Reculver; Excavations on the Roman Fort in 1957

RECULVER ; EXCAVATIONS ON THE ROMAN FORT IN 1957 By BRIAN J. PHTLP, A.C.C.S. Reculver Excavation Group : First Research Report INTRODLCTION THE north Kent coast has long suffered from the ravages of sea erosion and many acres of land have been lost. At Reculver this encroachment has destroyed nearly half of the Roman fort and washed away almost a mUe of land which existed to the north when the fort was buUt. Fortunately, however, most of the surviving portion of the fort is now protected by sea-aprons but a section some 180 feet in length stiU remains open to the sea (Fig. 1). The writer appreciating that more of the fort would be destroyed at this point, resolved to carry out superficial excavations. Six years rescue-work on the Foreshore to the west of the fort had resulted in the gaining of much useful information about the Prehistoric and Roman settlements.1 It seemed likely that the exposed section of the fort (here under review) would reveal definite evidence of Iron Age occupation and throw new light on the date of the fort's construction. Application was therefore made to the Ministry of Works (as Guardians of the fort), for permission to clean down the unprotected section. This was readUy granted. The work was carried out from 5th to the 12th of October, 1957 when 100 feet of the fort's interior levels were examined and recorded. The difficulty of cleaning-down the near vertical cliff-face some 20 feet above the beach was overcome by the use of simple scaffolding and numerous safety hnes. The section examined cuts the east waU of the fort at about 70 degrees at a point approximately 460 feet from the south-east corner. A cknowledgments The writer wishes to thank the foUowing for thek most valued support. In particular Mr. S. S. Frere, M.A., F.S.A., who has Idndly advised and guided the writer with the report ; Mr. N. C. Cook, B.A., F.S.A., who has continued to aUow his facUities at the GuUdhaU Museum to be used and to Mr. R. Merrifield, F.S.A., who has kindly dated the coin finds. Mr. M. R. Hull, M.A., F.S.A., has commented 1 For a report on the findings for 1962-54 see Arch. Cant., LXXI (1957), 167 and for 1956-57 see Arch. Cant., LXXII (1968), 160. 96 RECULVER PROXIMITY OF BATTELY'S 'CISTERNS' AND 'FOUNDATIONS' tOO 10 0 30 0 L OH WtMOT.- LOW OK HUM WMEK M&«X OF ORDINARY 1IBES SPOT HEIGHTS SHOWN THUS'- + Z4 UtAftJ ^aSBfU. AREA OF btSCOVCMES 5 ISS2-I957 S AREA OF DISCOVERIES MI-H57 l»«.o» M J WEST GATE ft i e f-SEcnoN 4 -p EXAMINED 1957. J ?.«"« KING tTHELB 2 ™ I ' l ' i i ' " " " " REPAIR CF .m^o-i FORfT WALL-* ^ " A ^ 0 0 ^ /,/""" HAIkliCt LOW LYING PASTURE LAND (SITE OF NATURAL ANCHORAGE) Fio. 1. Plan of Reculver, Kent, showing sites excavated, 1952-57. RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 on three key coarse ware potsherds ; Mr. B. R. Hartley, F.S.A., has dated the samian and Lt. Col. G. W. Meates, F.S.A., has examined some of the coarse ware. The work was carried out by members of the Reculver Excavation Group under the direction of the writer. In particular Messrs. Michael KeUaway, Brian KeweU and Derek Garrod are to be congratulated for their strenuous work throughout the excavation. Thanks are also due to Mr. and Mrs. H. E. Gough, Miss T. PoweU-Cotton, Messrs. Frank Harvey, David Mould and Bruce HaU for their occasional assistance. Mr. KeweU is responsible for the exceUent sections and plan and Mr. KeUaway for some of the pottery drawings. Mr. Dennis Hicks, Ministry of Works Custodian at Reculver greatly assisted the work by willingly offering all the facilities at his disposal. SUMMARY OF EARLIER FINDINGS Little evidence relating to the fort had previously been uncovered but over the past 300 years a number of general finds have been recorded. These included numbers of coins of considerable date range.1 In the late 17th century the Rev. J. Battely2 noticed foundations (perhaps of a bath house or villa), Cisterns (wells) and other remains to the north of the fort which was then intact. Sections cut against the fort walls by George Dowker in 1877s and Major Gordon Home in 1927* supplied a few structural details whilst the last named found evidence of a buried south gate to the fort in 1931. In 1951 Mr. F. H. Thompson, working for the Ministry of Works, caused a section to be cut against the south waU of the fort.5 In the lowest occupation levels he found potsherds of a native culture, possessing Iron Age A traditions and under direct Belgic Influence. Since 1952 the writer has been carrying out rescue operations over a large area of the site. The findings indicated (a) an occupation during the earlier Iron Age ; (b) a total lack of first-century Roman material ; (c) that a fixed form of settlement had probably existed on the site of the fort in the later second century and there had been an intense occupation during this same period ; (d) that finds of the third and fourth centuries were less prolific. 1 V.C.H. (Kent) III. p. 19, for details of finds up to 1932. 2 Antiquitates Rutupince (Published posthumously) 1711. Translated version The Antiquities of Richborough and Reculver, 1774. 3 Arch. Cant., XII, I. 1 Arch. Jnl., LXXXVI, 260. 