The winery complex sits at the top of ridge at the centre of the estate, a location now occupied by several nineteenth-century buildings set around a courtyard. Successive investigations in 2006–2010 excavated as much of the complex as the nineteenth-century buildings permitted. This has meant that in certain areas our understanding of the building and its function has been frustrated by a lack of access to the archaeology due to modern buldings and intrusions. Fig. 1 shows the trench locations and the locations of the standing buildings, while Fig. 51 gives an overview of the excavation of Area A.
This section gives a stratigraphic summary of all of the trenches excavated in the areas which are considered to constitute part of the Winery and aims at providing an overview of the stratigraphy of the complex. Stratigraphy from the following trenches has been considered: AI–V, GI–VI.  This report is based directly on the trench summaries produced by the respective trench supervisors (AI, Andrea Di Miceli; AII, GI, II and V, myself; AIII, Federica Romiti; AIV, Raffaele Laino; GIII, Dirk Booms; GVI, Seth Bernard) . Where possible, contexts have been grouped into sets of stratigraphically similar contexts and these have been phased and given an approximate date based on pottery, finds, or just their relative stratigraphy.

The site datum point was located at the threshold of the granary, at 212.39 m above sea level. It is roughly 20 cm above the main Roman floor levels of the building.

Eight basic phases have been identified, and are discussed in chronological order below. Lack of accurate dating has meant the phases are wide and based primarily on relative stratigraphy, brick stamps and construction technique rather than on finds. Phase I consists of the original construction and operation of the winery complex. This was then followed, in Phase II, by fairly widespread alterations to the wine production operation; some represent minor improvements but some significant changes in the production and storage of the wine are also seen. Phase III saw the destruction and robbing of the marble facings of the complex and the storage dolia followed by the destruction and collapse of the building. The post-Roman occupation of the site is evidenced by various postholes and several structures across the area: it has been broken down into two very broad phases according to the limited dating evidence available: Phase IV, the early ninth century; and Phase V, between the eleventh and the thirteenth centuries.

Finally a substantial body of evidence was identified associated with the construction of the casale and granaio in the mid-nineteenth century. This activity has been grouped into Phase VI.

Phase I: Original construction

Phase I sees the original construction of the complex of the winery, presumably contemporary with that of the wider villa complex (Fig. 2). Evidence from brick stamps suggests that this phase can be reliably dated to the second or third decade of the second century AD.¹ The walls form a complex, which consists of 11 rooms on a broadly symmetrical layout. At the central axis lies the cella vinaria (Room 1). Each room is briefly discussed in turn here,  together with its dimensions, wall and floor covering, entranceways and any structures dating to this phase.

The winery complex represents the earliest evidence for activity in the area. No indication of a natural clay was apparent anywhere and almost no remains predating the construction of the complex were encountered, except for stray finds (an antefix and a brickstamp), residual in context. The exception is a thin layer of opus signinum 1318 found ca. 60 cm below the floor level of Room 2, which may have been a building layer but might just represent some earlier structure.

Across the area a fairly comprehensive wall plan of the original building has been revealed, consisting of  a series of  more than 20 orthogonal walls, not individually numbered here (Group 1). These walls are in general 70 cm wide and constructed in opus mixtum, with courses of 5 bricks alternating with opus reticulatum of limestone blocks 8–9 cm in length. These walls, bonded to each other, form the main architectural framework of phase one of the complex.  Consistently across the complex these walls survive to no more than foundation or floor height. In places we have in situ evidence for their decoration or facing but in general only the core of the wall survives. Where corners are evident these finish in an opus vittatum quoin consisting of bipedales alternating with regular blocks of tuff.

Excavation at the base of the calcatorium [1752] in Room 1 indicates that the walls were built with the reticulatum facing starting at -2.19 m below datum (Fig. 3). Below this level the walls were constructed of large rectangular blocks. No foundation trenches for these walls were apparent anywhere, which leads us to believe that the whole area was terraced away, and the walls constructed from the same level as those of the vaults to the north and west. A layer of clean, stiff yellow clay abuts the foundation walls in most rooms and provides a base for the floor surface preparation; the same clay is consistently seen across the complex as a construction layer. In certain areas, specifically those in which production was taking place, the construction make-up of the floors consists of further layers and has a more complicated make up.

Room 1

Room 1 (Fig. 4; Fig. 5) measures 18 m x 8.5 m giving an internal area of 144 m².  Evidence for several masonry structures, which have been assigned to this original phase of construction, survive within this room.

On the north side of the room abutting the northern wall lies a flat structure built over a low vault [1116] which runs almost the width of the room, the core is made of rubble blocks and was covered by cocciopesto. To the east and west the structure is incomplete and has been destroyed in later periods but evidence can be seen for a change in level which has been interpreted as a possible step on the east side of the structure. The structure is interpreted as a calcatorium, on which the grapes were trodden at harvest time. Built against the centre of this structure is a three-sided vat lined in marble with a small hollow at its base, presumably for the settling of impurities [1020] (Fig. 6). The vat would have been used for the collection of the must from the calcatorium. Evidence from trench G1 suggests that the east side of the calcatorium was open onto the terrace beyond, as the wall separating Room 14 from Room 1 did not continue north, but rather returned to form the north wall of Room 14.

Two further structures abut the western wall of the room and rest directly on the construction clay. To the north, against the calcatorium, is a block of concrete 1.05 m x 1.05 m faced with tufa blocks and reticulate masonry [1751]. This was found largely destroyed. It could possibly have supported the base of a superstructure, perhaps the counterweight of a winch associated with a wine press structure on the calcatorium. Alternatively, and perhaps more probably, it might be interpreted as an altar, on analogy with that found abutting the winepress at the villa of N. Popidius Narcissus Maior.²

Also abutting the western wall [1250]=[7170] is [1251], a rectangular structure constructed out of tufa blocks, faced with reticulate masonry and bricks on its northern corner (Fig. 7). One single piece of marble facing survives on the southern face. The structure is built around a break in the wall formed by two bipedales laid a cappuccina to protect a lead fistula which appears to have run through the wall, running west (Fig. 8). The structure appears to represent another vat, with an outlet to the west through the pipe.

 

A further structure [1117]=[1120] runs parallel to and bonds with the southern wall of the room. This consists of a bedding of bipedales bonded to a small foundation wall [1120] constructed in masonry of slabs of limestone and tufa without surface treatment (Fig. 9). The preserved height of this wall varies throughout its length, but is generally around 20 cm above the floor level; it ends 1.2 m before the western wall of the room. The bipedales have slumped to remain at an angle between the two walls. Three breaks in this wall were identified at regular intervals along it. The infill of each of these breaks formed the back of a very deep square hole, from west to east numbered 1228/(1229 and 1234), 1238/(1239), 1242/(1240) and 1277/(1278) respectively (Fig. 10). Contained within the fills of these ‘drains’ were securely datable coins, pottery and glass which firmly place these in association with the original use of the room rather than as later intrusions. (1278) contained a sherd of an amphora dating to mid- second century, as well as two coins, one dated to AD 140–44 (C21) and the other to AD 164–9 (C11). The holes might be interpreted as soak-away drains used when the floor of the room was cleaned, or for drainage of rainwater,. The whole structure might be interpreted as a flight of steps leading up to Room 2 to the south, an interpretation strengthened by the fact that it did not run the full length of the south wall, and was thus certainly not a drain.

As part of this first phase of construction it seems that a series of dolia were built into the foundations of the room. Cuts were made for the placement of dolia in the lowest clay layer (1218). The cuts for 29 dolia were encountered during excavation. One dolium (1047) survived, broken but largely intact, and a fragment of another one (1123) remains in situ (Fig. 11). As only the construction cuts of the other 27 dolia survive in the clay, these have been given a single group number (Group 2). Some of their fragments were also found in later fills and are discussed within the dolia report.³ The construction cuts survive in varying degrees of preservation and are often distorted by the subsequent robber cuts. The bases are fairly regular with a diameter of ca. 25 cm.

