The Lullingstone Roman Villa

rcfeatogia: gantiana THE LULLINGSTONE ROMAN VILLA By LIEUT.-COLONEL G. W. MEATES, F.S.A., E. GEEENSTELD and EDWYN BlROHENOtTGH, M.A. INTRODUCTION THE Lullingstone Roman Villa is situated half a mile north of Lullingstone Castle, on the west bank of the River Darent. (Kent Sheet XVII S.W. 6 ins. to 1 mile. Long. 0° 11' 52"; Lat. 51° 21' 48" Fig. 1). Here the slope descends steeply to the flood-plain. Considerable accretion has raised the slope since Roman times, but the gradient was even then sufficiently steep to allow excavations to be made into the natural hill-wash for the insertion of low-level rooms. Excavation has been made possible by the permission of the joint owners, the Kent County Council and Messrs. The Kemp Town Brewery Company, of Brighton, and the work has been done by a large number of volunteers, a full list of whose names will appear in the Final Report, but thanks are especially due to Mr. A. G. Bell, I.S.O., F.G.S., and a few informed persons for having undertaken the constant public lectures on the site. Reports by the following persons are gratefully acknowledged : to Miss J. M. C. Toynbee, M.A., D.Phil., F.S.A., for the marble busts, to Mr. C. D. P. Nicholson, F.R.Hist.S., F.S.A.(Scot.)., F.S.G., for the mosaics and the wall plaster, to Mr. B. W. Pearce, M.A., F.S.A., for the coinage, and to Mr. B. C. Middleton for the glass. The plans and sections were prepared for publication by Messrs. D. A. Broodbank, A.R.I.B.A., and C. B. Mears, A.R.I.B.A., the photography was carried out by Mr. Cookson, the pottery was drawn by Mr. and Mrs. B. L. Mackrory, while Mr. R. J. Rook undertook to reproduce on paper the vast amount of wall plaster still in situ. Mr. M. R. Hull, M.A., F.S.A., supplied a note on the two fibulae and Mr. A. W. G. Lowther, M.A., F.S.A., made a prehminary investigation into the patterned flue tiles. Mr. J. H. Evans, F.S.A., has most kindly provided a revetment for the Basement Room, without which much of the investigation at this point could not have been made. To these ladies and gentlemen grateful acknowledgment is made. Finally acknowledgment is made to the many eminent archaeologists who have visited the site, to the Director of the Institute of Archseology for having 1 THE LULLINGSTONE ROMAN VILLA made possible the reassembly of the painted wall plaster by the provision of accommodation at the Institute, and to members of the Institute staff, for their advice and constructive help, which has cleared up difficult points of interpretation and rendered the task of research productive and pleasant. It is also pleasing to record the following grants in aid of the work. From the Society of Antiquaries, 7—1 7 y i i . . NORTH A » i tvn i n tbmch V* / X? £ *n O*J- urfj* , A Ift f Ntwbarn \ net • I I » <* »a «> s ^ *t* a>« X (3 *•*o'*• S? * •« AI. o? «fef... A°- .** J wn o ;no 1000 m r Reproduced from the Ordnance Survey Map with the sanction of tin- Controller of B.M. Stationery Office FIG. 1. of London—£60 ; from the Haverfield Bequest Administrators—£50 ; from the Kent Archseological Society—£25. This last sum has already been expended on the expert work of making safe the mosaic floors. The purpose of this first Interim Report, on the 1949 season's work, is to place on record the layout of the villa so far known, and to present the evidence so far obtained. The discovery of the villa was the result of several contributory factors. In the middle of the eighteenth century, when the deer-fence was erected round the new enclosure of 2 THE LULLINGSTONE ROMAN VILLA LuUingstone Park, holes for two of the wooden uprights revealed the existence of a mosaic floor of Roman date, and the information was recorded. At this time the chapel marked on the Ordnance Map was still visible to the north of the villa, and its ruined walls are said to have been largely composed of Roman materials. It is further reported that Roman coins and small objects had been picked up in the neighbourhood from time to time. (Hasted, and Arch. Cant., XXXIX, p. 158.) During the next two centuries the exact positions both of chapel and villa were forgotten until Messrs. E. Greenfield and E. Birchenough detected a Roman wall under the roots of a fallen tree in the summer of 1939. This was confirmed by the authors who, with a few interested friends, carried out a field survey of Romano-British sites in the Darent Valley in 1947-8. (Arch. Cant., LXI, p. 180.) The exact position of the chapel has yet to be discovered, but the medieval scatter on the villa site suggests that it was still in use in the thirteenth-fourteenth centuries. The site has not previously been excavated, and the work is being carried out by the Darent Valley Archaeological Research Group, which is a volunteer working party whose purpose is to add more to the known history of the Darent Valley, and to endeavour to tie up the Romano- British culture here with the rest of the south-eastern region of Britain. Except for robbing of the top levels for building purposes in medieval times, and some disturbance due to the digging of post holes and the planting of trees, the villa remained untouched after its final destruction late in the fourth century. The evidence of ploughing over .the northern portion of the site in the fourteenth century is conclusive, and is helpful in the study of the geological problem. There is at present no threat to the site, and it is hoped that support will become available to preserve the villa for public inspection in the future. It is unfortunate that the road running along the eastern edge of the villa cannot yet be removed or diverted ; but such action in the future will be essential if all the evidence is to be recovered. The objects from the Roman levels will remain, with certain exceptions, at the Working Centre of the Group at LuUingstone Castle until the pubUcation of the final report. Thereafter, they wUl be deposited in local museums, and wiU in no case form part of a private coUection. The exceptions are the group of metal ingots, which have been deposited in the British Museum, together with the marble busts. SUMMARY The cultural sequence found on the site as the result of the excavations carried out in 1949-50, gives the foUowing picture. The natural slope on which the viUa was sited, has been dug into in some parts and levelled in others ; and it thus shows floors whose edges 3 R O M A N V I L L A • L U L L I N G S T O N E E X T E N T O F P L A N A T E N D O F l s 4 « S E A S ON err es I IOO 55 S C A L E O F F E E T O.Broodbank AMI FIG. 2. THE LULLINGSTONE ROMAN VILLA descend here and there into deep rooms or staircase wells. This threedimensional effect is a feature of the earhest period and is continued throughout the later periods, although the later levels tend to obscure the earUer plan by being carried over the deep compartments and staircases, while at the same time making use of them. Period I. First Century, A.D. The natural graveUy slope has been partially excavated and then levelled by a platform of orange sandy clay, containing flints (Fig. 3), and from this level the following objects were recovered : Portions of a hand-made pottery vessel of native manufacture that probably runs parallel with the intrusive Belgic culture (Fig. 5, No. 1). Bead rims (Fig. 5). Hod Hill type fibula (Tiberius-Nero) (Fig. 3 and PI. Vic, No. 3). Hoard of ingots with associated cooking pot of " Patch grove " type, probably hand-made (PL VIS, and Fig. 5, No. 6). Pottery of Belgic affinity (Fig. 5, Nos. 7, 8, 9). This level has only been sectioned to date ; an investigation of its surface may reveal traces of post-holes, if they have not been obliterated by the building of the later masonry walls. The evidence certainly points to an occupation of the site in the first century A.D., and the affinities of the pottery indicate the earlier rather than the later half of that century. Period II. Flavian-Antonine. Very late in the first century or early in the succeeding one, the first flint and mortar waUs were buflt. The earUer level of Period I was trenched and a brooch of " dolphin " type (PI. Vic, No. 2 and Fig. 3) was recovered from the footing trench of the east wall. This brooch ceases in the reign of Vespasian, and it is unlikely that it was manufactured subsequent to that Emperor, though it may have had some persistence. I t is, however, supported by pottery sherds from the same trench ; they are of first century date and have strong relationship to the thick soapy native ware (Fig. 5). The general plan (Fig. 2), as it exists to-day, was adopted in this first bunding, subsequent periods being concerned mainly with repairs and cappings of the walls. The yard level to the west, exterior to the house, is clearly indicated by a stretch of broken tegulae sealing pottery sherds of Antonine type (Fig. 3). The level of Period I was destroyed to some extent by the insertion at a lower level of the Bath Wing, and at the north end of the site a Basement Room was created by cutting away the natural slope of hfll-wash and butting the waUs against the baulks so made, a flight of steps being added to afford easy access from the living levels above. This flight of steps appears, in the first phase, to have led down into a narrow corridor 5 THE LULLINGSTONE ROMAN VILLA cut deeply into the ground, but the significance of this is not yet clear. The three steps leading directly into the Basement Room were originaUy assigned to the fourth century on the evidence of a clay level containing Constantinian coins that appeared to extend beneath them. On fresh constructional evidence, however, these steps cannot be later than the second century, and the clay level does not now seem to be homogeneous throughout the room. As this room abuts upon the present road, the opportunity for making a further cut for confirmatory evidence may not appear for some time. With the exception of the Apse of the Mosaic Room, which projects beyond the west waU, and certain other late additions of the third and fourth centuries, the walls are weU buUt, the flints being laid in level courses and rendered smooth on both faces except where they are built against the baulks of natural hill-wash when no attempt has been made to render them off. The full extent of the Flavian-Antonine house has not yet been ascertained ; its foundations appear to continue to the north and north-east. Period III. Third century, A.D. There is no evidence that the Flavian-Antonine house continued in use into the thhd century. From constructional and pottery evidence (to be incorporated in the next report) a period of abandonment probably covered the final years of the second century and continued to the middle of the third century. The coin series begins to become symptomatic of occupation in the last half of the thhd century, and coins of Claudius II and AUectus (noted in the next report) have recently been recovered from the sealed tesseUated floor of Room 5. The Bath Wing was reconditioned at this time, and the cistern (Room 4) was sealed by a floor of rammed chalk at a high level: the pottery thus buried is exclusively second century in type. The condition and siting of the group of marble busts in the Basement Room has evidential value. These busts are undoubtedly of second century date, and when placed in the position in which they were found, they were already in a worn and dilapidated condition. The suggestion may be justifiably advanced that, if they are held to have decorated the Antonine house, they might have been brought to light among the decayed levels of the viUa and placed in the Basement Room by the rebuUders c. A.D. 250. Two of the four vessels discovered in the concrete floor of this room, the Rhenish " SVAVIS " pot (Fig. 8, No. 23), and a cooking pot (Fig. 7, No. 24), can both be assigned to the middle of the thhd century. Both of these were completely covered by the clay floor (Fig. 4) and there was probably some connection between the deposit of the busts and the pots buried in front of them. 6 THE LULLINGSTONE ROMAN VILLA Period IV. Fourth Century, A.D. It is probable that there was no further cessation of occupation though habitation does not seem to have got into fuU swing until the Constantinian period. This era shows intense occupation, the evidence for which is plentiful, not only in regard to the reconstruction of the house, but especially in the abundance of pottery and coins. The curve of coin finds rises acutely, and dechnes through the Valentinian period into that of Theodosius, the latter being sparsely represented, ending with a single Arcadius. It is to Period IV, early in the century, that the Mosaic floors belong, together with associated levels constructed over Period II. The Bath Wing continued in use during most of this period. The tips filling it contain coins of the House of Valentinian and quantities of sherds of typical fourth century pottery, the material being exclusively Romano-British. It seems likely that the extensive robbing of the Wing took place during the latest phase of Umited occupation that succeeded the main destruction c. A.D. 367. The latest floor of Room 5 was constructed about the time of Constantine II as Augustus (A.D. 337-40), as a coin attributable to that Emperor was found mixed in the concrete of that floor (noted in the next Report). It was during this period that a highly decorated room was buUt over the earlier Basement Room. Its level agrees with that of the neighbouring Mosaic Room, while a coin of Constans, as Augustus (337-50), which had been