6 Arch. Cant., LXVI, 62.. The seotion there recorded may usefully be compared with the present one. 98 RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 DESCRIPTION OE THE EXCAVATIONS SECTION A (Fig. 2) The Prehistoric Layer The natural sub-soil at Reculver is Thanet Sand reaching to a height of 10 or more feet above the present beach level. Covering the natural deposit was a 10-12 inch layer of shghtly darker sand containing some 50 scattered heavUy-gritted potsherds of Iron Age A date. These represented parts of at least eight separate vessels. Numerous chalk particles in the upper part of this layer may represent an attempt at liming for agricultural purposes. A shaUow pit or gulley cut into this layer yielded nothing. Slight indications of a native population under direct Belgic influence, yet with hngering Iron Age A traditions were found in the excavations of 1951 in this same layer at a point some 450 feet to the south of the present section. In 1953 the writer1 discovered a shaUow pit containing early Iron Age pottery and daub associated with this same occupation level. (The position of this pit has been projected onto Section B at its correct relative level.) Clearly this Prehistoric layer extends over the whole site though little can be said of the degree of occupation. Construction of the Fort Wall Only a broken and badly damaged portion of the east waU of the fort survives at the point examined and of its two known internal offsets only one and the first few courses of flints were visible. None of the external facing stones remain in place whUst barely eight feet of the waU's original base-width of ten feet survives. The Ministry of Works excavation in 1951 had revealed a very distinct mortar-mixing floor, some three inches thick, behind the south waU of the fort. Owing however to the introduction of a circular oven (see below), such a continuous layer was absent although certam patches were found. The fact that no occupation layers or debris existed between the prehistoric layer and the mortar droppings indicates that no Roman pre-fort settlement existed on the site, a view confirmed by the excavations of 1951. The fort construction-period is therefore regarded as the first (Phase I) Roman settlement of the site. The sequence of the construction of the fort waU could be followed. First, a footing trench some three feet deep was dug and the upcast thrown on the inside (indicated in the section as a sharp rise). Next, large beach pebbles were pUed up close to the trench to be used as footings. Not aU these pebbles were so used at this point as the base of 1 Arch. Cant., LXXI (1957), 167. Pit 4. 99 RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 one of the pUes was found. The lower courses were then laid and the sides of the trench fiUed to that level. The lowest of the two mortar levels indicates more courses being laid before a further foot of soU was packed in. The upper layer of mortar some three inches in thickness may indicate the last stage of construction. There were no associated finds. Barbed-wire and beer bottle fragments proved the shght excavation on the inside face of the waU to be of recent date. Oven 1 Close to the fort waU was a circular oven some six feet wide composed of tUe and stone. This was the structure noted in 1953 as being situated below the rampart bank and therefore earher in date than the fort, suggesting pre-fort Roman settlement. It is now clear that the oven and fort are contemporary. First, the oven was buUt upon the leveUed upcast from the footing-trench, and above the remainder of the pUe of footing pebbles. Clearly therefore the structure did not ante-date the buUding of the fort. Secondly, the lack of a continuous layer of mortar droppings either above or below the oven indicates that the oven had to be avoided by the buUders and cement carriers. Thirdly, patches of mortar were found under and on top of the ashes raked out of the oven, and finally the blocks of Kentish ragstone used in the oven are simUar to those used to face the fort waU. The oven was eventually buried under the layers comprising the rampart bank, the final stage in the completion of the defences. SimUar ovens (or hearths) have been found in several other forts as at Newstead,1 Malton2, Cappuck3 and Birrens.4 The oven was neatly constructed. Upon the leveUed upcast from the wall trench had been placed a bed of roughly squared blocks set in clay ; these formed a solid base and also the support for one wall. Over this had been laid beach pebbles set in a fine yeUow sand and rectangular tUes (measuring 11 by 13 inches) held by a fine red clay, completed the floor. The waUs had originaUy consisted of irregular blocks set in clay and stiU survived to a height of about 20 inches. The oven was stoked from the west for clearly there had not been a wall on this side. The ashes, mostly of brushwood, had been raked back through the stoke-hole to faU away below the oven floor level, each time being sealed by a thin band of clay. This most conspicuous layer, although now greatly compressed still remained some 12 inches deep. The only find from this oven was a small rim-sherd of late secondcentury date (Key Deposit No. 1). i P.S.A.S., LXXXIV, 1. 2 The Defences of the Roman Fort at Malton, 1930. Philip Corder. 3 P.S.A.S., LXXXV, 138. * P.S.A.S., XXX, 172. 