Once the dolia were in place, the room was built up to floor level around them using a light grey gravely layer containing frequent small chips of marble (1291), presumably a construction deposit intended to provide a non sticky surface between two layers of clay to aid drainage. The dolia were then packed around with a layer of yellow clay mixed with mortar lumps and rubble (1704), which appears to have the dual purpose of cladding the dolia and providing a base for the preparation of the final pavement. Again, due to the nature of destruction and repeated robbing during later periods this layer survived in varying states of preservation across the trench but it is clear that it originally covered most of the room.

In the northwestern corner of the trench where this rubble layer occurs in the area between [1251] and [1751] a difference was noted in the size of the marble chip inclusions. Sitting on this layer in this corner was [1756], a feature consisting of large bipedales resting on a layer of blue cement. This feature had been largely destroyed; only three tile fragments survive and they are not in situ. The feature is suggestive of a drain or channel structure although it is difficult to understand what function this drain would have had.

1041 forms the preparation for the opus spicatum pavement and sits directly on the top clay layer. It consists of mortar spread across a layer of tuff and limestone fragments. In the western limit it is unclear to what extent the original pavement will have reached, however there is no evidence to suggest it ever covered the square base in the northwest corner [1751] and it abuts the edge of the western vat [1251]. The pavement in opus spicatum, 1015, is laid in a herringbone pattern with tesserae of rectangular pieces of Portasanta marble of variable width and sits directly on 1041. Built into the pavement are a series of regular square holes, each measuring ca. 20 cm square and edged with spicatum pieces (Fig. 12). Excavation showed these are no more than 50 cm deep with a base of solid masonry. These may be interpreted as permanent postholes for the removable supports for a velarium, or awning covering for the room, used during the late spring and summer to protect the maturing wine.

It is envisaged that this room formed the main wine processing room with both pressing and storage taking place within the structures discussed above.

Room 2 

Room 2 (Fig. 13; see also Fig. 5) has a width of 18.5 m and length of 12.5 m, with the southern wall forming a hemicycle. The eastern part of the room was not exposed during excavation but it is possible to project the layout of this side.

The southern wall [1305]=[1306] sits on a foundation [1342] almost half as large again as the wall, something which is not seen elsewhere across the complex. This foundation projects 14 cm to the north of the wall and is finished with a smooth coat of mortar, which is scored with a series of lines presumably made with a stylus into the mortar. These incisions are found at regular distances of 2.20 m and 0.75 m. The cuts are thought to form guidelines to delimit features in the wall above, most likely pilasters. A central set of these markings was noted along the axis of the room, forming a break between walls [1305] and [1306] and marking the southern doorway measuring 2.20 m in width. On the east and west sides, there must have been doors between Room 2 and the corridors Room 16 and Room 11. However, the foundation is continuous across the west side and the east side is obscured by the nineteenth-century granary. There is no trace of walling above this foundation and it may simply have been intended as a structural link between the two walls.

No additional structures dating to the first phase were noted in the room, and no evidence survives to suggest what type of floor surface was used in this space. However, a layer of compacted clay and lumps of mortar (1307) covered the whole of the excavated area. This presumably represents the preparation for the construction of the floor above, which was probably in opus sectile.

Room 2 has been interpreted  as a coenatio or dining room, used in conjunction with Room 1. Its position would have enabled the diners to watch the grapes being pressed in Room 1 while eating.⁴

Room 3

Room 3 takes the form of a corridor parallel to Room 2,  2.45 m in width. The southern wall [1304] is narrower than other walls in the complex suggesting that its roof did not bear any significant weight.

The break between walls [1305] and [1306] provides an entrance into Room 2 from the corridor. The corridor itself could probably also be accessed from the western and eastern entrances to both rooms. The external face of [1304] was the outer wall of the building, and some plaster with red paint survives on its exterior. A small portion of a rough mosaic 1441 formed with large white tesserae 1.5 cm square survives in the northern part of the room covering a surface area of 0.40 m² and indicating this whole corridor was originally paved with this crude mosaic, overlaying the preparation layer 1433.The narrower wall and rough mosaic suggests that this corridor was a service corridor, serving Room 2.

Room 4

Room 4 measures 3.5 m x 3.5 m forming a space of 12.25 m². There are no breaks in the walls of the room to suggest how the room was accessed, the continuous foundation again masking the doorway, which was probably on the north side of the room. The floor surface does not survive, although the floor preparation 1084 consisting of mortar and cement survives in fragmentary condition across the room. Traces on the preparation (Fig. 14) and some remaining pieces of opus sectile have enabled us to reconstruct the pattern of the original floor.⁵

The function of such a small room is unclear. It has been suggested on the grounds of finds from the vicinity of the Room 4 that it was a room for a bookkeeper or an accountant. In the absence of any other ideas this seems as good an interpretation as any.⁶

Room 5

Room 5 is not so much a room as a stair, 2.70 m wide and 12 m long (Fig. 15). It consists of four landings, the upper two 2.8 m in length and the lower 3.8 m. These landings (not separately numbered) are separated by three sets of steps [1030]. Each set consists of three steps (ca. 30 cm long) which abut the bearing walls. The stairs [1030] and the risers are veneered with luna marble; the marble survives intact on several steps and is fragmentary on others (Fig. 16). Fragments of a fine white mosaic, 1110, survive on each of the landings. The mosaic is again well-preserved in some areas and fragmentary in others, particularly at the top of the stair. The mosaic is bedded in a mortar pavement preparation 1036 similar to that seen in other rooms of the complex.

Fragments of the original marble facing of the corridor walls [1040] and [1070] survive at a height of no more than 3 cm, but are enough to indicate that the walls of the corridor would have been faced in luna marble separated by bands of pavonazetto to at least 3 m.⁷ A single fragment of curved plaster might indicate that the corridor was vaulted, although if this were the case it is odd that no fragments of the vault survived. Further, the destruction deposits contain multiple roof tiles, suggesting a timber roof: the curved plaster may have derived from the border of a hung ceiling.

At the top of the stairs an opening of ca. 1.2 m indicates an original entrance from the stairs into Room 11. This entrance survives in cement only: no threshold is visible. The southern entrance to the stair is obscured under the garden wall of the casale, but the stipites of the door establish its position directly beneath the modern wall. The stairs clearly continue into the corridor to the south, Room 17. Their gentle slope would have allowed a litter bearing the emperor to be carried up them. Considering this characteristic coupled with their lavish decoration, the stairs of Room 5 are interpreted as the ceremonial entrance to the cella vinaria/coenatio complex.

Room 6

Room 6 also consists primarily of a set of stairs which lead from the ground level of Room 10 to the vault below, Room 27. The stairs [1130] are narrow, constructed from cement and tiles and have steps 35 cm deep and 22 cm high, bonded with the masonry of the adjoining walls (Fig. 17; Fig. 18). The stairs are reached through a doorframe indicated by a limestone jamb inserted into wall [1090].

These steps served as the passage from the vaulted structures mentioned above and Room 10; their size and nature suggest a service stair.

Space 8

In this phase Space 8 behind Room 3 was paved with a rough cement 1853 which sloped dramatically away from the outside of Room 3 towards wall [1040] and the stairs (Fig. 19). Running north–south along the stairs was a substantial covered drain [1834], 1.20 cm high, 54 cm wide and covered with two tegulae forming a peaked roof. The structure is covered by stone and mortar surface 1838, which runs alongside wall [1040], although later interventions make the original extent difficult to determine. The function of the structure may have been to house one or more fistulae running away from the building, towards the south. This is suggested by the presence of an opening in wall [1060], again protected by two tiles set at an angle: from here the fistula would have been channelled into the covered space. A simple function as a drain to remove water from the roof is not excluded by this hypothesis.

Room 9

This room is located on the lower level and will be described with the stratigraphy of the bath complex.

Room 10

This small room to the west of the stair is 4.2 m long and 6.4 m wide, symmetrical to Room 4 to the east of the stair. Two small fragments of marble 1103 on a pavement preparation 1045 survive as evidence for the original pavement of the room (Fig. 20). The mortar lines clearly indicate the location of the marble slabs.⁸

Room 10 is connected to the stairs of Room 6 through a door frame cut into wall [1090]. There is no other apparent entrance into the room, although we must presume that a door opened onto the corridor (Room 11) on its north side. The function of the room is not clear: it led to the stairs which lead to the vaults below. It may, however, also have had a door leading west onto the terrace above the vaults below the west side of the building, allowing passage by the servants into that space and enabling them to move around the structure without traversing the main rooms. It is not impossible that this room, like Room 4, had an administrative function, with the movements of the slaves being tracked from this point.

