The Roman Villa in Cobham Park, near Rochester

THE ROMAN VILLA IN COBHAM PARK, NEAR ROCHESTER By P. J. TESTER, F.S.A. THE Cobham villa was discovered in April 1959, during the course of some digging undertaken by the Kent Archaeological Society for the purpose of determining the age of the Cobham Park earthworks.1 During the following August, a fortnight was spent in investigating the associated Roman features, and in 1960 three weeks were given to uncovering the main building. All the excavation was done by volunteer diggers working under the supervision of the present writer. Interim notes on the progress of the work have appeared in Arch. Cant., LXXIII and LXXIV. THE SITE The National Grid Reference of the villa is TQ 68326932. The building lay along a low ridge, running N.W.-S.E., composed of Blackheath and Woolwich Beds overlying Thanet Sand. To the south is a wide chalk valley and northward an extensive area of London Clay. The ridge is defined generally by the 350 ft. contour and the ground east of the villa rises to form a low hill reaching 390 O.D. Watling Street ran 300 yards to the north, its course being slightly south of the present main road (A.2).2 Along this highway 3£ miles to the east was the Medway crossing and the town of Durobrivae (Rochester), while 4 | miles in the opposite direction there existed a Roman settlement at Springhead, identified by the Ordnance Survey as Vagniacae. Part of the villa was found to have been cut through by a wide ditch which contains the boundary fence between the wooded hill to the east and the adjoining field (Fig. 1). This partial destruction probably took place as long ago as the seventeenth or eighteenth century and no record of what was found appears to have been made. CONSTRUCTION Three phases of construction were evident in the villa, as indicated on the accompanying plan (Fig. 2). The original building consisted of five rooms flanked by a corridor on the N.E. side, these being shown on the plan in sohd black. In constructing the footings, flint and chalk had been laid in the foundation trenches with clay in lieu of mortar, and upon 1 Arch. Cant., LXXIII, 224-5. 2 Victoria County History (Kent), III, 137. 88 R O M A N V I L LA C O B H A M PARK UNEXCAVATED TRACES OF COBBLED FORECOURT PIT PIT IV SITE OF DESTROYED *>*3 HEARTH HYPOCAUST PIT V PIT III SXZZZZZZZZZZZZZL FLINT & IRON SANDSTONE FOOTINGS tWOOOO MORTARED FLINT WALLS V/////A UNMORTARED FLINT FOOTINGS SCALE OF FEET | I M l I I I I I | IO IO 20 30 40 50 H.A.JAMES 8, P.J.TESTER I960 FIG. 2. [face p. SS THE ROMAN VILLA IN COBHAM PARK this, mortared flint, pebble conglomerate and iron sandstone were set, the last-mentioned material forming the outer faces (PI. I A). Nothing remained to show the original height of these walls for stone-robbing and ploughing had reduced them to or below the Roman ground-level. Possibly they merely formed the base for a timber-framed structure, though the amount of flint rubble covering the interior of Room 3 suggested that the adjoining walls had been carried up some way in this material. 1 -(NV 4" COIN HOARD 100 o WELL ^ > OUTBUILDING i 1 IOO jui B. Room 8. Inner face of S.E. wall. PLATE III A. Room 8. N.W. side showing (right) brick pier forming one side of hypocaust aperture. * % • ,;:' -\ B. Room 8. Remaining side of hypocaust opening, viewed from the north. PLATE IV. I I • 1 | | A. Examples of timbers used to form lining of well. B. Boards from bottom of well. THE ROMAN VILLA IN COBHAM PARK one period by a thin layer of mortar, and from the ash below came several sherds of coarse pottery (Fig. 5, 28-31). Just outside the S.E. wall of Room 8 there was discovered an iron joint-ring of a type used in Roman times for hnking together wooden water pipes. Most probably, therefore, the water for the bath was conveyed by this means from a source on the hillside to the east, possibly that which still exists in the form of a pond 80 yards east of the villa, as shown on the O.S. map (25 in. Kent Sheet XVIII, 4). Wall plaster was found in scattered fragments, the main concentrations being on the S.W. side of Room 4, and immediately adjoining the N.W. side of Room 8, the latter having obviously been related to the destroyed hypocaust-heated room. One or two fragments with a plain green surface were found in the area of Room 2. Room 4 produced numerous pieces with parallel bands painted in red, yellow, green, light blue and purple-brown, mainly on a cream background. There were several pieces suggesting polygonal panels defined by broad purplebrown borders, and also lines of dark green on a lighter background of the same colour which seemed to represent long, narrow foliage. The assemblage recovered from the N.W. side of Room 8 included many pieces with a pink background splashed with blood-red. Parallel bands of red and green on a cream background were common. One fragment had what appeared to be part of a palm tree trunk in yellow and black. Another bore small open flowers—possibly roses—in dull crimson and pink, set among amorphous foliage. In both main groups the plaster was thick and contained much coarse sand, except on the painted faces. A few fragments of window glass recovered may be divided into two classes : (a) Translucent, greenish-blue glass, rough on one side and about •j- in. thick. (b) Transparent, streaky glass, -f% in. or less in thickness. A piece of (a) was found in Room 2 and fragments of both types came from just N.W. of Room 8. THE PITS Pit I occurred in the corridor and was 7 feet deep and 3 feet in diameter. Its filling was relatively clean and it was