of Pit; 4. Flint Cobbles; 5. Dirty Brickearth and Flints; 6. Brown Earth filling wide, flat-bottomed medieval Trench. a layer of black soil of a similar depth but of a denser and stickier consistency and with many fragments of Roman tile and flints. This yielded a considerable quantity of late Romano-British pottery and numerous late Roman coins (p. 44). At the eastern end of the section an oven or kiln had been cut into the gravel subsoil and this, in its turn, had been cut by the trench of a Roman wall which extended diagonally across the site. This wall had been rendered on its western side and a 36 ROCHESTER CASTLE, 1976 thin layer of plaster debris lay across the remains of the oven or kiln, which, like the others subsequently found further south, had been truncated when the site was levelled for the construction of the wall. A fairly well worn coin of Allectus in the fill of oven A suggests that this may have been early in the fourth century. The earliest feature was the shallow gully which extended N-S across the site and contained pottery not later than c. A.D. 200. The stratification over the rest of the site was broadly similar though the upper levels were more disturbed. It has, however, been thought worth while to give the most southerly section C-D (Fig. 5) to show a medieval feature tentatively identified as the bottom of a siege-trench or tunnel. This feature had been cut through the Roman levels and into the sub-soil and consisted of a flat-bottomed trench approximately 4 ft. in width approaching the castle wall at an oblique angle. There had been no silting and the trench had been filled immediately with brown earth containing pottery identifiable as being late-twelfth or early thirteenth-century in date. The Ovens. Remains of the six ovens or kilns, varying in size and shape (Fig. 4, and Plate III) were found. These were of a very simple construction consisting of holes dug into the sub-soil and lined with clay with slight traces of a flue-trench in two cases. The best preserved ( oven F) had been reinforced with lumps of chalk (Plate IV). No wasters or slag suggestive of potting or smelting were found and this fact, together with their small size, suggests that they were for domestic rather than industrial use, though it is curious that they should be so numerous. The Roman Wall. This was well constructed of layers of coursed flints 2 ft. thick and survived to a height of 1 ft. 3 in. (Plate V). It was traced for 55 ft. in the direction of the High Street to which it was approximately at right-angles. At the other end had been a T-junction and in the arm of this nearer to the castle wall was a well defined slot for a wooden cill (Plate VI). This wall enclosed an area of black humusrich soil containing late Romano-British pottery and some 750 coins ranging from Claudius II to Honorius, which are discussed below (p. 53). This high concentration of low denomination coins suggests chance losses in a locality where they were in constant use and makes it likely that this wall formed part of the site of a market or was closely adjacent to it.23 It is noteworthy also that activity evidently continued on the site right up to the end of the Roman period. 23 The concentration of fourth-century coins in the area of the theatre at Verulamium is a possible parallel. 