The Lullingstone Roman Villa

THE LULLINGSTONE ROMAN VILLA THIRD INTERIM REPORT By LIETJT.-COLONEL G. W. MEATES, F.S.A. EXCAVATIONS during the 1952 season included preliminary work on a large basilican-type building towards the north-east, of which a full account is reserved for a subsequent report; and further investigation was made into the residence proper, particularly at the northern extremity and in the region of the Bath Block, where a well was discovered in close association with the latter. The present report is concerned with the complex of walls and floors forming the northern extremity, and with the well (plan at Fig. 1). In the case of the Northern Complex, which was found to have been reduced to floor level in Period III, it was necessary to remove all that was left in process of its investigation, two sections (Fig. 7) and two plans (Fig. 8) being recorded. This part of the site was then filled in. The well, however, was left open, although it was found necessary to remove the remains of the oak frame at the bottom. My thanks are accorded to Dr. F. Oswald, D.Sc, F.S.A., and to Mr. Eric Birley, M.B.E., F.S.A., who kindly reported on a fragment of Samian and a mortarium rim respectively, both from the well fillings and of vital importance to the dating problem. I am also most grateful to Mr. A. W. G. Lowther, M.A., F.S.A., who examined specimens of the timber framework from the well, and also the roller pattern impression on one of the clay walls of the Northern Complex, upon which he has commented at Appendix 3. Once again Mr. D. A. Broodbank, A.R.I.B.A., very kindly prepared the plan, while Mr. Ian Noble undertook to prepare the plans and sections for publication. My thanks are due to these gentlemen, as also to the team of helpers who once again rendered the greatest assistance throughout the season, among whom must be specially mentioned Mr. R. J. Rook for his careful section drawing and planning of the Northern Complex. Finally, my thanks are due to Mr. E. Greenfield for his co-operation in the work under report, and to Mrs. Macrory for drawing the pottery. THE WELL The problem of water supply to the Bath Block having previously remained unsolved, a close examination of the Roman level southwards of the Bath Block revealed a well, completely filled in in Roman times 15 THE LULLINGSTONE ROMAN VILLA EXCAVATIONS UP TO THE END OF I9S2 L-31£ Vlesl i d e c! read from Famln-him to LuUlpptone FIG 1 THE LULLINGSTONE ROMAN VILLA and situated 6 ft. south-west of the small square hot bath (Fig. 1). It was reahzed that more secure dating of the Bath Block might possibly be obtained by a careful dating of the adjacent well, and a methodical excavation of its contents was carried out over a period of weeks. This dating remains, however, inconclusive. The well seems to have been dug and used before the Period II occupation. Its use was probably short, and it remained open and in a disused state before the Antonine/ Third century upper filling was inserted. Its flint walls collapsed inwards during this time, and at the Period III reoccupation it was filled up and levelled. If the Bath Block was not supplied from the well during the Period II occupation, the source of supply has yet to be discovered, as also for the subsequent late third and fourth century occupation. The absence of Antonine rubbish in the lower levels, which might have been expected by its abundance elsewhere on the site, does seem to infer a lack of use of the well during that period, though access for cleaning out would have been easy at any time. The well was constructed as follows : (a) A circular shaft 6 ft. in diameter was excavated through the natural clay and flint hillwash to a depth of 9 ft. until the underlying chalk was reached. (6) A basin was cut in the chalk approximately 2 ft. in depth in the centre. (c) A framework of four interlocking oak planks was bedded down so that the four corners rested firmly on the chalk. These planks formed a square 4 ft. either way, the north and south planks being tenoned and the west and east planks being mortised. The former panaveraged 2 to 3 in. in thickness, while the latter pair were less than half as thick. This caused weakness in the framework, and it was found on excavation that the latter pair of planks had largely perished and could not be saved. The thicker pah were successfully removed, but it was found that they had been cut in such a way as to be useless for tree-ring analysis. All that could be said was that the tree from which they had been cut had grown on a very wet site. (d) A lining of flints, mortared directly on to the clay and flint, was built upon this wooden framework. It reached upwards for just under 3 ft., but for some reason was not continued to the surface. This lining was probably square, but all the flints had come away and formed a loose mass, and it was not possible to be sure of its shape. In many cases individual flints showed mortar on one side and blackening by contact with water on the other. (e) The top 45,- ft. of the well showed no sign of any lining, and it must be presumed that the whole was covered by a heavy wooden floor to prevent the destruction by weather of the clay and flint sides, and no doubt provided with a trap in the centre. There was, however, no 17 THE LULLINGSTONE ROMAN VILLA sign remaining of such a construction either on the Roman ground surface or in the fillings of the well. A study of the fillings (Fig. 2) indicates a gradual silting during use and a subsequent deposit of brown sandy clay, probably the result of storm water. The primary silting contained a fragment of Samian Form 33 (0. & P., LI, 10), which Dr. Oswald says is probably Domitianic in date (c. A.D. 90-100) ; and the brown sandy