evidently not a rubbish pit, though it could have been a storage pit or a sump. At the bottom was a quantity of large chalk lumps and a tegula, the latter proving that the pit was not filled before the building of the villa. Pit I I was similar to the last and had the same mass of chalk at the bottom. A few small sherds occurred in the upper filling, probably of late first or early second-century age. A portion of the chalk floor of the corridor was noted as sealing this pit, but this does not necessarily imply 91 THE ROMAN VILLA IN COBHAM PARK that the pit was filled before the erection of the villa as the chalk floor may well have been a secondary feature, as it certainly was in Room 3 adjoining. Pit III contained domestic rubbish including animal bones and pottery of the second century. The association of this in the filling with loose tesserae and tile fragments is of significance and is discussed below. Pit IV occurred on the N. side of the furnace room and may have served as a sump for the adjoining bath. The filling was clean and in the upper part was the rim of a third or fourth-century mortarium (Fig. 6, 36). Pit V was 2 feet in diameter and 4 feet deep. It was tightly packed with chalk lumps except for a hole 9 inches in diameter through the centre. It had the characteristics of a post-hole with packing but its precise purpose in relation to the building could not be determined. COBBLED AREA Limited excavation on the north side of the villa showed that outside the central portion there was a wide area of closely packed flints which could be regarded as a cobbled forecourt, possibly at the head of a trackway leading from the house to the Roman road 300 yards to the north. We were not able to define the lateral limits of this flint spread but it seemed to end about 17 feet N.E. of the corridor. THE OUTBUILDING1 This was a detached rectangular building, 45 feet by 19£ feet externally, with unmortared flint footings very similar to those at the N.W. end of the villa itself and most probably of the same period. A photograph of a typical length of these footings was reproduced in Arch. Cant., LXXIII, 224, and may be compared with PI. IB of this report. The outbuilding had no floor except the sandy soil, which was strewn with a few tile fragments and potsherds. The latter seem to be of the second and third centuries and there was an entire absence of earlier material. Several pieces of broken Roman tile were observed bedded into the footings, as in the case of Rooms 5, 6, and 7, and there were traces of clay between the flints. The purpose of this outbuilding is uncertain but the absence of hearths or concentrated domestic rubbish suggests that it was a storehouse or cattle-shed rather than a human habitation. THE WELL A brief note on this interesting feature with a photograph of the timber lining in situ has already been given in Arch. Cant., LXXIII, 224. 1 The correct Nat. Grid Reference of this building is TQ 68296933 and not that given in Arch. Cant., LXXIV, 177. 92 THE ROMAN VILLA IN COBHAM PARK Its depth was no more than *1\ feet and when its contents were removed the lowest part rapidly filled with about a foot of water. The upper part of the shaft was choked with a mass of material composed of flints, clay, animal bones and potsherds. If the flints and clay were evidence of a collapsed steining, similar to that of the Lullingstone well (Arch. Cant., LXVI, 15), it must be assumed that its disintegration took place pari passu with the use of the abandoned well as a rubbish pit, as otherwise it is difficult to account for the domestic rubbish interspersed with the flints. At the bottom there was a framework of oak1 timbers forming a lining approximately 2 | feet square. These planks were triangular in cross-section, having been split radially from a trunk or branch and not sawn (PI. IV A). They were half-jointed at the corners of the lining, a generous overlap being allowed in both directions at the points of intersection. The lining occupied the lowest 2 feet of the well and in general form was very similar to that illustrated in Camulodunum, 127, Fig. 38. At the bottom were three sawn oak boards (PI. IV B) placed against the inside of the lining formed by the other planks, as though to reinforce them. They were half-jointed at the corners but the fourth board necessary to complete the construction was unaccountably missing. In the silt which filled the bottom 2\ feet of the well were found four vessels, one unbroken and the rest shattered but capable of reassembly2 (Fig. 7,38-41). These were most likely pots used for drawing water and lost down the well accidentally. Their forms are not closely datable and the most that can be said is that they indicate a second-century use of the well. The matter is discussed in more detail below and it is only necessary to add that the rubbish-sherds in the upper filling, which included some Samian, were also typically second-century. Pieces of roof- and hypocaust-tile in the filling confirmed that the neighbouring buildings were in existence by that period. DATING The pottery recovered from the site indicates occupation from the second half of the first to the mid-fourth century, but owing to the slight depth of soil covering the floors and footings of the viha—less than a foot in places—no value can be attached to the relative stratigraphical positions of any coins or pottery found in the upper levels, particularly as the site has been repeatedly ploughed over a long period. Detailed dating must depend, therefore, on finds from a small number of sealed contexts which are here described. 