During fieldwork, trench supervisors were responsible for recording data. Given the fact that excavation is destructive, it is necessary to provide for careful and effective documentation so that the greatest amount of information can be captured and saved. The greater the amount of this information, the more complex becomes its management and synthesis. But the act of documenting an excavation is not limited to collecting data in the field; it also concerns the digitization of data in the field or, as in our case, immediately afterwards in the laboratory, with the aid of an appropriate relational database.

The Excavation Notes were filled out every day by each trenchmaster (fig. 15). Here the archaeologist’s first impressions were recorded, along with all the activities carried out in the work area and including the finding of any noteworthy objects. The Excavation Notes form also included room for sketches and measured drawings. This kind of documentation, written in a discursive style, might seem old-fashioned, but it has proven to be a tool that retains its usefulness today, both for the interpretation of data recorded elsewhere in the information system (generally by various experts), and as an aide-memoire to the trenchmaster for the doubts, second thoughts, and the day-to-day unfolding of the excavation—all of which becomes important to review when the time comes to write a final report.

To register the data pertaining to the stratigraphic units (SU), we used as a model the forms developed by the Italian Ministero per i Beni e le Attività Culturali and approved by the ICCD (Istituto Centrale per il Catalogo e la Documentazione, a unit of the Ministry; figs. 16-17). A similar form was used for registering walls (Mural Stratigraphic Unit, or MSU). For the numbering system used to identify each SU or MSU, we decided to build into the code an indication of the excavation sector from which the SU or MSU came. The archaeological site was divided into the following sectors: I, bath complex; II, southern branch of the quadriporticus; III, residence; IV, western branch of the quadriporticus; V, pool; VI, central area of the garden; VII, northern area of the garden; VIII, eastern branch of the quadriporticus; IX, zone to the northeast behind the structures currently visible on the site and near the entrance from the parking lot; X, sector used to denote studies of the walls (see fig. 5). Each SU or MSU is identified by a number of four figures (five for Sector X) composed of the sector number at the head and then by a progressive numeration of the SU found in it. For example, if the sector is IV, the numeric series will run from 4,000 to 4,999; if VII, from 7,000 to 7,999; if X, from 10,000 to 10,999. Sector I has available all the numbers from 0 to 1,999.

By SU we mean every recognizable action changing the surface of the earth, whether human or natural.The registration of all the elements useful for identifying a SU is an indispensable means for being able to reach a comprehensive interpretation of a site, by means of—wherever possible—the recognition of activities. Thus, both in the moment of excavating and in the act of recording the data, particular attention must be paid to the characteristics and stratigraphic relationships so that the sequence of individual actions or activities can be reconstructed in the matrix. The design of the forms we utilized satisfies these requirements, obligating the person responsible to fill out all the fields on the records. In this way, next to the discursive and sometimes rather informal comments appropriate to the Excavation Notes, the trenchmaster also must record the data in a precise and concise way.

As the excavation proceeded, the trenchmaster had the responsibility of creating a catalogue of all the individual SU he or she found, giving the reference number and position for each.

During the excavation, once a SU had been identified, the trenchmaster prepared an overlay (or single-context plan) and documented it photographically. Survey was done by a hybrid method utilizing a total station and traditional surveying tools (measuring tape, plumb line, drawing frames). Each photograph was entered into the photo list and annotated when shot. The photo lists were then matched with the photographs once they were developed and contact sheets printed, and an inventory of the photographs was created.

Photography was both chemical and digital. Black and white 35 mm chemical photographs were taken of all stratigraphic units and small finds (Kodak 5052 TMX was the preferred product). Important stratigraphic units and all small finds were also photographed with 35 mm slide film (Kodak Ektachrome was the preferred product). Small finds were also recorded on 35 mm color print film (Kodak Gold 200-6 was the preferred product). During the course of the project, consumer digital photography made great strides. In 1997 and 1998, digital photography was only used for informal shots to be used on the project’s Internet site. By 1999 and thereafter, a Nikon Coolpix 2.1 megapixel camera was purchased and used for supplementary documentation of stratigraphic units and finds. All non-digital photographs were digitized at high resolution and included in the photographic database described below.

In 1999, an aerial photographic survey of the site was carried out by a small radio-controlled helicopter, under the supervision of Robert Ajtai of VE.DO.

At the end of each season, reports were produced by each trenchmaster, synthesizing the results for his sector and including, as an appendix, the filled-out forms (Excavation Notes, SU and MSU forms, photo lists, etc.).

Once archaeological material left the site, it was taken to the project’s nearby laboratory in the Mulino of Licenza for inventory and study. All the material taken to the laboratory was cleaned and classified and entered into a related catalogue. For the noteworthy materials, a special form was developed (fig. 18), which reflects the main fields of the so-called “pre-catalogue.” The noteworthy material, after being photographed, was then inventoried according to the rules and procedures of the Archaeological Superintendency for Lazio.

Throughout the project, written forms were digitized and included in the project’s information management system. The processing of information from the catalogues, inventories, and forms was carried out with the aid of commercial software such as Microsoft Word, Access, and Excel. In addition, all the slides taken of excavated material were digitalized with the Epson GT7000 scanner at a resolution of 1200 DPI (dots per inch). The scans were saved in JPEG (Joint Photographic Experts Group) format at the medium level of quality. In this phase of the work, our priority was to maximize efficiency in managing data and not to prepare publication-quality images. We knew that, in the end, few of our thousands of images would be published, and those could be digitalized anew in uncompressed format at the highest level of quality in the production phase of publication. All of the overlays were digitalized and then vectorialized with AutoCad.