clay contained a mortarium rim (Fig. 4, No. 96) which Mr. Birley suggests may be attributable to the third century (see Appendix 2). A zone of grey sludge separated the primary silting from the brown sandy clay deposit, and this again was succeeded by grey sludge. This latter deposit of grey sludge appears to indicate the beginning of disuse, as it contained broken pieces from the wooden framework, and many fragments of pine cone, which seem to have fallen from a nearby tree, probably after the postulated wooden cover of the well had been removed. The objects from this grey sludge consisted of two bead rims of first century native ware (Fig. 4, Nos. 94 and 95) ; two coins, one burnt and indeterminate, the other indeterminate and much worn, though with a radiate head the features of which recall those of Hadrian ; and the iron hoop of a bucket of diameters 13£ in. and 14£ in. and height If in. A further fragment of the small bead rim vessel (Fig. 4, No. 94) was also found in the primary silting, which suggests a short period of use of the well. Immediately above this grey sludge, the filling consisted for some 3 ft. of flints fallen from the lining, and this filling contained the mouth and handle of a small globular amphora (Fig. 4, No. 97), a type which seems to have persisted into the Antonine period (Collingwood type 94). The filling was increased by some slipping of the natural clay and flint sides, and the whole was sealed down by a thick blackish layer of surface scrapings, mainly of late Antonine date (pottery illustrated at Figs. 4 and 5). This layer also contained a few fragments of third century pottery, which suggests that this final filling did not occur until the Period III occupation. Two of these fragments of third century pottery are well paralleled by two of three vessels from the roadside grave near the walled cemetery at Colchester (Arch. Journal, Vol. CI for 1944, p. 89, Fig. 10, Nos. 2 and 3), though these two vessels are there given a somewhat later date. The sparse material listed above from the lower fillings is all early in date, and the absence of Antonine material in the lower fillings suggests lack of use at that time and is not inconsistent with perhaps a Flavian date for the digging of the well. The point cannot yet be determined, and a further survey of all the dating material for the Bath Block must be made. The upper blackish and predominantly Antonine layer was sealed down in its turn by a filling of dry flints and brown clay and flint, above 18 THE LULLINGSTONE ROMAN VILLA which was constructed a rough basin of yellow mortar. This contained fragments of a pie dish (Fig. 4, No. 107), probably Antonine. The purpose of this rough mortar basin is not clear, but it is likely to have been associated with the reoccupation of the house in Period I I I , and W E L L S E C T I ON I'l'I'I'I'I'I'I'I'I'ITI'I'ITITITITn'I'I'I'I'ri'l'I'l' . HUMUS -CLAY AND FLINT -DARK ZONE WITH GEORGIAN TILE . CLAY AND FLINT ^ ^ ~ E PERIOD g SURFACE PERIOD E DEPOSIT PERIOD m/EARLY PERIOD DT SURFACE PIE DISH No 119 PIE DISH No. 107 NATURAL NATURAL CLAY FLNT CLAY FLINT F L I N T !ij!jl " t o WOOD FRAME AMPHORA RIM No 97 BEAD RIMS Nu.94-.9S MORTARIUM RIM No 96 SAMIAN 33C.AD.90-IOO '~^&yjf:f":k BEAD RIM No 94 PERIOD 12 GULLY PERIOD TX PIT EARLY PERIOD ET GULLY PERIOO D SURFACE FAULT GREY BROW FILLING CLAY ( FLINT FILLING MORTAR RUBBLE WITH ROUGH CONCRETE BASE SWEEPINGS CONTAINING £• GROUPS POTTERY GREY SLUDGE BROWN SANDY CLAY GRAVEL & SAND PRIMSILT -ARY P T ^ NATURAL CHALK Era. 2 in particular with the rebuilding of the adjacent Bath Block at that time. The levels above the well surface are associated with the long period of intensive occupation during the fourth century and call for no comment. It was noted on complete excavation of the well that water 19 THE LULLINGSTONE ROMAN VILLA entered quickly from all sides at the junction of the chalk with the clay and flint, this water being clear and pure, its level rising and falling in response to the rainfall. The well was dug to the approximate level of the bed of the river, which now flows some 200 ft. to the east, and this may have effect on the water level in the well. THE NORTHERN COMPLEX The destruction of this part of the earlier house by rebuilders late in the third century was a thorough affair, so thorough that little can be known of its original elevation, though its ground plan is sufficiently clear. Fortunately, floor surfaces were to some extent intact, and the foundations of the flint and mortar walls remained, together with the lowest parts of clay walls, decorated with painted plaster, to which detailed reference is made later in this paper, where comparisons are adduced from other sites. It will be remembered that the ground upon which the house was constructed slopes down at a gentle gradient towards the east. This has resulted in some destruction of the higher, western levels, but those to the east are capable of planning, interpretation, and a fairly close dating. The planning of the corridors, slopes, and rooms of which this complex consists were conditioned in both Periods I and II by the necessity of approach down to the Basement Room from the north. This complex ceased to exist late in the third century when the construction of the heated apartment (Room 15) resulted in the closing of the northern approach to the Basement Room ; so it is with Periods I and II that we are now mainly concerned, covering roughly from c. A.D. 100 to A.D. 200, when Period I I occupation came to an end, not to be re-established until late in the century. There seems to have been a lessening in the intensity of occupation between Periods I and II, owing to the paucity of material of Hadrianic and early Antonine date, but it