37 C. FLIGHT AND A. C. HARRISON Medieval Pit. In addition to pottery of the late-eleventh or early twelfth-century (p. 43) this contained five complete loom-weights and several fragments. These are in every way similar to those found at East Gate in I 969, 24 associated with pottery of the same date. 'Siege-trench.' This trench, 4 ft. wide and with steeply sloping sides was traced for a distance of 18 ft. While its identification as a siegetrench can only be tentative, it should be noted that its bottom is approximately 20 ft. below the base of the fourteenth-century wall and 4 ft. below the bottom of the piers that fonned its foundations and must therefore have been well situated to undermine that wall's predecessor. Of the two thirteenth-century sieges the pottery would suggest King John's in 1215 rather than Simon de Montfort's in 1264. CONCLUSION The evidence of this paper suffices to show that the castle site is surrounded on all sides except towards the river by earthworks of considerable size. In addition, these earthworks are demonstrably earlier than the curtain-wall which is agreed to be Gundulfian; and Norman pottery has been recovered from a sealed context beneath the rampart. We therefore propose to identify the primary earthworks as the pre-Gundulfian castle which Livett failed to find, and which Wheatley sought to locate on a site outside the Roman wall. We suggest that the earliest castle at Rochester was built-probably soon after 1066-in the south-west corner of the Roman town. On the landward sides it was defended by massive earthworks, the rampart being composed of chalk rubble on the south and of gravel on the east and north. The depth of the ditch has only been recorded once: a cesspit dug through the filling near tower 1 reached chalk at a depth of 20 ft.25 On the west, overlooking the river, the Roman wall was incorporated into the defences of the castle. In digging the ditch, on the other hand, two stretches of the Roman wall would have had to be demolished. There are slight indications of a motte on the site of the existing keep; but this corner of the castle would seem to have been altered considerably when the keep was built, and again during the thirteenth century, so that the original plan is difficult to reconstruct. Bishop Gundulf, in or after 1088, built a curtain-wall of stone to replace the original stockade. The keep was built by Archbishop William de Corbeil after 1127, and the foundation which underlies tower 2 may also be of twelfth-century date. After the siege of 1215, repairs and alterations to the keep and curtain-wall were carried out 24 Arch. Cant., lxxxvii (1972), Plate I and Fig. 20, No. 12 and pp. 155-6. 25 G. Payne, Arch. Cant., xxix (1911), p. lxxxiv. 38 ROCHESTER CASTLE, 1976 during the first twenty years of the reign of Henry III. Since Boley Hill can no longer be dated to the eleventh century, some other interpretation will have to be found; and, perhaps, it may be identified as King John's siege-works or as a thirteenth-century out-work designed to reinforce the defences on the side which had been proved vulnerable in 1215. If the latter, it has a counterpart at Dover Castle in the earthworks surrounding the church and pharos. A thorough reconstruction of the east curtain was undertaken by Prior John of Hartlip, on behalf of Edward III, between 1367 and 1370. His clerk of works was clearly apprehensive about building a wall on top of a bank of gravel-and the cracked condition of the foundation beneath tower 2 suggests that he was right to feel uneasy. 