1 Samples of the timber were identified by The Timber Development Association and the B.M. (Nat. Hist). An attempt to date the wood by tree-ring analysis was unsuccessful. 2 Our member, Mr. L. C. Dale, kindly undertook the restoration of these vessels. 93 THE ROMAN VILLA IN COBHAM PARK r * r > / p 13 15 *ri j 10 1 A ^ 12 \ : 14 >////////////// 16 / 17 18 /I 19 INCHES 20 PJT Era. 3 First-century pottery. 94 THE ROMAN VILLA IN COBHAM PARK Plate I IA shows a section cut on a line N.E.-S.W. through Room 4, which provides an important key to the early history of the site. At the base, resting on natural Thanet Sand, was a dark sandy layer (D) up to 10 inches thick, containing abundant pottery, animal bones, charcoal and pot-boilers. This early occupation-layer was partly below the level of the footings of Room 4 and must belong to a period before the villa was built. Above this, and sealing it, was a band of clean sand (C) which can be regarded as spoil dug out in cutting the foundation trenches of the adjoining walls and spread about to make up the level preparatory to laying the floor of the room. Upon this rested a band of dark soil (B) containing loose tesserae and fragments of mortar from the ploughedout floor. This was in turn sealed by about 1| feet of material (A) thrown up in digging the boundary ditch previously referred to, and containing hypocaust flue-tiles, plaster, burnt material and other remains of the building destroyed by this operation. The pottery from the lowest stratum of this section (D) is described in detail under separate heading (Fig. 3, 4-16) and the conclusion is drawn that it belongs to the latter part of the first century A.D. NO tiles or other indications of a Roman-type building were noted in this early level and it might be interpreted as part of the floor of a nativetype hut or farmhouse standing before the villa was built and destroyed to make way for it. Similar evidence has come from several other villas in S.E. England.1 From the sealed ash of the small hearth in Room 3 came a rim of native ware (Fig. 3, 19) very similar to one found in the pre-villa layer under Room 4 (Fig. 3, 7), a discovery which suggests continuity of occupation between the native settlement and the villa which replaced it, and this conclusion is supported by a coarse rim (Fig. 3, 18) from the foundation trench of the wall between Rooms 1 and 2, associated with part of a cordoned butt-beaker (Fig. 3, 17). Pit III, just outside the wall of Room 6, contained near the bottom, 4 feet down, a Samian stamp of CARANTINUS whom Oswald <& Pryce date A.D. 75-110. There was also coarse ware which can be referred to the second century (Fig. 4, 21-27). An important fact is that this pit contained throughout its entire filling a number of loose tesserae and tile fragments indicating plainly that the villa was in existence by the second century. Rhenish ware and an unworn silver coin of Severus Alexander from the junction of Room 8 with the vanished bath-house show that this was in use in the third century. From under the sealing of the ash-pit mentioned earlier came several rims (Fig. 5, 28-31) typical of the late second century, which may put the first use of the hypocaust back to that period. 1 Notably Park Street (Arch. Jour., CII) and Lockleys (Ant. Jour., XVIII). 95 THE ROMAN VILLA IN COBHAM PARK VI INCHES FIG. 4. Pottery from Pit III. 28 ^ 2 9 ^ 30 31 f 32 33 34 i INCHES FIG. 5. Pottery from Room 8. Fourth-century pottery came from the upper levels in Rooms 1 to 7 and in the upper filling of Pit IV. Coins of the early part of this same century turned up in Rooms 1 and 4. Room 1 also yielded a coin of Victorinus (A.D. 265-8) and Room 2 contained one of Faustina II (A.D. 161-176). The latest coin on the site came from a point a few yards north of the villa in an unrelated context and is not earlier than A.D. 354. THE ROMAN VILLA IN COBHAM PARK Uncertainty exists as to when Rooms 5, 6 and 7 were added, as pottery found in association was not sealed or well stratified and covered the whole period of occupation of the site, the rubbish-survivals being indistinguishable from contemporary sherds. The footings marked 7 were 2-3 feet from the present surface and consisted of one course of flints laid upon chalk and broken tile (PI. I B). No continuation of the N.W. wall could be traced beyond the point indicated on the plan. The footings of 5 were set about a foot higher than those of 7, though the stratigraphical significance of this is uncertain. Possibly the footings of 7 were those of an uncompleted wing, either planned to replace 5, or alternatively abandoned before completion in favour of a less ambitious extension at this end of the house represented by 5 and 6. GENERAL OBSERVATIONS The overall picture obtained from the archaeological evidence is of an unpretentious Romano-British farmhouse, built c. A.D. 100 on the site of a first-century native settlement, and continuing in use for a further two and a half centuries. In the course of this period the original simple corridor-house received additions in the form of a small bath-house on the S.E., and humble apartments—possibly slaves' quarters—on the N.W. Coin evidence would suggest its abandonment before A.D. 360, as it is noteworthy that the latest coin in the hoard found only 100 yards away in 1883 (Fig. 1) was of Decentius (A.D. 350-353). Roach Smith .expressed the opinion (Arch. Cant., XV) that the hoard was probably buried for safety during the upheavals of that period, and one might extend this conclusion by suggesting that the