Luca Passalacqua and Stefano Camaiani, then two laureandi in the Department of Archaeology at the University of Siena, were responsible for our data management and for designing an information system that allows the user to make queries across most of the data categories (fig. 19).

Initial plans called for all the individual databases to be combined in a composite Geographic Information System (GIS) utilizing ArcView. In the event, this part of the data management project was not implemented for lack of personnel and funds. On the other hand, the lack of a GIS interface, while regrettable, did not materially compromise the usefulness of the system, which greatly facilitated analysis and interpretation during our study seasons and, afterward, during the production of this volume (fig. 20).

From a landscape architectural perspective, the overall siting and design of the villa take advantage of a variety of topographic features to create the architectural settings popular in gardens of the first centuries B.C. and A.D.: terraces, cryptoporticoes, and viewing pavilions (diaetae). Although Horace does not describe the contents and layout of the gardens of his villa, his poetry is set in the steep natural topography and such cultivated landscape as the rugged hills permitted (fig. 1). This is the landscape celebrated in the remarkable siting of the villa. The response of the achitecture to the spectacular vistas marks the villa as a type of “view villa,” popular in the first century B.C. and highly developed a century later. The original building must have been sited, at least in part, for the views, though we have few certain architectural remains from which to judge how the scenes were experienced. For the Flavian period, the peak of “view mania,” the siting and the archaeological discoveries at Licenza suggest that the architecture was designed to engage the surrounding landscape.

The siting of the villa’s residential and bath complex is unusual. The land was terraced to form a saddle between a small knoll to the east, known today as the Castagneto, and the lower slopes of Monte Rotondo to the west. As observers have noted, the setting was quite strategic both for views and for pleasant climatic conditions. Horace describes it as an arx enclosed by montes (Sat. 2.6.16), phrasing that well describes the site of current excavations. By terracing the saddle, the designer created broad level surfaces for gardens and working areas connected to the architecture, while providing extraordinary views from these terraced gardens to the north and south (fig. 2).

The Castagneto is a knoll rising about 50 m above the villa. At the summit, the 1998 survey team noted a remarkably level surface with a clearing in the woods. Today, the overgrown spools of coppiced chestnuts obscure the view, but in winter it is possible to appreciate a 360° view of the countryside surrounding the villa from this clearing (fig. 3). The knoll’s lower slope has now eroded over the east wall of the villa, and the overburden has prevented excavations of buried walls off the east portico. There can be little doubt that the architecture of the east side of the villa was carefully planned in relationship to this topographic feature, as the slope would have required retention. A retaining wall and a means of egress to the knoll could have been combined with either interior architectural features or exterior ones.

Villas of the first centuries B.C. and A.D. around the Mediterranean often featured such viewing knolls. For example, Herod the Great’s palace at Jericho had an artificial knoll with a pavilion at its summit (fig. 4). The pavilion provided views of the estate with its famous palm and balsam plantations. Closer to home, the Villa dei Papyri at Herculaneum and the Villa Iovis on Capri had viewing pavilions. It is difficult to believe that this knoll at Licenza was not incorporated into the villa’s architectural scheme; remains of a pavilion, however, have yet to be detected. Cores in the area demonstrated that the soil layers there have been very stable for tens of thousands of years (see Foss et al., ). This preliminary evidence suggests that the knoll may have had a gently rounded crest naturally, rather than a constructed terrace. It is likely that any disturbance for an architectural structure may thus be represented archaeologically as a minimal intrusion into the natural soil horizons, as with posthole construction. In March 2000, a ground penetrating radar survey was conducted on the knoll’s summit, but lack of time prevented a sufficient number of transects to be run. Further survey is warranted in this area to determine if there are archaeological remains of a diaeta or other structures associated with the villa.

On the north and south, the saddle, with its terraced platform for the villa, drops away about 50 m, quite steeply to the north, more gradually to the south. The view to the north, of the prominent hill at Civitella, is almost directly on axis with the villa. This view will be explored in more detail below, but it is worth suggesting here that when the as yet unexcavated northern side of the villa is investigated, archaeologists may consider the possibility of a garden, an ambulatio, and/or diaetae that address not only this hill, but the extraordinary panorama of the valley. The view to the south is no less impressive, though less visible now. Today the panorama begins with the view of S. Maria Delle Case on the slope below Roccagiovine, where the Temple of Vacuna may have been visible in Horace’s day (Epist. 1.10.49).

On the west, the long slope rising 530 m to the summit of the Colle Rotondo has also eroded, burying this side of the villa. How this slope was integrated into the villa is a more complex issue. It is possible that a peristyle at the northwest opened onto a courtyard that is now substantially buried under the eroded slope. Coring of the accessible portion of this courtyard has not produced garden soils; however, too little of the overall area is available for study to be sure that it was not planted, if only with shade trees. Also on this side of the villa, the recent excavations identified an atrium of the late Republican period (See Camaiani et al., ).

Lugli describes the garden at the northern end of the villa as one of two courts that furnished light and air to the rooms around it; the second is the so-called atrium roughly on axis to the south. Lugli regards Area 8 (his cortile A) as opening onto the countryside to the north; room 12 (Lugli’s cortile B) was an interior court with a water feature, but recent excavations revealed an earlier phase beneath it (fig. 5; and see De Simone, )

Lugli reports that the original appearance of Area 8 is not known, but that changes were made to the courtyard itself in the imperial period. Within the courtyard space is a low, rectangular construction, with small semicircular niches in the middle of each side. The construction consisted of a channel surrounding the central space, which was evidently occupied by a garden. This highly architectonic form of garden bed or planter was popular by the mid-first century A.D., as seen in the Domus Augustana in Rome, the Templum Pacis in Rome, and further afield, at Conimbriga in Portugal.