26 He therefore decided to sink a series of massive stone piers linked by arches through the gravel until he reached solid ground. Later, when the front of the bank was dug away and thrown back into the ditch, this produced the appearance of gravel-filled arches which baffled even the perspicacious eye of Canon Livett. If this reconstruction of events can be accepted, a number of difficulties disappear, for example those connected with the construction of the Norman curtain-wall and of the fourteenth-century east curtain, but, above all, it obviates the improbable expedient of placing the earliest castle upon the Boley Hill site outside the Roman walls in a manner entirely contrary to N onnan practice. 2 7 'Entities are not to be multiplied without necessity', said William of Occam, and this paper has endeavoured to show that his principle is also valid for the sites of castles. THE FINDS I. POTTERY (i) Romano-British28 In view of the abundant dating evidence from coins it was not thought necessary to illustrate any of the Romano-British pottery. From the main layer of black soil {Fig. 5, layer 8) besides much residual earlier material there were late colour-coated wares, stamped wares, 26 The necessity of rebuilding Gundulfs curtain-wall in the fourteenth-century was no doubt due to the settlement of the gravel bank over a Jong period with slipping of the bank material into the moat. 27 E.g. London, Canterbury, Colchester, Portchester, Pevensey etc. 28 Mr. A. P. Detsicas, M.A., F.S.A., kindly provided the information contained in this note. 39 C. FLIGHT AND A. C. HARRISON F 􀀁 2 l . I ( 7 (􀀈􀀉":::--___ [ ____________ ) I 􀀂 lU©J 􀀃 f" V' y􀀅 r; INCHES PJT Fig. 6. Medieval Pottery (Scale:¼). 40 ROCHESTER CASTLE, 1976 Oxfordshire wares and white-slipped wares, all well established types of the third and fourth centuries. From the shallow gully (layer 11) came pottery of the late second or early third century including samian Forms 37 and 31R, the latter with the potter's stamp SINTVRV·F. Pottery from the fill of the ovens and the plaster layer was scanty, but suggested a date not earlier than the second half of the third century, and the pit in the south-western corner of the excavation was very late second or third century. (ii) Medieval (Fig. 6) P. J. Tester, F.S.A. Group I (Nos. 1-3) This group is mainly composed of body sherds from layer 7 (Fig. 5, Section a-b) observed to run under the red gravel known to underlay Bishop Gundulfs curtain-wall began between A.D.1087 and 1089. This is strong evidence for dating the pottery before that period although the three rims figured were not actually sealed by the gravel. Half the total assemblage is grey sandy ware and the rest contains varying amounts of crushed shell. No. 1 is sandy with only slight shell admixture. Group II (Nos. 4-6) Pottery from pit containing loom-weights. Of the 48 sherds recovered a third are granulated grey ware and the rest shelly; the rims numbered 5 and 6 are in the former category. There is no independent evidence for dating these sherds but they are probably twelfth-century or perhaps a little earlier. Group III (Nos. 7-26) Sherds from the large medieval pit dug through part of the Roman wall. There are 96 