100 RECULVER (1957) 5tCllOW A Ft ML SfcHO 32-56 P U t t U S DARK >UWM CIA UIH wuu» sin ASU RU6BIE t MIDDEN ^ ^ BLACK ASM fclACH PUf t lM SHK SAND I k t C E N l . «y&> IE D/BROWN EAR1VI A FUN" FOOTINGS URCitD i I ORIGINAL Will THICKNESS— -sit Uve* /Ti e W->£ / „ * M RUBBISH PIT Z' ^ "vy^r-^ *> : J ^ . UPCAST FROM - OF fcAST WILL JuORTaC. OR0FP1KC4 / . F0O1IUCS 1UUCH OP KM IA0RUR DWPFIUG5 PRE r.F-ORT' (IROU A&t) LAYER • D U Ott"»»» IRON MNO PIT GULltY 9Ct ABOVE BCACU LPVtl WFllli OP MACU FtMl t t FOK PQM Wi l l FOOTINGS/ HE FLOOR OF OVEN I NATURAL (THANET SAND) PE&BLE FOOTINGS BttWEU PIG. 2. Plan—Seotion A. RECULVER (1957) 5EC1I0N D (WE5T) H U M U S RlClMT) OUTER FACIMC STONES RUBBLE CORE IUNF.R FACING ST (RECENT) COIN CONSTANTINE T AB.3S& WORN CON COUMODU5 AD 160-190 ROAD METALLING COIM L«t 3-i CENTURY TILE PAL LAVtR OP Will PIASTPR WW R.OOP PAH COIN. CARAUSIUS C A.O. I&7 «CtH ••fi •iiT«vitT-ifw; II FT ABOVE PRoTtCTtD fosinou \ HALLS OF PHASE TJ BEACH LEVEL OP PU(4) CONTAINING BUILDING PARLY IRON ACE FOlTtRY MORTAR BUFFINGS OVEN I kUD DAUQ. DISCOVERED IBS* (TILE FLOOR MISSING) 50 55 fcO ts 70 NATURAL (THANET SANb) 75 ao 65 V) 95 D ktWtll 100 Fra. 3. Plan Section B. [face p. 101 RECULVER : EXCAVATIONS ON THE ROMAN PORT IN 1957 The Rampart Bank The section cut in 1951 had revealed 30 feet of the rampart bankbehind the fort wall. The bank was again encountered but here the stratigraphy differed owing to the presence of the oven. The oven had eventually become buried under a three feet bank of sterile earth and clay thrown down to form a basis for the rampart bank and extending from the inner face of the fort wall for some 50 feet. Over this had been deposited a rubbish layer (Rubbish Layer I) containing glass, pottery, nails, shells, tiles and bones including the skeleton of a small dog (Key Deposit II). The whole bank was then covered with a band of clay giving a depth of five feet. A shallow rubbish pit (Key Deposit III) had been dug into the bank near its tail-end and produced pottery, nails, bone, glass, shell and a piece of lava millstone. Part of this pit had been examined by the writer in 19531 when it yielded more pottery, a bone needle, some minor bronze fittings and a fragment of daub bearing a keying impression to hold plaster. Sealing and levelling the pit was a narrow layer of oyster and mussel shells mixed with kitchen refuse and pottery (Key Deposit IV). A band of clay covered the higher part of the shell layer (Key Deposit V). A second layer of rubbish (Rubbish Layer II) had then collected over the entire bank (Key Deposit VI). A continuous but badly disturbed layer completed the bank giving an overall depth of six feet. Exactly what arrangement existed immediately adjacent to the fort wall was difficult to determine owing to extreme weathering on this very exposed corner. However the rampart is known from previous cuttings to have originally extended some way up the inside of the wall. In late Roman times a footing composed of heavy blocks and peach pebbles was cut into the bank 11 feet from the fort wall and had destroyed some of the stratification at this point. No attempt is made to interpret this feature owing to the disturbed nature of the adjacent levels.2 A layer of fine silt covered the tail of the bank and also a few feet of the road metalling. The dating evidence from the rampart bank is discussed separately (see also Appendix I). SECTION B (Fig. 3) This section studies the interior stratification of the fort between 50 and 100 feet from the east wall of the fort. The prehistoric layer 1 Arch. Cant., LXXI, 167. Pit 5. 2 In April 1949 Mr. H. E. Gough, of the Heme Bay Library and Museum, observed a rectangular shaft some 78 inches deep and 39 inches wide situated below and sealed by the rampart bank some 15 feet from the east wall of the fort. No datable objects were recovered and the pit was washed away shortly after. No indication of the oven was then visible although the edge of the ashes were just apparent. 101 11 RECULVER : EXCAVATIONS ON THE ROMAN PORT IN 1957 was traced for the whole length of the section but with increasing difficulty. The pit containing the Iron Age pottery which was found in 1953 is projected into the section as previously mentioned. Intervallum road At the foot of the rampart bank ran the intervallum road. This had been laid in conjunction with the rampart bank and therefore at a date earher than the digging of the rubbish pit. It was composed of compacted gravel and small pebbles varying in thickness from one to six inches. Where it abutted onto the tail of the bank its depth increased to nearly a foot ; here a corroded iron ring (with a diameter of seven inches) was found. The road covered a patch of mortar which again lay directly upon the prehistoric layer. The early third-century layer of ash and silt covered the first few feet of the road metalling and contained at this point a thin band of wall-plaster showing faint signs of red and white paint. The layers of clay above the road were barely touched by the excavations. Oven II The remains of a second oven were discovered 74 feet from the fort wall and 56 feet from Oven I. Like the other it had probably been circular, with a tiled floor set in a fine red clay and with a footing of fine gravel covering a 12 inch bed of chalk blocks. Only seven feet of the floor remained, the westerly part having been destroyed by the foundations of the later building (described below). The ashes from the fire had been raked back to the east but no damping layers of clay had been employed. When this oven was first observed in 1953 several of its floor tiles were still in position ; mention of this structure was made in the first report on the findings at Reculver. The Phase II Stone Building The most prominent features of Section B were two lengths of heavy walling representing the corner of a substantial building. Both walls (26 inches thick) were composed of squared blocks (each 7 by 9 inches) and