is not yet possible to be sure of this, though the concentration of material on the one hand to the Flavian period and on the other to late Antonine times may be significant. It should be said that there is no clear destruction or abandonment level dividing the one from the other. The ensuing analysis of the Northern Complex should be read in conjunction with the plans at Fig. 8 and the sections LM and NP at Fig. 7. Period I I t will be remembered that the Basement Room was approached in this period by three ramps, the two leading down on the south side having had some sort of wooden erections fitted above them (Second Interim Report, Arch. Cant., Vol. LXV, p. 31), and having been pro- 20 THE LULLINGSTONE ROMAN VILLA vided with rammed chalk surfaces. The third ramp, under present discussion, was composed of rough concrete, but, unlike its two southern companions, its gentle gradient, slightly uneven (approximately 1 in 5), gives it the appearance of a walking way downwards from the north, much in the manner of the ramp approach to the cellar at Verulamium (Verulamium, p. 78). It was flanked by retaining walls, of which that on the east was only evident by a remnant beneath the third century cross-wall. The west retaining wall, however, was for the most part in situ and contained two large post-holes (Fig. 8), the only survivors of a line of probably five. The southern post-hole (No. 1) was found to have been modified at the Period II rebuild, and the evidence of the northern one (No. 5) was inconclusive, as at that point the ramp was joined by a continuation-ramp of Period I directed upwards towards the west. Post-hole No. 2 was destroyed when the footing trench was dug for the Period I I I cross-wall, but Nos. 3 and 4 remained in good condition. It may be that during Period I the walking ramp led directly north from the Basement Room and was confined as a corridor along its entire length by retaining walls provided with sockets for heavy uprights, although, if they had formed part of the east wall construction, there is now no evidence of them there. A wide and shallow Period III gully passes over the site of the possible exit from the corridor into the exterior, but the natural surface slopes up and away from the north side of this later gully and shows evidence of a rough surface of sandy gravel covered with a blackish, mixed occupation spread containing Period I material. It is possible that before the Period I I I gully was dug, this surface continued northwards directly from the presumed exit from the corridor. An iron brooch with fragments of poppyhead beakers, Patchgrove vessels and small native bead rims were recovered from this surface (Fig. 5). The brooch is of first century type (Camulodunum, PL LXXXIX, No.4, c. A.D.43 /44-48; Swarling, PL XII, No. 1; Maiden Castle, Fig. 85, No. 34, c. A.D.25 /50), and while it may have persisted, it tends to tie up the Period I ramp-corridor with the exterior level on which it was found. It is illustrated by Mr. Paul Rook's drawing at Fig. 3. Its associated bead rim pottery supports it as evidence for a Period I level in origin, but the Samian DIVIXTUS Form 37 fragment near it and a few Antonine sherds of coarse pottery suggest a level still open at their date. The western retaining walls were constructed of the usual flint and mortar, very solidly made, and that part of the natural slope upwards towards the west that had been cut away to take the footings was reinstated on the exterior of the wall with a packing of greenish clay, from which several fragments of coarse pottery of Flavian date were recovered, including sherds of poppyhead beakers and Patchgrove 21 THE LULLINGSTONE ROMAN VILLA vessels. There was a notable absence of Antonine material from this clay packing, and the post-holes were found to be devoid of datable material. There was no exterior gully here, and it is not certain that the curious rectangular enclosure on the west, with its steeply descending and rough surface, was or was not covered by any kind of roof. The best preserved of the post-holes is No. 3. It held an upright 9 in. square, its back having received a packing of flint and mortar with some of the green clay, thus filling the break in the wall, the total thickness of which is 15 in. w Eio. 3 (|) It would seem, therefore, that the Period I construction is of Flavian date, and consisted of a walled corridor leading downwards from the north into the Basement Room, the retaining walls being provided with slots to receive stout wooden uprights carrying a roof of some considerable weight. A covered entrance to the Basement Room was thus provided with a gentle gradient as a walking way. Period II It would seem that the reconstruction of the Period I corridor and ramp was contemporary with the insertion of the four-fold tiled stairways, which latter on other evidence appear to have been inserted late in the second century. The reconstruction involved the following : (a) The provision of two corridors with concrete floors, roughly level, leading respectively from the north and east steps, the former 22 THE LULLINGSTONE ROMAN VILLA overlying the disused ramp, with its surface merging into the earlier surface further along the corridor to the north. (6) The provision of a third corridor, also with a concrete floor, at the north and running parallel to that leading from the east steps, and turning at right angles into the northern corridor. (c) The enclosing of an apartment between these two corridors running east. This apartment was provided with a floor of crushed brick and is numbered 17 on the plan. (d) The possible provision of a concrete ramp, or possibly the reconstruction of one of Period I—this point is indeterminate—leading down from the west