mixed in the mortar of the wall of the upper room was recovered from the destruction layers that formed the Basement filling. The technique of the waU paintings, theh composition and colour, give the impression that they were painted rather later, perhaps towards the middle of the century. The final destruction by fire which involved both the Mosaic Room and the upper room and the Bath Wing can probably be associated with the events of the year A.D. 367, on the pottery evidence and cessation of coinage thereabouts, in the Basement Room, its fillings, and elsewhere. The quantity of coins of the House of Valentinian and Theodosius is too smaU to warrant the conclusion that occupation continued on the same scale as at the time of Constantine I and his sons. Such occupation was strictly limited in area. The Bath Wing filling contains six of these coins, and similar coins appeared upon the floors of rooms 7, 8, and 8a, and the corridor south of the Mosaic Room, mingled with a scatter of habitation refuse. The Arcadius coin lay in contact with the rough concrete of the floor of Room 8 (Fig. 3). On excavation the bad state of the floors mentioned above appeared in the strongest contrast with the cleanliness and preservation of the Mosaic Floor which lay only a waU's width away. The latter was sealed by a thick layer of broken the and carbonized wood, but 7 THE LULLINGSTONE ROMAN VILLA otherwise was devoid of domestic debris. It is likely that the occupation post A.D. 367 was concentrated on or about Rooms 7, 8, and 8a. Period V. Post-Roman. There is no evidence of occupation from the end of the fourth century until medieval days. The existence of a chapel has been mentioned, and there is a general scatter of medieval pottery, in places mixing with the top Roman levels, no doubt in connection with " robbing " and agriculture. The apsidal portion of the Mosaic Rooms shows a series of furrows in the coarse red tesserse towards the northwest edge, and a silver halfpenny of Edward I I found in direct association with these furrows, indicates that they were caused by ploughing in the fourteenth century. A comparison of the probable ground level over the Mosaic Floor then and now, shows that most of the hfll-wash accretion has occurred since medieval times, and has doubtless been accelerated by many years of strip ploughing on the slopes west of the site. The erection of the eighteenth-century deer-fence has already been mentioned, and since then the .site has remained undisturbed. DESCRIPTION OF THE EXCAVATIONS Unit 1. Bath Wing. (1 to 6) Unit 2. Concrete Room and W. Corridor (7, 8, and 8a) Unit 3. Mosaic Rooms. (9 and 10) Unit 4. Basement Room. (11) Two sections are iUustrated (Figs. 3 and 4, A-B and C-D). A-B (Fig. 3) is a transverse section through the vflla from west to east and shows clearly not only the successive periods but also the accretion of hill-wash which is a feature of the site. C-D (Fig. 4) is a north to south section across the Basement Room and shows the steps upon which the marble busts were found and the destruction layers of the fire of c. A.D. 367. Unit 1. Bath Wing (Rooms 1 to 6) The natural slope was excavated for the reception of these rooms, which are thus protected from extremes of climate. The combustion chamber (1) is a shaUow bowl of rammed chaUi, and a section of its filling showed two zones of charcoal where raking back had occurred. Sealed beneath the lower of these zones was a sherd of red coated ware stamped with rosettes (Fig. 6, No. 10), which indicates the use of the furnace in the fourth century. A flue, fourteen feet long and two feet wide issues from (1) in an easterly direction. This is also floored with rammed chalk and has a semi-circular arrangement of five deep holes which probably held upright bars of a movable griUe. Fragments of a 8 THE LULLINGSTONE ROMAN VILLA pie dish with interior lattice work decoration (Fig. 6, No. 11) sealed beneath the earhest charcoal layer in this flue helps to confirm its use in the fourth century. The flue is flanked by packings of chalk blocks, and leads directly forwards to Room 2. This is a corridor where the bases of the waUs that were exposed to heat between floors, have been insulated by clay ramps between the piers and the waUs. The piers are rectangular and are constructed of square bricks bound by clay and the lower floor is composed of concrete with pounded brick. A section across this corridor shows a thick layer of carbonized material between the lowest courses of the piers. The upper floor has disappeared, and a few large fragments of pinkish concrete in the debris provide the only evidence of its construction. The fiUing contained much building debris including many fragments of patterned flue the. An account of these will appear at a later date when it may be possible to relate them to some building as yet undiscovered. The pottery recovered was all of fourth century date. The heat from the furnace was also drawn aside beneath the floor of Hoom 3 which was made of coarse red tesserae and rested upon columns of brick, eleven inches square, with an estimated height of thirty inches. A narrow guUy, ending in a shaUow sump, lies below the lowest floor level and may have some connection with the cold water cistern above it (Room 4). Insets for two vertical flues exist in the east wall, while a horizontal flue was inserted beneath the later (Period III) floor at the foot of the stairs and turned verticaUy within the waU, the lining consisting of flanged roofing tiles. The stairs that lead down directly into this room are bounded on the south by the cistern and on the north by the wall of Unit 2. They can be dated to the first half of the second century by pottery recovered from below them. The bottom two treads of the eight that compose these stairs were hidden when the tesseUated floor in Room 3 was laid down in Period I I I and they retain theh facing of tiles undamaged though the remaining six were bereft of theh tiles when Unit 1 was wrecked late in Period IV. The north waU of these stairs was plastered in Period II by smaU panels in perspective, coloured deep red with borders of blue and yellow, and in Period I I I by a pinkish wash, stippled in small splashes of pinkish-red. In the Period I I occupation a short corridor led forward from the foot of the stairs to Room 5. This room is bisected by the modern road, and it possesses three floors, the two earUer being of coarse red tessera?, completely sealed by the later floor, a Constantinian one, which is of thick concrete with quarter round mouldings of large dimensions. The short corridor was roughly blocked before the insertion of the second floor, thus separating Rooms 3 and 5 completely. The walls of Room 5 were decorated with plaster stippled hi red and blue, and the two holes in the floor were made before the final destruction of Unit 1, 9 THE LULLINGSTONE ROMAN VILLA and contained the mortary rubble that, alternating with black earthy layers, formed the fillings of the rooms in this Unit. The cistern (Room 4), is lined with heavy pinkish plaster but the source of its water supply is not apparent. It appears to have been floored with thick red tiles resting on a bed of opus signinum, which remains in situ round the edges. The pottery sealed beneath the chalk floor consists of native made cordoned ware, and a sherd of Samian (Drag. 36), the indications being of an exclusively second century character. The west waU of this cistern was removed late in Period IV, coinciding with the final destruction of Unit 1. Unit 1 remained undisturbed after its destruction and subsequent robbing is absent, the fillings over and above the waUs being entirely Roman. The early medieval scatter hes on an average one to two feet above the undisturbed fillings and there is no intermingling of periods except on the level of medieval plough lines. Unit 2. West Corridor and Concrete Room (7, 8, and 8a) The corridor (7) leads from the top of the steps northwards to the Mosaic Room, and may not have been interrupted in its course until the insertion of the later Apse. The floor of Period II was composed of pebble concrete, finished with a facing of rubbed down tile fragments, but in the Period I I I reconstruction the Concrete Room seems to have had a floor of opus signinum, whUe coarse red brick tesserae were laid in the corridor. This tesseUated floor seems to have served throughout the remaining occupation; it shows much wear when compared with the tesserae in the Mosaic Room and was covered with fragments of the decorated wattle and daub waUs that coUapsed in the fire. This painted plaster consisted of orange and red vertical bands divided by thin white lines, though, as most of it is clearly from the splays of doorways and perhaps windows it may have been used to decorate these apertures only. The few coins that came to Ught in the destruction layer over the tesseUated floor of the corridor are indeterminate, but by size are almost certainly of the Valentinian or Theodosian period. Coins in the Concrete Rooms (8 and 8a) immediately adjoining, and at the final destruction level, include one each of Gratian and Arcadius, and bear out the possibility of such occupation after the main fire which seems likely to be associated with the events of the year A.D. 367 (vide account of the Basement Room). It would seem possible that in Period IV the corridor was arcaded and open on the east into the Concrete Room which may have been a yard at that time. A row of brick piers was inserted on the exterior of the east waU of the corridor and clearly helped to support its roof; but one pier by its cant towards the east points to an outward thrust which in turn suggests an unroofed yard. Further there is evidence 10 THE LULLINGSTONE ROMAN VILLA that the pier replaced a wooden post which could not bear the thrust. The latter room was also devoid of any destruction layers, which again suggests an absence of roof, and its concrete floor was found to have suffered great disintegration from long contact with the weather, the final opus signinum finish being almost enthely destroyed, and the whole much patched with chalk and rubble. In its original state, the floor of the Concrete Room stretched over 8 and 8a and sealed the middle wall whose original purpose is obscure. This is an interesting waU, as not only is its top surface smoothly finished in yellow mortar slotted and bevelled to take a large horizontal beam, but this mortar also shows many impressions of studded sandals. The floor hes dhectly upon clay and flint which contained in places a thin layer of burning and sealed the brooch (Fig. 3) and pottery sherds of Period I (Fig. 5). Whether this floor is of Flavian or Antonine date is uncertain ; but it cannot be later than Period II, and the absence of Antonine pottery beneath it, and its abundance elsewhere make the former date likely. In Period IV a pit was dug through it to contain a large storage jar (Fig. 8, No. 12) which was found in situ. Coins scattered at all levels in the pit and in the jar put the digging of the pit at c. A.D. 341 and the completion of its filling by c. A.D. 350. The associated pottery and glass are iUustrated in Figs. 6 and 9, and the bronze finial in PI. Vic, No. 1. This was a second use of the doUum, as it was placed in the pit at an angle of thhty degrees to the vertical, while its interior showed a circular stain that had been made when it had been previously standing elsewhere in an upright position. A hoard of four metal ingots and a roughly circular casting (PI. VI6) was found beneath the level of the concrete of the floor in Room 8a. They were grouped with a native hand-made cooking pot of " Patchgrove " type (Fig. 5, No. 6) which may, on future consideration of such ware, place the group rather in the early first century than relate it to a sub-Roman context. Analysis of the ingots and casting, carried out at the British Museum by the kindness of Dr. H. J. Plenderleith, F.S.A., shows a composition of 60 per cent, copper and 40 per cent, lead in each of the five items. AU four ingots have the same dimensions (fourteen inches by five inches by one and a half inches) and the casting is twenty inches in diameter. There are no markings and the reason for theh deposit remains unknown. Unit 3. Mosaic Rooms (9 and 10) These rooms, whose mosaics are described in Appendix VI, are bounded on the north and south by short corridors. That on the south is an extension eastwards of the west corridor (7) from which a shaUow step leads down into it. The painted plaster is preserved in situ along the base of the waUs and shows the same colour scheme as 11 THE LULLINGSTONE ROMAN VILLA in (7), but the arrangement here is composed of orange and red vertical bands with ovoid bases enclosing panels decorated in barbaric scroUwork painted in purple on a cream background. The flooring of these short corridors is of the same period as the west corridor and the Concrete Room, consisting of a heavy pebble concrete faced with rubbed down the fragments and later covered with a surface of coarse red brick tesserae. This floor was laid over the level of Period I I which here shows patches of burning. The floors have been patched and theh dUapidation indicates lengthy occupation and exposure to weathering. The mosaic rooms were constructed later than the west corridor, both floors of which project a few inches beneath the wall of the Apse. The Apse waU is composed of flint and mortar footings with short returns to the Mosaic step ; but the sides of the rectangular room