a rubble core set in a stiff clay. The nature of the two walls clearly indicated that they had formed part of the same structure, rectangular in shape, with the minimum internal dimensions of 13 by 8 feet (calculated by projecting the lines of the wall into the cliff). The size of the walls suggests a building considerably larger, perhaps a barrack block. Numerous broken tiles lay in close association and indicate the type of roofing. Above the lower courses the walls were probably of clay held in a timber frame as a heavy stone wall of full size would not survive with purely clay bonding. The lack of any 102 RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 trace of mortar or clay flooring suggests that wood was used for that purpose. The building clearly post-dates the oven through which its footings were cut ; a worn coin of COMMODUS (A.D. 180-193) was found in the clay of the footing trench. Its worn condition suggests that it was not dropped until well into the third century. The rubbish and debris from inside the walls contained two coins, one of CARAUSIUS (A.D. 287-293) and a barbarous radiate (late third century) ; pottery dating c. A.D. 300-350 suggested the period of occupation. A coin of CONSTANTINE I (A.D. 330-335), in good condition was found outside but associated with one of the' walls. These finds and the lack of earlier material suggests that the building dates from the late third century.1 EAST GATE OE THE POUT Of the gates of the fort very little is known. Early maps of the site show a gap in the centre of the west wall of the fort which has long been accepted as the main entrance and logically so when the local topography is considered. Shght traces of a southern gate were discovered in 1931. As the opportunity to examine the surviving part of the fort walls presented itself during the excavations an attempt was made to determine whether or not an eastern gate had existed. The evidence strongly suggests that one had. First, the modern path cuts through the east wall at about 285 feet from the south-east angle of the fort to coincide with the exact centre of this side.2 Secondly, at this same spot are massed a quantity of large stone blocks several times larger than the ashlar blocks employed in the facing of the fort wall and certainly suggesting some structural change in the wall here. One of the blocks,3 of considerable size, measures 48 inches by 36 inches and is 12 inches deep, shows distinct mouldings on one edge and it bears a marked resemblance to masonry incorporated in the south gate at Caerwent ; the east gate at Birdoswald and the north gate at Melandra Castle. Yet another point of importance is that close to the footpath a relatively modern cess-pit had been sunk directly on the line of the wall. Had the wah existed at this spot as it does for its entire surviving length of 460 feet on this side, then the pit would have had to be cut through particularly hard Roman concrete many feet thick. By 1 Such construction may be regarded as the Phase II Roman occupation of the site. 2 A measurement taken in 1781 when the north-east angle of the fort still remained indicated internal dimensions of 570 by 585 feet. 8 Mention of this jamb-stone (?) is made in V.G.H. (Kent) III, 21. 103 RECULVER : EXCAVATIONS ON THE ROMAN PORT IN 1957 siting the pit just a few feet away any such toil could easily have been avoided. The pit's present position suggests therefore that here was an original break in the wall, perhaps the carriage-way of the gate. With this in mind the undergrowth and loose earth next to the path were cleared away revealing a short length of curved wall set some four feet back from the extended line of the wall facing. It was found to be composed of the larger type block aheady noted, but no archseological evidence was available to prove that it was of Roman construction. A gate at this point could not have a width greater than about 35 feet, a more than adequate figure. The break in the centre of the south wall gives a maximum width of about 41 feet for a gate at that point. REPAIR OE THE EAST WALL OE THE FORT At a point 127-138 feet from the south-east angle of the fort an 11 feet section of the wall can be detected as a later addition. This patches a large breach in the wall from three feet six inches above the footings to at least the present height of about eight feet. It is roughly rectangular and is composed of a variety of material which contrasts with the composition of the original wall. The mortar used in the repair differs from that used along the rest of the wall as it bears a high percentage of crushed shell. The repair includes a broken piece of tile, irregularly shaped lumps of masonry and a portion of a fine cement floor 30 inches in length and 12 inches deep. Several external facing-stones remain in place. It is not clear if the repair extends through the entire thickness of the wall. This hasty method of patching, clearly associated with one of the later phases of occupation, suggests a deliberate destruction of the wall at this point. The possibility of a bastion having fallen out here must be considered despite the total lack of references to bastions by earher observers. THE DATING OE THE CONSTRUCTION OP THE FORT No assistance as regards the date of the fort's construction is available from epigraphical or literary sources. / . Archceological Evidence In order to interpret correctly the dating material from the rampart bank it is first necessary to prove that the