and turning at right angles into the northern corridor opposite the northernmost of the east corridors. (e) The provision, along the line of the east wall of the north corridor and turning east and along the north exterior wall, of clay walls upon puddled chalk and mortar foundations faced with flints. These clay walls were decorated with painted plaster and form an unusual feature of this part of the house. THE CLAY WALLS (PL II (b)) The existence of clay partition walls with painted wall plaster keyed on to their surfaces has long been suspected in the planning of Romano-British houses of the earlier period in Britain. The appearance upon the floors of some rooms in such houses of large quantities of clay, especially when found mixed with many fragments of painted wall plaster, could in some cases only be explained on the supposition that clay partition walls had collapsed into such rooms. But the discovery in the second Roman house at Farningham (partially excavated in 1948 and awaiting publication), and here in the Lullingstone house, of the lower portions of such clay partition walls still in situ on their flint and mortar bases, has rendered the supposition extremely likely, while the discovery by Mr. S. S. Frere, F.S.A., of another example in situ at Canterbury confirms the point. Such collapsed clay would no doubt have proved useful to subsequent builders as packing beneath later floors at a higher level, which certainly took place in the Farningham house. The example from the Farningham site was an especially good one and is illustrated at PL I I (a) where the construction is very clear, being of the same type as at Lullingstone, though at the latter place the walls were found to be very fragmentary and delicate by comparison with those at Farningham where, indeed, most of the Flavian house was partitioned in this way. None of these clay walls could have taken any great weight, and presumably they were fitted in from floor to rafter within the wood framework of the house. Some are thicker than others, those at Farningham averaging 15 to 16 in. in thickness, while the Lullingstone examples are only 23 THE LULLINGSTONE ROMAN VILLA 9 in. thick. The latter are illustrated at P1.II (b) and are indicated by the lines of pegs on the photograph, the fragmentary wall plaster being visible adhering to the clay. At Lullingstone the method of keying is especially interesting, consisting as it does of a roller pattern impression. This has been submitted to Mr. A. W. G. Lowther, M.A., F.S.A., who has very kindly commented upon it at Appendix 3 and PL IV. The green and yellow colours on the plaster are repeated on the Period II plaster that adorned the Basement Room, and which decorated the southern walls of the east and west tiled steps, and it is likely that it is contemporary with the Period II fourfold stairway complex, thus suggesting a mid to late Antonine date ; this particular use of the roller pattern does not in this instance appear to go back to Flavian times. The clay is of a light brown sandy nature and is remarkably free from fragments of flint, but its source is at present unknown. The keying at Farningham was a simple scoring by diagonal lines intersecting to form a rough diamond lattice pattern, and no roller pattern was found there ; and at Farningham its date was undoubtedly Flavian and not later. At both sites, the clay was placed upon a flint and mortar base, with some puddled chalk at Lullingstone, and quarter-round concrete mouldings were provided, those at Lullingstone actually embracing the bottom inch or two of the clay. At Farningham the clay did not start at floor level, but a few inches above it. The suggestion has been advanced that the walls may have, been made in the first place by unbaked clay bricks, but there is no evidence that this was the case, even on the closest scrutiny. Neither was certain evidence found that any wattle was incorporated either at Farningham or Lullingstone. It is likely at Lullingstone and certain at Farningham, that the walls were inserted between timber uprights, thus strengthening the walls and preventing distortion. The uprights at Farningham were well suited to this purpose, being thick rectangular planks, providing the largest possible surfaces to hold the clay and also being stout enough to support a thatched roof. Thirty-five post-holes of the appropriate shape were found, forming an integral part of the construction of the flint and mortar wall footings. After the collapse of such walls and the spreading of the clay over the floors of neighbouring rooms, the only indication of their existence is the appearance during careful trowel-work of straight lines of plaster in the clay, any one of which may indicate the remains of a wall face. The Lullingstone clay walls collapsed during the period of abandonment of the house c. A.D. 200-250, and their fallen clay sealed fragments of Castor ware, Antonine pie dishes, Patchgrove vessels, with a few sherds of somewhat earher roughcast and poppyhead beakers, together with part of a carrot-shaped amphora ; this last is roughly made and is creamy-yellow in colour, being probably a pre-Flavian survival. 