Recent investigations have shown that the north enclosure wall of the building was reconstructed incorrectly by Pasqui’s team: the boundary wall visible today should not be regarded as the north wall of the villa and the complex clearly continues northward (see Frischer, ). The implication of this discovery is that Area 8 was an introverted space, not facing the north view as indicated by Lugli, although the doors onto it along the north-south central axis may have formed part of a series of frames through it to the vista of the valley beyond.

Lugli also suggests that the entire area around this feature was a garden (un cortile fiorito) that surrounded one of the rooms (room 7) projecting onto it, which he interprets as the summer triclinium. The present ground surface of the area disguises any hint of a garden, and cores of the central feature revealed that the early excavators had removed all of the original soil before refilling and replanting it. While the area clearly bears further examination, we did not make it a priority for the feasibility study.

The garden excavations focused entirely on the area within the large quadriporticus, discovered during the 1911-1914 excavations by Pasqui and further investigated by Lugli and Price during the American Academy in Rome excavations of 1930-1931 (fig. 6). The architecture of the quadriporticus itself was the focus of further work during the 1997-1999 seasons and is reported separately (see De Simone et al., ). Lugli reports the dimensions of the quadriporticus garden as 34 x 76 m. It is comparable in size to the peristyle of the Villa dei Papyri (28 x 94 m), the House of Octavius Quartius at Pompeii (30 x 86 m) or the central peristyle of the Villa at Sirmione (50 x 80 m). Lugli thought that the garden was at the front of the building, with the entrance to the villa at the south end of the quadriporticus. Excavations in 1997 showed the south wall to be buttressed in such a way as to make an entrance here unlikely (see Passalacqua, ).

Within the quadriporticus was a spacious garden focusing on a large piscina in the center. Lugli describes it as follows:

“[The garden was] surrounded on all four sides by a crytoporticus or corridor with large windows opening on the garden. In the center is a large tank with two pilasters on the southern side, which perhaps supported statues, and with a drain for the outflow. Edible fish were probably kept in this tank, while the xystus or garden was laid out with avenues of box and borders of flowers, and rustic benches were conveniently placed in the shade of the fruit trees.”

This is the image that guided Thomas Drees Price’s 1932 reconstruction and has influenced the plantings installed at the site since the 1950s, as well as the reconstructions shown at the Museum in Licenza today. No actual archaeological evidence, however, was reported for fish-raising features in the pool, for walks, planting patterns, or benches. A flower pot was found during the Pasqui excavations and exhibited in the Museum, but its findspot is not known, nor is it clear if its function was recognized. Lugli and Price undertook excavations primarily in the northeastern area of the quadriporticus, where they discovered a niched basin within the portico. It was placed, apparently in a later phase, to terminate the east-west axis through the piscina. It is possible that the basin was visible from the garden, and its construction compromised the passage through the porticus at that point. This suggests that the facade was opened up to the garden. The Lugli-Price team also excavated the northeastern corner of the piscina wall and along the north edge, revealing that it had four piers rather than two as described above. No other results from their excavation in the area are known.

In summary, the features of the quadriporticus garden securely known archaeologically at the start of our project were: • the central piscina with its four piers, a tank and conduit leading to the drain in the southwest area of the quadriporticus; • a set of steps leading into the garden on the central N/S axis; • a flower pot from the Pasqui excavations; • the surrounding porticus with the articulation of its facade, seen to be a series of alternating doors and panels separated by pilasters. Within the porticus was a fountain feature on the east side of the E/W cross-axis through the pool; • an understanding of the slope of the garden (Price calculated a slope of 0.9% east-west, 1.2% north-south).

The challenge for the new excavations was to locate features within the garden soil itself, under the unexcavated central area of the quadriporticus.

To detect the fine features of a relict garden, as well as to reconstruct as much as possible the plantings, decoration, activities and habitats of ancient garden areas, we planned an integrated and coordinated series of specialist activities.

A geophysical surveying (remote sensing) team, a pedologist, and an archaeologist began the study of the quadriporticus garden and other gardens within the building complex. First, the remote sensing team located possible subsurface features. In 1997, a geophysical survey was conducted using a magnetometer. Further studies with resistivity and magnetometer equipment were conducted in 1998, and the area was checked with soil cores to test the readings (see Foss et al., ). By and large, debris blocked the cores taken for this purpose. Studies were also done of the larger area of the villa to describe the geological, geomorphological and pedological conditions at the site (see Foss et al., ).

Ideally, field work on the larger landscape of the villa estate goes hand in hand with garden excavations, especially with this type of villa. A team from Sheffield University worked at the site in 1998, conducted some promising preliminary field work, and proposed a feasibility study that would have integrated the environmental retrieval studies with an assessment of habitat and resources in the area. The project would also have identified, through systematic field walking, any archaeological remains of outlying features of the villa, such as other gardens, roads, and outbuildings, providing important information for the interpretation of the garden, villa views, and environmental evidence from the excavations. Unfortunately, the funding for this study was not available. Neverthless, this is an important dimension for future work at the site.

Cores are key tools for the early detection of garden soils, revealing which levels were formed through erosion, redeposition of fill, cultivation or natural processes of soil formation. Many of J. Foss’ initial cores were blocked by debris, but, after stratigraphic excavation through rubble, new cores “periscoped” down, permitting more accurate removal and readings of the fine soil layers and features of the garden by the archaeologists. This coordinated effort is worth planning into any garden or landscape project from the onset. In 1998, it permitted us to determine the presence of two levels of ancient cultivated soils, although only one could be reached by the archaeologists in the time available that season. Without the cores, the feasibility of a garden excavation project at the site would have appeared less promising.