were made of wattle and daub partitions, presumably inserted between wooden uprights, though post-holes have not come to Ught. These partitions were first some twelve inches in thickness, in a trench cut back into the edge of the pre-existing concrete floor, but a further reconstruction in Period IV reduced theh width to six inches. The daub was plastered on both faces, that towards the short corridor as aheady described, and that facing inwards to the room rendered plain, probably cream, though now burnt to a grey tint. A small door is inferred in the south waU by the presence of two hon door hinges, a number of heavy hon bolts, and a fragment of the plaster splay of a door in the destruction layer of the daub waU. The main entrance to the room was in the middle of the east waU which shows a flint footing, the mortar of which has mostly disappeared. At first the entrance was ten and a half feet wide and was set in the middle of the waU, but later it appears to have been narrowed to five feet. A ramp of opus signinum on rough concrete leads up to this entrance. The fiUings below this ramp consist of tips of gravel and large dry flints, from which came the pottery illustrated at Fig. 5, Nos. 7 and 8, and which, being of first century date, indicates the period of the waU upon which they abut and on which the entrance was subsequently built. The short corridor on the north side has the same characteristics as its southern counterpart and details will apear in the Second Interim Report. This corridor was entered by a thhd door at the north-east corner of Room 10. Unit 4. Basement Room (11) This unit was constructed by cutting away portions of the hiU slope and bunding flint waUs against the vertical faces thus made. The result was a room whose floor roughly agreed with the river level and into which a flight of steps led down from the north-west corner. Previous to the opening out of this room in an eastward direction, the 12 THE LULLINGSTONE ROMAN VILLA excavation of the hiU slope had been hmited to provide a narrow corridor running north and south, the foundations of whose eastern wall remained in situ when the room was made, being overlaid by its concrete floors. This narrow corridor was probably constructed late in Period I, as the stairs and room are of Antonine date. The evidence for this is recent and an account of it will appear in the Second Interim Report. At Period II the room probably served as a loggia or garden room ; it was certainly not meant to be a cellar as we understand the term, and should not be described as such. The walls were plastered, and decorated with a fine design of panels outhned in thin lines of orange, green and red within a framework of thick orange bands. This framework rests on a deep dado of rectangles and lozenges containing scroUs in the same colours, and is interrupted by orange coloured vines in an upright position, from which hang clusters of red grapes. This decoration remains in one place to a height of six feet, and indicates a height for the room of nearly eight feet. It does not appear upon the face of the wall block above the steps and thus demonstrates a later date for the block. The period of abandonment inferred from the most recent evidence, extended from c. A.D. 190 to c. A.D. 250 and particularly affected the Basement Room. The.tUes were removed from the steps and much of the wall plaster stripped off the walls ; and at this time the marble busts (Appendix V, Pis. II, III and IV) were not present in this room ; they were probably elsewhere in the derehct house. It was only at the rebuild c. A.D. 250 that they were placed on the steps where they were found. The waU block had aheady been inserted above the steps in the second century, possibly to complete a compartment on the north side of the Basement Room, and the busts were deposited in a damaged condition with theh backs to this wall block (PI. Via). A new wall decoration was keyed on to the original wall plaster. This consisted of red and yellow bands on a white ground running horizontally along the walls, the red ones running verticaUy up the corners of the room. This decoration was roughly executed and the red paint has dripped down here and there, no attempt having been made to remove the blemishes. Two pots (Fig. 7, No. 24, and Fig. 8, No. 23) were inserted in the concrete floor at this time. Both pots conform in date, and are likely to have had a votive significance. One of them, a smaU indented Rhenish beaker bearing the word " SVAVIS " (Fig. 8, No. 23), contained part of a rib bone to which a snail adhered. A room was subsequently placed over the Basement Room and formed a protective roof for the busts, and the magnificent wall plaster decoration of this upper room, which is described at Appendix VII and PI. VII belongs to Period IV. 13 5 THE LULLINGSTONE ROMAN VILLA About the time that the decoration was placed upon the waUs of the upper room, a rededication took place in the " cenotaph " room below. A clay floor was laid down over the concrete floor, sealing the two pots, and two further pots were placed in the floor. The larger (Fig. 8, No. 26) was deposited in line with the original pah and a large fragment of mortarium (Fig. 7, No. 27) was placed upon it to serve as a Ud. The second pot (Fig. 7, No. 25) was inserted out of line with the other three but as nearly in front of the easternmost bust as the old waU foundation would allow. It was accompanied by a House of Constantine coin of Fel. Temp. Rep. type and thus dates its insertion as later than c. A.D. 345. Both these pots were of sufficient height to aUow theh Ups to protrude above the surface of the clay floor, which itself contained in its upper surface coins of Constantine I, Crispus and Urbs Roma, dating its use, if not its construction, to a period subsequent to c. A.D. 330. At the destruction by fire of the viUa, which may tentatively be placed to c. A.D. 367 on the evidence of the lack of post-Constantinian coins in the destruction layers filling the Basement Room, the floor of the room above feU with pottery and waU plaster into the Basement Room and was succeeded after a short time by parts of the roof, more waU plaster and sections of the walls, in that order. The carbonized wood was in places of sufficient thickness for analysis, which was kindly undertaken by Mr. A. W. G. Lowther, M.A., F.S.A., with the result that in one case the oak from which the beam came was shown to have begun its life in the thhd century A.D. The pottery associated with this destruction is iUustrated in Fig. 7. A worn coin, probably of Constans as Augustus, Two Victories type, was recovered from the mortar of the faUen waU debris, and helps to date that part of the watting to after c. A.D. 341 (Fig. 4). A skeleton of a cat was found upon the surface of the clay floor and sealed beneath the destruction layer. It will be described in a subsequent report. The fining of the Basement Room remained undisturbed during the subsequent limited occupation of the viUa until excavation in 1949. APPENDIX I THE COARSE POTTERY GROUP I. PERIOD I. (Fig. 5.) 1. Handmade Olla, soft soapy orangy ware, with brown slip on interior and exterior. Bead rim, angular carination at top of shoulder. Horizontal band of vertical cuts one inch below rim. Probably four applied cones placed equi-distant upon cut zone. Native ware, running paraUel with Belgic. c. 50 B.O.-O. A.D. 50. Ref. Maiden Castle, p. 234, Fig. 75, No. 191. 14 LULLINGSTONE VILLA - KENT. inches COIN OF GRATIAN. TANTINOPOUS BROOCH ( T I B E R I U S - N ERO.) I'CI'iwn C O I N Of THE HOUSE OF CONSTANTINE M I N I M . C O I N OF ARCADIUS. BROOCH (VESPASIAN) STUR ^ ^ " " f s S R S Es MORTAR U N E X C A V A T ED 'TRCNOflg-y.Rt,' U N E X C A V A T E D . U N E X C A V A T E D . S E C Tl ON A - B PLAN JcaLeoflsSm m$ m Oronqe Brown Cloy Pockinq containinq native p o t t e r y circo 5 0 B C - 5 O A0 L i q h t Brown C l a y e. Gravel Floor containlnq at Junction with 13. p o t t e r y circa 2 S - I O O A .D Roman Black I I A circa I O O - I SO A.D. Broken T i l e Spread circa I O O - I SO A.D Roman Block TJ B circa 3 O O - 3 S O A . 0. White Cement Floor with Tesserae. Pottery c l r c o 2 S O - 3 5 0 A . D . on surfoce While Cement Floor with Fourth Century p o t t e r y on surface. Opus Signinum Floor with Fourth Century P o t t e r y on surface Burnt Loyer P l a s t e r Fall 'ifty/,401 Roman Block I Mixed Second to Fourth Cen t u r i e s 1 L i q h t brown sandy c l a y . t flint. 2 Greyish sandy clay 3 Fine sandy qravel A Ginqer sandy c l a y i f l i r r t . (hillwash.) 5 Brown sand y qravel 6 Oronqe sandy qravel 7 Rusty sondy q r a v e l v f l i n t ( h i l l w a s h . ). 8 F l i n t t d e c o m p o s e d chalk stained qinqer (^topstone) 9 Natural chalk 10 Brown sandy qravel t rnor to r I I Liqht brown clay t f l i nt 12 Greyish brown e a r th 13 Oranqe yellow clay i f l i nt 14 Reddish cloy t ( l i nt FIG. 3. LULLINGSTONE VILLA - KENT. C D RECENT , u / C O I N Of « /CONSTANS FOURTH CENTURV POTS THIRO CENTURY POTS. S E C T I O N C-D Try1)?*/>*r/. ' J f-r- I o ;o so / c o l a of uTcit ( T ) Post Roman Fillinu. (Black with lorqe (lints t scattered Fourth Century pottery). K (AUC' Q Wall FoM (Flints, mortor, c tome tile). ( 3 J Unburnt Roof Tiles with creamy ^ " ^ mortor (A) Clayish layer mixed with mortar £ ^ ^ charcoal scatter. ( T ) Flood Plain Grovel M FaHen Wall Ploster Charcoal loyer. Rouqh concrete. SS|§| Burnt Layer containinq corbonlsed woodJplaster/ tile, opus siqninum • with some puddled chalk, ot bo s e. Sandy Cloy Floor unburnt beneath containing Constantinian coins in top. FIG. 4 THE LULLINGSTONE ROMAN VILLA f < IA P. 7 ; ? GROUP I. FIG. 5. (\) 15 THE LULLINGSTONE ROMAN VILLA "'ww/iuummmtwmim k CROUPU miiiffi-aiiraiiirrj •21 •: '••''r,f.'U''*' GROUPIE FIG. 6. (i) THE LULLINGSTONE ROMAN VILLA 1A. Handmade vessel of soft soapy brown ware, carinated shoulder with cordons on neck. Probably four applied cones below carinations. Ref. Pilgrims Hill, St. Pauls Cray, Kent. Romano-British hut site (unpublished). Note. Both vessels are native products and are probably forerunners of later Patchgrove fabric. 2. Wheelmade Bead rim Bowl, soapy brown-grey fabric with crushed shell. Flavian. 3. Wheelmade Bead rim Bowl, slightly larger bead than No. 2. Same fabric. Flavian. 4. Wheelmade Bead rim Bowl, well fired light grey fabric, with dark grey slip. Groove along top of bead. Late first-early second centuries, A.D. 5. Wheelmade flat topped Bead rim Bowl, well fired light grey fabric with dark grey mottled slip on exterior. First half second century, A.D. 6. Handmade ovoid vessel. Patchgrove fabric. Outcurling rim. Interior roughly finished, exterior smooth. Upper half of vessel fired dark brown, lower half orange. Probably first century, A.D. 7. Handmade large Bowl of grey brown soapy Belgic type Patchgrove ware. Smoothly surfaced interior and exterior. Boldly cordoned angular shoulder. Outcurling rim. First half first century, A.D. Ref. Verulamium, p. 165, Fig. 17, No. 51. 8. Handmade large Bowl. Same fabric as No. 7, but with bolder cordons on shoulder. Outcurling rim. Same date as No. 7. Ref. Verulamium, p. 165, Fig. 17, No. 51. 9. Handmade large Bowl of brown-orange Patchgrove type fabric with dark grey surfacing on exterior. Horizontal band of tooling on angular shoulder. Side of vessel rough, with acute tooled lattice work. Similar to Nos. 7 and 8 in shape and fabric and date. GKOTJP II. POTTERY EROM FLTTB IN UNIT 1. (Fig. 6.) 10. Carinated Bowl with moulded lip. Two horizontal rows of stamped rosettes with narrow band of rouletting below lip. Hard light brown fabric with blue core. Red colour coat on exterior, well worn. Fourth century, A.D. Refs. Richborough IV, p. 270, No. 479, and Heywood Sumner, Excavations in New Forest, pis. V and VI. 11. Flanged Bowl. Hard well fired dark grey ware, smooth surfacing. Horizontal band of tooling below exterior of flange. Three bands of horizontal tooling overlying tooled acute lattice on interior. Fourth century, A.D. Refs. Alice Holt, p. 43, Fig. 8, and Park Street, Fig. 20, No. I. 17 6 THE LULLINGSTONE ROMAN VILLA GROUP III. POTTERY FROM PIT IN ROOM 8. (Figs. 6 and 8.) 12. Dolium or large Storage Jar. Hard grey ware, well fired. Heavy bead rim. Band of incised feathering between parallel grooves on shoulder. Two bands of interlacing wavy lines between parallel tooled grooves round girth at widest part. First half of fourth century, A.D. Ref. Alice Holt, p. 12, Pl.l. 13. Small Storage Jar of hard well fired light grey ware. Roll and bead rim. Two pohshed zones decorated with band of incised multiwavy lines between parallel grooves on shoulder. Spaced oblique cuts on neck ring. Two tooled horizontal lines on lower part of vessel. Akin in type and fabric to No. 12. Second quarter of fourth century, A.D. 14. Bowl of fine sandy grey ware, well fired. Everted rim thickening towards lip. First half of fourth century, A.D. 15. Bowl of dark grey ware, well fired. Bead rim sloping towards interior. Horizontal tooled groove on neck below rim. First half of fourth century, A.D. Ref. Lockleys, p. 369, Fig. 10, No. 24. 16. Pedestal Bowl with Bead rim. Fine light red ware with orange red colour coat on interior and exterior. Two parallel grooves on lower edge of rim. Probably East Gaulish. Second quarter of fourth century, A.D. 17. Castor ware Box Lid. Cream ware with bronze-coloured metallic glaze on exterior. Interior silver-green metallic glaze. Multi bands of rouletting on flange. The vertical flange is here internal as opposed to the more usual external flange (see No. 18). Lid perforated in top to facilitate handling. First half of fourth century, A.D. Ref. Park Street, p. 89, Fig. 19, No. 42. 