coins (836 in all) were buried by the last occupier of the villa shortly before disaster overwhelmed the settlement. A farm on the side of a main highway would fall an easy prey to bands of marauders. An unusual feature is the apparent N.E. aspect of the house. No trace of a corridor could be found on the opposite side and had the villa faced S.W. it would have fronted slightly uphill—an unlikely arrangement— as the building lay just N.E. of the crest of the ridge. In the N.E. direction the house commanded an uninterrupted view down a gentle slope to the Roman road. The central room (3) lacked all evidence of the luxurious appointments commonly associated with the main room in this position in many other Roman villas. It had no traces of wall plaster and its original primitive clay floor bore marks of fires and a scatter of domestic refuse. Possibly it served as a combined living-room and kitchen such as might be expected to occur in the simpler forms of Romano-British dwelling.1 1 See Miss Joan Liversidge's remarks on " Kitchens in Roman Britain " in Archaeological News Letter, Vol. 6, No. 4 (1957). 97 12 THE ROMAN VILLA IN COBHAM PARK At the S.E. end of the range were the " best " rooms with elaborately decorated wall plaster and adjoining baths, in contrast to the unfloored simplicity of 1, 5, 6 and 7 at the opposite end. The plan of the original house at Cobham may be compared with the first-century villas at Park Street near St. Albans, and Lockleys near Welwyn, both of which overlay indications of earlier settlements, and were laid out in the first instance without wings. A significant difference is, however, that whereas Lockleys originally had only a simple verandah supported on timber posts, the Cobham villa apparently had a fully developed corridor as an integral part of its primary layout. This accords with the relative dating, for Park Street and Lockleys were slightly earher than Cobham. At Lullingstone the first villa seems to have been a range with corridors on both sides, and has been dated c. A.D. 90.1 All the materials needed for the buildings could have been obtained from the immediate vicinity of the villa, the only probable exceptions being iron nails and window glass. Flint, chalk, clay and timber were all to hand within a few hundred yards of the site. The iron sandstone and pebble conglomerate used extensively in the footings of the earhest part of the villa came from the local Tertiary beds. Samples were kindly examined by the Geological Survey and Museum who confirmed that ferruginous pebbly conglomerate has been observed in sections at Shorne Common Rough and actually in Cobham Park itself, and is known to form a constituent of the Blackheath (or Oldhaven) Beds. Similarly, the iron sandstone is obtainable from the Woolwich Beds and occurs in seams up to a foot thick in the Shorne outlier not far north of the villa. The economic basis of the settlement was probably agricultural as there were no signs of industrial activity. From the fact that the villa succeeded a native settlement it might be inferred that the light soils of the ridge on which it lay and the valley-slope to the south were cultivated on the Celtic principle of small square fields, but no trace of these can now be detected. The proximity of the main road and the consequent accessibihty of Rochester would enable the products of the estate to find a ready market. Evidence of the enlargement of the building and the addition of the bath-house would suggest that a moderate degree of prosperity was enjoyed by the owners of the villa. EVIDENCE OF MEDIEVAL ACTIVITY Several unstratified sherds of undoubted medieval pottery were found in the area of the villa. These are most likely evidence of sporadic stone-robbing which appears to have been directed mainly at recovering the tiles and sandstone. The earhest datable sherd is a cooking-pot rim 1 G. W. Meates, The Lullingstone Roman Villa (1965), 63. 98 THE ROMAN VILLA IN COBHAM PARK of reddish-brown, shell-filled ware, in form very similar to twelfthcentury examples from Canterbury figured in Arch. Cant., LXVIII, 132, with the additional feature of finger-tip impressions around the top edge, as on the rim from Dover shown in Arch. Cant., LXIV, 144, No. 3. A stabbed jug handle of unglazed grey ware may be assigned to the late thirteenth century together with a cooking-pot rim of similar material, of the same form as the Canterbury example shown in Arch. Cant., LXVIII, 133, No. 29. In the soil over the remains of the outbuilding, a small bronze buckle was found, resembhng one figured in the London Museum Medieval Catalogue (1940) PI. LXXV, 4. DESCRIPTION OP THE FINDS The landowners have kindly consented to the finds being added to the Society's collection at Maidstone Museum. COINS1 1. Marcus ^E as. Undated issue of FAUSTINA I I . A.D. Aurelius Obv. Bust of Faustina, r. 161-176 Rev. Fecunditas draped stg. f. B.M.C.R.E. IV Marcus No. 980 (PI. 74, 8). Found unstratified in Room 2. 2. Severus AR den. Obv. IMP C M AVR SEV ALEXAND AVG. 222-235 Alexander Rev. AEQVITAS AVG stg. 1. Under debris on N.W. of Room 8. 3. Victorinus M ant. Obv. IMP C VICT(ORINVS). Head rad. r. Rev. PROVIDENTIA AVG stg. 1. 265-268 Room 1, unstratified. 4. Maximinus I I M fol. Obv. MAXIMINVS NOBILISSIMUS CAES. 305-307 Bust draped, laur. r. Rev. GENIO POPVLI ROMANI. Genius stg. facing, head to 1., holding patera r. and cornucopia 1. No mm. Attributed to London. Room 4, layer B. 5. Constantino I M Obv. CONSTANTINVS P F AVG. 308-337 Head laur. r. Rev. SOL INVICTO COMITI. Sun god stg. 1. mm PLN (London). Unstratified in area of Room 1. 