fort wall and the bank are contemporary constructions. A number of Roman forts originally built with earth and clay ramparts had stone walls added to them at later dates. Clearly this had not occurred at Reculver. The section cut through the south wall in 1951 showed the mortar droppings, 104 RECULVER : EXCAVATIONS ON THE ROMAN PORT LN 1957 associated with the wall building, to extend under the length of bank examined. Similar though not so marked traces of this dropping layer were forthcoming from the present section. The lack of occupation material between the mortar spread and the rampart bank in both sections clearly points to the latter having been thrown up very soon after the construction of the fort wall. As described the Oven (No. 1) is likewise contemporary with the construction of the wall. I t now remains to examine the archseological material. Over 500 stratified potsherds were recovered from the several layers and rubbish pit comprising the rampart bank (Key Deposits I-VI). The similarity between the Samian, pie-dishes, jars and rouletted beakers from all levels indicates that the bank had been formed within a short period. In all probability this section of the bank had been used for domestic rubbish during the construction and opening phases of the fort. The bulk of the pottery appears to date A.D. 160-200 suggesting a construction date of c. A.D. 200. However Mr. Sheppard Frere who has kindly examined some of the coarse pottery has also found similar material in third-century levels in Canterbury and is in favour of a construction date nearer A.D. 225. The presence of a few sherds of Rhenish pottery indicates that the bank was not formed before A.D. 175. On the basis of the archseological evidence therefore a provisional dating of c. A.D. 200-225 may be assinged to the construction of the fort (See Appendix I). Confirmation of this dating was obtained from the silt and ash layer covering the tail of the rampart bank and also the first few feet of the road. This clearly formed after the construction of the bank and the road and is dated by pottery and a coin A.D. 200-230 (Key Deposit VII). Supporting evidence for an occupation at about this date is obtained from a quantity analysis of the writers finds over a period of six years from an area 2,000 feet in length. This reveals an intensive settlement during the later second and early third centuries. Most of the wells and rubbish pits date from about this time. II. Structural Implications Roman forts in this country may be broadly divided into two groups each having its own structural characteristics. The earher group is dated to the first and second centuries and the later group, evolving from the earlier, to the later third and fourth centuries. The known structural details of the fort at Reculver are certainly more characteristic of the earher group. That Reculver had early-looking defences has long been known, but such structural evidence was disregarded through lack of archseological data and the historical implications, so that until the present 105 RECULVER : EXCAVATIONS ON THE ROMAN PORT IN 1957 excavations the fort was always considered to be of late third-century foundation. Indeed R. G. Collingwood1 admits the early characteristics of the fort and remarks that the men who designed Reculver and Brancaster had such very divergent ideas about the principles of fortification from those who designed Richborough, Porchester and Cardiff (Late third century), that it would be rash to assume, without definite evidence, that they were contemporaries co-operating in a single scheme. However, he still assigns Reculver to the later type of fort on the basis of few and narrow entrances and waU thickness. It must be pointed out, however, that even today no measurement is known for any gate at Reculver, and also there exists evidence for listing three gates whilst it is not altogether unlikely that a fourth existed. It now remains to discuss the wall thickness criterion. The walls at Reculver were found to be 10 feet thick at the base and reduced by at least two internal offsets each 12 inches wide. The forts of the Saxon Shore (typical of Group II), have walls 10-14 feet thick whereas the forts of Hadrian's waU (typical of Group I) usually have a wall thickness of 4-5 feet. The transition is not difficult to trace. Brough Castle2 (c. A.D. 158) had walls 5 | feet thick whilst Balmuildy3 and Castle Cary4 on the Antonine wall had wall-base thicknesses of 1\ and 8 feet respectively. Even thicker walls were being introduced by the second half of the second century as at Malton,5 where the walls were certainly 10 feet in thickness by A.D. 182. The undated forts at Templeborough (8 feet 6 inches) and Elsack (9 feet) must surely date to about this time. Clearly therefore, the construction of a fort at Reculver c. A.D. 200-225 should not be discounted merely through its walls being 10 feet in thickness. Indeed the structural characteristics of the fort are consistent with this dating. Ill Historical Considerations Reculver certainly formed part of the Saxon Shore fort system during the late third century but the evidence set out above for an earher foundation demands a brief consideration of the historical context. During the last decade of the second century many Romano- British towns were receiving their walls,6 a measure possibly attributed to ALBINUS. The building of a fort at Reculver during this period is a possibility consistent with the insecurity of the time. With the removal of ALBINUS, Britain was effectively reorganized by the 1 The Arch, of Rom. Brit., London. 