fragments, mainly shelly cooking-pots. Also included are 8 body sherds of a grey sandy vessel, insufficient for illustration, with finger-impressed strip decoration-probably part of a pitcher. One vessel is of particular interest: no. 7 is part of a red-painted beaker of imported Pingsdorf ware or one of its derivatives. Light-buff ware with fine sand, wheel-turned and well fired. The dark-red decoration resembles that on vessels from Normandy and the Ile de France belonging to the Pingsdorf group and dating from the eleventh 41 C. FLIGHT AND A. C. HARRISON and twelfth centuries (cf. Medieval Archaeology, iii (1959), no. 13, and 63, nos. 4-8). Pingsdorf wares and their derivatives occur in England at sites along or near the south and east coasts, and the occurrence of this example at Rochester near the Medway estuary is within this pattern of distribution based on water-borne trade. It is remarkable that hitherto Rochester has produced comparatively little imported medieval pottery despite the considerable amount of excavation that has taken place in recent years. This group fits comfortably into the range of wares and rim forms judged on other Kentish sites to belong to the twelfth century. Group IV (Nos. 27-37) Pottery from the filling of the medieval trench. About 60 sherds, nearly all coarse shelly ware. The general characteristics are twelfth-century, perhaps extending slightly into the thirteenth in some cases, e.g. no. 35. It is probable that the material in the back-filling or silting of this trench would contain a high proportion of sherds, to some extent earlier than its excavation and abandonment. This is compatible with the excavator's suggestion that the trench was related to the siege of the castle in 1215. There is no trace of vessels with flattopped flanges characteristic of the late thirteenth century. No. 27 is part of a sandy grey pitcher with strip decoration similar to an example mentioned previously from Group III. II. SMALL FINDS (Fig. 7) Roman period I. Bronze penannular brooch, with turned-over terminals, cf. R. G. Collingwood and I. A. Richmond, The Archaeology of Roman Britain, Fig. 106, no. I 15, and Richborough, l, Pl. XV, no. 26. This is a firstcentury type and must therefore be regarded as out of context. 2. Bronze toilet set, consisting of tweezers, nail cleaner and earscoop, cf. London Museum Catalogue No. 3, Pl. XXXIX. 3. Bronze harness fitting. 4, 5 and 6. Bronze terminals from furniture, cf. London Museum Catalogue No. 3, Pl. XLVII A. 7 and 8. Bone spindles, cf. London Museum Catalogue No. 3, Pl. XLVI. 9 and 10. Bone hair-pins. 11. Bone casing for wooden box or casket, decorated with· incised geometric pattern executed with a compass, cf. Richborough, IV, Pl. LVII. 42 ROCHESTER CASTLE, 1976 􀀄,· \f\, - ttl . 3 : t. ' ' '' 7 ' ' .... ,, '.: g· (:..:. .... .... .' .. ..:----· ' .. - \ 4 l-• -- -• 8 10 9 C •· .4>,-􀀊:__.-􀀋 Fig. 7. Sma1 1 Finds (Scale:¼), 43 ' ' 5 6 \ \-- , ., ,􀀊- 􀀅 . 