24 1'lJ.TK I TIIK WELL. HEAKTH X SECTIOX ' far, p. 21 PLATE II (a) FARNINGHAM. Clav Wall. (b) LULLINGSTONE. Clav Walls. PLATE III t-ir % - __ "*» M£ (a) NORTH-SOUTH CORRIDOR, Showing (1. to r.) clay wall, concrete corridor with hearth, Period II wall and Period I slotted wall. (b) PERIOD II GULLY. Looking north. PLATE IV VERTICAL. G R O O V E S S. r PATTERNED WALL PLASTER. THE LULLINGSTONE ROMAN VILLA The Castor ware could scarcely have been in use before the last quarter of the second century (see Dr. Kenyon's valuable note in Jewry Wall, p. 120), and thus helps towards dating the Period I I occupation, though we still cannot determine with any exactitude the building date of the clay partition walls themselves. But this Castor ware is the latest dating material yet found in association with the clay walls. A glance at the plan (Fig. 8) will show a large rectangular post-hole of this period at the north end of the west wall of the corridor. This west wall probably did not incorporate a clay partition, but the post supported the end of a clay partition wall along the north, and the small gully on the exterior of this may have drained water away from the clay partition wall, though its position seems unsuitable for this purpose. It may have been constructed to take water flowing down from the steep upward slope of the ground on the exterior, thereby affording protection to the clay wall (PL I I I (b)). Period III The reconstruction of the house in the second half of the third century rendered the fourfold stairway complex redundant, and the exterior wall was drawn back some 13 ft. to the south, where a long and strongly built flint and mortar wall was carried from west to east across the earlier north-south walls, across the corridor, and across the crushed brick surface of Room 17, into which it was set, being also stepped up the slope to the west. Everything exterior to this new wall was levelled to the floors and footings, and a rough chalk and concrete surface was spread over them, sloping down to the east. At the same time, this part of the house was protected on the exterior by a wide and shallow drainage gully cxirving round the northwest corner of the house and continuing downwards towards the east at an average distance of some 10 ft. from the new wall. It overlay the earlier plan and contained both pottery of Antonine date and also a large proportion of third century sherds, notably fragments of Rhenish metallic glazed vessels of the mid-third century. The slope upwards from the north edge of this gully was given a rough surface of pebbles and chalk in which was found a denarius of Septimus Severus (Second Interim Report, Arch. Cant., Vol. LXV, p. 68). The outflow of this gully is yet to be found as it probably lies beneath the modern road. It remained open for the rest of the occupation of the house, and in course of time its silt content rose and spread laterally, containing in its higher levels fragments of fourth century pottery, including sherds of colour-coated and rosette-stamped ware, together with a few coins of the later periods. The whole of the Northern Complex was finally covered by the downward creep of clay and flint hillwash. 25 THE LULLINGSTONE ROMAN VILLA APPENDIX 1 THE POTTERY POTTERY EROM THE WELL FILLINGS Group I 94. Bead Rim of native ware, the rim sharply inbent, wheelturned and of soapy orange-brown texture, with bluish core and having interior grooves on bead, the topmost of which may have a functional significance (see Well Section). Such inbent rims appear characteristic of North-West Kent (Ward Perkins, Oldbury, p. 163). First century A.D. 95. Heavily gritted native ware vessel, greyish buff with pronounced inbent rim and hard rough surface. Wheel turned. This is paralleled in Mr. A. W. G. Lowther's Report on Purberry Shot, Fig. 18, No. F.24, which in its turn was found identical with a similar vessel at Otford, figured in Oldbury, Fig. 14, No. 15. (see Well Section). First century A.D. 96. Mortarium. This is discussed by Mr. Eric Birley at Appendix 2, and can probably be dated end second/third century. The form foreshadows the latter period while the rough texture is of second century type, being of a hard, buff-coloured consistency (see Well Section). 97. Rim and portion of a handle of a globular amphora. Light buff ware, the surface eroded by prolonged contact with water. This type appears to have persisted well into the Antonine period (Collingwood, type 94) (see Well Section). Group II The following examples indicate the predominantly Antonine character of the upper, blackish filling which sealed the earlier deposits in the weh. There is a small percentage of third century material also included, which suggests the deposit of this filling in the well at the beginning of the third century occupation. S75. Samian Form 31 (Sb). Antonine. This is the same vessel as S24 (Second Interim Report, p. 46), and fragments of which were found stratified in a late Antonine level in the Northern Complex (see Section LM.). S76 and S77. Two further examples of Form 31, similar to S75. Antonine. 98. Flask. Soft grey ware, pitted on exterior and decorated with a band of rouletting near the neck. Mouth wanting. Ospringe, No. 433, PL XXXVHI. A.D. 160-190. 26 THE LULLINGSTONE ROMAN VILLA 9+ 'T 95 V ST7S~ Tfc ^7 F ^ 101 % 107 * t o a ^109 Cm \Ti2 S~ l0< r 103 r \ FTG. 4 d) 27 10 THE LULLINGSTONE ROMAN VILLA 19 7 s\r\r\r\ VENT—-, \Yw-VENI VENT-= 120^— TOP v: 128 ^ 134 35 r r r f 1ZI 122 125 123 124 T 127 ftff?l'Sa*K¥SSilSSiS*te 129 r 132 I r 131 Fio. 6 (J) 28 126 w?w/mmwm»)Wi>m )\uuimnp 133 THE LULLINGSTONE ROMAN VILLA 99. Flask. Soft grey ware, decorated on the exterior with two incised horizontal lines. Mouth wanting. Ospringe, No. 133, PL XV. A.D. 140-190. The Ospringe Report contains numerous examples of these flasks, slightly differing in shape but all allocated to the second half of the second century. 100. Cooking Pot of hard grey ware, lip and top of the shoulder burnished, and roughly decorated with thirteen lightly scored groups of four vertical lines each. A late Antonine type. 101. 102, 103 and 106. Cooking Pots of hard grey ware, of late Antonine type. 104. Rim of large Patchgrove vessel. Many fragments of Patchgrove vessels have been found in association with Antonine material on this site. 105. Coarse dark grey rim, recessed for lid. It has a somewhat soft texture and is of Patchgrove type. Wheel made. 107. Pie Dish of light brown ware, smooth and well made. Antonine (see Well Section). 