18. Castor ware Box Lid. Hard buff ware with silver-green metallic glaze on exterior running below edge. Red brown surfacing on interior." Multi rouletting on exterior. First half of fourth century, A.D. Ref. Park Street, p. 89, Fig. 19, No. 42. 19. Castor ware circular Dish. Fine hard cream ware. Deep bronze metallic glaze on exterior and interior. First half of fourth century, A.D. Ref. Jewry Wall, p. 122, Fig. 32, No. 2. 20. Conical necked Beaker of fine well-fired Hght red ware with brown slip. Two parallel grooves on neck. Slight cordon at base of neck which is rouletted. First half of fourth century, A.D. Ref. Richborough I, PI. 29, No. 120. 18 THE LULLINGSTONE ROMAN VILLA 21. Mortarium. Hard well fired orange-cream ware with quartzite grit on interior. Square flange well below bead. First half of fourth century, A.D. Ref. Richborough I, PL 28, No. 100. 22. Crucible. Fine sandy grey ware with multi-coloured vitrification on point of base. First half of fourth century, A.D. GROUP IV. POTTERY PROM BASEMENT ROOM. (Figs. 7 and 8.) 23. Rhenish indented Beaker. Fine well-fired hght red ware with green-bronze metallic glaze on exterior. Six equally spaced indentations each containing a letter in cream barbotine composing the word SVAVIS. Between the letters are clusters each of four applied berries in barbotine. Above and below the indentations are parallel scrolls in barbotine superimposed upon bands of fine rouletting. Mid third century, A.D. Ref. T. May, Silchester Pottery, PI. XLII, No. 4. 24. Cooking Vessel. Well-fired sandy grey ware with cavetto rim. Broad band of tooled acute lattice lines equidistant between rim and base. Second half of third century, A.D. Ref. for types : Park Street, p. 89, Fig. 19, Nos. 30 and 39. 25. Squat Jar. Pale biscuit, fine sandy fabric. Cavetto rim. Tooled groove below shoulder. Three horizontal tooled lines that divide and merge again at intervals between neck groove and base. Mid fourth century, A.D. 26. Large Cooking Vessel. Rough dark grey fabric, hard and wellfired. Roll and Bead rim. Mid fourth century, A.D. 27. Mortarium. Cover for No. 26. Orange-buff well-fired ware. Quartzite grit on interior. Rising flange ending in a drooping lobe with pronounced bead. Mid fourth century, A.D. Ref. Richborough I, PL 28, No. 101. 28. Flanged Bowl. Heavy well-fired dark grey fabric. Multi tooled lines on exterior below rim. The interior is decorated with an acute lattice design with parallel bands of horizontal lines applied over the lattice. The interior of the base is decorated with a tooled spiral pattern. Third quarter of fourth century, A.D. T&efB. Alice Holt, p. 43, Fig. 8, and Park Street, p. 92, Fig. 20, No. 1. 29. Shallow Dish with incurved sides. Orange-buff coarse sandy fabric. Two parallel grooves below lip on exterior. Fourth century, A.D. Ref. Alice Holt, p. 42, Fig. 7A. 30. Flanged Pedestal Bowl. Hard well-fired red ware with brownred colour coat on exterior. Incised ring on interior of base. Lower edge of flange decorated with shallow notching at quarter-inch intervals. Probably New Forest ware. Mid fourth century, A.D. Ref. Heywood Sumner, Excavations in New Forest, PL VII, No. 4. 19 THE LULLINGSTONE ROMAN VILLA /f^^WfliH^a!nKO<,^li.f/s(lM GROUP IY FIG. 7. (£) 20 THE LULLINGSTONB ROMAN VILLA < i.g^Cs^^^b.CSrf fl <§ff wi £kflrW X HI? for * •Si 25 <*) FIG. 8. s 0 21 THE LULLINGSTONE ROMAN VILLA 31. Mortarium. Large, orange-cream ware, with quartzite grit on interior. Right angled flange, the downward sloping edge of which is decorated with a single band of shallow notching. Mid fourth century, A.D. Ref. Heywood Sumner, Excavations in New Forest, PL XA. 32. Mortarium. Small, light-red ware with blue core. Cream slip on interior and exterior, quartzite grit on interior. Top of bead and edge of flange are decorated with shallow grooves. Right angled flange with downward rounded edge. Mid fourth century, A.D. Ref. Richborough I, PL XXVIII, No. 97. APPENDIX II GLASS By BERNARD C. MIDDLETON The first season's work has resulted in a fine harvest of fragmentary glassware, but nothing complete, or nearly so, has been revealed. All the glass finds (with the exception of No. 2) have by coin evidence and/or other associated finds been dated to the fourth century, A.D., thereby enabling a fairly clear picture to be formed of the kind of glass wares in use in well-to-do circles of that period. It has been possible to carry out a certain amount of restoration work on the beakers and other vessels, in some cases making it less difficult to determine their original size and form; unfortunately, no vessel has been completely restored. Generally speaking, very httle remains of any particular object. As obtains on most other Romano-British sites in Britain the glass remains fall into two distinct categories, (a) vessels, (b) window-panes. The season's yield was 225 fragments of the former and 639 of the latter class ; 17 fragments were unstratified, making a total of 881 pieces in all. Room 8 produced the greatest amount of both vessel and window-glass fragments, the numbers being 93 and 289 respectively. Much of the vessel glass from this room was found in the pit containing the dolium, and has been closely dated by associated coins. Although such evidence is invaluable in that we are provided with the latest possible use-period, judgment of the vessels' date of fabrication must be held in reserve. Glass seems to have been in common use in most villas throughout the Occupation but it must have been expensive and difficult to replace and was probably preserved with no httle care. For the purpose of obviating wearying and space-wasting reiteration of certain words in the Catalogue the following abbreviations have been 22 THE LULLINGSTONE ROMAN VILLA adopted : " ADR "—Approx. diam. at rim ; " ADB "—Approx. diam. at base ; " (a) "—colour of glass ; " (b) "—degree of iridescence or type of weathering ; "(c)"—decoration; "(d)"—technique of manufacture. Other abbreviations are self-evident. Rim measurements marked with an asterisk should be regarded as a closer approximation to the original dimension than those not so marked. The designation of some vessel forms must necessarily be arbitrary owing to the smallness of the fragments. CATALOGUE VESSEL GLASS UNIT 1. ROOM 5. 1. Bowl,fr. of side, (a) hght yell.-gr. (b) considerable (c) one part of frag, stepped in (d) prob. mould-blown. Late fourth century. (Fig. 9, No. 1.) ROOM 3. 2. Bowl, fr. of rim. ADR. 4§~f in.* (a) hght yell.-gr. (b) milky weathering. Probably first half of second century. (Fig. 9, No. 2.) 3. Bowl, frr. of rim and sides. ADR. 4 in * (a) very shght gr. tinge (b) incipient (c) several shallow fines cut concentrically around rim, the furthest being J in. from same. Second half of fourth century. (Fig. 9, No. 3.) 4. Bowl, fr. of rim. ADR. 3 in.* (a) light bl.-gr. (b) incipient. Fourth century. (Fig. 9, No. 4.) 5. Bowl, frr. of rim and sides. ADR. 3 | in. (a) fight gr. (b) mod. U.S. Probably fourth century. (Fig. 9, No. 5.) 6. Beaker, frr. of sides and base. ADB. If in., kick £ in., pointed. (a) mod. bl.-gr. (b) incipient. Second half of fourth century. (Not illustrated.) 7. Beaker, frr. of rim, sides and base. ADR. 3 ^ in.* ADB. 1£ in., kick -ji. in., pointed, (a) mod. gr. (b) mod. Second half of fourth century. (Fig. 9, No. 7.) 8. Beaker, fr. of rim. ADR. 3 in. (a) light yell.-gr. (b) mod. Late fourth century. (Fig. 9, No. 8.) 9. Beaker, fr. of rim. ADR. 2f in. (a) light gr. (b) mod. Late fourth century. (Fig. 9, No. 9.) 10. Beaker, fr. of side at turn of base. ADB. 2J in. (a) light gr. (b) mod. Second half of fourth century. (Not illustrated.) 11. Jug, fr. of rim of orifice. ADR. 2J¥ in.* (a) mod. gr. (b) incipient (c) large thread of same colour as jug applied around outside £ in. from rim. Second half of fourth century. (Fig. 9, No. 11.) 12. Bottle or Jug, fr. of reeded handle, (a) mod. yell.-gr. (b) incipient (c) vertical striations. Second half of fourth century. (Not illustrated.) 23 THE LULLINGSTONE ROMAN VILLA ( ^ r T ^ 7 \ \x X » 17 I 1 + Z6 X (4) 23 20 El V I \ < \ 36 3 32 38 B 4-3 » fO fa 19 FIG. 9. Glass Vessels (§) 24 THE LULLINGSTONE ROMAN VILLA 13. Jar or Jug. frr. of sides and base showing corners, (a) colourless (b) considerable. Massed small bubbles. Second half of fourth century. (Not illustrated.) 14. Two frr. of fiat glass, each with rim raised on either side, (a) mod. yell.-gr. (b) incipient. Fourth century. (Fig. 9, No. 14.) 16. Cup or Bowl, fr. of side. Orig. diam. not less than 5 in. (a) tinted yell.-gr. (b) mod. (c) two parallel, applied threads of glass greener than that of vessel, 7 mm. apart, unmarvered. Numerous bubbles. Third quarter of fourth century. (Not illustrated.) ROOM 4. 16. Bowl, fr. of base, (a) mod. gr. (b) incipient. Second half of fourth century. (Not illustrated.) 17. Beaker, fr. of rim. ADR. 3 in. (a) tinged yell.-gr. (b) considerable. Second half of fourth century. (Fig. 9, No. 17.) 18. Bowl or Beaker, fr. of base. Slight kick, prob. rounded. (a) mod. gr. (b) incipient. Second half of fourth century. (Not illustrated.) ROOM 2. 19. Bowl, frr. of sides, (a) tinged gr. (b) mod. Heavily encrusted with mortar-like substance. Second half of fourth century. (Not illustrated.) 20. Beaker, fr. of rim. ADR. 3 - ^ in.* (a) tinged yell.-gr. (b) mod. Second half of fourth century (possibly last quarter). (Fig. 9, No. 20.) 21. Bowl, fr. of rim and sides. Diam. not more than 3 in. (a) mod. gr. (b) incipient. Second half of fourth century. (Fig. 9, No. 21. The angle of the side cannot with certainty be determined.) UNIT 2. EXTERIOR OP CONCRETE ROOM (8). 22. Bowl, fr. of turn of side and base. (?) Approx. diam. at that point, 5 in. (a) mod. bl.-gr. (b) mod. Fourth century (?). (Not illustrated.) ROOM 7. 23. Beaker, fr. of rim. ADR. 3-if in.* (a) tinted yell.-gr. (b) incipient. Very thin-blown. Fourth century (possibly second half). (Fig. 9, No. 23.) 24. Bowl or large Cup, frr. of sides and base. ADB. If in., shallow kick, rounded, (a) mod. gr. (b) incipient. Late fourth century. (Not illustrated.) 25. Flagon (?) frr. of rim of orifice. ADR. 2 in. (a) tmted yell.-gr. (b) incipient. Late fourth century. (Fig. 9, No. 25.) 25 THE LULLINGSTONE ROMAN VILLA ROOM 8. 26. Beaker, frr. of rim, sides and base, comprising greater portion of vessel. ADR. 3^- in.* ADB. 1£ in. Kick very shallow, rounded. (a) mod. yell.-gr. (b) mod. Massed small bubbles in patches. Second quarter of fourth century. (Fig. 9, No. 26.) 27. Beaker, frr. of rim, sides and base. ADR. 3£ in.* (a), (b) and bubbles as above, but thicker glass. Kick shallow and rounded. Second quarter of fourth century. (Fig. 9, No. 27.) 28. Beaker, frr. of sides and base. Orig. diam. not determinable but appears to have been appreciably larger than either of foregoing vessels from this room. ADB. 1 | in. (a) and (b) as above. Kick - ^ in., rounded. Second quarter of fourth century. (Not illustrated.) 29. Beaker, fr. of rim. ADR. 2§ in. (a) and (b) as above. Second quarter of fourth century. (Fig. 9, No. 29.) 30. Beaker, fr. of rim. ADR. 3 in. (a) as above (b) incipient. Mid-fourth century. (Fig. 9, No. 30.) 31. Beaker, frr. of rim and sides. ADR. 3-g$ in. (a) as above (b) heavy. Second quarter of fourth century. (Fig. 9, No. 31.) 32. Bowl, frr. of rim and sides. ADR. 4i£ in. (a) tinged yell.-gr. (b) incipient (c) two shallow, parallel, incised grooves running concentrically around rim, -jL in. and -fc in. from same. Close under the grooves are unpolished abrasions made vertically with a wheel to form a picture of a tree and a right arm and hand. The remainder of the figure and scene is missing. Prob. Rhenish. Good glass. Second quarter of fourth century. (Fig. 9, No. 32.) 33. Bowl, fr. of side, (a) tinted yell.-gr. (b) incipient (c) one applied thread, unmarvered, running around body, with trace of second one -jL in. away. Same colour as vessel. Second quarter of fourth century. (Not illustrated.) 34. Bottle, circular, frr. of handle, sides and base. Handle 1-jL in. at narrowest point, almost complete, striated, (a) mod. gr. (b) incipient. Second quarter of fourth century. (Not illustrated.) 35. Bowl, fr. of side, fairly large; (a) mod. gr. (b) mod. Second quarter of fourth century. (Not illustrated.) 36. Vessel, fr. of side, (a) colourless (b) mod. (c) two applied threads of bright blue glass, cross-tooled (Snake Thread). (For discussion on this technique see W. A. Thorpe, English Glass (1935) ; illustrations of analogous examples, op. cit. pp. 34, 35, three examples from Darenth Villa, Kent, p. 35, Figs. 2e, 2/ and 2g.) Late fourth century. (Fig. 9, No. 36.) UNIT 3. ROOM 9. 37. Bottle, fr. of side. Orig. diam. not less than 4 in. Thickness 26 THE LULLINGSTONE ROMAN VILLA -i- in. (a) deep bl.-gr. (b) none, but surface cracked by heat. First half of fourth century. (Not illustrated.) UNIT 4. ROOM 11. 38. Beaker, fr. of rim and side. ADR. 2fi in.* (a) mod. gr. (b) heavy. Fourth century. (Fig. 9, No. 38.) 39. Bowl, frr. of rim and sides. ADR. 4 | in.* (a) tinted yell.-gr. (b) incipient. Good glass. Fourth century (Fig. 9, No. 39.) 40. Bowl, thick, fr. of rim. ADR. 3 i | in.* (a) yell.-gr. (b) mod. Fourth century. (Fig. 9, No. 40.) 41. Beaker, frr. of sides, less than -2 mm. thick, (a) colourless (b) incipient (c) very fine, faint ribbing. Fourth century. (Not illustrated.) 42. Bowl, fr. of rim. ADR. 4£ in. (a) tinted bl.