13 C. FLIGHT AND A. C. HARRISON Viking period 12. Iron axe head. This is a symmetrical axe with a moderately expanded blade and spur projections above and below the socket. It corresponds exactly with type IV, as listed in London Museum Catalogue No. 1, 25, and Fig. 10, no. 1, which is described as the characteristic form of the tenth century. Medieval period 13. Small whetstone of grey micaceous schist, pierced with a hole for suspension from a belt. These hones are thought to have been derived from Brittany and have been found in a number of sites of the Norman period in south-east England, cf. London Museum Medieval Catalogue, Pl. XCIV. 14. Loom weight. An example of the bun-shaped weight in common use from the ninth to the twelfth century. This weight is 5} in. in diameter with a central hole I¾ in. in diameter. It weighs 3½ lb. III. COINS A. Roman. Reported by E. H. Redfern The condition of the coins varied considerably; a few being well preserved and easily readable, but the majority were corroded-many so badly as to obliterate the design and legend. Of the 355 coins listed as obverse and reverse not distinguishable many showed some indication of being VICTORIA or SAL VS types of the late fourth century. It is possible that the proportion of barbarous coins is higher than shown in the list. Mariy of the coins were so corroded that even die-axis checks were impossible and only coins showing clearly barbarous features have been so listed. The list indicates that the coins were chance losses, the small coins and large issues being generally well represented with very few of the larger, more valuable coins. The mint marks, where these can be distinguished, show the expected preponderance of north-western mints. Apart from the Antoninus Pius outliers the coins cover the period A.D. 269 to a date after A.D. 392. The coins of Antoninus Pius showed very different amounts of wear. The sestertius was very worn and like many second century sestertii may have been in circulation up to the third quarter of the third century. The as was only slightly worn, indicating either that it was lost after a comparatively short time in circulation or that it had for some reason been hoarded. 44 ROCHESTER CASTLE, 1976 COIN LIST Reference: LRBC -- R. A. G. Carson, P. V. Hill and J.P. C. Kent, Late Roman Bronze Coinage. Flan Coin diam. No. mm. Antoninus Plus: Ae 31 (ANTONIN)VS AVG PIVS PP T-Fortuna stg. I. sacrificing at altar, rudder in I. hand 2 Ae 25 (ANTONINVS A V)G PIUS PP TRP COS. III ANNONA A VG S C 3 4 5 6 7 8 Annona holding cornucopia and ears of corn over modius. Claudius II: Ae 18 (IMP C CLA V)DIVS A VG GEN Genius stg. I. Ae 17 -- IVS -- Radiate head of Claudius r. Ae 19 IMP CL -- Radiate head of Claudius r. Ae 15 Radiate head of Claudius r. CONSECR(ATIO) Altar Ae 18 Radiate head of Claudius r. (C)O(N)S(E)C(RA TIO) Altar Ae\4 Radiate head of Claudius r. Spread eagle. Victorinus: Mint marks 9 Ae 17 - ICTORI - Radiate head of Victorious r Pax (?) stg. I. holding branch 􀂘 and sceptre Tetricus I 10 Ae20 (IMP) C TETRICVS PF A(VG) LAET(ITIA A VGG) 11 Ae 18 (lM)P C TETRIC(VS PF A VG) PA(X A)VG 12 Ae20 IMP C TETRI(CVS PF A VG) Tetricus II: 13 Ae 21 C PIV(ESV TETRICVS) CABS S(PES) AVGG 14 Ae 21 C PIV ESV T(ETRICVS CAES) S(PES P)VBLICA 15 Ae 19 Radiate bust of Tetricus II r. Sacrificial implements (?) 45 Type RIC 48 type RIC 261 type RIC 261 type RIC 266 type RIC 118 type RIC 87 RIC 100 RIC 270 RIC 272 C. FLIGHT AND A. C. HARRISON Carausius: 16 Ae 23 IMP CARA VSIVS PF A VG RIC IOI PAX AVG ML 17 Ae 21 IMP CARA VSIVS PF A VG 18 Ae 21 Radiate bust of Carausius r. Animal stg. r. RSR Allect11s: 19 Ae 20 IMP C ALLECTVS PF A VG VIRTVS AVG QC RIC 128 Barbarous Radiate copies: 20 Ael3 Radiate head r. Altar 21 Ael4 IMP T-- 22 Ae 16 --ICVS AVG 23 Ae15 --RICVS P-- 24 Ael4 -TETRO 25 Ae 13 to to } RReavdeiarstee nbouts td ri.s tinguishable 29 Ael7 Radiate. unallributable: 30 Ae18 Radiate bust r. --CONS AVG Animal advg. I. 31 Ael4 Radiate bust r. Spread eagle 32 Ae 20 Radiate bust r .. SPES AVGG 33 Ae 20 Radiate bust r. -- PVBL- Spes (?) advg. 1. 