Room 11

Room 11 takes the form of a U-shaped corridor or ambulatory which encircles Rooms 7 and 8. Its western extent coincides with the east wall of the casale. The width of the corridor is 3.5 m.

Evidence from the preparation suggests that all of the corridor was paved with rectangular marble slabs, measuring 0.50 m x 0.80 m, although only two pieces of white marble with yellow veining survive. The preparation for this pavement is preserved across this area: in the southern corridor 1052, in the western corridor 7004 and in the northern corridor 7031=7069 (Fig. 21). Traces left in the preparation indicate the original position of the marble slabs, particularly in the southern corridor.

This corridor was entered from the top of the stairs (Room 5) through an opening in wall [1050], as discussed above. A break, subsequently blocked, in wall [1150] directly opposite the stairs forms an entrance to Room 12, which seems to have been operational in this phase. No other entrances were identified, although presumably the corridor will have also opened onto Room 2, and doors onto the western and northern terraces are likely. The connection between the northeast end of the corridor and the calcatorium has been lost: we do not know if they communicated. A later robber trench severed the corridor from any connection with the calcatorium that might have existed.

Room 12

Room 12, 8.70 m long and 6.9 m wide, was only partially excavated due to the presence of a large tree, although all of the perimeter walls were identified. The inside of this room contained fragments of cocciopesto pavement 7123 indicating a service function.

As noted above, there is an access point to this room directly opposite the top of the monumental stair during Phase I, which suggests that the room was used in conjunction with the ceremonial stair.

Set within this room is a structure which can best be described as a quadripartite basin (Fig. 22; Fig. 23). It is separated from the walls by a narrow walk-way, just over 1 m wide. It consists of four small walls [7121], each 38 cm wide, forming a rectangular basin divided into four segments by two dividing walls [7128], each 28 cm wide. The walls are no more than 12 cm in height. The whole of the basin, including the walls, appears to have been lined with opus signinum 7551. Where the dividing walls intersect at the centre of the basin lies a flat, square cement platform [7559] measuring 1.18 m x 1.18 m (Fig. 24). A later robber cut 7556=1541 at the centre of this platform suggests the presence of a fistula here, and in fact its trace is visible in the base of the cut heading towards the southeast corner of the structure. This fistula must have served as a drain for a vessel, perhaps of marble, in the centre of the basin. The robber cut runs out of the southeast corner of the room and east along the corridor of Room 11: its trace is lost at the entrance to Room 4, but the fistula probably ran under this and out the southern wall [1030], where two bipedales clearly mark the exit of a pipe and a robber trench continues its line to the south.

A second cut 7554 into the masonry of this platform, thought to also date to this phase, takes the form of a circular hole ca. 35 cm deep. Its curved base seems to suggest that it was not a posthole, but had a specific function in the use of the basin, perhaps to hold a vase.

The function of this room is connected to the production of wine, which presumably ran through the fistula, although its exact purpose is not entirely certain.⁹

Room 13

Room 13 lies in between Room 1 and Room 7. It measures 8.7 m long by 7.2 m wide. No obvious entrances survive.

In the northern part of the room the light yellow sandy construction clay (7125) which is seen across the complex is very much in evidence. Cut into the construction clay is the cut 7090, from the robbing of a fistula which ran east from the vat in Room 1 directly towards the quadripartite basin in Room 12, passing underneath it (Fig. 25). This trace of the fistula is clearly visible in the clay and is interpreted as part of the original operational phase of the area.

It is unclear what the original floor of this room was like, however substantial parts of the preparation for this are preserved in the southeast part of the room. This preparation layer 7165 is underlain by construction clay (7125). It most probably prepared a pavement in opus signinum, although no evidence survives for this. Further north another portion of pavement preparation 7172 survives, again above the construction clay.

---

To the east of the complex the room plan was more difficult to identify due to the position of the 19th century granary. We believe that the mid point of Room 1 formed a central axis point for what was largely a symmetrical complex. Several trenches were opened around the building in order to confirm this, and the walls identified within these trenches show that at least the northern part of the eastern half of the building was almost symmetrical with that on the west. Several further rooms have been identified, although little is known about these rooms apart from their locations and presumed dimensions.

Room 14

The western and northern limits of this room were identified and indicate a room apparently identical in proportions to Room 13, 8.7 m by 7.2 m (Fig. 26). None of the original floor of Room 14 survives within excavated trenches, although a fragment of pavement preparation 6012 was recorded which appears to have overlain the upper clay, (6037). The construction method below this pavement preparation is similar to that seen in both Rooms 13 and 1. The first construction layer was of very stiff sandy clay (6039) and was assumed to cover the whole room. Overlaying this in the east of the trench was (6038), a thin dark brown layer with frequent white inclusions. Elsewhere, a gravelly layer containing inclusions of small marble chips (6047) covered layer (6039).

A large regular cut 2.2 m wide, 6044, was identified in the north wall of the room. The base of this cut revealed a clay preparation (6043) which is very similar to the construction make-up elsewhere in the complex, and was therefore assumed to belong to this phase of construction. The cut could perhaps have held a large threshold. Although it is significantly wider than the other doorways of the building this may reflect greater traffic at this point, where the room opened directly onto the external terrace rather than onto the corridor—a significant departure from the overall symmetry of the building. As elsewhere in the building the terrace lies above the level of the floors, and thus cut 6044 might also represent the removal of a short stair.

Room 15

To the east, Room 15 was identified by its northern, eastern and southern walls and measures 8.70 m long by 6.9 m wide (Fig. 27). It is identical in shape to Room 12. No trace of the original pavement survived but a large area of pavement preparation covered most of the exposed part of the room. This preparation was a bluish mortar set over hard white stones, 6175=6176, overlying the construction clay (6177). The small part of the room exposed did not contain any entrances.

Contemporary with the wall construction is a U-shaped structure [6173] measuring 1.02 m by 1.12 m. Constructed vertically in opus vittatum, this structure forms the framing for what appears to be a small drain measuring internally 50 by 50 cm. This structure can be seen to line up directly with the drain features identified in both Rooms 1 and 13 and presumably served some sort of function as a conduit for wine. This suggestion is supported by a cappucina tile arrangement in the east wall of the structure, identical to those which protected the western fistula. The fistula may thus have entered Room 15 at a higher level, although still under the pavement, and then run down into the drain and out the eastern wall of the room, running towards a storage area to the east.

Room 16

Only a tiny portion of Room 16 was excavated, however it has been assumed that it will have formed an ambulatory similar to that on the western side of the complex. Only the construction clay and two walls of the southern arm were visible in the small area which was excavated in trench GV; both date to the first phase of the complex.

The northeastern corner of this room—and of building as a whole—was identified in trench GVI to the east of the complex (Fig. 28). A wall of opus mixtum [6158] was covered with a white plaster surface, of which only a trace survives (6308). It is notable that the north branch of this corridor is shorter than that on the western edge; wall [6048] blocks off the space at the western edge of Room 15. This blockage was probably cut by a door, although we have no direct evidence for this. Such a door would have allowed access from the corridor to the terrace over the vaults.

The exterior: pavements and terraces

Room 43

Room 43 is the terrace that presumably covered Rooms 27–39. It was glimpsed only for a short stretch at the north end of trench AIV where a pavement in opus signinum covering the vault over Room 34 was recorded outside the north wall of the room, at 20 cm above the floor level. We assume that this terrace ran along the north and west sides of the building, allowing communication outside the ‘reception’ spaces of the winery. The connection between the northern and western terraces at the northwest corner of the building is problematic, as there is no real evidence for a vault in that position. However, the north wall of the building does extend beyond the corner and may have defined one further room, covered by a vault but not communicating with the western range.

The terrace was 11.51 m wide on the west and 8.14 m wide on the east, except where it widened between Room 15 and Room 1. This widening, one of the only asymmetries in the building, invites the suggestion that the terrace could have been reached by a stair that ran up from the vault Room 37. Access to the terrace would thus have been hypothetically possible from Room 10 (as suggested above), from the northeast side of the corridor Room 11, through a door in the north side of Room 13 and, perhaps, from the east side of the calcatorium, where no wall was observed.