-gr. (b) slight milky weathering. U.S. Probably fourth century. (Fig. 9, No. 42.) Few of the beaker and bowl rims have been finished with a firepolish ; most are roughly ground and square in section. WINDOW GLASS. A close study of the multitudinous remains of glass in this category seems to show that there are extant remnants of at least 38 different window panes—but it is not necessarily evidence of the same number of windows. This is the result of a grid-by-grid examination of the fragments. In a few instances it has seemed possible that pieces from one grid have belonged to a group from an adjacent trench, and in such cases due allowance has been made. Calculations of this kind are made more difficult by the fact that when several fragments from the same pane fall in different places and come into contact with different acids, etc., the iridescent and other effeots may be quite diverse. Further difficulty has been caused by the distortion of scores of pieces (Rooms 2, 5 and 8) which occurred in the last destruction period. However, a certain amount of identification has been possible by an examination of iridescence, weathering, density and shape of air bubbles and variation of tone and shade of colour. In the case of colour it has been found better to study the fragments in section or transversely ; variation of thickness throughout the area of one pane can give the appearance of more than one shade of colour when that pane is looked through, and may be misleading when it is in fragments. No Roman parallel is known to me of the triangular pane (Fig. 9, No. 43) of which fragments have been found in Room 8. The three corners are represented ; one edge (A) was moulded, the other two edges (B) and (C) were chipped to the requisite shape. Originally, no doubt, it was cast in the normal rectangular shape. From the same room there is another fragment with moulded and chipped rims similar 27 THE LULLINGSTONE ROMAN VILLA to the upper left piece in the diagram, and one also from the Bellerophon Mosaic Room (10). From Room 8 there are many other fragments of moulded and chipped rims, but only one other has a combination of both; in this case the two rims form a right-angle in the common manner. Most of the glass is of the thick, green variety, but it ranges to thin, Hght blue metal. Several fragments from Room 5 exhibit the most brilliant iridescence to be seen from this or most other sites. A few other pieces clearly show the marks made by tools when the soft metal was pushed to the extremities of the mould. I am indebted to Mr. W. A. Thorpe and to Mr. D. B. Harden for information relating to the foregoing glass ; also to Miss Dora M. Davies for assistance with the onerous task of marking many hundreds of fragments. APPENDIX III SMALL FINDS—BRONZE. PL Vic, No. 1. Finial. From pit in Room 8. Heavy bronze casting depicting a lion's mask on a hollow cylindrical mount with two raised horizontal flanges. Two rivet holes near the end. Analysis : 71-6 per cent. Copper, 16-6 per cent. Lead, 11-8 per cent. Tin. Refs. Wroxeter 1912, Fig. 10, No. 15 and Wroxeter 1914, PL XX, No. 1. PL Vic, No. 2, and Fig. 3. Fibula, (Communicated by Mr. M. R. Hull, M.A., F.S.A.). A well-known type derived from the " Colchester " brooch, but in this case ending in a kind of neck with a small knob. This is a rare feature in this type and there appears to be no parallel. Found in the Flavian period and this specimen may be placed in the reign of Vespasian. PL Vic, No. 3, and Fig. 3. Hinged, " Hod Hill " type. A characteristic specimen of the group which has a more or less parallel-sided bow and no lateral knobs, except on the head. The chief incidence of these brooches is under Claudius and Nero, though some may have been Tiberian. They probably do not survive into Flavian times. APPENDIX IV THE COINS By BERTRAM W. PEARCE, M.A., F.S.A. The coins under review in this article are those found at LuUingstone in the year 1949 only and number 83 of which 2 are post-Roman and 15 indeterminate or having collapsed under treatment. Of the indeterminate coins 5 are minims ranging from 4 to 7 mm. in diameter— these are perfectly plain. Evidence is accumulating that such very 28 THE LULLINGSTONE ROMAN VILLA small coins need not be of post-Roman date, but may have been used for small change in the later Roman times. Three coins only are Theodosian A.D. 379-95, 6 of the Valentinian period A.D. 364-83, while 6 belong to the second half of the third century. The rest were issued during the reigns of Constantine I and his sons, divided nearly equally between the two periods A.D. 320-37 and 337-61. Usually more of the latter group are found, the others being larger and heavier, and tending to be withdrawn from circulation and melted down as soon as the smaller type became current. The greatest density of occupation therefore appears to have been from A.D. 320-30, allowing a few more years for the coin drift period, i.e. the time needed for the coins to get into general circulation. Note that several are from the London Mint. None of the Constantinian coins appear to have been in circulation long, the general condition being good. On the reverse of the Crispus, Cohen 13 coin with a BE AT A TRANQVILLITAS reverse, the three stars are found inside the inscription, whereas usually they form part of the same circle. Maurice (Numismatiques Constantiniennes) figures both arrangements cf. Vol. II, PL IV, but as far as I know does not comment on the matter. I have noted two similar coins from Verulamium (Numismatic Chronicle, 1948, Parts I-II, pp. 88-9), all three coins being of Lugdunum minting. The coin finds give no proof of occupation earlier than the reign of Carausius, as smallish third century radiates seem to have circulated during his reign. There was probably an active occupation during the period A.D. 320-30, diminishing gradually till the beginning of the reign of Valentinian I after which it became intermittent, the Theodosian and smaller coins being possibly due to the use of an abandoned site by occasional squatters. The references are R.I.O = Roman Imperial Coinage—Mattingly and Sydenham and others. Cohen=Metafiles Imperiales, Cohen. Emperor Gallienus Claudius II Date A.D. 260-8 268-70 Coin Antons. Obverse and Reverse GALLIENUS AVG. Head r. radiate. SECVRIT PERPET. Sst. 1. with sceptre, leaning on column M.M. | H IMP. CLAVDIVS AVG. Head r. rad. LAETITIA AVG. L. st. 1. with wreath and anchor. M.M. | XII Ref. R.I.O. 280(s) 1 56 1 29 THE LULLINGSTONE ROMAN VILLA Emperor Tetricus 1 Carausius Constantine I Date A.D. 268-73 287-93 307-37 1 1 i 1 I i Coin Antons. M 3 ,, n n it Obverse and Reverse IMP. C. TETRICVS P.F. AVG. Bust r. rad. dr. AEQVITASAVG. M st. 1. with scales and cornucopiae. Rev. FIDES MILITVM. F. st. 1. holding two ensigns. Rev. PAX AVG. (2) CONSTANTINVS AVG. Bust r. helm. cuir. BEATA TRANQVILLITAS. Globe on altar with VOTIS XX; 3 stars above -PTRA. D. 320-4 do. M.M. STR. CONSTANTINVS AVG. Head r. laur. SARMATIA DEVICTA. Victory adv. r. with trophy and palm : in front, a bound captive. M.M.STRiy, -PLG-(2) CONSTANTINVS AVG. Bust r. helm. cuir. VIRTVS EXERCIT. Two captives at foot of standard inscr. VOT XX. M.M. PLN. CONSTANTINVS AVG. Bust r. helm. cuir. VIRTVS EXERCIT, as above but trophy M.M. STR. CONSTANTINVS AVG. Head r. laur. D.N. CONSTANTINI MAX AVG in wreath. M.M. S*TR. CONSTANTINVS AVG. Bust r. laur. dr. PROVIDENTIAE AVGG. Camp Gate M.M. PTR-A.D. 324-6. VRBS ROMA. Bust 1. helm. cuir. (2) Wolf and twins M.M. TRS w PLC. A.D. 330-7 Ref. 52 1 cf. 33 3 C20 8 J* 487 690 697 123 454 2 30 THE LULLINGSTONE ROMAN VILLA Emperor Date Coin Obverse and Reverse Ref. I A.D. Helena Theodora Constantine I (con) Crispus (c] MS (small) M 3 Constantine II as Caesar M 3 CONSTANTINOPOLIS. Bust 1. helm with sceptre. Rev. Victory on prow. A.D. 330-7. FL. HELENA AVG- 12 VSTA. Bust r. diad. dr. SECVRITAS REIPVBLICAE. S. with olive branch. M.M. SARL. A.D. 324-6. FL. MAX. THEODORAE AVG. Bust r. dr. PIETAS ROMANA. Empress holding child. M.M. TRS. A.D. 330-7. j i CRISPVS NOB. CAES. j C 5 Head r. laur. BEATA TRANQVILLI- ; TAS. As above but j stars inside inscription, j M.M. 2. H A.D. 320-4 C. PLG CRISPVS NOBIL. Bust 1. helm. cuir. as above. P.LON. IVL CRISPVS NOB. CAES. Head r. laur. CAESARVM NOSTRORVM in wreath round VOT X MM 1 SCON A.D. 320-4 FL IVL CRISPVS NOB. C. Bust r. laur. dr. VICTORIAE LAETAE PRINC. PERP. Two victories placing shield with VOT P R on altar M.M. PLN CONSTANTINVS IVN. N.C. Bust 1. rad. dr. Rev. BEATA TRANQVILLITAS as above M.M. F I D PLON A.D. 320-4 CONSTANTINVS IVN. NOB. C. Bust r. laur. cuir. C 13 44 cf. 153 C5 C 164 31 THE LULLINGSTONE ROMAN VILLA Emperor Constantius II as Caesar Constantius II as Augustus Constans as Augustus House of Constantine Date A.D. 337-361 337-50 i j 1 Coin M 3 ft JJ M2 JE 3 Obverse and Reverse PROVIDENTIAE CAESS. Camp gate PLON 324-6. As above. Rev. GLORIA EXERCITVS. Soldiers guarding 2 standards. MM TR-S A.D. 330-7. FL. IVL. CONSTANTIVS NOB. C. Bustr. laur. cuir. As above •PLG. FEL. TEMP. REPARATIO types CONSTANTIVS. P.F. AVG. Bust r. diad. dr. A in 1. field. Legionary spearmg fallen horseman. MM. LVG P. r in field. Legionary spearing fallen horseman. One M.M. D | (4) PCON Rev. GLORIA EXERCITVS. Soldiers guarding labarum. FL. IVL. CONSTANTIVS. AVG. Bust r. | laur. cuir. \ VIRTVS AVGGNN. V st front with shield and i E 4 M 3 reversed spear. Rev. GLORIA EXERCITVS, standard, barb. CONSTANS P.F. AVG. Bust r. diad. dr. VICTORIAE DDAVGGQNN, 2 Victories (4) D • M.M. TRS (2), TRS, Indeterminate (1) A.D. 341-5. Rev. FEL. TEMP. REPARATIO Legionary spearing fallen horseman : one large, the rest l l | t o 15 mm. (10) do. GLORIA EXERCITVS. 2 standards (1) Ref. C 122 C 104 1 7 C 44 C45ff C 92 C303 5 Cohen 179 15 32 THE LULLINGSTONE ROMAN VILLA Emperor Magnentius Valentinian I Valens Gratian Date A.D. 350-3 364-75 364-78 367-83 Coin M2 i E 3 M 3 M 3 J 1 3 M 3 Obverse and Reverse Rev. GLORIA EXERTVS, 1 standard (2) do. VICTORIAE DD AVGGQNN (2) M.M. D , M TRS / / / / // D. N. MAGNENTIVS P. F. AVG. Bust r. bareheaded, dr. VICTORIAE D. D. N. N. AVG. ET. CAE. Two victories with a wreath VOT.V. MVLT. X. M.M. AMB. As above M.M. R.P.L.G. D.N. VALENTINIANVSP. F. AVG. Bust r. diad. dr. GLORIA ROMANORVM. Emp. holding labarum and dragging captive r. MM. 0 | F II LVGSP Obv. as above. Rev. SECVRITAS REIPVBLICAE. Victory moving 1. OF | I SCON D.N. VALENS P.F. AVG. Bust r. diad. dr. SECVRITAS RELPVBLICAE as above OF | I CON D.N. GRATIANVS AVGG. AVG. Bust r. diad dr. GLORIANOVI SAECVLI. Emp. st. front, with labarum and shield. M.M. OF | I I I 367—79 (2) TCON D.N. GRATIANVS P. F. AVG. Bust r. diad. dr. SECVRITAS REIPVBLICAE as for Valentinian. M.M. PCON. A.D. 388-96. Ref. C68ff 2 C 12 2 C 37 C 47 1 C 13 3 C34 33 THE LULLINGSTONE ROMAN VILLA Emperor Valentinian II Theodosius Arcadius Indeterminate Date A.D. 383-92 379-95 395-408 ! Coin & 4 M4 M, 4 THE COINS—STTMM Roman Third Century Constantine I House of Constantine House of Valentinian House of Theodosius Total identifiable Roman indeterminate Post-Roman. M. Edwar Jetton. Anglo-E d I I Ha rench. Obverse and Reverse D.N. VALENTINIANVS P.F. AVG. Bust r. diad. dr. VICTORIA AVGGG. Victory moving 1. MM. PCON. A.D. 388-95. D.N. THEODOSrVS P.F. AVG. Bust r. diad. dr. In a wreath VOT V MVLT X D.N. ARCADrVS P.F. AVG. Bust r. diad. dr. SALVS REIPVBLICAE. Victory with trophy moving 1. and dragging captive g | Ref. C 46 1 C 68 1 Saba- 1 tier 41 ////// All probably C 4, mostly j 15 late including 5 minims &RY AND PERCENTAGES _ ..... Totals Percentages 6 9-1 22 33 -3 29 43 -9 6 9-1 3 4-6 66 100 15 lfpenny. London Mint. Fourteenth-fifteenth centi try. 34 THE LULLINGSTONB ROMAN VILLA APPENDIX V THE LULLINGSTONE BUSTS By Miss J. M. C. TOYNBEE, M.A., D.Phil., F.S.A. Among the most important results of the 1949 season's work was the discovery, in the Basement Room, of two large, male, Roman portrait-busts. Not only have no works of art comparable to them been yielded by any other Roman villa so far excavated in this country, but they are also remarkable for the number of questions which they raise concerning the history, social status, and mentality of the villa's owners during the main phases of its occupation. Both busts are of Greek, probably Pentelic, marble1 and are over life-size. They were clearly meant to stand in recesses or niches and to be viewed only from the front; for in each case the body is concave behind, the weight of head and neck being supported on a rectangular shaft which runs up the centre of the hollow and is firmly based upon a circular, moulded pedestal. In front, between the upper surface of this pedestal and the terminal line of the bust, is a horizontal plaque, with a moulded border above and below and a £>eZfo-shaped indentation at either end. Such plaques are occasionally inscribed with the name of the person portrayed2 ; but here, as in the case of the great majority of Roman portrait-busts, the marble is tantalizingly dumb. That the Lullingstone busts were not made as pendants is evident from the fact that they are of shghtly different sizes and exhibit divergencies in technique and style. Both were obviously intended to be realistic likenesses. BUST I. (PL II.) The smaller and more perfect of the two busts measures c. 28 in. in height, from the crown of the head to the bottom of the pedestal, c. 8 in. from the back of the head to the greatest point of projection of the brow, and c. 18 in. across the upper arms and chest at their greatest extent. The head and body are carved from a single block. Apart from the loss of the nose and abrasures in the beard between the right-hand corner of the mouth and the tip of the chin and again on the right side of the face, the piece is comparatively well preserved. The 1 Mr. Bernard Ashmole, Keeper of Greek and Roman Antiquities in the British Museum, compared the busts, when they were temporarily lodged in the Museum, \\ ith samples of Greek marbles and came to the oonclusion that they are of Pentelic marble. The present writer has had no opportunity of checking this comparison personally ; but the appearance of the marble in which the busts are carved does, in her opinion, fully support Mr. Ashmole's view. a E.g. the bust of a young Greek, C. Voloacius Myropnous, in the Ostia museum (R. Oalza, Museo Ostiense (Itinerari dei Musei e Monumenti d'ltalia, 79) 1947, No. 38, PI. 43). 