108, 109, 110, 111 and 112. Bead Rim Pie Dishes of grey ware. Antonine. Nos. I l l and 112 are probably earlier than the others and may be Hadrianic. 113, 114, 115, 116 and 117. Straight sided Pie Dishes of grey ware, except No. 116 which is light brown. Antonine. 118. Wide-mouthed bowl of smooth grey ware. Slight cordons and curvilinear decoration between the cordons. It is paralleled in the grave group at Colchester, illustrated on p. 89 of The Archaeological Journal, Vol. CI, but is here unlikely to be of fourth century date. Third century. 119. Flanged bowl of greyish ware, of typical fourth century type (see Well Section). POTTERY FROM THE NORTHERN COMPLEX From Period I levels 120. Portion of a lid pierced by three vent-holes. This unusual object is of very coarse native ware, greyish black in colour. It has the soapy surface of Patchgrove ware with which it seems allied. It must be placed not later than the first century A.D., and is probably early in that century. 121. Bead rim of hard, gritty texture, reddish brown in colour. It is wheel made and is probably not later than Flavian. A tendency to flatten the top of the bead is paralleled in a complete vessel of this type found in the Faminglutm Villa II (not yet published), which there was securely dated c. A.D. 60-80. 122. Small bead rim of slightly gritty texture, greyish buff and with a soft surface. It shows no sign of Romanizing influence, and is 29 THE LULLINGSTONE ROMAN VILLA clearly of purely native technique. It cannot be later than the first century A.D. 123. Bead rim of hard, gritty texture, reddish brown with greyish blue core. It is wheel-turned and is similar to No. 121, except that the top of the bead is not flattened to the same extent. It is probably not later than Flavian. 124. Small bead rim of gritty texture and orange brown colour. This is__ purely native ware and is unlikely to have persisted beyond the first century A.D. 125. Hard, dark grey bead rim with no grit, the rim brought to a point and flattened downwards internally. This shows evidence of a Romanizing tendency and is probably of Flavian date. 126. Hard, gritty bead rim, brownish orange on interior and dark grey on top of bead and on the exterior. The top of the bead is not flattened, and this example, while wheel-turned and of hard fabric, still shows no Romanizing influence. It probably does not persist into the Flavian period. 127. Rim of girth beaker, probably Gallo-Belgic, in Roman polished grey ware, rouletted in three bands. These beakers are characteristic of the first half of the first century A.D. (see Camulodunum, p. 232, and PL LV). From Period II levels 128. Platter with moulded base. Pale brown clay, coated inside and to five-eighths of an inch below the lip on the outside with a deep rich red slip (" Pompeian Red "). The base is blackened by fire owing to use in the kitchen. The curved rim shows a greyish core. An example occurs in Richborough II, p. 102,.No. 161, illustrated at PL XXXI. This Richborough example is unstratified, but other fragments there have been found in first century deposits. There is a detailed note on these red-coated platters in Camulodunum, p. 221, and this example appears to be referable to Camulodunum Form 17B. Its range there is from just after the conquest to c. A.D. 65. Here at Lullingstone it is probably a survival, appearing in a few large fragments in the late Antonine (Period II) levels. 129. A fragment of a smaller platter, exactly similar in type to No. 128. The diameter of this platter is approximately 8 in. 130. A fragment, towards the pointed base, of a carrot-shaped amphora. The fabric is a coarse, creamy buff, the incised lines being poorly and irregularly placed round the body. There is an example in Richborough III, p. 167, No. 196, illustrated at PL XXXII, where it appears to be pre-Flavian. A note on this type of vessel appears in Camulodunum Form 189, p. 253, and PL LXXII, where it is dated normally Claudian-Neronian, Flavian survivals being judged rare. 30 THE LULLINGSTONE ROMAN VILLA 136 pm -V « * 143 142 145 ^_ ^ / / C ^ r 144 ib * * 146 H »PI47 r i S 151 I Eio. 6 (J) 31 149 148 5* tftMHIWtlWMnm THE LULLINGSTONE ROMAN VILLA Both at Richborough and Camulodunum the ware is reddish or tile-red. As in the ease of Nos. 128 and 129, this example must be a survival. 131. Bowl or olla with lip projecting to an internal bead. Hard greyish brown polished ware, fully Romanized. 132. Jar with slightly everted rim, similar to Nos. 137,138 and 141. Grey polished ware with no indication of barbotine dots or other decoration. Antonine. 133. Roughly made Patchgrove type vessel, greyish brown and not wheel-turned. Very slight incipient cordons above shoulder, soapy surface. This ware might easily be mistaken for similar vessels of the late Roman periods, and proper stratification is essential before deciding their true period. 134. Straight sided pie dish of hard grey ware. This is typical of the second century, and in this case may be Hadrianic to mid-Antonine. 135. Bead rim pie dish of hard dark grey ware. A second century type, in this case possibly Hadrianic to mid-Antonine. From the top filling of small Period II Gully 136. Mortarium. Orangy buff with wide flange and up-standing rim. The texture is reminiscent of second century examples, but the form is certainly later. A near parallel is in Wroxeter, 1912, p. 80, Fig. 20, No. 146, where its type is said to belong to the later Roman periods. The present example may tentatively be placed in the late third century, the design being more ponderous than the lighter designs of the fourth century. From the clay bottom of the Period III Gully 137. Jar with slightly everted rim, no neck. Dark grey polished ware with vertical bands of barbotine dots. This is an Antonine type and may have persisted to the end of the second century. 