Western Vaults 27–33

No excavation was carried out in these areas, which remain as relatively intact vaulted substructures.  In brief Rooms 27–33 consist of seven vaulted rooms, each c. 11.5 m long and 3.5 m wide.  Each is open to the outside on its western face and this appears to have been originally been the case.

The vaults are built in opus mixtum with rubble barrel vaults and opus vittatum quoining.  Room 27 leads to the stair case which links to Room 6 and was presumably the main passage to the vaults from above.

Rooms 34–36; Space 42

An east–west wall in opus mixtum [7401], plastered and painted red on its north face, is presumed to be the outside north wall of the winery complex, corresponding to [6067] and [6069] to the east (Fig. 29). The western limit of this wall was not exposed, suggesting the structure will have continued to the west where [7401] becomes wider by 15 cm.

To the north of [7401] is found a conduit [7445] which runs along the outside of the structure. Visible for 5.80 m, it measures 60 cm wide by 1.10 m deep, built in phase and bonded with the wall. While the conduit could have served as a drain running around the exterior of the villa it is also possible that it functioned as a cavity for the purpose of absorbing humidity from the walls and thus protecting the foundations from infiltration.

Space 42 is comprised of the area north of the east–west wall [7401], at the exterior of the building. This space had an opus signinum floor 7451 constructed over the standard yellow construction clay (7423). The mortar is 40 cm thick around the drain, suggesting that this area was used as some sort of production area as well despite the assumption that it was to the outside of the winery.

To the south of external wall [7401], two north–south walls were exposed coupled with the remains of three vaults (Rooms 34–36) which are visibly incorporated into the northern substructures of the modern casale, cut away at the time of its construction (Fig. 30). These remains indicate the presence of a series of vaulted rooms along the west side of the original complex. The height to the springing of the vaults in each room is 2.40 m, and each is 4.20 m wide. No floor surface survived within these vaults.

Excavated Room 34 has two doorways containing thresholds, [7436] and [7434], each consisting of a solid mortar surface set with red bipedale tiles (Fig. 31). Together, these form what seems to be the space for a corridor or passageway running east–west along the northern edge of the vaulted rooms for the whole length of the building and eventually joining up with the door recorded in Room 37. The door in the east was 2.40 m wide, while in the west an additional wall forms a smaller opening of 1.3 m.

Room 34 has been reconstructed as forming the northwest corner of the building. We should note that the north wall of Room 33 does not show a scar indicating a bond with a previous wall, nor any openings. We would have to imagine that the vaults were entered from the western side, and that there was no communication between the northern range and the western one.

Due to the proximity of the vaults to the pressing room and wine production areas it is likely that these were used for the storage/fermentation of wine, probably in barrels or dolia. They were covered by the terrace, Room 40.

Rooms 37 and 38

Two walls dating to the original phase were identified as the most northerly extent of the winery complex, [6067]=[6048] and [6069] (Fig. 32). To the south of [6069] is a gap with a flat plastered threshold, which presumably represents a door. To the east, the pier [6091] defines the other side of the only known entrance to the building on the north side. South of this pier, [6090] probably forms the other side of the door and the east wall of the space. Room 38 is the space defined by these walls, perhaps a sort of vestibule open on the north and east, while to the west it would have led into the vaulted spaces along the north side of the building. The original floor level was not reached during excavations. A drain [6106] was built into the northern part of wall [6069] and continues to the north (Fig. 33).

Only a tiny corner Room 37, to the west of Room 38, was excavated. However, the configuration of the rooms above, and the necessity of having some access to the terrace and the calcatorium, suggests that this room contained a staircase leading to the upper storey—a suggestion which also explains the drain, which would have drained rainwater coming down such a stair.

Space 40–41

A concrete pavement 6278 defined the exterior at the northeast corner of the winery, east of wall [6158]. This perhaps formed an exterior eastward walkway adjacent to the winery. The pavement was set on yellow clay (6284) with a layer of chipped limestone followed by chipped red-orange brick or roof tile. Two pieces of slate in situ indicate that the pavement was finished with inset pieces of slate, presumably to back a paving of limestone or marble.

Phase Ib: Minor expansion

Although most of the activities subsequent to the original building have been grouped together in a general Phase II, the opus vittatum mixtum construction of Room 41 appears to have occurred earlier than this second phase, coming some time before the end of the second century.

Room 39

Wall [6163] was built northward from the northeast corner of the building, defining a new room to the west of it—Room 39 (Fig. 33, above). The wall preserves the same alignment of the original eastern wall of the winery. The rough reticulate construction of the wall, without quoining, resembles a similar masonry technique found on the southern face of nearby [6068], which blocked the northern entrance to the vaults. It is reasonable to presume that this wall was part of the same project, extending from the northeast corner of the cella vinaria complex. The new east wall cuts into the first phase pavement of the terraced area. The presence of multiple mosaic tesserae and painted wall plaster in the destruction fills of Room 39 suggests that the two new walls supported a new room or rooms on the upper terrace, overlooking the working entrance to the building. This entrance would have shifted one room to the east, lying now outside the pier [6091] and its southward extension, [6090]. The presence of much ceiling plaster in the lowest level of the destruction deposit suggests that the lower storey of Room 39 was not vaulted, but rather roofed with timber.

Phase II: Alterations to the complex

After the original construction of the winery a series of alterations are made across the complex (See Fig. 2). Some of these seem to represent simple upgrades to the existing structures, pavement repair, etc. However some seem to be more fundamental changes in the use and method of operation of the winery overall. In general, due to the lack of pottery from the site, most of these changes cannot be individually dated accurately. It is imagined that an ongoing system of updating and repair was being carried out across the complex, over a period of 100 years or so. However, it does seem apparent that at some point, perhaps in the Severan Period, the winery underwent a fundamental alteration in usage. For the sake of this report these changes have all been grouped together under the ‘second phase’ which has been given a date range of 190–220. It is likely that more than one phase existed but in the absence of more detailed dating this is difficult to define. The second phase activity is discussed in brief below room by room.

Room 1

Evidence within Room 1 suggests that at some point in the late 2nd or early 3rd century some new structures were inserted into the room altering the system of operation of the main part of the winery.

In the northwestern corner a very thick layer of black tufa lumps mixed with a distinctive dark yellow clay (1775) covered the tile feature 1758 and survived to the height of the nearby structures, presumably deposited to raise up the area to the level of the surrounding structures after the original feature had been cut away (no evidence for this cut survives).

Two parallel robber trenches, 1202/(1203 and 1213) and 1219/(1220) seem to indicate the presence of linear structures or walls running north–south between the calcatorium and the stairs [1120] (Fig. 34 and Fig. 3, above). The fact that these were the only walls robbed in the trench suggests that they were built using a different construction technique, perhaps with small-block tufelli, which would have made their stones easier to extract and reuse than those of the opus mixtum. No traces of the original structures survive. The construction of the westernmost of these walls resulted in the removal of four dolia, whose cuts underlie the foundation cut, but both walls apparently respected the calcatorium and the step, while the opus spicatum floor was repaired along the line of 1202 (Fig. 12, above). We might suggest that they were walls supporting channels running from vats cut into the ends of the calcatorium. In fact, the west end of the calcatorium is cut away 5.59 m from its western side, 1752, which would work well with the idea of a vat. A similar cut on the eastern side is only 3.85 m from the east wall, and lines up with the center of the presumed channel. The must would have thus flowed from the calcatorium into the lateral vats, and from there into the channels from which it would have been distributed into the remaining dolia and into the western and eastern vats. This system of distribution is described by Palladio, but clearly dates to a much earlier period: it is discussed in greater detail in the published text.¹⁰

Room 2

The door at the back of Room 2 was blocked at some point and, despite the lack of firm dating, it might be suggested that this took place during this second phase when the room was no longer in use as a banqueting hall.