Garden soils are identified on the basis on several characteristics. Roman soils were typically cultivated with fertilizer, which consisted of debris gathered from kitchens, barns, and even privies (if Virgil is to be believed). This material was gathered into large piles, in order to be used later and worked into garden soils and fields. As a result of this deposition process and continued cultivation, the artifacts from cultivated layers are highly abraded, often preventing identification. Finds within the soil include potsherds, bones, carbonized plant and wood remains, and other artifacts, all randomly scattered.

A team dedicated to the excavation of the garden soils is typically needed on Mediterranean sites. K. Gleason planned and carried out the strategy, coordinating the work with B. Frischer and G. Ponti. J. Schryver supervised the excavation and interpretation of the medieval levels. After various stratigraphic units in the garden had been identified, flotation and wet-sieving of soil samples were undertaken in 1999 and 2000. Recovery of plant remains was successful, despite rather poor conditions for preservation and sieving (see Ramsay, ). Animal bones, eggshells and land snails were also retrieved. No feasibility studies were carried out for insects, pollen or phytoliths.

Excavation focused first on the identification of the nature of the strata. It is evident from the levels of the 1911-1914 excavations that the quadriporticus lay under two meters of later deposits. The first meter consists of almost continuously cultivated soils. Around the perimeter of the courtyard, these preserve a rubble layer, dated to medieval times (fig. 7). As has been observed at other sites, a layer of rubble debris is often ideal for preserving garden features, and this proved to be the case in the quadriporticus. The best preservation occurred under the thickest portions of this layer, and, during excavation, we paid particular attention to the identification of cultivated surfaces. Once these were established, and the surfaces exposed, excavators turned to the identification and documentation of every small feature and change in soil color. Due to the short work seasons, we exposed small areas at the time (fig. 14). Only at the end of the three seasons was it possible to put the plans together and observe a distinct pattern to the finds. That said, the discovery of three planting pots and several pits in 1999, all parallel to the central N/S axis, gave us some immediate clues as to the design of the garden.

The archaeological stratigraphy is sufficiently clear and data-rich to reconstruct the general chronology, with the main periods of use dated by coins and pottery. The following discussion presents the garden excavation in two groups: Sectors V and VI, the central area of the garden with its pool; and Sector VII, a single large trench at the north end of the garden, at the base of the central steps. The excavation of Sectors V and VI is first discussed trench by trench; thereafter the activities they represent are related to each other. The various activities of each period are initially described with stratigraphic unit numbers, artifact descriptions, and soil context; then they are interpreted, within the limits possible for a feasibility study.

The Pasqui team first excavated the central piscina and its water pipe and small tank on the western side. They also identified two large piers along the south side of the pool. Lugli and Price pursued the pool’s outlines further on the north and east sides and located the northern two piers. In the published study of the excavations, Lugli describes the pool as having a spurting central jet (fontana zampillante) that provided an attractive focus within the well-tended gardens for those strolling in the porticoes. This jet is purely speculative: neither campaign fully excavated the bottom of the pool.

To clarify aspects of the construction of the pool and the contents of its fill, teams conducted remote sensing surveys over the pool and excavations were undertaken inside it (northwest corner, Sector V) and outside it (east side and near the southeast pier, Sectors VI.1 and VI.2, respectively). The pool walls seen today are heavily consolidated with a cement cap. No mortar or clamps survive inside the pool to hint as to the nature of the original coating, nor is there any indication of hydraulic mortar or cocciopesto to suggest that the pool was lined with waterproof material. Lugli and Ponti assumed that the walls visible today are foundations, the former on the basis of the materials of construction, the latter on the basis of garden levels to the north of the pool, and both on the assumption that the original garden surface was a short step down from the thresholds between the portico and the garden. The garden levels in cores and trench VI.2, however, place this hypothesis in some question. It may simply be the case that the limestone chip and cement walls seen today are but the core of walls once finished in finer materials. This issue is linked to phasing: the core suggests garden levels well below the top of the walls. The results of the 1999 excavation in Sector VI.2 address this relationship inconclusively (see below). The excavations to date do not establish the original date of the pool’s construction; nor do they clarify the interpretative issues discussed above. If anything, our studies have suggested new possibilities and raised more questions to guide future excavation strategy.

Sounding in the piscina (Area 25, Sector V)

Area 25, the 12.5 x 25 m rectangular structure located in the center of the garden, is interpreted as a pool (fig. 8). The excavation was undertaken to understand the chronology of its destruction, to verify the quota of the floor, and to re-examine its function. The original plan was to undertake several stratigraphic explorations and then to expose the entire pool with the help of a backhoe. Because of structural problems with the perimeter walls of the pool, however, the project had to be reduced in scope and was limited to a single trench in the northwest corner (fig. 6).

Study of the fill (SU 5001) in this trench revealed a surprisingly large amount of marble rubble, which probably resulted from the stripping of the pool’s revetment panels. Ceramic finds, however, were sparse, and no statuary or decorative elements were encountered. Thus it is difficult to date the destruction of the pool, beyond noting that in late antiquity we often find cisterns or large basins used as dumps for rubbish and debris. That the structure was a pool seems beyond doubt, even if in the area investigated no traces of decorative surface treatment of the walls or floor were preserved.

In this sounding, all that remained was the preparatory bedding for the floor of the pool, at a quota 2 m below the top of the perimeter walls. The bedding consisted of a thick stratum of opus caementicium (SU 5002). No hydraulic mortar or cramps were found on the walls.