34 Ae 14 to to } RRaedveiarstee hnoeat dd iosrti nbguusti srh. able 41 Ae20 Constantine I: 42 Ae 24 IMP CONST A(NTINVS PF A VG) SOLI INVICTO COMITI PTR 43 Ae23 {IMP) CONSTA(NTINVS PF AVG) (SOL! INVIC)TO(COMl)TI J 44 Ae l9 CONST ANTINVS AVG SOLI INVICTO (P)TR 45 Ae 20 CONST AN TINVS AVG PROVIDEN TIAE AVGG PT( R) 46 Ael7 CONSTANT! NVS MAX AVG GLOR IAEXERC ITVS (2 standards} TRS· 47 Ae 18 CONST ANTI NVS MAX A VG GLOR IA EXERC ITVS (2 standards) TR·S 46 ROCHESTER CASTLE, 1976 48 Ae 16 DN CONST AN TINVS PF AVG Gloria Exercitus, 2 standards type Constantine ll: 49 Ael7 CONST ANTINVS IVN NOB C GLOR IA EXERC ITVS •(2 standards) 50 Ae 17 (CO)NST ANTINVS IVN (NOB C) (GLOR) IA EXERC (ITVS) (I standard) Constans: 51 Ae 16 CONST ANS PF A VG GLORI A EXER CITVS (1 standard) 52 Ae 14 CONST AN (S PF A VG) Gloria Exercitus, 1 standard type 53 Ae 17 CONST ANS PF A VG VICTORIAE DD AVGGQ NN 54 Ae l 6 (CONSTAN) S PF AVG (VICTORIAE DD) A VGGQ NN 55 } to Ae 16 CONSTAN S PF AVG VICTORIAE DD AVGGQ NN 58 59 Ae 17 DN CONST A N(S PF A VG) FEL TEMP REPARATIO (Fallen horseman) 60 Ae 16 (DN CO)NST A NS (PF A VG) Constantius JI: 61 Ae 16 FL IVL CONST ANT(IVS A VG) GLOR (IA EXERC) ITVS (I standard) Constantinian: 62 Ae 20 CON-- 63 Ae 14 CONST -- Diad. bust I. Decentius: 64 Ael5 (DN DECENT)IVS NOB CABS Victoriae DD NN A V.G ET CAE (LRBC Type 1 - but HI/IN in shield) Valentinian I: 65 Ae 17 DN VALENTIN(! ANVS PF AVG) GLORIA ROMANORVM (LRBC Type 8) 66 Ae l 7 DN VALEN -- Valentinian I, diad. r. Securitas Reipublicae type 67 Ae 16 DN VA-- Valentinian I, diad. r. Valens: 68 Ael6 DN VALEN (S PF A VG) GLORIA RO (MANORVM) (LRBC Gratian: Type 8) 69 Ael8} DN GRATIANVS AVGG AVG 70 Ae 20 GLORIA NO VI SAECVLI 47 ·PLG (TMP)? TRS (PLG)? ·TR(S?) (AMB)'! olF.1 LVGS COPCON C. FLIGHT AND A. C. HARRISON 71 Ae 19 ON GRATIANVS AVGG AVG GLORIA NO VI SAECVLI 72 Ael6 DN GRATIA NVS PF AVG VOT{X.V/MVLT/XX in wreath Valentinian 11: 73 Ae 14 (DN)VALENTINI (ANVS PF A VG) SAL VS REI PUBLICAE 74 Aei4 DN VAL(ENTINI ANVS PF A VG) (SAL VS REI PUBL)ICAE 75 Ael4 DN VALENTINI (ANVS PF A VG) SAL VS REI PUBL(ICAE) 76 Ae 13} ON VALENTI (NIANVS PF A VG} 77 AeJ4 VICTOR IA AVGGG 78 Ae 15 (ON V ALENT}IN IANVS PF A VG VICTOR IA A VGGG 79 Ae 14 } to DN VA LENTINI ANVS PF AV G (VICTOR I)A A VGGG 81 Ae!3 82 Ael5 ON VALENTIN -- VOT/X/MVLT/XX in wreath 83 Ae 13} DN VALENTIN -- Valentinian II, 84 Ae 14 diad. r. Theodosius I: 85 Ae 14 DN THEODO (SIVS PF A VG) VICTOR IA AVGGG (LRBC Type !) 86} Ae 14 87 DN THEODO SIVS PF A VG VICTOR IA A VGGG (LRBC Type !} 88} Ae 13 89 DN THEO DO (SIVS PF A VG) VICTOR IA A VGGG (LRBC Type 2) 90 Ae 14 DN THEO(DO SIVS PF A VG) SAL VS REI (PVBLICAB) 91 Ae 14 DN THEODO SlVS PF AVG SA(L VS REI PVB)LICAE 92 Ae 13 to lo } DSANL TV HS EROEDI POV SBILVISC PAFE A VG 95 Ae 14 96 Ae 14 ON THEODO SIVS PF A VG Victory advg. I. holding wreath 97 Ae 12 to to } RDeNv eTrsHe EnOotD dOis tSinIVguSi sPhFab lAe V G 105 Ae!5 Magnus Maximus: 106 Ae 13 to ;􀂋 } DN MAG MAXIMVS PF AVG 109 15 SPES RO MA NORVM (LRBC Type I) 48 CON SISC AQP (A)QS fl LVGP -- PCON SMAQS SCON AQP .lL AQS (l ROCHESTER CASTLE, 1976 