Room 7 and Space 8

This room seems to have been entirely constructed during this phase (Figs. 35–36). Two curving walls [1580] and [1100] 0.46 m wide form a semi-circular corridor, ca. 2 m wide, which encloses an interior space, Room 7 (Fig. 37). The walls are constructed a tufelli, of irregular tufo blocks set in a lime mortar and roughly faced. The eastern wall of the upper landing of the ceremonial stair was cut to create an entrance to the corridor. A loose construction fill (1852), composed of broken tile and mortar, lies to the east of the corridor, indicating that this space was never formally used.

This area was paved in a very well-preserved mosaic 1826 consisting of large black and white tesserae, the majority in white with a black band along the perimeter. The later tile and plaster collapses which were preserved on the floor—(1809) and (1845)—as well as those remaining on the wall—(1849) and (1850)—indicate that the corridor was decorated with painted plaster imitating marble panels.¹¹ To the south, a small wall [1820] divides the corridor, perhaps forming a small doorway. The mosaic pattern respects this and continues as 1842=1843 within the new division. To the west the corridor widens at this point by 52 cm [1810], corresponding to an opening in the floor with a masonry block on its east side. This opening was bottomed at ca. 60 cm below the level of the mosaic, with a cement surface. As this void ran into the southern limit of the section we do not know how it continued, but it seems very likely that it represents a drain, which would have been covered with seating. The masonry walls of the structure suggest that it supported the seat of a latrine, which would have continued for some distance toward the south and drained in that direction.

The interpretation of the corridor as access to a latrine or lasanum is strengthened to some extent by the discovery within it of an almost intact jar whose parallels elsewhere are identified as chamber pots (Fig. 38).¹²

Much less easy to interpret is Space 8. This is the area enclosed between Room 7 and the ceremonial stair, Room 5. No floor is associated with this space, although the original concrete skim was covered, with 1847 a layer of earth containing a certain amount of pottery and finds. It is clear that the area did not constitute an internal space, but it may have been used for planting, possibly a viridarium onto which the exedra opened. Protruding into the space is a platform [1855] measuring 1.75 m east–west, and again covered by a white mosaic, 1856. The platform bonds with the walls of the exedra, but it was apparently not accessible from the latrine itself, and thus remains in the space between the curved corridor and the stair. Its proportions suggest that it might have served as the base of a statue, but it is hard to know where it would have been visible from—an observation that applies to the supposed viridarium as well. The walls of the apse may have had windows on the western side, which would have given light and air and allowed the space to be viewed.

Room 10

On the north edge of Room 10 an oval vat with vertical sizes [1089] was created, cut into the pavement (Fig. 39). Its construction was careful, a larger hole having been cut and then lined with layers of masonry and plaster. Its small size (.92 by .74 m) and shallow depth (.40 m) make its interpretation difficult, but it was certainly a functional structure, suggesting that Room 10 was now a work space.

Room 11

The northern part of Corridor 11 underwent fairly radical changes during this phase. A linear cut  in the pavement, 7038, runs north–south into the northern wall of Room 12. This cut is filled with the broken pavement preparation (7039), suggesting that the area was repaved to the original level and activity continued on top of it. Into this, cut 7080 was made to construct a drain [7163] which is square and constructed with brick and stone courses (Fig. 40). It leads from the quadripartite basin into the vaults located on the north of the complex. It is probable that this served to channel the wine pressed on the quadripartite basin into the northern vaults rather than to the southeast. It is probable, but not certain, that the wine would have been carried in a fistula. The drain was covered with a pavement preparation, 7041, at the same level as that of the corridor: it is not impossible that the marble paving was replaced at this time. Probably contemporary with these activities was the blocking of the door between Room 11 and Room 12, at the head of the stairs [1155]=[1570]. It is not clear whether a new door was opened to the west of it. All of this activity seems to have redirected the wine produced in the quadripartite basin from the fistula running south towards an unknown destination in the first phase, now channelling the must towards the vault to the north of it, Room 35.

Room 34 and Space 42

During this phase the cocciopesto floor outside the external wall was cut through and the conduit was partially blocked with a large stone (not numbered). Pieces of the mortar floor were then replaced, this seems to have taken place prior to the construction of the mosaic (see below) but is possibly part of the same construction alterations.

In this phase a preparation layer 7406 was laid over the earlier floor 7451 and a mosaic 7405 was laid on top (Fig. 41). The mosaic is very coarse, made of large white marble tesserae (4×4 in a 10cm square) with an occasional tessera of serpentino. In conjunction with the mosaic and abutted by it a shallow linear structure [7420] was built that seems to have been formed of bipedales, as there are 60 cm x 60 cm impressions in the cement. The function of this structure is not known: it could have been either decorative or functional. The mosaic respects it and they seem contemporary. It is rather too close to the façade of the building to represent the base of a portico, and therefore remains mysterious.

In a second phase, a small brick wall [7442] running east–west was added at the southern limit of trench AV abutting wall [7431](Fig. 42). This presumably formed a doorway with [7438]. When this door was closed, access to the southern half of the vault would have been impossible. The offset on this wall is slightly higher than those built with the original walls. This is the only structure in opus latericium found in the winery and it is unclear how we should date it: perhaps later than the interventions that increased the wine storage capacities of the building, it may in fact represent a transformation of the building into storage of a different sort.

Rooms 38 and 39

New floors were constructed in Rooms 38 and 39 during this phase. Floor 6110, a tile floor composed of reddish yellow tiles, abuts the existing walls and clearly provides the floor surface for this phase. Further east is the preparation for a brick spicatum floor 6115, presumably also dating to this phase. Imprints of the bricks were visible but no actual blocks were found. Only a very small area was exposed here and as such interpretation of these walls and floors is difficult other than noting the relatively rustic nature of these floors compared to those elsewhere.

Room 41 and Space 40

No new pavement was laid in this area before a new wall [6297] was built, extending east from the original corner of the building and returning to the south, [6259]. These walls defined an interior space 8.1 m long and were constructed in almost undressed limestone laid in courses and only roughly faced. The stones were laid into very thick mortar beds which were then leveled off giving the wall flat horizontal surfaces; the mortar was also occasionally smeared over the facing surface of the laid stones, but not with any regularity. The space on both sides of the wall was paved with a friable white lime pavement bedded on black soil, 6260 and 6279 respectively, and containing broken fragments of wall fresco in its makeup. A shallow hole for a storage dolium, 6283, was cut into the pavement and stabilised with clay (6280). On top of the clay, a final paving was laid of hard mortar 6260 and 6296. Access from one side of the wall to the other was provided by a door in the wall 2 m wide.

This activity would appear to be aimed at the construction of new storage structures, similar to those to the west and north. There are no traces of vaults and it seems more probable that the rooms were roofed in wood and tile. There is also no evidence for a terrace over them, which would not, in any case, have joined up with the terrace of Room 43. The construction technique, even rougher than that used in Room 7, suggests that this was the latest addition to the winery. It is unclear whether Space 41 was also covered and used for storage at this time: the similarity of its floor might suggest this.

Several minute changes occurred to the structure during the occupation of this storage structure. A posthole of oval shape 6285 was sunk to the yellow clay. At some point, the uppermost courses of the original Roman wall were remade. Just before the point where the wall in opus mixtum vittatum was eventually destroyed, the quoin was remade with bricks inset vertically on either side of a tuff block.

Phase III: Spoliation and abandonment

Following the fairly significant alterations of Phase II, the next phase in the life of the winery is a destruction phase. Phase III is evidenced primarily by the robbing of the dolia, seen in the production rooms 1 and the late storage space. The dolia were clearly still considered to be useful as they were quite carefully robbed: it is entirely possible that they were removed for reuse in a new structure. The extent of marble in the backfills suggests that many of the wall facings had already been destroyed and collected. Pottery from this phase gives it a fairly secure date in the mid 3rd century.

Room 1

The first Phase III activity in Room 1 consists of three very large regular cuts—1056/(1039), 1046/(1018) and 1125—in the spicatum pavement 1015 and its preparation 1041. These cuts follow roughly the lines of the location of the dolia bases and it seems safe to assume that they were made in order to rob the dolia. Each individual dolium pit excavated showed signs of robbing. Robber cuts 1292–1750 (Group 3) all showed signs of attempted careful excavation, some preserving exactly the shape of the orginal dolium with only a few, such as 1718 and 1716, showing signs of less careful extraction.