Sounding in Sector VI.1

In 1998, a small trench was set out that extended from just inside the pool across the pool wall and into a baulk remaining from the earlier excavations. The aim of the trench was to reveal the nature of the soil layers above the Roman levels and to ascertain the preservation of the stratigraphy along the pool wall. The earlier excavators had “chased” the pool walls, but they did not dig deeply enough in this area to sever the connection between the original garden levels and the pool wall. The stratigraphy of this trench is correlated with that of Sector VI.2 (see the following discussion). Although no garden soils were encountered during the brief excavation, a core taken a few meters southeast of the pool indicated that they were at a lower level than anticipated (figs. 9 and 10).

Sounding in Sector VI.2

In 1999, a 4 x 6 m trench was laid out just east of the southeast pier of the pool with the following objectives: • to locate and examine the garden soils identified by cores taken in 1998; • to ascertain the relationship of the cultivated levels to the pool walls; • to examine the stratigraphy associated with the pool and perhaps date the construction of the pool; • to determine if the piers of the pool were original to the construction or added as support in a later phase.

The phases for Sector VI generally correlate with the phases given for Sector VII below.

Period I (Activities 1-5)

This Period cannot yet be correlated with other Periods at the site, as no datable pottery has been found, there are no trenches that link the pool area to stratigraphy from the enclosing porticoes, and we were not able to reach the natural soil in either trench. While we have information on phasing and construction techniques, we cannot yet offer a date for the original construction of the pool. For the present, the assignation of the pool to the “original building” is neither proven nor disproven, pending further excavation.

While this first Period has no confirmed date, later Periods and activities around the pool do correlate with other areas of the garden and villa. Activity 1: The first activity at the site is identified by soil cores, which attest to a redeposition of subsoil (See Foss et al., ). The redeposition suggests a probable leveling or raising of the ground. Such grooming of the surface may represent the leveling of the entire terrace for the construction of the villa as a whole, or a more specific preparation for the construction of the pool. There is no ceramic material to date this activity. Activity 2: A cut (SU 6024 in VI.2) just outside the outer pool wall is seen in both VI.1 and VI.2. This cut may be evidence of a construction trench, in which the walls are built up against standing soil. If, however, the site was graded as a terrace before construction of the portico and courtyard, an interpretation of a construction set against earth (controterra) suggests that the pool was built once the garden soils had already been deposited. This would imply that the pool was part of the imperial phase of the villa. If the pool was original to the platform, however, it would have typically been constructed first into the subsoils, and then the fills and garden soils would have been brought in and spread around the pool. There simply is not enough stratigraphic information to further inform us on this issue. Activity 3: The wall of the pool (SU 6026) is 6m thick and constructed of concrete and limestone chips. The excavation showed that the four 2m2 piers are integrated into the walls of the pool (fig. 11). It is notable that a mortar coating (SU 6025) was found in a highly decayed state on the outside of the pool wall, as it has been difficult to detect such material elsewhere.

Activity 3 is the construction of the pool wall. It is likely that the walls seen today are but the cores of walls with a finer original finish. It is difficult to conclude from the mortar found in the excavations that the walls were simply plastered, and no cramps or other evidence guide us further in interpreting the treatment of the walls. It is also possible, though speculative, that the walls are foundations for a structure at a higher level, or for a reused feature, such as a cistern, associated with an earlier period at the villa.

Activity 4: Immediately outside the pool walls, set within the cut, are fills SU 6006 and 6023. SU 6006, in Sector VI.1, is within the cut (SU 6009) of the earlier excavations of Pasqui and Lugli. SU 6008 is at the base of this cut. SU 6023, in Sector VI.2, is of a lighter color than the surrounding fill of SU 6022. No diagnostic material was recovered to date the fills of these cuts, nor were plant remains or other evidence of fertilizer recovered from the soil samples.

Activity 4 is problematic to interpret from test trenches. The clearest explanation is that the cuts are for the construction of the pool. This may suggest a construction set against earth. If so, the stratigraphy of VI.2 indicates that the earth cut into to create the pool was already layered with cultivated soil levels of earlier periods. This would indicate that the pool was installed later in the history of the villa.

The difference in soil color suggesting a cut might indicate the presence of decayed plaster.

Activity 5: SU 6019 and SU 6022, in Sector VI. 2, are redeposited fills around the pool. The placement of this activity in the chronology is speculative, as there has not been enough excavation to determine if the pool was cut into these soils or if the soils were laid down after the construction of the pool, as indicated above. SU 6022 should be examinated further as a possible cultivated soil.

We cannot rule out the possibility that a garden surface began at the pool edge, as no pavement or substrate for a pavement was seen. However, organic soil, amended with debris, was seen in a core sample just over one meter beneath this surface (the level at which the Pasqui and Lugli/Price excavations stopped, south east of the corner of the pool). This activity is placed in Period I. Either a garden was laid out at the same time as the pool construction, or there was already a garden in place, and the construction trench was cut into it.