35 THE LULLINGSTONE ROMAN VILLA form of the bust, terminating, not at a point, but in a roughly horizontal line, and comprising part of the upper arms with most of the chest, but finishing off well above waist-level, is generally characteristic of the Hadrianie and early- and mid-Antonine periods. The drapery consists of tunic and cloak (sagum), the latter being secured by a large, round brooch on the right shoulder and showing on the left shoulder a fringed border. The head, slightly turned towards the spectator's left, reveals a benign, if serious and somewhat dreamy, countenance, characterized by a slightly protruding brow, deep-set eyes, the pupils of which are each rendered by two very shallow, linked depressions, a longish upper lip, and a thin, wide, and firmly-set mouth, almost straight at the centre, but drooping shghtly at its corners. The hair is combed over the crown of the head in thick, curling locks, which grow luxuriantly upon the nape of the neck and form a neat fringe low over the brow. The beard, also curly, is short and cut to a blunt point on the chin. Neither hair nor beard is heavily drilled or undercut. The head, as a whole, exhibits the soft texture and gently modulated surfaces which we associate with Roman portraits of east-Mediterranean provenance1 : we miss here the hard, metallic brilliance, characteristic of so much Italian and western second-century work.2 Who is the serene and kindly individual here portrayed ? At the time of its discovery, attempts were made to identify our portrait, immediately recognizable as dating from between the early twenties and late seventies of the second century, with a contemporary Emperor, with Antoninus Pius3 or Marcus Aurefius.4 But these equations, while understandable as an instinctive reaction to the dignified and majestic gravity of the subject, cannot actually be maintained. If we place the Lullingstone bust beside a representative series of portraits of Antonius Pius,5 we observe that in the former, as compared with the latter, the face is longer and narrower, the hair over the brow grows considerably lower and flatter and is handled in a much less rich and plastic way, the beard is longer and more pointed, and the crown of the head is more convex, while the pupils of the eyes are less strongly marked. Even the cuirassed bust of Pius from the Palatine,6 to which our bust approximates most closely, differs from it in these respects. The resemblance is, indeed, a purely general one, confined to the mild and 1 E.g. the bust of Hadrian in the National Museum, Athens (A. Hekler, Greek and Roman Portraits, 1912, PI. 258a). 2 E.g. the early head of Hadrian recently discovered near the Stazioni Termini in Rome and now in the Museo Nazionale Romano delle Terme (B. M. Felletti Maj, " Un nuovo ritratto d'Adriano " (Arli Figurative ii, 1946, p. 22 f£., Pis. 9, 10)). 3 The Illustrated London News, 30th July, 1949, p. 171. 4 The Times, 1st August, 1949. 6 E.G. M. Wegner, Die Herrscherbildnisse in antoninischer Zeit, 1939, Pis. 1-9. 8 Ibid., PI. 8. 36 THE LULLINGSTONE ROMAN VILLA almost deprecating expression and slight turn of the head. Still less convincing is the confrontation of the Lullingstone personage with portraits of Marcus Aurehus.1 Again we glimpse a fleeting likeness between the former and the philosopher-Emperor in the shape of the face and meditative air. But we do not find here either that pronounced arching of the eyebrows, which stamps the face of Marcus with its suggestion of perpetual surprise, or the deep drilling out of the pupils of the eyes and tumbled, bushy appearance of hair and beard which are the hall-mark of mid-to-late-Antonine imperial portraiture. Aelius Csesar and Lucius Verus (to name other less likely, if possible, candidates of the Antonine House) are both ruled out by their crisper, curlier hair and longer, more luxuriant beards. The features of the Lullingstone bust are patently not those of Hadrian ; but it is to his time, rather than to that of his two immediate successors, that it must, on stylistic and technical grounds, be assigned. The absence of heavy drilling and of undercutting, and the shy, tentative experimenting in the plastic rendering of the eyes, all point to the early Hadrianic period. Another pre-Antonine trait is the flat, plastered-down effect of the locks on the brow. We meet this on the bust of an unknown Roman, labelled " Aelius Caesar ", in the Ny Carlsberg Glyptotek in Copenhagen* and on two figures in the Hadrianic " tondi " on the Arch of Constantino, that of the bearded man immediately to the right of the Emperor in the lion-hunt scene,8 and that of the bearded man on the extreme left in the scene of sacrifice to Apollo.4 We discover the same feature on the marble imago clipeata of an elderly, beardless man, of Trajanic date, recently found at Ostia.5 We may, then, describe Bust I, carved probably not later than c. 125 to 135, as depicting an unknown Roman of distinction, aged about forty-five or fifty, and wearing semi-military dress. The fringed sagum, while frequently worn by Emperors, is not their monopoly. It is to be seen, for instance, adorning the persons of officers, or officials, in the train of Septimius Severus on a frieze from that Emperor's four-way arch at Lepcis Magna.6 BUST II. (Pis. I l l and IV.) The second of the Lullingstone busts is decidedly larger than the first and is much less well preserved. Not only are the surfaces of its head more weathered, but the head and body had parted company in 1 E. G. M. Wegner, Die Herrscherbildnisse in anioniniseher Zeit, 1939. Pis. 19-33. 2 R. West, Bdmische Portratplastik, ii, 1941, PI. 42, 156; Ny Carlsberg Qlyptotek : Billedtavler til Kataloget over antike Kunstvaerker, 1907, PI. 57, 687. 3 Revue Archiologique xv, 1. 1910, PI. 8, No. 10. 4 Ibid., PI. 4, No. 4. 6 R. Calza, op. cit., No. 56, p. 46. 0 Africa Italiana iv, 1931, p. 97, Fig. 133. 37 THE LULLINGSTONE ROMAN VILLA ancient times. The flat surface of the shaft at the base of the neck on the head appears to fit the corresponding flat surface on the body exactly, portions on either side, where the drapery passed across the top of the shoulders, having been lost, while the upper edge of the drapery passing round the back of the neck is preserved on the head. It looks as if the head had been severed from the body by a single, straight blow. On both flat surfaces there are traces of weathering. That the two parts did originally belong together is demonstrated by PL IV.1 The head measures c. \6\ in. in height, c. 9 in. from the back of the head to the most prominent point of the brow, and c. 8 | in. from ear to ear. The nose is lost. The body, as it has survived, is c. 19£ in. high and c. 17 in. wide at its greatest extent. It shows an ancient break down the right side of the chest, while the slope of both shoulders had been deliberately and symmetrically shaved down by tools, also in ancient times, but before the break took place, since, unlike the latter, the tooled surfaces show signs of wear. It is this narrowing of the shoulders which gives the body its present effect, as seen from the front, of being too small for the head. The drapery of Bust II is of a purely civilian character, consisting of tunic and toga, the latter worn with a stiff, vertical fold falling from the left shoulder and a twist in the transverse fold across the chest.2 The face is distinctly less genial than that of its companion and portrays a man about five years the latter's senior. The hair is straighter and, even allowing for the greater weathering of its surfaces, somewhat less luxuriant, and it does not fall so low over the brow, which is furrowed by two deeply incised, horizontal hnes. The beard is square-cut, and both it and the hair are heavily drilled. Deeply drilled holes mark the pupils of the eyes and lend a somewhat severe and peremptory expression to the countenance. The fact that shghtly more of the chest and upper arms are shown than in Bust I, the use of heavy drilling to pick out the locks of hair and beard, and the more realistic and pictorial rendering of the eyes, all combine to indicate a later date.3 We might assign Bust LI to c. 155-165, or a little later, at any rate to a generation later than Bust I. The texture of the carving of the head closely resembles that aheady noted on the earher bust. 1 The writer is greatly indebted to Mr. Richard Holworthy, Archivist to the Kent County Council, who arranged for the photographirg of the reconstruction of Bust II, and to Mr. Reginald Davis of Maidstone, who prepared the drawing of Bust II with the missing portions conjecturally restored. 2 For this style of toga cf. an Antonine bust in the Capitoline Museum (ed. H. Stuart Jones, Catalogue of the Museo Capitolino, 1912, p. 132, No. 57, PI. 32) ani the figures of Hadrian in the adlocutio and adventus reliefs in the Palazzo dei Oonservatori (ed. H. Stuart Jones, Catalogue of the Palazzo dei Conservatori, 1926, I'l. 105, ScalalV, 1 ; PI. 12, Scalall, 12). 3 Cf., for style and technique, portraits of Marcus Aurelius (M. Wegner, op. cit., Pis. 20, 25, 28, 29 ; Ny Carlsberg Olyptotek, etc., PI. 58, 7). 38 THE LULLINGSTONE ROMAN VILLA If Bust I is not to be identified as the portrait of any member of the imperial family, this is still more obviously the case with Bust II, the features and expression of which bear not the slightest resemblance to those of any Emperor or prince of the Antonine House. Between the profile view of Bust II and the coin-portraits of Clodius Albinus (193-7)1 there is, indeed, some likeness. But we miss in the marble Albinus' markedly protruding brow and more flowing and luxuriant beard, quite apart from the fact that the style of the bust, while not completely excluding an early-Severan date, is more appropriate to the third quarter of the second century On the other hand, although the head of Bust I is distinctly rounder than that of Bust II, and its brow somewhat squarer, the repetition in the later bust of certain facial traits already noted in the earher one suggests that a bloodrelationship may have existed between the two persons portrayed. We observe once more, in Bust II, the deep-set eyes, the shghtly prominent forehead, the long upper lip, the same modelhng of the cheeks, and, above all, the same wide, straight-set mouth with drooping corners. It is tempting to see in the Lullingstone busts family portraits dating from two successive generations—perhaps of father or uncle, in the one case (Bust I), and of son or nephew, in the other (Bust II), with a difference of about twenty-five years in age between them. We have, then, in a Roman villa in west Kent, less than twenty miles from London, two large marble busts portraying private individuals, probably related, wrought in Greek marble, and exhibiting a technique which is at any rate suggestive of east-Mediterranean workmanship. It cannot be too strongly emphasized that we have as yet no absolute proof that the busts reached Lullingstone before the mid-third century (vide infra p. 41) ; and that the theory of their earher history and movements here put forward represents no more than the most likely working hypothesis. But it would seem to be, at the least, improbable that they were carved in Britain, for romanized natives, by imported artists working on imported marble. The subjects are distinguished persons of a Mediterranean type, wearing Roman official costume ; and it would appear reasonable to suppose that the busts represent part of the family portrait-gallery of some person in the imperial service, stationed in Britain for a longish term, and imported by him into the province. Such a person might have worked in Londinium and used the Lulhngstone villa as a holidayresidence, or place of retirement after his term had expired. As we have seen, the later of the two busts was, at some point, modified by the shaving down of its shoulders, as though to adapt it to a niche or recess for whiohithadnot been originally intended. Whetherthis modification 1 H. Mattingly and E. A. Sydenham, The Roman Imperial Coinage, iv, i, 1936, pis. 2, Nos. 9-22 ; 3, Nos. 1-3 ; 4, Nos. 3-10. 