Each dolium robber cut contained a distinct individual fill (Group 4). Several of these—1728, 1718, 1738 and  1748—had charcoally black fills at the base, whilst others contained a mixture of clay, mortar and spicatum blocks from the pavement representing backfill with materials already present in the area. Cut 1729 in the northwestern corner contained a distinct layer of tile and rubble forming a secondary fill of the dolium cut.

The lower fills within the rectangular cut 1125 were then covered by a large, very distinct layer (1204). This fill consisted of a loose sandy silt densely packed with very large marble architectural fragments, tile and spicatum blocks from the pavement. Analysis of the marble fragments from this fill revealed, among other things, the presence of over 20 different types of cornice, indicating that this backfill was a purposeful deposition of marble collected from the winery as a whole. Pottery from this fill, notably a fragment of ARS Hayes 50A from (1018), dates this spoliation to the second half of the third century.

A similar layer (1215) covered the dolia robber fills in the eastern section of the trench within cut 1056. This layer had been cut repeatedly by modern intrusions and was as such less well preserved.

In the western corner of the room a layer of crumbly white mortar (1241) overlay (1251) and (1273). This layer was not well preserved and only survived in a narrow stretch, making it difficult to interpret. It could represent the possible preparation for a late floor surface, or is perhaps the result of the robbing of the possible altar. Immediately to the east was a small (ca. 12 cm by 13 cm), distinct layer which contained many white tesserae (1235).

Room 4

It is worth noting here that the ceiling collapse in Room 4, associated with the next phase of actual destruction, lies directly on the pavement preparation, suggesting that the marble floor had been removed prior to the collapse of the ceiling—perhaps as part of the same phase as the robbing of the dolia.

Room 13

Layer (7173) identified in this room was similar to (1206) in Room 1. It contains many marble fragments and directly covers the preparation clay. This seems to represent the same backfilling action as immediately followed the robbing of the dolia in Room 1.  It is possible that this room originally contained dolia too, which were robbed in this period, leaving less archaeological trace than those in the adjacent room.

Rooms 14–16

The floors of each of these rooms also survived only to the foundation clay. It is possible that something below ground level was robbed and cuts in the pavement preparation were made in order to carry out this robbing. Alternatively, and more likely, this destruction could have taken place when the granary was built.

Room 41

The storage dolium in this area was also robbed at this time. The robbing activity removed the dolium whole, leaving no traces of the pot itself within its cut. The robbing activity filled by (6281) was extensive, destroying all trace of previous layers down to the first Roman pavement in the excavated portion of the stratum. Again, the latest sherds were of ARS C wares, dating at the earliest to AD 230.

Abandonment

Following the purposeful robbing of the dolia and other useful features of the winery complex, the sequence of events then becomes somewhat more difficult to define. Evidence for collapse and destruction survives in very few areas and absolute dating is difficult.

Rooms 38 and 39

In Rooms 38 and 39 a large destruction layer survives (6098)=(6103). This layer sits directly on the original pavements and beaten earth floor, 6118 and 6110 respectively. The many tesserae evident in this layer are interpreted as the floor of an upper level, which fell onto the lower floor level as the building collapsed. It is also significant that large sections of ceiling plaster formed the lowest level of destruction, which suggests part of the ceiling of the rooms at the lower level onto which then fell the remains of the upper level. Significantly, large amounts of painted fresco were excavated in this layer.

Further to the east was a large cut 6277/(6263) into the fills left by the earlier robbing action. This cut destroyed what was left of the later Roman pavements and also contributed to the destruction of the walls, breaking the door jamb. This layer contained only pottery dating to the 6th or 7th century, suggesting that the deposit represents some subsequent but ill-defined occupation of the area.

---

To the south of the complex in Rooms 4, 5, 7 and 8, the depth of the rooms seems to have ensured the survival of the destruction levels, and each of these rooms revealed large intact destruction layers.

Room 4

Layer (1547)=(1072) covered the floor preparation within most of Room 4, containing primarily wall plaster and marble fragments. This layer was interpreted as the collapse of the wall and ceiling plaster of the room. No colours or pattern were observed in the plaster fragments.

Room 5

In Room 5, above the landing and the steps, part of a layer (1557) was preserved which contained many fragments of marble mouldings. This layer also contained several bronze small finds. Overlying this layer was (1556), which contained primarily roof tile but also further marble mouldings. The two layers are likely to represent the wall and roof collapse of the area.

Room 7

In Room 7, a large layer of roof tile and wall plaster was recorded directly on top of the mosaic floor (1579), representing the roof collapse (Fig. 43). Further layers of collapse include (1809), (1845), (1849) and (1850). Layer (1809) was dated to the second half of the third century by a fragment of ARS 50A.

Room 8

The same destruction phase of Room 7 is also identified in Room 8, where evidence of the tile collapse (1589) sits directly on the original pavement. The large drain filled with a layer of clay and earth (1846), containing relatively few finds.

Dating

The abandonment and destruction of the building is dated by a very few sherds of pottery, but consistently falls within the period of the second half of the third century. This is the last trace of occupation in the area for a considerable time.

A single exception is the area of the northeast corner of the building. The destruction levels of storage space Room 41, layer (6258), contained a sherd of ARS Hayes 62, dating between AD 360 and 440, while even later pottery, in the sixth-century steccata tradition, was found in the upper layers of Rooms 38 and 39. It is unclear what sort of occupation these fragments represent: it clearly follows the collapse of the building as, in particular, deposit (6259) covered the destruction of the walls.

Phase IV: Reoccupation

Following this limited evidence for destruction, it seems the site was unoccupied until at least the 9th century. The evidence that survives from this date onwards is equally limited in this area, again due to construction of the later buildings which razed much of the stratigraphy to the level of the spoliated Roman pavements. In general, evidence for re-occupation consists of posthole cuts within the core of the Roman walls and pavements, with more substantial evidence surviving only in pockets across the area.

The stratigraphy is better preserved on the lower levels to the north and south of the winery complex where evidence for medieval occupation was also recovered.

Ninth century

The earliest datable evidence for occupation after the abandonment of the winery was found to the north west of the winery complex, where evidence for what has been interpreted as a bread oven was uncovered (Fig. 44).

In this area the Roman pavement was cut away and replaced by two consecutive pavements, 7429 and 7430, marked by sporadic Roman tiles. At the level of the tile pavements were several layers of burned clay (7444) cut by a large circular pit. These layers extended into a conduit behind the blockage. The tiles at the back of the conduit were also burnt, indicating that this conduit contained a fire. It is possible that the conduit was re-used as a bread oven during this period. The pavement layer inside the conduit 7450 contained charred hornbeam, oak and prunus. Ash from the oven may have been scraped out of the conduit into the pit, burning the clay in the process, while the pit itself would have facilitated access to the conduit/oven, which was very low on the wall. To the west of the cut was a section of tile covered with mortar 7432, which was left in place and which possibly served as a work surface in conjunction with the oven. A large fragment of Forum Ware was found on pavement 7429. This occupation seems to coincide with the formation of black earth in trench F, (8175)=(6253), also characterized by Forum Ware and a few fragments of globular amphorae.

The blocking of the service stair in Room 6 with a cross wall [1516], along the lines of the main west wall of the winery, is a further known activity on the lower level of the building in this period. Behind this cross wall a layer of earth (1517) contained pottery of the 9th century, particularly globular amphorae and a fragment of a chafing dish in Forum Ware. The wall must represent a deliberate separation of the storage spaces below from occupation levels above, blocking an entrance to the upper part of the building.

Finally, a layer 20 cm thick of dark earth with numerous bone and pottery fragments (8175)=(6253) lay over the ruins of the atrium of the bath building to the south of the winery. This seems to represent occupation refuse, possibly mixed with earth brought in for the cultivation of a garden, and is further discussed in the stratigraphy of the medieval period in trenches FI–FIII.