1930, 56. 2 Jnl. Derby A. & N.H. Soc, LIX, 63. 3 Roman Wall in Scotland, 1934, p. 315. Sir George Maedonald. * P.S.A.S., XXXVII, 271. 6 The Defences of the Roman Fort at Malton. 1930. Philip Corder. ? Arch. Jnl, CXII, 20. Philip Corder. 106 RECULVER : EXCAVATIONS ON THE ROMAN PORT IN 1957 Emperor SEPTIMIUS SEVERUS (A.D. 197-211) and it is equaUy possible that the fort formed part of this scheme. Large scale Saxon raids were certainly prevalent in A.D. 287 as we learn from the career of CARAUSIUS. The ditches of the temporary earth-fort at Richborough1 (recognized as an ante-pirate measure,2) were dug and filled before that date and in Sussex certain indications exist of destruction, perhaps by Saxon raiders, at about A.D. 270.3 It is possible that these raids commenced considerably earlier. If the raids were known at the beginning of the third century then the reason for the construction of the fort at Reculver, guarding as it does the Thames estuary and adjacent shipping routes, is apparent. If this is the case the Saxon Shore system of coastal defence commenced early in the third century and was considerably extended at a later date. The present findings have an important bearing on the dating of the fort at Brancaster, Norfolk (previously mentioned) ; to use the words of its excavator,4 if the station at Reculver were precisely dated it would suggest a date for the construction of the fort at Brancaster. This statement is well founded as structurally the two forts are very similar. Like Reculver, the fort at Brancaster was used as a base against the Saxon raiders in the late third century and this phase may be regarded as the later of two shown by excavation to exist. The earlier phase, representing the fort construction period, has received a tentative mid-third century date. The finds certainly indicate that the fort at Brancaster was garrisoned about the middle of the third century. If then we are to accept that the forts belong to a single scheme then Brancaster may likewise be assigned to the first quarter of the third century.5 SLMMARY PREHISTORIC SETTLEMENT The prehistoric layer was found to extend along the entire length of the section examined and it is now known to extend over the whole site. A shallow pit of gulley and a scattering of heavily gritted potsherds were the only discoveries made of the Iron Age settlement. Other Iron Age pottery had been recovered from this layer in 1953. There is no evidence to suggest an intensive prehistoric occupation 1 Rich., IV, 60. 2 Roman Britain, 1955, p. 60. I. A. Richmond. 3 Arch, of Sussex, p. 305. A small villa at Preston (Nr. Brighton) and the village at Park Brow were burnt down about A.D. 270. 1 Norfolk Arch., XXX (1949-52), 145 also V.C.H. (Norfolk) I, 303. 6 In the same manner as Reculver guarded'the shipping routes of the Thames estuary, Brancaster protected those of the Wash with its rivers leading into the heart of the country. 107 RECULVER : EXCAVATIONS ON THE ROMAN PORT IN 1957 of the site. No trace of the known pre-Roman Belgic settlement was encountered. ROMAN SETTLEMENT Since excavations commenced in 1952 not a single shred of evidence to support a first century A.D. Roman settlement has been recovered. A few first century coins are listed as having been found during the eighteenth century but contemporary pottery illustrations show typical second century types. It is probable, therefore, that the native occupation failed to survive the conquest. A small amount of early second century pottery has been recovered, but the lack of associated structures suggests only temporary use of the site at that date. Phase I The evidence indicates that the permanent occupation of the site commenced with the construction of the fort.1 The wall mortar-droppings immediately upon the prehistoric level provides evidence of this. In the absence of conclusive evidence as to the precise date of the construction of the fort a provisional dating of A.D. 200-225 may be assigned. This dating is obtained from the archseological material2 recovered from the rampart-bank and is also consistent with the fort's structural characteristics. Historical considerations favour the reigns of ALBINUS or SEVERUS as the building period. The function of the fort appears to have been the protection of important shipping routes, possibly against Saxon pirates, in the same manner as did the very similar fort at Brancaster (which appears to have formed part of the same scheme).8 The walls of the fort were 10 feet thick at the base and reduced by at least two internal offsets each 12 in. deep. The walls were rounded at the corners and contained about seven and a half acres ; a rampart bank 50 ft. wide reinforced the wall and the intervaUum road, 18 ft. wide, ran along the foot of the bank. Three gateways and possibly a fourth gave access and at least two ditches (but no external bastions or bonding tiles) are known. Associated with the fort's construction were two ovens. 1 When Oven I was first found in 1953 the writer regarded it as evidence of pre-fort Roman settlement on the grounds that it was covered by the rampart bank. It is now clear that the oven and bank are contemporary with the fort wall. 2 Although the pottery from the several layers comprising the bank is uniform in date an excavation elsewhere inside the fort, to endorse the findings, is desirable. 