110 Ae14 Diad. bust r. VO{T'IS/V in wreath Arcadius: Ill} to Ae 14 DN ARCA-- 113 VICTOR IA AVGGG (LRBC Type I) 114 Ae 13 to to } VDINC TAORRC AIAD IAV VSG PGFG A V( LGR BC Type I) 116 AelS 117 } DN ARCAD-- to Ael4 VIC(TOR IA)AVGGG (LRBC Type 1) 119 120 .Ae 14 } DN ARCAD-- to to VICTOR IA AVG - 127 Ae IS Victory l. bearing wreath 128 Ae 12 } DN ARCADI VS PF A VG to to SAL VS REI PVBLICAE 131 AelS (LRBC Type 2) 132 Ae 10 to to } RDeNveArRseC noAtD d-isti-nguishable 142 Ae 15 Honor/us: 143 Ae 13 DN HONORI VS PF A VG VICTOR (IA A VGG) (LRBC Type 1) 144 Ae14} DN HONORIV -- 145 Ae 12 VICT --Victory I. bearing wreath 146 Ae 13 DN HON-- SAL VS(REI PV)BLICAE 147 Ae 15} ON HON-- 148 Ae 13 --VBLICAE 149 Ae 14 DN HONORIVS P(F A VG) Rever se not distinguishable EMPEROR NOT DISTINGUISHABLE Gloria Exercilus, 2 standards: 150} Ae 18 Diad. bust r. 151 GLOR (IA EXERC) ITVS Gloria Exercitus, I standard: 152 Ae 15 FL IVL CONS -- 2 soldier s, 1 st and a rd 153 Ae 14 CONSTA-- (GLOR IA)EXERC !TVS 154 Ae 17 FL IV-- GLOR IA EX(ERC ITVS) 49 LVGP PCON TR (T)RS"' (P)LG -CONS 155 lo 157 C. FLIGHT AND A. C. HARRISON Ae 14} Diad. bust r. 􀂢􀂣 I 6 GLOR (IA EXERC ITVS) Urbs Roma: 158 to 162 Ae 13 } VRBS ROMA 􀂢􀂤 17 Wolf and twins, 2 stars above Constantinopo/is: 163 to 167 1 Ae 14} Helmeted head I. A􀂤 16 Constantinopolis stg. I. on prow 168 Star in Wreath: Ae I 4 Diad. head I. Star/CONSS in wreath Victoriae DD Augg Q NN: 169 Ae 15 (CON)STAN --PFAVG VICTORIAE DD AVGG(QNN) {LRBC Type 1) 170 CONSTAN ---- to } Ae 15 172 VICTORIAE DD A VGG Q NN (LRBC Type 2) 173 to 175 Ae 14} Diad. bust r. to (VICTORIAE D)D A VGGQ NN Ae 16 {LRBC Type 2) Diad. bust r. 176} Ae 15 177 {VICTO)RIAE DD A(VGGQ NN) (Type uncertain) Fe/. Temp. Reparatio: I 78 Ae 15 Diad. bust r. FEL TEM(P REPARATIO) Fallen horseman type 179 Ae 20 DN CON--VS PF AVG (FEL TEMP REP)ARATIO Fallen horseman type 180 Ae 14 Diad. bust r. Fallen horseman type Gloria Romanorum: 181 Ae 16 Diad. bust r. Gloria Romanorum LRBC Type 8 I 82 Ae 19 l Diad. bust r. 183 Ae 15 J G ---- M (LRBC Type 8) Securitas Reipub/icae: I 84 Ae 17 Diad. bust r. SECVRITAS REIPVB(LICAE) 50 TRS PCON OfF(A)? (L)VGo} FOFj! LVGROCHESTER CASTLE, 1976 185 Ae 17 Diad. bust r. SECVRIT AS REIPVBLICAE SMAQ 186 Ae18 Diad. bust r. SECVRIT AS REIPVBLICAE -CON 187 Ae 18 DN VALEN SPF AVG (SECV)RITAS REIPVBLI(CAE) Victoria Auggg (LRBC Type 1): 188 Ae l4 ANVS PF AVG (VICTOR) IA A VGGG (L)VGP 189 Ael4} Diad. bust r. 190 Ae 13 VICTOR IA A VGGG PCON 191 Ae 13 Diad. bust r. VICTOR JA A VGGG SCON 192 Ae 14 Diad. bust r. VICTOR IA A VGGG -CON 193 Ae 12 to to } Diad. bust r. 202 Ae 16 VICTOR IA AV GGG Victoria Augg (LRBC Type 1): 203 Ae 14 Diad. bust r. (VICTOR IA A)VGG PCON 204 Ae 13 Diad. bust r. (VICTOR IA)A VGG Victoria Types: 205 Ae 15 Diad. bust r. VICTOR PCON 206 Ae 13 } Diad. bust r. to to Reverse lettering: fragments of 225 Ae 15 VICTORIA AVGGG (where visible) Victory advg. I. holding wreath Salus Reipublicae: 226 Ae 13 fJ } Diad. bust r. 227 Ael2 SAL VS R(EI PVBLICAE) AQS 228 Ae 12 to to } Diad. bust r. .J:L 234 Ae 14 (SAL VS)REI PVBLIC(AE) 235 Ae 12 } to Diad. bust r. rL to SAL VS REI (PVBLICAE) 247 Ael4 248 Ae 14 Diad. bust r. (SALVS REI P)VBLICAE SM(RP'?) 249 Ael3 to to } (DSiAadL.V bSu)sRt Er.I PVBL(ICAE) 254 Ael4 51 255 256 2S7 2S9 to 362 C. FLIGHT AND A. C. HARRISON Vora Coins: Ae 13 Diad. bust r. VOT/V/MVLT/X in wreath Ae 15 --(RATIA?)