138. Jar with slightly everted rim, no neck. Hard creamy white ware, colour-coated with a deep bronze metallic-type slip, and roughcast on the body. This again is an Antonine type and may have persisted to the end of the second century. (For Nos. 132, 137, 138 and 141, see Dr. Kenyon's comments on the dating of these vessels in Jewry Wall, pp. 104-108.) 139. Bowl of reddish-pink ware with bluish-grey core. The fabric is slightly soft and is of a delicate " biscuit "-like consistency. The inside of the slightly beaded rim is recessed, possibly for a lid, and the exterior is decorated with horizontal bands of finely incised lines. There is the appearance of a thin reddish-pink wash on both the exterior and interior. The base is missing, but may have stood on tripod feet (vide Camulodunum, PL LIII, No. 63a, where examples are quoted on p. 228, but roughcast in addition and dated c. A.D. 49-65). 32 THE LULLINGSTONE ROMAN VILLA NORTHERN COMPLEX. ROAD SECTION L-M. PERIOD BTA SURFACE HUMUS - RECENTJff^ fflTi'irn'M'iTrro GULLY OVERFLOW • •COIN OF ARCADIUS -COIN OF HOUSE OF THEODOSIUS •SAMIAN CURLE lie. HADRIANIC •SAMIAN MOTTO BEAKER S2 cAQ200 wwiTi'i'i'i'i'i'i'i'nTn'i'i'i'i'ri'i'i'i'rm DRAIN- [UPPER FILLINGLOWER FILLING .MORTAR FLOOR ^ r r w f f l w m f f l ^ HU mfm MEDI ROAD FLINT SPREAD BLACK EARTH WITH CHALK \ RECENT BLACK EARTH WITH C L A Y ? \ & FLINTS TILE & MORTAR MEDIEVA ROBBER PERIOD HT GULLY R E C E N T WALL RECENT UPPER FILLING MORTA PERIOO J5T WALL PERIOD SURFACE PERIODS IMODERNI CLAY WALLS NATURAL YELLOW SANDY GRAVEL CLAY SAMIAN 31 ANTONINE ' LOWER FILLINGPRIMARY FILLING PUDDLED 'CHALK & FLINT WALL BASE HEARTH WALL FOOTING COIN OF HOUSE OF CONSTANTINE NATURAL COOMBE DEPOSIT NORTHERN COMPLEX SECTION N-R 1 2 9 6 3 0 GSBESHKS INCHES s i — - - H= F E E T SAMIAN 18/31 SACRILLUS c A.D. I 5 0 - I 6C COIN OF VALENS COIN OF CONSTANTINE E. AVG Wl'IWI'IWIWtlWI'l1^^ COIN OF HONORIUS OIN OF SEPTIMIUS SEVE IRON BR H U M U S M O D E R N D I S T U R B E D BY ROOTS WALL R O O M CHALK & FLINT LINE PERIOD CLAYISH ^ ^ L L Y UPPER-fr^—JLy T^~—U L LING //•//. PERIODIC HEARTH CLAY WALL ROTTED WOOD PUDDLED CHALK & FLINT WALL BASE. / OCCUPATION SAMIAN 45 EARLY I I CENTURY PEBBLE SURFACE MORTAR TO PERIOD D STEPS-* SAMIAN 37 CERIALIS PERIOD n / S t /WAr>- « » , c. AD. I40-ISQ SAMIAN 37 OIVIXTUS % r : > c. A.D. I 4 0 - I 5Q N A T U R A L YELLOW SANDY GRAVEL g g ) ROOF & WALL FALL f H RUBBLE MORTAR a CLAY H I CLAY • BLACK ASH g ^ SANDY GRAVEL MAKE UP § H BROWN EARTH BRICK & OP. SIG § j § BROWN CLAY & CHARCOAL • B PERIOD E CONCRETE FLOORS ^m. PERIOD E CRUSHED BRICK FLOOR ESS PERIOD I OCCUPATION LEVEL [face p . 82 THE LULLINGSTONE ROMAN VILL N O R T H E R N C O M P L E X . PERIODS I & n . INCHES 1 PERIOD I RAMP mmu^ PERIOD n TILED STEPS PERIOD H CORRIDOR PERIOD E LANDING iPERIODn ! TILED ; w STEPS i T PERIOD I RAMP UNDER DOOR SLOT PERIOD H '//////////////. ROOM 17 PERIOD H STEPS CORR PERIOD a GULLY CORRIDOR; PERIOD I WALLS PERIOD H WALLS PERIOD n CLAY WALLS PERIOD m WALLS FIG. 8 M C O M P L E X . PERIOD TJJ. PER l o om GULLY x PERIOD HI HEARTH [face p. 32 THE LULLINGSTONE BOMAN VILLA The present specimen appears to be a rare example and is here clearly a rubbish survival. 140. Platter of grey ware with rough grey-brown core, the exterior and interior polished to a soft, somewhat soapy degree. It is decorated on the exterior with an irregular, incised wavy line. Second century, probably Hadrianic ; a survival of the Period I I occupation. Nos. 137, 138, 139 and 140 are clearly rubbish survivals from Period I and Period I I occupations. From the Period III Gully Filling The period of this filling is attested by a number of sherds of metallic Rhenish vessels, while No. 141 is an Antonine survival, No. 143 being probably a third century vessel. Nos. 142 and 144 are fourth century, the gully remaining in use throughout Periods I I I and IV. 141. Small jar with slightly everted rim, no neck. Very thin ware with grey core, rather poorly fired. It has a strongly metallic bronze slip and is roughcast on the body. It is a survival from the Period II occupation and probably persisted to the end of the second century. 142. Flanged pie dish of light brown polished ware and decorated with a lightly incised curvilinear design. Fourth century. 143. Small platter of hard grey black polished ware. It has a very slight carination near the base and can probably be allocated to the late third century. 144. Small open dish of heavy, roughly made dark grey ware. The poor technique of this example is similar to quantities of cooking pots of like technique which have been recovered from the latest levels of this site, attesting to the deterioration of manufacture of such vessels late in the Roman period. Fourth century. From unstratified upper levels 145. Mortarium of heavy, rather coarse buff ware with slight traces of a yellowish slip. This is similar to No. 136 and is probably of late third century date. It is unlike the examples from Lullingstone of the large fourth century mortaria, both in coarseness and fabric, which in this specimen is reminiscent of second century types. 146. Mortarium of similar fabric to No. 145, also with traces of a yellowish slip. It has a slightly drooping reeded flange and is beginning to approximate to the reeded hammerhead type of the third/early fourth centuries. This example is probably of third century date. (Wroxeter, 1912, Nos. 170, 174, 178, p. 80.) 