3 The presence of a fort at Reculver from this date may well explain why the Roman town at Canterbury remained unwalled until the late third century. The construction of the fort may also have effected the degree of occupation at the nearby Richborough site. 108 RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 The area west of the fort has yielded large quantities of pottery and other debris, including many coins. No structural remains have been revealed, however, but twelve weUs, hearths and numerous rubbish-pits and other scatter indicate that this area was extensively used. The remains revealed centuries ago to the north of the fort probably represents the fort's external bath-house ; whilst the other remains are similar to those discovered west of the fort in recent years. Indeed the similarity existing between the recently discovered weUs and Battely's Cisterns has aheady been noted. It is fairly certain, therefore, that an extra-mural settlement existed at Reculver on the two landward sides of the fort. The lack of structures, other than the probable bathhouse, indicates that such settlement was not of a substantial or permanent nature. Phase II There is evidence for a period of renewed activity on the site dating from about the middle of the third century. An analysis of the earlier findings had suggested a period of inactivity during some part of the third century A.D. The building in Section B was probably constructed during the later part of the century and waUs uncovered by Major Gordon Home1 indicated two building phases. It is important to note that the fort at Brancaster also witnessed a period of activity at about this time. Such activity suggests some degree of reorganization ; a reorganization which may have been responsible for the temporary earth-fort at Richborough and later the full development of the coastal defence system. The evidence indicates that the fort was occupied during the fourth century when the repair to the east waU was probably carried out. Unlike the fort at Richborough, Reculver has produced very few coins of late-Roman date ; perhaps troop economies caused the fort to be vacated at about the time the garrison of the north and west were being withdrawn, sometime before the close of the fourth century. 1 Letter to The Times, 18th July, 1924. 109 APPENDIX I DATING OE KEY DEPOSITS Key Deposits I A.D. 160-190 I I A.D. 160-200 I I I (1957) (1953) A.D. 160-200 IV A.D. 160-200 V A.D. 160-200 VI A.D. 160-200 VII A.D. 200-230 VIII A.D. 300-360 Datable finds in deposit J a r A.D. 160-190. Ash from Oven I Samian P.37 Late 2nd century Samian F.37 Late 2nd century Samian F.31 Antonine Samian F.31 Antonine Jar A.D. 160-190 Rhenish beaker Not before A.D. 175 Pie-dishes (6) Antonine/late 2nd century Jar 2nd half 2nd century Glass A.D. 160-190 Beaker c. A.D. 200 Jar Antonine Rubbish Layer I Samian F.37 Late 2nd century Rhenish beakers (2) Not before A.D. 175 Pie-dishes (6) Antonine/late 2nd century J a r 2nd half 2nd century Samian F.37 A.D. 160-180 Samian F.33 Antonine Samian F.31 Antonine Pie-dishes (10) Antonine/late 2nd century Needle 2nd century Rubbish Pit Samian F.31 Antonine Beaker 2nd/early third century Pie-dishes (5) Antonine/late 2nd century Oyster and mussel shell layer Jar 2nd half 2nd century Pie-dishes (2) Antonine/late 2nd century Clay above shell layer Beaker 2nd-early 3rd century Rubbish Layer II Coin, Elagabalus A.D. 219-222 Samian F.45 Late 2nd/early 3rd century Rhenish beaker c. A.D. 200 Jar Late 2nd/3rd century Mortarium Ash and silt sealing tail of rampart bank and intervallum road. Coin, Carausius A.D. 287-293 Coin, barbarous radiate Late 3rd century Coin, Constantine c. A.D. 330-335 Pie-dishes (4) 4th century Bowl Cooking vessel Beaker Castor ware (3) Late 3rd/4th century Phase I I Stone Building Reference Fig. 4, No. 1. Fig. 4. No. 2 Fig. 4. Nos. 3,4 Fig. 4. No. 9 App. I I I . No. 6 Fig. 4. No. 18 Fig. 4. No. 13 Fig. 4. Nos. 3,4 Fig. 4. No. 10 Cinnamus or associate Fig. 4. Nos. 3,4 App. III. No. 11 Fig. 4. No. 15 Fig. 4. Nos. 3,4 Fig. 4. Nos. 11, 12 Fig. 4. Nos. 3,4 Fig. 4. No. 16 App. I I I . No. 2 Fig. 4. No. 8. Fig. 4. No. 7 App. III. No. 3 App. I I I . No. 4 App. III. No. 5 Fig. 4. No. 5 Fig. 4. No. 6 Fig. 4. No. 14 Fig. 4. No. 17 110 RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 APPENDIX II COARSE POTTERY (FIG. 4) Truncated rim of hard grey ware. Very similar to a vessel from sealed levels in WeU G, dated A.D. 140-180. Fairly simUar types occur in late-Antonine deposits at Lullingstone (Arch. Cant., LXVI (1953), 26. Nos. 101-103) and at St. Albans (Verulamium, p. 184. No. 17), dated A.D. 160-190. (Key deposit I—A.D. 160-190.) Rimsherd of jar simUar to No. 1, but heavily undercut. Grey ware. (Key Deposit II—A.D. 160-200.) Pie-dish of burnished grey ware. Heavy roUed rim. Chamfered base. (Key Deposits II-V—A.D. 160-200.) Straight sided pie-dish, grey-black ware. Slight demarcation below rim. (Key Deposits II-V—A.D. 160-200.) Pie-dishes are the most frequent coarse ware type found at Reculver. Some 29 examples were recovered from the rampart bank, 19 having the pronounced bead-rim and the other 10 ending with a plain lip. All are straight sided and are black or brown in colour. Only one has a burnished treUis pattern on its exterior. Most of the wells west of the fort contained these dishes and many other examples have come from the beach. WeU G produced several in sealed layers dated A.D. 140-180. At Canterbury (Roman Canterbury 3, Fig. 7, No. 2. and Fig. 15, Nos. 11-13), these dishes are