Period II (Activities 6-10)

This correlates with a period of remodeling or alteration throughout the villa. The stratigraphy of the soils to the southeast of the pool is intact. It is clear from this trench and related cores that cultivated soils surrounded the pool, though no particular design could be ascertained in the small area of the trench, nor were any soil discolorations or fine features observed. No pots or garden artifacts were discovered, but limited ceramic evidence indicates a date in the second century A.D. SU 6013 overlies another fill level (SU 6017), with pottery dated to the first century A.D. Activity 6: This is the deposition of SU 6017. This fill is a clayey loam with visible amendments in the form of pottery fragments and occasional carbonized plant material. No identifiable plant remains were recovered from this level, but the soil has all the attributes of a cultivated soil. Depending upon the following interpretation of the structures, the deposition of this soil may have preceded the construction of the piers as a second cultivated soil layer. SU 6107 is dated by its scant ceramic remains to a time from the end of the first century B.C. to the end of the first century A.D.; the later date is more likely (see Angelelli, ). If the footings (activity 7) are associated with this level, they would have projected above the garden surface. Activity 7: This is the construction of two crude brick rubble-core footings (fig. 12). The two features, made of loosely compacted, degraded brick fragments, may be foundation supports for columns or trellis posts. They are spaced ca. 0.9 m apart, and a core taken to the east at the same interval produced more such brick fragments. The line of these features is parallel to a wall extending westward off the east portico into the garden. Further excavation is needed to reveal their extent to the west, as cores were inconclusive. If the footings are associated with SU 6017, they date to the first century A.D., while if they are associated with SU 6013, they may be dated to the end of the first century or in the second century A.D.

It is conceivable that these footings are associated with the surface of garden level SU 6019, as their bases are at the surface. It seems more likely, however, that they are either substructures for SU 6017 or for a feature associated with SU 6013 (activities 6 and 8). Further archaeological exploration, either by geophysical survey or excavation, should investigate the possibility that they are supports for a trellis or other light architectural feature associated with the extension of the aforementioned wall, as hypothetically reconstructed in fig. 13.

This is the level into which it seems most likely that the brick footings were set; however, the composition of this layer would not have presented a useful or pleasing surface to a garden or courtyard.

Activity 8: This is the deposition of SU 6013, a layer ca. 26 cm thick. It is a redeposited yellowish clay fill, lacking organic inclusions but with occasional pottery, bits of brick, tufa, and pockets of sand. We interpret this layer as fill brought onto the site, but not apparently for cultivation, as it does not have the necessary organic content. Ceramic remains date this level to the end of the first century or beginning of the second century A.D. Activity 9: This is the deposition of SU 6012, a clay-rich layer below the root zone, barely distinguishable from SU 6010 in color. Its presence was marked by the absence of tree roots and other recent organic material. The fill contained roof-tile fragments, a slab of pavement, and large tesserae. Environmental retrieval produced carbonized wheat grains, culm nodes, chenopodium, buttercup, primrose, and allium. Ceramics in the fill are dated to the end of the first century/beginning of the second century A.D. This appears to be a fill taken from the immediately surrounding area to level the ground around the pool, and it is strikingly similar to the eroded soils that buried the nymphaeum in the east portico. It seems unlikely that the plant remains represent cultivation in the garden, although it is premature to rule out this possibility. It is more probable that this is soil originating from the nearby slopes. Activity 10: This is the deposition of SU 6010, dated by a single potsherd to the first or second century A.D. It is the ancient level at which the early excavators stopped, and it has been infiltrated by roots of the modern sod grass and other organic surface materials. A soil sample from the lower levels of SU 6010 produced a carbonized seed of Galium (bedstraw), a local weed associated with cultivated areas.

This may be the ancient garden level of the area around the pool; the early excavators evidently thought so. If so, it may be related to SU 6008 from Sector VI.1. It is not possible to offer a definitive intepretation from the evidence—it will be necessary to correlate this test trench with excavation of areas protected by unexcavated overburden.

Period III (Activity 11)

This is a general designation for the period that represents the decline of the garden, and probably of the adjacent quadriporticus. Much of the area of Sector VI was excavated during the Pasqui excavations. This period is clearly preserved in the upper levels of Sector VI.1 (excavation conducted by G. Ponti).

Period IV

This Period has no activities preserved in Sector VI.2, as the levels were removed during the Pasqui excavations. It is represented in Sector VI.1 as the layer of rubble excavated as SU 6005. The rubble is made up of construction materials that seem to represent either neglect or active dumping. It appears that the materials were sifted through to remove pieces suitable for reuse.

This Period correlates with the corresponding Period in Sector VII.1, as well as with the test pits excavated for soil cores along the western edge of the quadriporticus. It is defined for the garden as a whole by the intentional dumping of debris in the northern and western areas of the garden, probably indicating the medieval reoccupation of the adjacent buildings. The excavated trenches of Sector VI show that the early excavations severed any relationship of this layer with the eastern wall of the pool (SU 6002).

Period V (Activity 12)

This Period, too, is represented mainly in Sector VI.1. It correlates well with the northern area of the garden, as a long, continuous period of cultivation on the site, with pottery from the seventeenth to the twentieth centuries. SU 6001 in Sector VI.1 is comparable in depth to the cultivated soils described in Sector VII.1. It is notable that these soils have considerably fewer potsherds and other inclusions than Sector VII. This may be due to a more sporadic cultivation history, and/or to the difference in the agricultural practices of the parcel owner, as Sector VI.1 lies on different property from both Sector VI.2 and the northern area of the garden. In Sector VI.2, the archaeological excavations removed most of the cultivated soils of Period V.

Period VI (Activities 13-16).

This includes the twentieth-century excavations and activities associated with the archaeological park. In the garden areas, generally, this is seen in photographs and records as excavation activity, as restoration activity and as cultivation taking place on the unrented portions of the site during the excavations. The pool itself was excavated in this period (see the previous discussion). Extensive restoration is seen around the quadriporticus. Regarding the pool itself, restoration efforts focused on the southwestern portions of the pool, while less restoration is seen on the northern and eastern portions. Photographs from the Lugli/Price excavations show that the earth removed in 1931 and 1932 was used to groom the slopes between the excavated and unexcavated areas to give the effect seen in Price’s model of 1932. The following discussion addresses only those activities encountered in Sectors VI.1 and VI.2 Activity 13: In Sector VI.1 this activity is represented by cut SU 6009 and the redeposition of possible garden soils SU 6006 and 6007 as one layer, which was later cut again (SU 6004) by the Lugli/Price’s excavation, as seen in the photographic record.