6. Constantius I I M Barbarous copy of FEL TEMP REPARATIO. 354 or Obv. Head of emperor, diademed r. later Rev. Horseman falling from horse, raising arm towards figure who spears him. Outside villa, just N. of corridor. POTTERY Only a small proportion of the abundant pottery recovered from the excavation can be illustrated and described in detail. Much of the 1 The British Museum Dept. of Coins and Medals kindly identified Nos. 1, 4 and 6, and supplied information incorporated in these notes. 99 THE ROMAN VILLA IN COBHAM PARK material was poorly stratified but the eight groups described below were recovered from well-defined contexts and provide important dating evidence for the features with which they were associated. The unsealed sherds found in the soil overlying the floor-level of the villa contained mainly second to fourth-century types, mostly of commonplace character. They included Patch Grove ware and Charlton bead-rims, flanged bowls (Richborough type 121-2) and red-coated wares with decoration in white slip (e.g. Lullingstone, 34 and 35). Samian ware was fragmentary and is only listed where it has a bearing on the age of the associated coarse pottery. The only potters' stamps recovered were those of LOGIRNUS and CARANTINUS. GROUP I (Figs. 3,1-3). Three rims of coarse native ware were found in a shallow, dark-filled depression just outside the corridor at its N.W. end. As a group they provide the earliest evidence of occupation of the site, before the erection of the villa, and should not be much later than mid-first century. Pitting on the surface indicates weathered-out shell-filling. All were either hand-made or turned on a slow wheel. 1. Bead-rim with lid recess. 2. Simple out-turned rim as found in E.I.A. ware at Bigbury (Arch. Cant., XLVIII) and Oldbury Hill (Arch. Cant., LI). 3. Slightly thickened rim with lid recess. GROUP II (Fig. 3, 4-16). All the sherds in this group came from the sealed pre-villa occupation-layer under Room 4. A few splinters of Samian ware were in association, and the probable age of the assemblage is c. A.D. 70-100. Although native wares and forms predominate, the inclusion of Romanized elements such as 13 and 14 make an earlier date unlikely, while on the other hand there is a complete absence of essentially second-century forms. 4. Bead-rim with lid recess. Grey ware with reddish surface and shell-filling. 5. Similar to the last but with no shell. Grey ware with pinkish surface. 6. Bead-rim of a form which Dr. Kathleen Kenyon includes in the Charlton type (Southwark, p. 57, 25). Grey ware with reddish surface and traces of shell-filling. 7. Rim with two parallel grooves below the internally thickened lip. Seemingly an E.I.A. South-eastern B derivative, comparable with Crayford, Fig. 9, 1-4. Black ware. 8. Part of everted rim with low cordon. Hard grey ware. 9. Upper part of globular vessel of a type represented in the Belgic cemetery at Stone (Stone, Fig. 3, 1-2). Grey ware with blackcoated surface. 10. Carinated bowl. Hard grey ware with reddish surface. Cf. Richborough 14: ; first-century. 100 THE ROMAN VILLA IN COBHAM PARK 11. Plain everted rim, rising above ridged shoulder and resembling the " necked jars " figured in Southwark, p. 54. Dark ware -with burnished surface. 12. A common form of first-century platter with internal step. Hard grey ware. Of. Camulodunum, form 21B and 21D. 13. Flagon neck. Orange-red ware with white slip. Cf. Richborough, 34-36. Mid- or late first-century. 14. Platter with black, burnished surface and decoration of scored oblique lines. At Colchester this type of vessel is not earlier than A.D. 64 and covers the period c. A.D. 70-180. (Colchester, p. 29.) 15. A native-type beaker imitating Samian form 30. A band of rouletting above the angle formed at junction of side and base, and another band below—both eroded. Hard grey ware. A further joining fragment of this beaker was found outside the villa associated with part of a Samian base bearing the stamp of the Flavian potter LOGIRNUS. 16. Bowl of red ware with polished surface in imitation of Samian. GROUP III (Fig. 3, 17-20). The four sherds comprising this group were found in situations which suggest that they only slightly antedated or were contemporary with the erection of the earliest part of the villa. When compared with Group II, they indicate that no appreciable interval occurred between the abandonment of the native settlement and the building of the villa on its site. 17. Butt-beaker of gritty, grey ware with reddish-brown surface. A common first-century Belgic form in S.E. England. Cf. several examples from Cheriton figured in Arch. Cant., LXII. Recovered from foundation trench of wall between Rooms 1 and 2. 18. Bead-rim with lid recess and parallel grooves. From same situation as the last. Black ware with shell-filling. 19. Very similar to 7 above. From the ash of the tile-edged hearth in Room 3, sealed by later chalk floor, and therefore contemporary with the first phase of the villa. 20. Bead-rim vessel with stabbed decoration. Found immediately under the clay floor of Room 3. Grey ware with brown surface and shell-filling. GROUP IV (Fig. 4, 21-27). Pit III produced a quantity of coarse ware of which the seven vessels figured are representative. At the bottom was a Samian base with the stamp CARANTINIM. According to Oswald & Pryce, Carantinus may be dated c. A.D. 75-110, but here the associated coarse wares seem to extend beyond this, though not necessarily outside the second century. The complete absence of the common third and fourth-century flanged bowls (e.g. Fig. 5, 34) is significant. 