39 THE LULLINGSTONE ROMAN VILLA took place at Lullingstone, or elsewhere in Britain, or before it reached the province, we cannot say. It is possible that Bust II portrays the onetime owner of the villa, who might have come to Britain c. 165-170 ; while Bust I portrays one of his relatives, who need never have visited Britain or Lullingstone in person at all. It would appear that by c. 190 the site was temporarily abandoned (p. 13), an occurrence which could have been connected with the mutinies among the British troops after Ulpius Marcellus' recall from the island c. 185. If we can credit Dio's story (lxxiii, 9) of a delegation of 1500 men sent from the British garrison to Commodus in Rome at this time, insubordinate soldiery may have been ranging the countryside of southern Britain; and we know that Pertinax, despatched by Commodus in 185 to restore order in the province, had the greatest difficulty in executing his task (Dio lxxiii, 9 ; Scriptores Historiae Augustse, Vit. Pert. 3). In such circumstances the owner of Lullingstone might have thought it prudent to quit hurriedly for the continent, leaving his bulky family portraits to their fate. In the abandoned villa these portraits would then have remained, exposed to the onslaughts of the weather and to the attentions of human intruders, until the house was re-occupied, apparently c. 250 (p. 13). The wear and facial damages displayed by both busts, and the rending assunder of head and body in the case of Bust II, could easily have taken place during this interval of roughly sixty years. The blow which severed the neck of Bust II may well have been deliberate ; but the fact that the features were not defaced would seem to rule out the idea of mutilation from pohtical motives. That these disasters were, indeed, sustained by the busts before their migration to the Basement Room is rendered highly probable by the fact that no trace whatsoever of marble fragments or chippings was found in or near the place of their discovery. Bust I and the body of Bust II were found lying face-downwards on the flight of three steps originally leading down into the Basement Room from the north, but blocked, during the second half of the second century, by a wall (p. 13). They had stood on the top and bottom step respectively, and were knocked over, Bust I in a westerly, Bust I I in an easterly, direction, when the ceiling of the Basement, and the contents of the fourth-century apartment built above it, collapsed. The righthand portion of the body of Bust I I was snapped off on the edge of the lowest step, a clean, sharp break being produced. The head of Bust II was carefully placed in an upright position on a small, square stand, or podium, immediately to the east of the steps, so as to rest against the north wall of the room. The fact that this head is somewhat more weathered than are its body and Bust I may mean that the podium was more exposed to the action of wind or water than were the steps themselves. There was no sign of any intention of hiding the marbles. 40 THE LULLINGSTONE ROMAN VILLA Not only had they obviously been arranged and displayed in these positions with dehberate care, but opposite the centre of the steps three votive pots were sunk into the floor of the room, in a straight row, one behind the other, and clearly aligned with the busts ; while a fourth pot was sunk as nearly opposite the head of Bust II as the early wallfoundation, which ran across the room from north to south, would seem to have aUowed (PL Via and Figs. 2 and 4). Of the three pots opposite the steps, the most southerly one, a third-century cooking-pot, and the central one, a Rhenish beaker of the mid-third century, with SVAVIS painted in its six indentations and the rib-bone of an animal inserted into it (p. 13, PL Via), were sunk into the original concrete floor of the room and sealed by the later clay floor of Constantinian date. It therefore seems probable that the deposition of the busts in the Basement Room took place contemporaneously with the apparent reoccupation of the villa c. 250, when the wall blocking the Basement stairs was repaired and the walls of the room were roughly redecorated with bands of colour on a cream ground, this second scheme of decoration, but not its predecessor, being extended to the blocking wall. The remaining two pots, found respectively between the SVAVIS pot and the steps and just to the east of the early wall-foundation, are of coarse fourth-century ware : one was associated with a Constantinian coin and both were sunk through the Constantinian clay floor. Why were the busts deposited with such circumspection in the Basement Room, and what purpose, after their deposition, did that room serve ? Any attempts to answer these questions must be regarded as extremely tentative, pending further exploration of the room and of the house as a whole. But it would seem to be a reasonable deduction that the pots, so symmetricaUy and purposefully related to the busts, were votive in character and expressed an attitude of religio towards the marbles in the minds of those who owned the villa in the late third- and fourth-century phases of its occupation. The thirdcentury re-occupiers, presumably members of quite a new family, may, when entering into possession, have found the busts standing or lying about in the house. They might not have cared to adorn their own living-rooms with portraits of total strangers, weathered, and, in the case of Bust II, broken and battered at that. Yet to those newcomers the busts would have remained impressive and awe-inspiring reminders of earlier owners, whose manes were spiritual powers still to be reckoned with. To destroy or neglect such images would have been to court disaster. Only one course remained, and that was to relegate them respectfully to honourable retirement in some kind of " cenotaphmausoleum ", where the spirits of the departed could be propitiated with worship and gifts. We cannot completely rule out the possibihty that the mid-third-century occupants brought these second-century 41 8 THE LULLINGSTONE ROMAN VILLA portraits to Lullingstone from elsewhere, for the purpose of venerating them in the Basement Room. But such a procedure would seem to be, on the face of it, unlikely. The fact that the head and body of Bust II were not juxtaposed suggests that the persons who placed them in these positions did not know that they had belonged to one another. How the Basement was entered during its " mausoleum " stage has yet to be determined. The significance attached to the portrait, or imago, in the cult of the dead is a well-known feature of Roman rehgious thought and practice. The whole development of reahstic portraiture as an art in Rome, from the second century B.C. onwards (as distinct from the more primitive use of conventionalized ancestral funerary masks), was bound up with the growth of belief in individual survival after death and the repercussions of this belief on family worship. The reahstic likenesses of the dead, whether made posthumously or taken during their lifetime and preserved after their departure, were held to reflect, and, in a sense, to perpetuate upon earth, the imperishable soul and personahty now enjoying other-worldly beatitude.1 Of the application of these ideas in imperial times only a few examples can be quoted by way of iUustration here. For instance, we learn from Tacitus (Ann. xvi, 7) that, in the course of his denouncement of the " republican " C. Cassius in 65, Nero " obiectavit Cassio quod inter imagines maiorum etiam C. Cassi effigiem coluisset " ; while the Younger Phny writes (Epp. i, 17) of his contemporary Titinius Capito : " est omnino Capitoni in usu claros viros colere : minim est qua religione, quo studio imagines Brutorum, Cassiorum, Catonum domi, ubi potest, habeat." Imagines clipeatae, containing busts of the departed in roundels, were especially prominent as the central features of carved sarcophagi, both Italian and provincial, in the third and fourth centuries.2 One such sarcophagus, dating from the first half of the third century and displaying the portrait-busts of a husband and wife in its central medalhon, is of particular interest in the present context. It stands in the centre of a barrel-vaulted family tomb found at Weyden-bei-Koln and still preserved in situ.3 The sarcophagus is flanked by reproductions in stone of two high-backed wicker chairs, of a type frequently met with on Rhineland grave-reliefs and presumably intended here for the use of the departed souls when visiting their tomb. Round the waUs of this 1 Cf. I. A. Richmond, Archaeology, and the After-Life in Pagan and Christian Imagery (Riddel! Memorial Lectures), 1950, pp. 39, 40. 2 I. A. Richmond, loc. cit.; F. Cumont, Recherches sur le symbolisme fune'raire des Romains, 1942, pp. 486-7. The motif also appears upon fourth-century Christian sarcophagi (P. Ducati. L'Arte in Roma dalV origini al sec. viii, 1938, pi. 236, 1, 2). 3 E. Esperandieu, Recueil giniral des bas-reliefs, statues et busies de la Oaule romaine, viii, 1922, pp. 375-7, Nos. 6484-5; F. Fremersdorf, Die Denkmaler des romischen Koln, II, 1950. Pis. 72-76. 42 THE LULLINGSTONE ROMAN VILLA tomb are small niches for cremation-burials, while in three large recesses, one at the back, and one on either side, are shelves each supporting a family portrait-bust. The LuUingstone Basement Room was, so far as we know at present, not a tomb. But the resemblance of its late-third- and fourth-century set-out to that of the Weydenfunerary chamber, if remote, is none the less striking.1 It remains to add a brief comment on the busts as domestic decorations. Some evidence exists for a taste in marble statuary among the owners of Roman-British country seats elsewhere. Samuel Lysons describes the discovery in various apartments of the Woodchester villa of marble figures and groups, more modest in scale than our busts, but no less classical in character.2 These include part of a replica of a well-known group of Cupid and Psyche3; the feet of a statue on a circular base4 ; the leg of a statue6; and the headless and armless, but otherwise complete, figure of a goddess, probably Diana Lucifera, eighteen inches high.6 The first and last of these items are in the British Museum and carved in Luna marble. But the possibility opened up by the LuUingstone excavations, of viUas in this province adorned with over life-size famUy portrait-busts of high artistic quality, is a completely new one.7 The find has made a notable contribution to our knowledge of the high standard of classical culture maintained in these provincial and rural residences, on the outskirts of the Empire, throughout the history of the Roman occupation—a standard to which, for the early-fourth century, the Lullingstone pavements (Appendix VI, Pis. 16 and V) also bear impressive testimony. 1 The present writer was reminded of this tomb when seeing a view of it thrown upon the screen in a recent lecture by Miss M. V. Taylor. 2 S. Lysons, An Account of Roman Antiquities discovered at Woodchester in the County of Gloucester, 1797. 3 Ibid. PL 37, 1. 4 Ibid. PI. 40, 1. 6 Ibid. PI. 36, 3. 6 Ibid. Pis. 38, 39 ; British Museum Guide to Roman Britain, 1933, PI. 28. A recent examination of this figure in the British Museum does not confirm Lysons' suggestion that it is of Parian marble. 7 Portrait-galleries in country villas are not unknown on the Continent. The most remarkable instance from the western provinces is that of the great villa at Chiragan, near Toulouse, where there came to light more than seventy marble heads and busts, ranging in date from the age of Augustus to that of Gallienus. A number of these are portraits of Emperors, Empresses, and princes, while the rest appear to be family portraits of unknown men and women. Most of them are over life-size ; and while the great majority are of Italian marble, a few are of Greek marble. See Mdmoires prisentis par divers savants a VAcaddmie des Inscriptions et Belles-Lettres, Ire Serie, xi, 1901, pp. 329-44, pis. 16-25. 43 THE LULLINGSTONE ROMAN VILLA APPENDIX VI THE MOSAIC FLOORS By CREGOE D. P. NICHOLSON, F.R.Hist.S., F.S.A.(Scot.), F.S.G. The Mosaic Floor found in Room 10 is the finest floor that has hitherto been discovered in the County of Kent, and is entitled to rank amongst the best mosaic floors which have been found in this country (PL 16). Its state of preservation is remarkably good, if we except two holes driven through it in the middle of the eighteenth century during the erection of a deer fence. At one point a mole, or some other small animal, burrowing beneath it, had caused displacement of the tesserae but fortunately the part damaged was not of great importance and the stones were not disturbed sufficiently to destroy the design ; consequently accurate repair has been possible. The only other damage was at the top of the step leading into the apse, where a part of the design on the riser and a portion of the inscription above were destroyed. A small portion remaining intact at each end of the step established what the design had been at the top of the riser and the missing letters of the inscription have been supplied by Professor R. E. Mortimer Wheeler. The Mosaic Room is in fact two distinct rooms, the lower, which is rectangular, measuring 20 ft. 6 in. by 15 ft. 9 in., and the upper, which is an apse, 16 ft. by 20 ft. 6 in. The entrance from the lower to the upper room is by means of a step 9 ft. 6 in. long and 9 in. in height. At the top of the step, on entering the upper room, there is a panel of mosaic work measuring 8 ft. by 8 ft. and apsidal in shape (PL V6). This panel is unusual in its arrangement since it is intended to be viewed with the chord of the arc at the top. It is enclosed by a band of twisted gufiloche of a very uncommon pattern. The twists are open and the enclosed spaces are filled alternately with red and white. This design is very similar to one found on the painted wall plaster which is believed to have formed the frieze of a room. It is possible that one design was suggested by the other. Within this band is a representation of the abduction of Europa, which for boldness of execution and artistic effect is the equal of any Roman mosaic work ever found in this country. The scene depicts Europa, clad in diaphanous drapery, seated upon the back of Jupiter, who in the form of a very spirited bull is in the act of plunging into the sea. Europa is seated facing the bull's tail, with her right leg raised, but her head is turned and she is looking over her left shoulder. In her hands she holds a scarf or veil which is floating in the air above her head. The bull is galloping towards the right with 44 PLATE I - - • :.