The cut SU 6015 and subsequent fills SU 6014 and 6016 in Sector VI.2 are commonly referred to as a “wall chasing” trench with subsequent fill.

Activity 14: Removal of soil down to the upper surface of SU 6010, the surface level of the 1999 excavation of Sector VI.2, after leaf removal. This fill is loam with occasional pottery fragments. Activity 15: In Sector VI.1, SU 6004 is the cut made through any eroded debris from the Pasqui excavation into presumably unexcavated fills in the narrow area between the pool and the quadriporticus. In Sector VI.2, SU 6011 is a linear compacted surface on top of SU 6010.

Activity 15 is the Lugli/Price excavation. There are no notes on the material removed in 1931, but the excavation of this cut is seen clearly in progress in a photograph. Sector VI.1 showed that the Lugli/Price excavation cleared the line of the pool wall (SU 6002) along its east and north sides, but did not excavate deeply into the stratigraphy. Careful study of various photographs taken during restoration of the pool over the years suggests that the Pasqui and then the Lugli/Price baulks were trimmed back on other occasions as well.

SU 6011 in Sector VI.2 appears to represent the compaction and deposition of mud on the wheelbarrow paths seen in the photographs of the Lugli/Price excavation.

Activity 16: Activity 16 is represented in Sector VI.1, where SU 6003 is the erosion of the soil (SU 6001) cut by SU 6004 into the “wall chasing” trench of the Lugli/Price excavation.

Sector VII was laid out as a 4 x 5 m trench in 1998, first by K. Gleason, who was able to excavate only the western half and a small area along the steps. Later in the season, the trench was brought down to bedrock on the eastern side under the supervision of L. Cerri. In 1999, the western side was completed and a backhoe was used to open a 5 x 4 m area, expanding the dig to the south and west. This was excavated as a series of steps, for safety reasons, and was not taken down to bedrock. It remains excavated only to the ancient garden level (Period II), with the southern steps remaining unexcavated entirely. In March 2000, a short excavation season of two weeks completed work on the trench prior to backfilling in June 2000. The trench was expanded by a meter to the east, by a meter to the west, and into the lowest of the terraced baulks to the south (fig. 14).

During the excavations we identified six distinct periods in this area of the garden. These are presented on the Harris Matrix (fig. 15), and described as follows.

Period I. The Early Garden (second century B.C. - first century A.D.)

This Period (Activities 1-2) can be dated to the late first century B.C. It is characterized by a cultivated soil layer, organically rich, but with few finds or features. Very little of this level has been exposed, but it may be interpreted as the garden associated with Period I of the villa (see Camaiani et al., , and De Simone, ). Activity 1: SU 7023, 7024, 7028. The earliest activity identified was a north-south cut in the yellow shale, just to the east of the current north-south axis of the central steps to the garden. No diagnostic artifacts or plant remains were recovered.

This activity is the shaping of the shale bedrock for the garden or terrace (see Foss et al., ).

Activity 2: SU 7042. The fill of this cut in the bedrock is a layer of redeposited brown shale-derived soil with characteristics of cultivation, erratic distribution of potsherds, carbonized plant material, and bone fragments. No discolorations or other features were observed in this soil, but the area exposed was limited. No plant remains were found in the soil sample processed (fig. 16). Diagnostic ceramics recovered date to the late Republican period.

The layer was exposed at the base of the trench excavated in 1998 and 1999, but only in the western part of it; to the east is the bedrock, reached in 1998 and in 2000. It continues to the south in the area excavated in 2000. In 1998, we observed that the garden soil appeared to run north under the steps. When we attempted to check the relationship in 1999, however, electrical wiring for new lighting in the archaeological park had been placed under the steps, destroying the stratigraphy and thus preventing confirmation of the continuation of SU 7042.

SU 7043 is a cultivated soil of the late Republican age. It is likely to be a garden of the first phase of the villa, but this cannot be confirmed without the discovery of further features or architectural associations. The cut of the bedrock is man-made, but not enough area is exposed to understand the intention. The line of this cut should be related to the Republican atrium identified beneath the bath complex and to other first century B.C. remains, as the interface between the soil level and the bedrock is along a clean north-south line. No specific relationships of this feature to the adjacent architecture are obvious at present; however, if this garden soil in fact continues for any distance under the steps or under the portico itself, it may suggest the phasing of those architectural features, possibly dating them to the Flavian period. This could be clarified with trenches anywhere along the wall, as the Pasqui excavations did not reach this level.

Period II. The Garden of the Flavian Age

Period II (Activities 3-8) represents a time of remodeling or alteration and is characterized by yellow, clayey, shale-derived fill, with inclusions of small quadrangular pieces of brick pavers, stones, and occasional pottery of the first century A.D. This fill was deposited to raise the level of the courtyard to form the feature that dominates this period, i.e. the Flavian garden. The cultivated soil contains pottery, coins, and other artifacts, as well as carbonized plants, bone and mollusc remains. The most exciting discovery was of a flower pot (VH 148, SAL inv. no. 114428, from SU 7040) and an amphora (VH 160, SAL inv. no. 114550, from SU 7040), reused as a flower pot. These, together with remarkably well-preserved pits, stake holes and post holes, indicate the design of the garden. The finer soil features are preserved only under the medieval rubble layer (see Period IV). It is clear from the remains in this small area that the potted plants date to at least the mid- to late first century A.D., but there are hints of alterations to the design during and after this time. Activity 3: SU 7021, 7022, 7027. This activity represents the burial of the original garden surface SU 7042 with 30 cm of redeposited shale-derived clay soils, perhaps from building activity elsewhere in the garden or villa, mixed with stones and fragments of small quadrangular pieces of brick for paving, found erratically distributed in the matrix of the fill. SU 7027 was sampled for plant remains, and a cereal grain fragment of indeterminate identity was retrieved.