dated to the mid and late second century. At Lullingstone (Arch. Cant., LXVI (1953), 27-29, Fig. 4, Nos. 107-112 and Fig. 5, Nos. 113-117) to the Antonine period (Fig. 5, Nos. 134- 135) and Hadrianic to mid-Antonine. Chalk (Arch. Cant., LXVIII (1954), 152, Nos. 11-13) produced examples dated Antonine, and in Joydens Wood (Arch. Cant., LXVIII (1954), 180, Nos. 28-31), the dishes were found in association with mid-second century samian. At Springhead (Arch. Cant., LXXI (1957), 71, Nos. 10, 11 and 13), they are dated to the Antonine period and noted as being very common A.D. 150-200. St. Albans (Verulamium, p. 193, Nos. 49-52 from the triangular temple), has examples dating to the late second century. Canterbury (Arch. Cant., LXVIII (1954), 121, Nos. 98 and 100), again has examples of the Hadrian-Antonine period. The Roman cemetery at Ospringe also produced a number of these dishes. Flanged dish of hard grey-brown ware. A typical fourth century type. (Key Deposit VIII—A.D. 300-350.) Flanged bowl with high rim. Red-brown ware. Not a common form. (Key Deposit VIII—A.D. 300-350.) I l l RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 Ti V K Ss LU-LULU ,K,7 8 T ^ 3?--\0 ? \2 r u \L 1 5 / ^ f~u16 In -1j7 is PIG. 4. Coarse Pottery. 112 RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 7. Mortarium of black ware with rounded rim. White slip on exterior. Unusual type. (Key Deposit VII—A.D. 200-230.) 8. Jar with cavetto-type rim. This is a common third century type but is known in late second century deposits. (Key Deposit VII— A.D. 200-230.) 9. Jar of grey-white ware. Burnishing on exterior, thickened outcurved rim. (Key Deposit II—A.D. 160-200.) 10. Jar with outcurved rim. Grey ware. A very simUar vessel was found in Group CLXIII of the Roman cemetery at Ospringe, No. 520 ; with it was a samian vessel dated A.D. 150-190. (Key Deposit III—A.D. 160-200.) 11. Jar, with roUed rim. Not unlike No. 10. (Key Deposit IV— A.D. 160-190.) Vessels very similar to Nos. 9-11 were found in numbers in the Roman cemetery at Ospringe where they are dated by association with samian pottery to the second half of the second and early third centuries 12. Wide-mouthed jar in grey ware. (Key Deposit IV—A.D. 160-200.) 13. Everted rim of small jar. Grey ware with black slip. Antonine. (Jewry Wall, Fig. 27, Nos. 42, 45, 50). (Key Deposit II—A.D. 160-200.) 14. Boldly outcurved rim of cooking vessel. Hard red ware. (Key Deposit VIII—A.D. 300-350.) 15. Rim of fine black beaker. Verticle rouletting on shoulder. Very simUar beakers occur in numbers in the Ospringe cemetery there dated from the second to early third centuries. (Key Deposit IV— A.D. 160-200.) 16. Rim of smaU beaker. Red ware with black glaze. Rim shghtly undercut. Zone of diagonal scoring on side. (Key Deposit VI— A.D. 160-200.) 17. Rim of small beaker. White ware with black slip. Profile simUar to No. 15. (Key Deposit VIII—A.D. 300-350.) 18. Indented beaker of soft brown ware. Examined and dated by Mr. M. R. HuU, M.A., F.S.A., c. A.D. 200. (Key Deposit II—A.D. 160-200.) APPENDIX III OTHER FINDS Coins 1. COMMODUS. A.D. 180-193. Sestertius. Worn. (Footingtrench of Phase I I BuUding.) 2. ELAGABALUS. A.D. 219-222. Denarius. R.I.C. 88. (Key Deposit VII—A.D. 200-230.) 113 RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 3. CARAUSIUS. A.D. 287-293. Antoninianus. R.I.C. 642 or 643. (Key Deposit VIII—A.D. 300-350.) 4. Barbarous radiate. Late third century. (Key Deposit VIII— A.D. 300-350.) 5. CONSTANTINE I. A.D. 330-335. M. 3. C.17. (Key Deposit VIII—A.D. 300-350.) Glass 6. Six fragments of fine glass from a beaker decorated with applique studs. Footring and possibly fine handle. Verulamium, Fig. 29, No. 26. A.D. 160-190. (Key Deposit II—A.D. 160-200.) 7. Green chip from large storage jar. (Key Deposit III—A.D. 160- 200.) 8. Green window-glass with beveUed edge for fitting in wooden frame. (Key Deposit VIII—A.D. 300-350.) Quern-stones 9. Fragment of scored lava-stone, 3 in. thick. (Key Deposit III— A.D. 160-200.) A simUar fragment of stone came from the topsoU layers and other traces have come from Wells K and M. and also Pit 9 from the west of the fort. 10. Upper stone of rotary quern. Sandstone. Concave grinding surface and lateral handle socket. Badly worn and damaged. Diameter greater than 15 in. (Key Deposit VIII—A.D. 300-350.) Miscellaneous 11. Needle. Bone, complete (4| in. long) with large eye. A typical second century type. (Key Deposit III—A.D. 160-200.) 12. Stylus. Bronze, length 4 in. From roof-fall of Phase I I Building. ABREVIATIONS Arch. Cant. Archozologia Cantiana. Arch. Jnl. Archaeological Journal. Arch, of Rom. Brit. R. G. CoUingwood. The Archceology of Roman Britain, 1930. Arch, of Sussex C. Curwen, Archozology of Sussex, 2nd Ed. 1954. Jewry Wall Excavations at the Jewry Wall Site, Leicester. Society of Antiquaries Research Report, XV, 1948. Jnl. Derby A. <& N. T.S. Journal of the Derbyshire Archceological and Natural History Society. 114 RECULVER : EXCAVATIONS ON THE ROMAN FORT IN 1957 Norfolk Arch. Ospringe P.S.A.S. Rich. V.C.H. Verulamium Norfolk Archceology. Excavation of the Roman Cemetery at Ospringe, Kent. Society of Antiquaries Research Report, VIII, 1931. Proceedings of the Society of Antiquaries of Scotland. Excavations of the Roman Fort at Richborough, Kent. Society of Antiquaries Research Reports, VI, 1926; VII, 1928 ; X, 1932 ; XVI, 1949. Victoria County Histories. Verulamium : A Belgic and Two Roman Cities. Society of Antiquaries Research Report, XI, 1936. 115