It is clear that ninth century occupation was in fact concentrated on the main level of the building, over the vaults, within the spaces defined by the ruins of the Roman walls (Fig. 45). Even though the lack of stratigraphy precludes a relative chronology for the multiple postholes that pierce the Roman floors, there is a very clear distinction between them. Most of those to the west of the excavated area lie along the walls of the building, as if doubling the walls; they have been grouped together for analysis as Group 05. We have interpreted these postholes as roof supports for the various existing rooms, or as timber supports shoring up the walls. This is most obvious in the case of Room 10 with postholes 1178/(1179), 1171/(1172), 1173/(1174), 1514/(1515); Room 11 and postholes 1163/(1138), 1563/(1564), 1166/(1167); and Room 12 with postholes 1558, 1531/(1532), 1518/(1519). Room 5, the grand staircase, certainly still served as the entrance to the structure, though the stairs were covered by a ramp formed of beaten earth and fallen marble (Fig. 46). At the base of this ramp, a wall [1550] visible in section partially blocked the doorway, presumably restricting entry. Associated with this wall is a pavement of beaten earth and tiles 1573. The mosaic at the top of the stairs was likely still exposed, as it is significantly more worn than elsewhere. A posthole 1175/(1176) situated at the eastern edge of the old doorway presumably supported a door to close off the top of the ramp.

To the east, two significant robber trenches seem to reflect newly constructed walls that delimited the structure. The first of these, 7167/(7168), is a straight cut 21.82 m long and ca. 1m wide, narrowing at its southern end where it runs along the east wall of Room 4 (Fig. 49) . This cut truncates walls [7028], the south wall of the northern extent of Corridor 11, and [1080], the south wall of Room 13. However, 7167 did not cut the north wall of Room 3, continuing beyond as a narrower and slightly shallower truncation. This leaves us in some doubt as to its real length; indeed, its structural role may have been taken over by the pre-existing east wall of Room 4. At its northern end, 7167 appears to have abutted the northern wall of the complex, returning south as feature 7243/(7243). The cut also seems here to flank an existing wall, although we cannot be sure of this.

The second robber trench, 1453/(1454)=1470/(1471), is slightly narrower than 7167 at .75 m (Fig. 48). Not entirely recognized during excavation, this trench is clearest on the eastern side where it runs for 8.11 m with two clear returns at its northern and southern ends. As 1453=1470 cut both hemicycles of the Roman exedra, these were certainly destroyed when a later wall, since robbed, was put into place. The southernmost wall of this later structure cut Room 7, continuing across the destruction layers, but then disappeared. The northern wall is traceable across the inner exedra wall, beyond which it appears to have widened to a pier: again, this is a later rationalization based on photographs taken from the tower of the casale (Fig. 49).

The two walls appear to have delimited an area occupied by roofed structures, leaving the enclosed sector of Room 13 as perhaps an open court. The western edge of the complex was presumably the western wall of the winery. The extension at the southeastern corner might be interpreted as a tower, commanding the entrance ramp and the area to the south, as well as the exit—or entrance—onto the razed part of the building to the east, which may have served for animals or other activities. On the reconstruction we have indicated a fence surrounding the tower, although this is merely a suggestion based on the position of a seemingly coherent, curving range of stakeholes.

There are many elements of this reconstruction that are very tentative: however, we are reasonably confident of its general lines. The blocking of the service stair (Room 6) will have left two possible entrances to the upper level of the structure: the ‘imperial stair’ (Room 5) and the stair at the northeast corner which we have suggested occupied Room 37. We know nothing of the fate of the latter—if of course it ever existed—but it works well with the idea that animals might have been kept in the courtyard to the west, as it would have given access to this area without passage through the main building. The tower, then, would have also protected the house against any attack from that quarter.

These component parts add up to a substantial building, occupying a commanding position on top of what must have been one of the strongest of the imperial ruins, elevated by its vaults on all sides. The only disadvantage of this position, rather than that chosen by the future monastery on top of the old imperial residence, was the lack of a good view over the valley. It might be suggested that anyone living there would have needed a watchtower at the north end of the ridge, but this is of course speculation.

Phase V: The tenth–thirteenth centuries

A series of small postholes and larger pits (ca. 300) were recorded across the area of the casale as a whole. Little diagnostic material were excavated from these postholes, although some do contain medieval ceramics. Nonetheless, there are meaningful distinctions between them. This section begins with the very few intrusions that can be dated with any confidence to the tenth or eleventh centuries.

The most notable activity dating to this period is a large sub-circular cut 1308/(1309) located just outside the exedra of Room 3 (Fig. 50). This feature cuts the Roman wall [1304] and surface 1474, and may have been covered by surface 1487. This pit yielded more ceramics than anywhere else in the area (ca. 1000 pieces), a majority of which are cooking wares. It is likely that this feature is a refuse pit or midden, and is also possible that simply through an accident of survival does more pottery survive here than elsewhere.Cut 1308 is located in an area containing many postholes and cuts. These remain largely un-interpreted, however, so it is difficult to determine a relationship between this pit and any nearby activity.

In the eastern corner of Room 14 a semi circular cut 6046/(6045) was identified. Two pieces of medieval lead glazed pottery date the upper fill of this cut (6013) to the tenth or eleventh centuries. However, the nature of the cut is not suggestive of occupation or related activity and it can be perhaps be interpreted as an exploratory hole cut in search of material to steal.

In the north of Room 11 was a circular cut 1141/(1139) with regular sides built against the side of wall [1570](Fig. 51). The fill (1139) of this feature contained medieval cooking wares and some early medieval liturgical furniture (AE1039). It is thought likely, given the size, location, regular shape and lack of midden refuse, that this pit was a silo for grain storage.

Postholes

The posthole evidence has been roughly divided into three groups according the physical nature of the holes.

The first group (Group 5) includes those postholes we have assigned to Phase IV—generally rectangular holes located directly alongside the villa walls. These can be seen on Figure 45, above, and have been excluded here.

The second group (Group 6) is seen largely in the vicinity of Room 2, which revealed the highest density of activity as evidenced by ca. 100 postholes in this area alone. These irregular postholes—found generally away from walls and cutting the pavement preparation, often with stone packing—are indicative of wooden stakes. Many of these postholes have yielded bande rosso and green glazed ceramic dating from 13th–14th centuries. Significantly, several large but shallow irregular cuts (1312, 1322, 1370 and 1416) in the Roman pavement preparation 1307 in the area of Room 2 also seem to be of the same phase. It is unclear what the relationship between these cuts and the surrounding stake holes. However, it has be postulated that these may represent slightly hollowed floor surfaces and possible hut structures, thus suggesting the presence of domestic occupation within this area. The clearest group of postholes, for example, seems to outline a long, sub-rectangular building straddling the old exedra which delimited Room 2 — but other interpretations are possible.

Finally, over 100 very small, very circular postholes (Group 7) occur across the whole area. These holes are ca. 10–15 cm in diameter and are generally cut directly into the foundation level of the villa walls. The seeming irregularity of the layout of these holes makes their function very difficult to determine. It is possible that they represent the agricultural use of the area, however there are a number of other possibilities. The irregularity of the postholes in trench F1, both in terms of their size and their depth, makes it possible that the many postholes that cannot be combined to make the outline of a plausible building or fence were used with other, shallower ones now disappeared.

Robbing cuts

Other activity dating to this period also occurs consistently across the winery complex. Most of this activity consists of large cuts, most easily explained as robber cuts—either taking out whole walls, robbing the fistulae, or exploratory cuts presumably searching for more features to rob. Two of these features have already been discussed: the two large parallel linear cuts 1202/(1203)(1213) and 1219/(1220) cutting into the dense fills associated with the robbing of the dolia in the west and eastern parts of Room 1 (Figs. 3 and 34, above). Both cuts are distinctive for their straight edges which cut through the sequence of construction deposits. The primary fill of the westerly cut (1203) contained fragments of medieval ceramics, which date the deposit to the 12th–15th century. The eastern fill (1220) contained no diagnostic pottery but its stratigraphy suggests a similar date.

In the southeastern corner of this room are further cuts that date to this period. Semicircular feature 1131 cut into (1204) and was distinguishable for its distinct blackish fill (1132), which contained inclusions of marble and tile similar to those contained within the surrounding fills as well as 12th–14th century pottery. Elsewhere, spicatum pavement 1015 is cut by several shallow superficial cuts which, due to their nature and fills, are considered contemporary to 1131. These features can only be explained as exploratory holes, due to the similarity of the fills with the original material.