-;--,!.-?'-^W!,'"ilSi •fl*. • ' ...:„,,,.- ~' *S6 "**" — "' £• ' ' ^ " " ^ •* '4r wmm a. View facing N.W. i . Mosaic Floor, General View LULLINGSTONE ROMAN VILLA [face p. 44 a. Bust I ft. Bust I LULLINGSTONE ROMAN VILLA a. Bust II : head. b. Bust II : body LULLINGSTONE ROMAN VILLA a. Bust II : restoration 6. Bust I I : showing junction of head and body LULLINGSTONE ROMAN VILLA PLATE V a. Mosaic. " Bellerophon" *VH$ytMttl*t*>xMitM* b. Mosaic. " Europa" LULLINGSTONE ROMAN VILLA PLATE VI «9«rfW3&S^ a. Basement room looking N.W. showing Busts as found . f-rT-P 6. Ingot Hoard c. Bronze Finial and Fibulae LULLINGSTONE ROMAN VILLA •tf a. Painted Wall Plaster 6. Painted Wall Plaster LULLINGSTONE ROMAN VILLA THE LULLINGSTONE ROMAN VILLA his fore-feet in the air and his hind hoofs in the sea. On the left stands a winged cupid, holding the bull's tail and apparently trying to puU it back. This figure may be symbolical of restraint or caution. On the right, in front of the bull, stands or dances another winged cupid, holding what appears to be a torch in the left hand and with the right hand beckoning on the lovers. This figure probably represents abandonment. All the figures are outlined in red, instead of the usual black, which is probably due to the fact that they lie partly over the sea, which is executed in dark blue tesserae, and a black outline would not have shown up against such a background. The hair and faces of the figures include dark purple and several shades of yellow; the latter perhaps Itahan marble. Above this picture is a two-line inscription running across the top of the panel. This had been damaged and a few letters had been lost, but Professor R. E. Mortimer Wheeler, who kindly examined the floor, was able to supply the missing word. The inscription, as restored, reads INVIDA SI (TAVRI V)IDISSET IVNO NATATVS IVSTIVS AEOLIAS ISSET AD VSQVE DOMOS ("If jealous Juno had seen the swimming of the bull, she might more justly have gone to the halls of JSolus.") Dr. Wheeler points out that this is a reference to the first book of the Mneid, " where the hostile Juno appeals to iEolus, god of the winds, to raise a storm in order to overwhelm JSneas on his voyage to Italy, with the implication that she might, with greater provocation, have made a similar appeal had she seen her wayward consort eloping to Crete with the fair Europa on his back ". This reference to the Mneid is of particular interest in view of the discovery at Otford, in 1927, of two fragments of painted waU plaster showing part of an inscription, which was identified as a quotation from the Mneid, and a portion of a human figure believed to represent iEneas or Achates (Arch. Cant., XXXIX, p. 153). Otford is situated 3 | nfiles south of Lullingstone. The recent discovery at Low Ham, in Somersetshire, of a mosaic floor entirely devoted to scenes from the Mneid may also be mentioned. The step joining the apse to the lower room had been covered with an embattled design in red and white, in two rows. The upper row had been destroyed except for a small portion at each end of the step. In the lower room, which is rectangular, mosaic work covers the greater part of the floor, with only a narrow surround of coarse red tesserae on three sides. The designs face towards the apse, which is on the western side, and are enclosed within a Greek-key border which runs around three sides of the floor but does not continue in front of the step. This floor is divided into two distinct portions, the eastern half 45 THE LULLINGSTONE ROMAN VILLA * containing pictorial panels and the western a series of small designs which are repeated so as to form a general patchwork. The centre of the eastern half contains a cushion-shaped panel formed by a band of guiUoche twist. This form of panel, which has also been found at Brislington, Somersetshire, is very uncommon. The guiUoche surround contains only two coloured threads—red and white—instead of the usual three. Within this panel is another spirited scene : Bellerophon killing the Chimsera (PL Va). BeUerophon, mounted upon Pegasus, is wearing a dark red garment and buskins. In his right hand he holds a long spear, made of alternating red and white tesserae, with which he is piercing the back of the Chimsera. With his left hand he holds the reins, which are of red tesserae ; perhaps to emphasize the fact that it is the magic bridle given to him by Minerva with which to capture Pegasus. The attitude of Pegasus, flying through the air, is very similar to that of the bull in the apsidal panel. The mosaic work is remarkably fine, some of the tesserae being only one-eighth of an inch square. The Chimsera is shown with the head of a lion breathing fire, a goat's head growing from the middle of its back where the spear is entering, and a serpent's head at the tip of the tail. Four porpoises or dolphins fill the four corners of the panel and are probably intended to signify the sea voyage which Bellerophon undertook from Argos to Lycia. Two small red " pillars " may be symbolical of his entrance into the wilderness in pursuit of the Chimsera. This cushion-shaped panel was surrounded by four medallions containing the four Seasons. Of these only three remain, the head of Summer having been destroyed by one of the posts of the deer fence already referred to. Spring is represented by the bust of a young woman with a swallow on her right shoulder. Autumn by a middleaged woman with corn in her hair, and Winter by an old woman wearing a cowl. The latter head is very similar to that found at Bignor. Autumn was damaged by fire and Winter was partly destroyed by one of the deer fence posts. On either side of this central group were strips of simple chequer pattern in red and white and black and white. In the middle of each strip was a small oblong panel which appears to have contained a cantharus, but only a small portion of one panel has escaped the post holes. The western half of the floor is a strange medley of crosses, hearts, swastikas and a number of other small devices contained in squares and octagons. The impression which this produces is that the mosaic workers, having completed the main picture, were only concerned with filling in the remaining space. The whole floor is enclosed on the north, east and south sides by an 46 THE LULLINGSTONE ROMAN VILLA unusual form of Greek-key border. The " key " is in white, edged with red, on a white background except on the south side, where the background is yeUow. This border design has strange breaks in its continuity which at first led to the behef that the floor had been repaired at some time, but a very careful study of the border has revealed the fact that the breaks in the design are almost certainly due to errors in the layout by the original mosaic workers. Portions of the floor, especiaUy the apsidal panel and the medalhon containing the bust of Autumn, were damaged by fire probably by a burning rafter falling during the final destruction of the villa. An interesting feature is the presence of a large number of white tesserse from an earher, probably first century, floor, which are re-used in the Greek-key border. The date of the floor, judged only from the design and the workmanship, may be put somewhere about the year A.D. 300. APPENDIX VII THE PAINTED WALL PLASTER EROM THE BASEMENT ROOM By CREGOE D. P. NICHOLSON, F.R.Hist.S., F.S.A.(Scot.), F.S.G. The painted wall plaster which is the subject of this appendix was found in the Basement Room (Unit 4). It would appear from the condition of the plaster itself, and from the stratification, that it has not been collected from other parts of the site, and thrown into the Basement as a filling, but had fallen directly into the position in which it was found when the walls of the room above collapsed during the final destruction of the villa by fire. The heat of this fire caused some of the fragments to change thencolour— reds and blues, for instance, to dark olive green, which has greatly increased the difficulty of reconstructing the original design. The burning has, however, thrown some hght upon the sequence of events in the final catastrophe. Fragments of the plaster untouched by the flames and retaining their original colours have been found to fit together with surrounding fragments which have been changed in colour by the heat. This could only have occurred if the plaster had first fallen into the Basement and some fragments been covered by the debris and protected from the flames whilst others were scorched by the burning timbers from the upper room. Much of the plaster has a backing of clay and it is probable that on two of its sides the upper room had timber and clay walls with a covering of painted plaster. When the timbers had burnt through, the 47 THE LULLINGSTONE ROMAIC VILLA waUs would faU into the Basement where the timber would continue to burn in contact with the plaster. So far as the reconstruction has at present been carried, it is not possible to give a complete picture of the decorative design. Numbers of groups of fragments have, however, been fitted together and these have revealed details of remarkable interest and enable a rough idea to be formed of the general appearance of the waUs. I t is important to bear in mind that plaster from at least three of the four waUs of the upper room, and perhaps some from the waUs of the Basement itself, which also had painted plaster, are mingled together in the confused mass of over five thousand fragments which has been recovered. This may mean that three distinct designs have to be recognized and separated, unless, as in the case of the fresco of the Dionysian Mysteries in the ViUa dei Misteri near Pompeii, the painting formed one continuous scene running round the four waUs of the room. A large number of fragments has been discovered of what appears to be the Dado. This consisted of rows of flowers, rather hke a wildrose, joined by a stem. The flowers ascend from floor level to a height which has not yet been established, but which was probably from 3 feet to 3 feet 6 inches. The rows were inclined to the left at an angle of about 45°, and between each row of flowers was a row of leaves or buds. The whole design is executed with a bold stroke in a dark brown pigment on a bluish-white background. The centres of the flowers are roughly painted in blue, purple, red, green, or yellow. At the top of the dado there is a dull green band, followed by a dark red band 2 inches deep. Above this there is a white line half an inch wide and then another band of red, which is of a fighter shade. The depth of this second band is 4 inches. Above this dado came the main pictorial zone. This would have been about 3 feet 6 inches to 4 feet high. It contained a number of pillars of the Ionic order (PL VII6). These differ in size, which may be due to their representing a colonnade receding into the distance. The columns are of various colours. Some are yeUow, shaded and fluted in brown and white, others are dark blue, shaded in black and white. Two have a fish scale pattern. Of these one is in red and the other is a dark olive green. The latter is undoubtedly a case of the colour having been changed by heat. Above the capitals were architraves in various colours, and between the pillars a coffered ceiling was visible. The spaces between the pillars contained human figures and animals. One human figure, which has been reconstructed down to the waist with the exception of the forearms, shows a young man dressed in blue and brown, wearing a deep sash. His apparel is heavUy trimmed with pearls or buttons (PL Vila). Another, much smaller figure, appears to be wearing a cuirass, but it is at present too imperfect to 48 THE LULLINGSTONE ROMAN VILLA permit more than a guess. There were probably two or three female figures. One of the most interesting features is a wide band with a leaf and bud pattern. This is executed in bright blue, brown, yellow, red and black. It has so far been impossible to decide whether it formed a circular medallion, an undulating band or an arch. It seems to have rested upon the capital of one of the pUlars, and upon a branch which springs from it is seated a bird, perhaps a pigeon, eating seeds. An interesting feature of this band is the fact that in many places there is visible a line marking the curve, which has been scratched with a pointed instrument to serve as a guide to the artist. Other features, at present only partly recovered, include a halftimbered house ; a garden containing lilies on the bank of a river ; a dog ; two more birds and a quantity of plaster which seems to have formed a splayed window or door-way. Above this pictorial zone there was probably a frieze and to this may be assigned a curious and very ugly design consisting of two bands of guiUoche work. Each twist of the guiUoche is nine inches long and the space between the bands is painted purple. The whole frieze was about 18 inches high. If these calculations of the height of the various parts of the design are correct, the whole wall must have been between 8 and 9 feet in height. 49 o