-VOT/ XV/MVLT/XX in wreath Ae 14 Diad. bust r. VOT/X/MVLT/XX in wreath Diademed bust, right. Reverse not distinguishable: Ae 12 to } Mintmarks distinguishable Ae 18 BARBAROUS COPIES Barbarous, Constantius II: 363 Ae 16 CONSTANTIVS AVG Gloria Exercitus, I standard type Barbarous, Valentinian JI: 364 Ae 14 DN VALENTINI ANVS(PF AVG) (VICTOR) IA AVGGG (LRBC Type 1) Barbarous, Theodosius I: 365 Ae 13 DN THEO -VICT---- Barbarous, Magnus Maximus: 366 Ae 14 DN MAG MA (XIMUS PF A VG) (SPES RO MA) NORVM Barbarous, Arcadius: 367 Ae IS (DN A)RCADI VS PF (AVG) VI(CTOR IA) AVGGG 368 Ae 13 DN AR --- • 369 to 371 372 VICTOR --- Victory advg. I. Barbarous, Gloria Exercitus (1 standard): Ae 13 } CONSTAN-- 􀂢􀂣 15 (GLOR!) AEXERC ITVS Barbarous, Urbs Roma: Ae 12 VRB(S) ROMA Wolf and twins Barbarous, Fel Temp Reparatio: 373 Ae 20 CONS --- 374 Ae 16 37S Ae 12 to to } 377 Ael4 PEL TEMP REPARATIO (Fallen horseman) Diad. bust r. Fallen horseman type Diad. bust r. Fallen horseman type 52 TRP {-TRP LVGP PCON -CON LVGP AMB AQS ROCHESTER CASTLE, 1976 Barbarous, Victoria Types: 378 Ae 13 Diad. bust r. VICTOR IA AVGGG (LRBC Type 2) 379 Ae 13 Diad. bust r. 380 to 384 385 386 387 388 to 398 VICTO(RIA A VGG) (LRBC Type 5) Ae 12 } Diad. bust r. to Various fragments of VICTORIA Ae 14 AVGGG Barbarous, Salus Reipub!icae: Ae I 2 } Diad. bust r. Ae 13 As LRBC, Type 2 Ae 10 Helmeted head r. Extremely barbarous figure, possibly Victory Barbarous, Diademed Bust r.: Ae 9 } Diad. bust r. to No reverse types or mintmarks Ae 15 distinguishable A further 355 coins (Ae 10 to Ae 20) were so corroded that neither obverse nor reverse could be distinguished. The technique developed by A. Ravetz (Numismatic Chronicle, 1964) and extended by J. Casey (British Archaeological Reports, No. 4, 1974) has been applied to the 175 coins which can be dated with sufficient precision. Starting from A.D. 260 the coins are grouped into periods of political and numismatic significance and the results are shown on a graph (see Fig. 8). Each column represents the coins of a period according to the expression: Coins of Period 1,000 X Length of Period Total Datable Coins The dark portions of the columns represent barbarous copies. The figures along the horizontal axis show the beginning and end of each period (A.D.). This method of representation gives a clear indication of the comparative rates of loss of coins in the various periods. The picture is of continued site activity from the middle of the third century to at least the end of the fourth century. The peaks (periods A.D. 260-273, A.D. 330-346, and A.D. 388-402) correspond with issues of large numbers of small or 'inflated' coins and do not necessarily indicate large surges of activity at the site. It is notable that the troubled period 53 C. FLIGHT AND A. C. HARRISON A.D. 260 21.1 286 296 :117 .J.10 346 364 J78 .J88 402 Fig. 8. Histogram illustrating Coin Distribution. A.D. 378-388 is comparatively well represented and the period A.D. 388-402 shows the highest rate of coin loss. Even allowing for rapid inflation, considerable activity must have continued after A.D. 392 at this site. Due to the long periods over which VICTORIA and SALVS reverse types were issued it is not possible to give a date for the latest coins other than 'after A.D. 392' - except for one coin which appears to be a barbarous copy of a coin struck in A.D. 410 or later. 54