101 THE ROMAN VILLA IN COBHAM PARK 21. Complete vessel found in the upper filling of the pit. Grey ware with polished zone above base and around neck and shoulder. Decoration of scored wavy line. Colchester type 299 which ranges from c. 150 to 350. 22. Cavetto-rim jar. Grey ware with polished brown surface and scored lattice decoration. Probably late Antonine. 23. Patch Grove storage jar. Grey ware with dull orange-buff surface and stabbed decoration. Very common ware in N.W. Kent during second century. Cf. Joyden's Wood, 2-4, from a second-century context. 24. Platter of grey ware with black surface. A common and longlived type. A second-century example is Joyden's Wood, 29. 25. Necked jar of grey ware. Cf. Southwark, Fig. 16, 3, 4 and 6, and there dated late first or second-century. 26. Platter, diameter c. 9 in., of grey ware with black surface. A close parallel is Southwark, Fig. 15, 25, which is described as probably third or fourth-century. On evidence of association this Cobbam example is almost certainly earlier. 27. Neck of grey-ware flask. Cf. Ospringe, 314, dated 100-180. GROUP V (Fig. 5, 28-31). Four rim-sherds were found in Room 8 at the firing-point of the hypocaust furnace. They were in black ash sealed by a thin layer of mortar. As they can be paralleled by secondcentury material they point to the hypocausts having been in existence by A.D. 200. 28. Moulded rim, similar to Richborough, 262-4, which are dated 80- 120. Black ware. 29 and 30. Examples of a type of bowl very common in the Antonine period. Dark ware. 31. Everted rim in black ware. Cf. Lullingstone, 102 and 103 ; late Antonine. GROUP VI (Fig. 5, 32-35). The continued use of the hypocaust furnace-room in the third century is attested by sherds on its floor, sealed by the debris of the collapsed walls. 32. Beaker of grey ware with traces of black coating. The rim has not quite reached the late form where it definitely overhangs the bulge. 33. Typical Charlton-type bead-rim, common in N.W. Kent. Although most frequent in the second century it survived into the third century in London and probably elsewhere (Southwark, p. 56). 34. Common type of third and fourth-century flanged bowl, many examples of which occurred at Cobham. This specimen had a black polished surface. At Colchester the form is not earher than A.D. 350 (Colchester type 305). 102 THE ROMAN VILLA IN COBHAM PARK 35. Rhenish ware indented beaker. Very hard, thin ware with greenish-grey metallic sheen. Third-century. 5^' : I « v • • 3 6 h=—^ _ » _ = • _ c = — INCHES FIG. 6. Pottery from Pit IV. GROUP VII (Fig. 6, 36 and 37). The comparatively clean filling of Pit IV produced the following vessels : 36. Red-coated mortar of late third or fourth-century type. Cf. Richborough, 98-9. 37. Diminutive, crudely-made tile-clay vessel. Either a toy or an unguent pot. GROUP VIII (Fig. 7, 38-41). At the bottom of the well were four complete pots figured herewith. 38. Hand-made vessel of grey-brown ware, polished around neck and base, and with wide lattice-decoration scored on body. 39. Pedestal-base vessel with poorly developed cordon on shoulder. Hard buff ware. No exact parallel has been traced but the Belgic-type pedestal-base suggests a date not later than c. A.D. 100. 40. A long-lived and common type, similar to 21 above. Gritty, dark ware. Probably second-century. 41. Thin-walled jar of black ware with decoration of scored vertical hnes. At Colchester the type may have originated as early as Flavian times. During the second and third centuries the form became progressively taller and narrower. The suggested date for this Cobham example is c. 150. MISCELLANEOUS FINDS Iron Joint-Ring. As already described, this was found on the S.E. side of Room 8. It is of the form illustrated in J. Ward's Romano- British Buildings and Earthworks (1911), p. 280, fig. 86, and consists of a hoop, internal diameter 3 inches, with a mid-rib or stop. As originally used, the sharp edges of the hoop were driven into the ends of the wooden pipes, the stop ensuring equal penetration. A further note on the use of these rings or collars, with a photograph showing them in situ in the ends of oak pipes recovered from the City, occurs in the London Museum catalogue London in Roman Times (1930). Roof Ventilator. A conical, tile-clay object, found in the tile scatter overlying Room 2, has been identified by Mr. A. W. G. Lowther, F.S.A., as the broken top of a roof ventilator. There are shght remaining 103 THE ROMAN VILLA IN COBHAM PARK ( 39 38 INCHES k I 40 ) 4 40 PJT FIG. 7. Pottery from well. indications that the sides of the ventilator were pierced with rosetteshaped openings and in this respect it resembles an example from Ashtead. Mr. Lowther intends to include a description of the Cobham fragment in a paper on the subject of roof ventilators which he is contributing to a future volume of Antiquaries Journal. 