147. Mortarium of reddish-pink fabric with a creamy pink slip. Typical of the mid to late fourth century examples of small mortaria found on this site. 33 THE LULLINGSTONE ROMAN VILLA 148. Pedestal bead rim bowl with short upright neck and bulging under towards the base. Round the bulge a band of scroll in white paint bordered with bands of rouletting. Hard pink ware with red slip. Late fourth century. 149. Bowl of hard pinkish-red ware with red slip and decorated with a double band of rouletting beneath the rim and with scroll work in white paint on the body. Late fourth century. 150. Coarse dark grey vessel roughly made with low cordon at neck. This is a late fourth century example showing much deterioration. 151. Storage jar or large cooking pot of hard grey fabric, burnished on the exterior which is orangy brown and black towards the rim. Slight cordon effect. A late fourth century example of similar deterioration as No. 150. APPENDIX 2 THE MORTARIUM FROM THE WELL (Fig. 4, No. 96) By EBIO BTJRLBY, M.B.E., M.A., F.S.A. 1. The type does not occur in the north of Britain, nor has it been noted in sufficiently clear stratification for a close dating to be put forward with any real confidence ; but on general grounds it seems likely to belong to the latter part of the third century or the beginning of the fourth. 2. In Bushe-Fox's series (Wroxeter I, Fig. 20, p. 80), the nearest analogy is type 158, representing a Caerwent specimen, " probably of the later period." Cf. also, for profile, Richborough I, PI. XXVIII, Nos. 97-102, especially No. 101. The Richborough specimens are assigned to the fourth-century levels of that site, and as far as the profile is concerned, it does not seem to be out of place in that period, or certainly attested in an earlier period. 3. But the fabric of the Lullingstone specimen seems to be quite unlike that of any of the fourth century examples at Richborough, and it reminds me rather of the southern British mortaria of the second century. 4. I am inclined, therefore, to suggest that it may be best to regard it as a forerunner of the fourth century type, produced at an old-established pottery in the third century. One might compare the 34 THE LULLINGSTONE ROMAN VILLA third century examples of hammer-head mortaria with reeded rims, in the Wall area, which anticipate the fourth century profile, but retain the characteristic fabric of the second and third centuries. (sd.) E.B, 26th May, 1953. APPENDIX 3 WALL-PLASTER PATTERNED BY MEANS OF A ROLLER. SHAPED DIE (See Plate IV) By A. W. G. LOWTHBB, M.A., F.S.A. A piece of wall-plaster bearing traces of impressed patterning, produced by means of a roller-shaped die, was removed from the foot of a wall and submitted to the writer for examination and report by Col. G. W. Meates. This piece of plaster, 15 in. long and from 2 to Z\ in. in height, is of about J in. average thickness, and had formed a " firstcoat " of coarse sand/lime plaster work and contains much small gravel and particles of chalk incorporated in the mixture. As has been explained by Col. Meates, the Avail to which the plaster was applied had been built of clay " daub " (probably originally with some wattle or timber work, incorporated in the structure, though no definite trace of. this was found). The patterning, intended as a key for the plaster, had been applied to the clay surface of the wall, and thus it was the inner face of the wall plaster (that against the face of the wall) which was found to preserve a cast of the pattern which had existed on the wall face to which it was applied. At several other Roman sites (Wroxeter and Silchester, to mention only two, described in A Study of the Patterns on Roman Flue-tiles and their Distribution, by A. W. G. Lowther, Research Paper No. 1 of the Surrey Archseological Society) it is the actual clay daub wall-surface which has been preserved, together with its impressed patterning, through the action of fire. The clay was burnt to a consistency of a soft brick, but the actual plaster had, through the action of the fire, broken away from it and disappeared. The roller-shaped die used for this work appears to have been similar to those used in the manufacture of some box-flue tiles (as described by the writer in the pubhcation mentioned) and of which examples, with some twelve different patterns, have already been 35 THE LULLINGSTONE ROMAN VILLA found at this site. The wall-plaster dies were, however, much larger and appear to have consisted of wooden rollers 12 to 15 in. in length. The pattern in this instance cannot be established for certain as too little of it is preserved, but it appears to have been a plain " zig-zag " or " chevron " design and if this was so, it would agree with that found at Silchester (preserved in Reading Museum, Publication op. cit., p. 9, Fig. 2). However, it could have been of a " diamond " type (as the writer's " Group 5 " for the flue-tile dies) and the fact that a slab of wall-plaster found at this site has such an impressed diamond pattern on the back surface (it has a painting of a fish on the outer surface) supports this possibility. As this strip of wall-plaster was at floor level, it preserves only the pattern produced when the die was first applied to the wall's surface, at the start of the operation if, as seems probable, the workman operated the roller in an upwards direction. Part- of two such impressions are preserved (6£ in. of one and about 1 in. of another) separated by a space, 2\ inches wide, containing some vertical grooves which may be due to wattle-work projecting slightly from the clay surface at the time when the plastering was carried out. It is to be hoped that some further specimens of this die-patterned work will be found and explaining some of the points that are at present conjectural. 36