A third robber trench 7167/(7168)=7242/(7243) also runs north–south, parallel to 1202 and 1219, from the north wall of the building, across Room 13 and into Room 2, cutting both the later clay and the walls at foundation level (Fig. 55, above). Its fill (7168)=(7243) contained over 30 pieces of medieval ceramics dating to the thirteenth or fourteenth centuries. It is far less regular and deep than cuts 1202 and 1219 in Room 1, and we suggest it removed the east wall of the ninth century structure.

In the northeastern corner of the complex similar activity was seen, with evidence for two almost rectangular cuts, 6088 and 6094, which cut into the preserved destruction layer. Their fills (6089) and (6095) are both similar and contain building material but it is unclear what, if anything, was being robbed.

The robbing of the various fistulae represent another important activity that can probably be attributed to this phase or the previous one. Indeed, it is clear that the inhabitants were prepared to go to considerable lengths to remove them. As we have seen above, the fistula which ran west from Room 1 was removed with the pavement of Room 13. The original fistula running from the quadripartite basin was ripped out from the basin, and its course can be traced through the cut 1541 in Corridor 11 (Fig. 51, above) From the door to Room 2 the robbing cut turned south through Room 4, the fistula presumably being pulled from under the pavement when this proved too hard to hack through. A very deep pit in Room 7, 1118, might represent a search for the same fistula (Fig. 52). The fistula in fact passed just to the east of 1118, emerging through an a cappucina hole in the wall of Room 7, where its line continues as a deep trench beside the drain 1851/(1847).

In the northwestern exterior area in this phase, after the putative oven went out of use, the conduit was blocked very thoroughly with mortar and stones (Fig. 29, above). Any structure from the oven was removed when the pit was cut and the area was filled with the same fill (7426) which covered the tile pavements. Fill 7426 contained quantities of medieval pottery, such as bande rosse and lead glazed, thus dating the layer to Phase VI; the midden deposit suggests that this area had gone out of use.

Phase VI: 19th century casale

There is no evidence to suggest any occupation of the area from the end of the fourteenth century until the late eighteenth century when the current standing buildings began to be constructed. The whole area of the winery was seemingly cleared and levelled at some point between the end of the eighteenth and beginning of the twentieth century for the construction of the casale and the granary.

Evidence for a large lime kiln was recorded in the southwest of the complex. This kiln burned stone and marble within the confines of Room 9, which we have interpreted as one of the hot rooms of the bath complex, with the entire room effectively being used for burning lime (Fig. 53). A slaking vat [1500] was built into the northeast corner of the room, using walls [1570] and [1070] as the support structures (Fig. 54). It is likely that this kiln dates to the first stages in the levelling of this area and was used to burn much of the surviving material. Also dating to this phase is a possible lime slaking pit located on the west side of the granary building, consisting of a very regular linear cut 6166 into the Roman construction clay (6164). This cut was filled with a very thin layer of flaky lime mortar (6170) suggestive perhaps of the remains of another lime slaking pit (Fig. 27, above). At this point all the walls were razed to Roman floor level or below. Some of them may have still been standing in the 1870s, when Henry Stevenson recorded a ‘theatre’ facing a structure that is clearly the calcatorium.¹³ This suggests that at least part of rear wall of Room 3 survived until that time; the inner wall was clearly destroyed by the early medieval period.

The foundations of the winery were then reused as foundations for the 19th century buildings, which consisted initially of a small farmhouse, gradually enlarged to include a fake medieval tower to the west and a granary building to the east. Construction on the casale complex commenced in 1870.

To the northeast, adjacent to the granary building, a sequence of linear cuts truncating earlier fills, 1223/(1224) and 1221/(1222), are clearly associated with the construction of the granary. A narrow cobbled surface 1207, probably a gutter channel abutting and contemporary with the granary, cuts into the earlier wall [1752] (Fig. 55). Trenches inside the granary building revealed a similar series of cuts relating to the granary construction, including an internal drain 6004/(6007)(6003) which was covered by deposits (6001) and (6000), respectively forming a levelling layer and rubble layer for the concrete floor of the granary.

Evidence for the construction of an electrical cabin abutting the granary was identified in linear cut 1210, dug for a cable recorded in fill (1206). A circular cut 1216 then cut deposit (1215), and seems to have been intended for the placement of a earthing grid which survived at the base of a deep circular hole cutting through the original dolium into the yellow clay. The fills in this area, (1217) and (1209), consisted of similar material to the earlier fills, but also contained many fragments of modern materials. They represent the backfill of these modern cuts with the same material that was excavated to create them. Overlying all of these features, including the cobbled gutter channel, was a shallow construction cut 1214 with rubble fill (1212) which formed the foundation of a concrete platform on which the electrical cabin must have been constructed.

Outside the courtyard to the north, where the land slopes northwards, a complex stratigraphy attests to the construction of the road running around the building and the related building up of the land for the road. Brownish, sandy-clay layer (6076) demonstrates this construction process, covering both the destruction layer (6081) and wall [6074]=[6063]. On top of (6076) was a layer of hard yellow clay, (6064)=(6073). Both layers have been interpreted as deposits for the construction of the road.

These build-up layers were then cut by several modern intrusions. Cut 6065, which contained a trench-built drain [6066], probably connected to the granary. A second modern circular cut 6111=6096 was so deep that its base was not reached. Its purpose is uncertain, although it is probable that it was dug to install a modern electricity pole—the only structure that would justify such a deep cut. A last modern intrusion in this area is 6071, a fairly large feature in the north of the trench containing large blocks of rubble.

Further east, reticulate Roman wall [6160] was cut and a deep trench 6307 was sunk against it. The cut was filled with black rubbly material including modern brick (6157). This was perhaps to shore up the next construction activity: the levelling of the entire area with a thick bed of yellow clay (6152), thought to be the same as (6064) above, and the final levelling layer for the road.

In the southeast portion of the courtyard a large layer (1319)=(1473)=(1665) covers the western portion of Room 2 and is datable to the 19th century by two coins to Vittorio Emanuele II, specifically 1865 and 1866 (C43). This is assumed to be a levelling layer associated with the construction period of the granary, which is dated by an inscription over its door to 1879. In turn, layer (1319) is covered by a layer of dark soil (1321) with traces of burning. Patches of mortar surviving in the area 1410=1602=1663 are perhaps suggestive of the original floor of the courtyard, although perhaps surprisingly these surfaces are quite unlike any of the pavements identified elsewhere in the courtyard from this phase

In the southern area of the courtyard, the 19th-century pavement 1026 overlies two rubble deposits, (1008) and (1035)=(7003). This pavement is given a terminus ante quem by the cut of the water channel [1027] which is covered with paving tiles dated 1894 and 1895. Another cut 1024, made parallel to this in order to hold a lead pipe, cuts the masonry of walls [1060] and [1100] as well as pavement 1026.

In the south west of the courtyard, under a thin covering layer (1016)=(7062), is a second pavement in gravel 1002, covered in turn by a pavement of pebbles 1003 and a thin layer of cement 1004. These must date from the first half of the 20th century as they seal the 19th-century layers. The topsoil of the courtyard (1001) covers the 20th-century pavement. To the north the stratigraphy is the same, although a modern gutter [7140] and a distinctive layer (7005) were also identified.

---

1 See: Bernard, print volume: 77–9.

2 See: Conticelli De’ Spagnolis 2002, 49.

3 For a discussion on the dolia, see: Rice, print volume: 119–121.

4 See: Fentress, print volume: 203–8.

5 See: Booms et al., print volume: 93.

6 See: Cernuta, print volume: 118–119.

7 See: Booms et al., print volume: 91–2.

8 See: Booms et al., print volume: 93–4.

9 See: Booms et al., print volume: 98–9.

10 See: Booms et al., print volume: 94–98.

11 See: Booms et al., print volume: 107.

12 From Knossos: Hayes 1983: 109 and fig. 14; from Poggio Gramignano, in the upper Tiber valley: Piraino 1999: 320 and fig. 250; cf. Martin 1999: 254 and fig. 183, 49.

13 See: Gatti, print volume: 7–9.