104 THE ROMAN VILLA IN COBHAM PARK ACKNOWLEDGEMENTS Permission for the digging was kindly given by the Rt. Hon. the Earl of Darnley and the Ministry of Works. Mrs. E. V. Piercy Fox lent the tools and Mrs. B. de Seyssel transported them. Mr. H. A. James did the surveying and supplied draft plans from which Figs. 1 and 2 have been prepared. It is impossible to acknowledge individually all the helpers who came for short periods, but the following must be mentioned for their sustained support : Mrs. J. D. Crawshaw, Misses G. Bath, J. S. Cross, M. Evans, P. Flood, D. M. Home, S. Horton, E. M. Mackinder, F. Marks, H. Middleton, Janet Parsons, E. Pirie, J. Pollard, R. H. Saunders, H. Smith ; Messrs. R. Acton, R. Chaplin, M. L. M. Clinch, J. D. Crawshaw, D. Dorrington, A. J. F. Dulley, C. R. Flight, J. Flood, R. G. Foord, I. Jackson, E. G. Lee, M. Ocock, R. Norris. ABBREVIATIONS Gamulodunum. C. F. C. Hawkes and M. R. Hull. Report of the Research Committee of the Society of Antiquaries, No. XIV. Colchester. Roman Colchester, M. R. Hull. Report of the Research Committee of the Soc. of Antiquaries, No. XX. Crayford. J. B. Ward Perkins. An Early Iron Age Site at Crayford, Kent. Proc. Prehist. Soc, 1938. Joyden's Wood. P. J. Tester and J. E. L. Caiger. A Romano-British Settlement in Joyden's Wood, near Bexley. Arch. Cant., LXVIII. Lullingstone. G. W. Meates. The Lullingstone Roman Villa. Arch. Cant., LXV and LXVI. Ospringe. W. Whiting, W. Hawley and T. May. The Roman Cemetery at Ospringe, Kent. Report of Research Committee of the Soc. of Antiquaries, No. VIII. Oswald & Pryce. Terra Sigillata. Richborough. J. P. Bushe Fox. The Roman Fort at Richborough (1st and 3rd Reports). Report of Research Committee of the Soc. of Antiquaries, Nos. VI and X. Southwark. Kathleen M. Kenyon. Excavations in Southwark. Research Papers of the Surrey Archaeological Society, No. 5. Stone. M. A. Cotton and K. M. Richardson. A Belgic Cremation Site at Stone, Kent. Proc. Prehist. Soc, 1941. APPENDIX A RECONSIDERATION OE THE COBHAM PARK OPPIDUM In 1877, Charles Roach Smith published in Arch. Cant., XI, 121-2, an account of his discovery of a " British oppidum " in Cobham Park, situated between the Hall and Watling Street. He considered that an 105 THE ROMAN VILLA IN COBHAM PARK area of about 20 acres had formerly been enclosed by a deep fosse and double vallum, remains of which were evident on the north and east sides for 300 paces. On the west they were also visible, but he observed that on the south they had been wholly levelled " for horticultural purposes ". This interpretation was amplified in Victoria County History (Kent),I (1908), where a plan was given, based on a survey undertaken by Colonel 0. E. Ruck in 1905. The Ordnance Survey have subsequently reproduced the features shown by Ruck, particularly on the 25 inch Kent Sheet XVIII, 4. As these are readily accessible, a plan of the area is not given here, but reference to either the V.C.H. plan or the O.S. map will be helpful in following the description below. North of Cobham Hall the land rises to form a hill, the summit of which is 390 feet above Ordnance Datum. Skirting the foot of the slope on the north and east is a wide and deep ditch bordered by banks on both sides. The north end of this crescent-shaped earthwork terminates on the edge of a large fish pond—one of a group which lies just inside the Park boundary south of the main road (A.2). The southern end of the crescent fades out as though its former extension had been destroyed by landscape gardening, as Roach Smith suggested. Across the widest part of the ditch, on the east, there is a causeway, closed at its outer end by iron gates probably erected within the last 200 years. On either side of the causeway the bottom of the ditch is deepened at intervals to form a row of pits—four to the north and three to the south. West of the hill, a straight ditch follows the boundary between the wooded Pleasure Ground1 and the field to the west. Ruck's plan shows an outer bank flanking the ditch at its north and south extremities. Westward from the straight line of this earthwork the plan indicates a semi-circular area defined by a downward slope and forming—according to the V.C.H. account—a possible annexe to the main defences. Although the " oppidum " theory seems to have gone unchallenged for many years, a cautionary note was entered by R. F. Jessup in his Archozology of Kent (1930) where he observed that nothing had ever been found in association with the earthworks to prove their prehistoric origin. But as recently as 1958 a popular work on the archaeology of south-east England retailed a suggestion that " the fort at Cobham was a Belgic tribal centre ". An investigation to determine the origin of the Cobham earthworks was undertaken by the present writer on behalf of our Society during the spring of 1959. The results are summarized herewith and, in-view of the conclusions formed, plans, drawmgs of the sections or detailed 1 Mr. A. J. A. Booth, Lord Darnley's agent, informs me that this is the name by which the hill is commonly known. 106 THE ROMAN VILLA IN COBHAM PARK presentation of the evidence afforded by the digging are considered superfluous. Two cuttings were made, as follows : (1) This was across the ditch and double banks of the crescent earthwork at a point about 100 yards from its junction with the pond previously mentioned. (2) The second cut was made through the bank and ditch on the west side of the hill, about 50 yards south of the northern extremity of the outer bank as shown on the 1939 revision of the 25 inch O.S. map. Cutting 1 was 73 feet long and 4 feet wide, intersecting the line of the earthwork at right-angles. Digging through the silt of the ditch had to be abandoned as the trench rapidly filled with water. Both banks were found to consist simply of dumps of clayey material apparently thrown up in digging the ditch, the outer bank being 5 | feet high above the line of the old surface buried beneath it, and 28 feet wide. The inner bank was only 2 feet high but owing to the steep natural slope of the ground its top was 2 feet higher than that of the outer bank. Its width was approximately 17 feet and the present floor of the ditch was 7 feet below its summit.1 In the body of both banks, and also sealed on the old surface below the outer bank, were numerous fragments of thin, red roof-tiles, of the plain rectangular type used from the thirteenth century to the present day. These are most certainly not Roman and are either medieval or later. One piece has remains of a tapering peg-hole, and others bear patches of glaze like that on the tiles from the Joyden's Wood medieval site described in Arch. Cant., LXXII. A clay tobacco pipe, poorly stratified under the tail of the outer bank, can be dated c. 1680-1720 by comparison with specimens from London.2 Nothing else of archseological significance was discovered. Cutting 2 immediately disclosed a quantity of Roman tiles and pottery in and under the banks, clearly indicating that this straight earthwork on the west side of the hill is not pre-Roman. This discovery led to the uncovering of the villa which, as described elsewhere, was revealed to have been cut through by the digging of the ditch, probably long after the building had ceased to present any surface indications. An inspection of the slope forming the west side of the " annexe " suggested that this was simply a natural feature accentuated by past ploughing on the hillside in such a way as to avoid the reservoir which still remains on the crest. DOCUMENTARY EVIDENCE The facts noted above disprove conclusively the supposed prehistoric origin of the Cobham Park earthworks, and the " oppidum " can 1 It should be noted that the earthworks increase in size in the southward direction. Typical profiles accompany Ruck's plan in V.C.H. 2 Archaeological News Letter, vol. 5, no. 12 (1955), p. 247, No. 8. 107 THE ROMAN VILLA IN COBHAM PARK accordingly be removed from the archseological record. We cannot, however, leave the consideration of this subject without reference to other evidence bearing upon the real age and purpose of the earthworks. A map of Cobham Park made by George Russell in 1718, and now preserved in the County Archives Office at Maidstone, shows a feature curving around the north and east sides of the hill on the line of the existing crescent earthwork. Along it are clearly written the words Conduite pipe. Its north end connected with the corner of the large pond, as does the earthwork, and its southward extension is shown terminating about 250 yards from Cobham Hall against the wall of an enclosed garden then situated north of the house. The pond at the north end is shown with a Fountain House on its banks, close to where the later pump-house (marked on the O.S. map) now stands. I am informed that this now-derelict pump-house, which is a red-brick structure of c. 1800, was formerly used to raise water through a pipe to the brick-lined reservoir in the field west of the hill, indicated by the words " Water Gauge " on the O.S. map. Iron pipes from this reservoir to Cobham Hall are still in position and were pointed out to me by one of the Ministry of Works engineers in 1960. The details of the matter are not at present fully understood, but the general evidence suggests that the ditch of the crescent earthwork served in comparatively modern times as a water channel, its flanking banks being incidental to its excavation and having no defensive purpose. Our digging showed that the ditch is still waterlogged and if scoured it would probably carry surface water from the hillside into the pond. Probably the causeway was made after the channel had gone out of use. A point of interest is that Thomas Norton's map of the Park (also at Maidstone), dated 1641, does not mark the conduit, a fact which accords very well with the 1680-1720 age of the clay tobacco pipe found in our excavation. Regarding the straight line of the western " defences ", these are obviously no more than a boundary ditch and bank, and old estate maps again provide a useful clue. The 1718 map of the Park shows broad, tree-hned avenues radiating from the front of Cobham Hall. All but one—that running to Cobham village—were later swept away by Humphry Repton when he transformed the Park for the 4th Earl of Darnley in the late eighteenth century. One of the vanished avenues ran northwards to Brewers Gate—the entrance to the park from Watling Street—and its east side followed exactly the line of the ditch once considered to form the western defence of the oppidum. The significance of this hardly requires further comment. A late seventeenthcentury date seems reasonable for the avenues as they were shown in 1718 but not by Norton in 1641. 108 THE ROMAN VILLA IN COBHAM PARK Ruck's original plan shows a feature of interest not reproduced in the V.C.H. version.1 On top of the hill he marks a mound which can still be made out among the trees close to the edge of one of the old sandpits. In case this should ever be mistaken for a barrow, it may be noted that it almost certainly formed a windmill-stead, for across this area on the 1718 map are written the words Windmill Hill. A windmill is shown in this position relative to Cobham Hall on P. Symonson's map of Kent published in 1596 (Arch. Cant., XXX, 85). Roach Smith suggested that traces of an old road running parallel to and just south of the main highway are indications of a pre-Roman track, but this seems unlikely. Some earthworks which lie along this hne west of Brewers Gate are probably only park boundary banks and are not evidence of the course of the Roman road or its hypothetical predecessor. 1 A blueprint of Ruck's plan was kindly lent me by Gravesend Library. This also shows the exact position of the 1883 Roman coin hoard described in Arch. Cant., XV. 109