Ceramics from this level contain identifiable types, but their chronology is uncertain (see Angelelli, ). A single rim fragment from the late second century/early first century B.C. is present, but this may be residual.

The original cultivated surface was buried to raise the level of the garden 30 cm, presumably as part of the Flavian renovation represented in the next activity. The contents of the fill may include elements of an earlier garden (the pavers, for example), but we cannot conclude this from the test trench.

Various small postholes and soil discolorations are evident as inclusions of grey clay or organic soil in the matrix of SU 7021, 7022 and 7027 (see below, and fig. 24). These are postholes and planting pits from the cultivated level above. Preservation conditions in the soil above made it difficult or impossible to see these features in the surface, so they are best recorded at this level. As yet, we cannot assign a chronology to the phasing of each act of setting a posthole or pit.

Activity 4: No SU was assigned—feature not removed. A line of chipped limestone was laid out just west of the north-south central axis. As the feature was uncovered in the southern area of the trench, it began to curve out slightly to the east. No significant pottery is associated with this stratum (figs. 17 and 19).

As this feature was first observed only under the bases of the flower pots, we initially interpreted it as gravel laid to aid in draining the plantings in this clay-rich soil. We were not able to excavate along the sides of this feature to find its bottom. The size and density of the stone chips look similar to the images of the pool walls prior to restoration; this may not be a level of loose gravel, but the disintegrated top of a masonry feature that separated the beds of the lower garden, or a phase of the upper garden, that we have too little evidence to identify. We simply do not know how substantial a feature it is.

Activity 5: SU 7019, 7026, 7029, 7040, 7041, 7043, 7047, 7063, 7070, 7081. This activity involved the deposition of cultivatable soil. Another layer of brown clay loam, richer in artifact content than the lower garden soil, was laid out and cultivated over the yellow clay fill. The horizon between the yellow clay fill and the cultivated soil is typically sharp in the sections. However, along the baulk in the southeast area of the trench dug in 2000, the upper cultivated soil formed a sharp, but furrowed, interface. Not enough evidence of such furrowing was seen in other baulks to allow a full interpretation. Ceramics from this level include residual Republican wares (second-first century B.C.) and fragments dating throughout the first century A.D., with the major finds centering on the Flavian era.

Remains of cultivated crop plants and related weeds were retrieved: an unidentifiable fragment of nut, Pisum sativum (common pea), milk vetch, Lolium sp., crane bill, Allium sp., and bulrush, a plant found in wet conditions. A flotation sample produced a range of cereal, cereal processing debris, and cultivated herbs and weeds (see Ramsay, ).

This is a cultivated soil layer from a Flavian era garden. Residual pottery fragments are typical of amended soils, and the soil of this level has clearly been amended, or fertilized with food processing remains, hearth sweepings, and other burnt debris from kitchens thrown in a compost pile. The features of this cultivated area that lead us to conclude it is a garden soil rather than a field are discussed below.

Activity 6: SU 7067. A line of pots is set into the amended soil. These may be separate activities over time (i.e., different seasons, different years), but the dating of the pottery puts all of the vessels into the same time frame, that of the mid- to late first century A.D. (fig. 18). The northernmost feature is the bottom portion of a small, purpose-made olla perforata (flower pot, VH 203, SAL inv. no. 114529) found bottomside up at the margin of the garden level and the layer of debris above. It is not in situ, but was located along the line of the other features and may be very near its original location. Also found at this margin was a large fragment of a well-preserved glass plate (VH 194, SAL inv. no. 114534, from SU 7061). Approximately one meter to the south, a complete perforated olla (VH 148, SAL inv. no. 114428, from SU 7040) was found set into the garden soil, its rim approximately 3 cm below the surface level (fig. 20; see Macaulay, ).

Soil retrieved from inside the pot (SU 7044) produced remains of horse bean, elderberry, an indeterminate fruit or nut and several plants that grow in wet places: cranebill, sorrell, and sedge. Clearly, the process by which the soil came to be in the pot was a separate and earlier activity that took place away from the garden itself, but it is most appropriate to mention it here.

Emanating from the complete pot is a series of seven small circular holes, 4-7 cm in diameter, appearing as dark soil discolorations on and just below the surface of SU 7040 (fig. 17). These were not excavated and thus have no SU number.

Parallel to the central axis, 0.94 m to the south, a small pit was identified by the looser consistency of the soil. A narrow amphora had been removed from the layer immediately above this location, and it may have caused the pit to form, or it may have been embedded in this location originally.

Finally, 1.1 m further south along the same line, the upper third of a cylindrical-ovoid amphora was found, placed upside down in reuse (VH 160, SAL inv. no. 114550, from SU 7040). It had been shattered, possibly before deposition (fig. 22; see Angelelli, ). The soil from inside the amphora (SU 7048)—again to be regarded as an earlier, off-site activity—contained a single grain of cultivated barley.

This line of features represents a series of planting pots, embedded in the garden soil of SU 7040 (figs. 17 and 21). The olla perforata is a type of purpose-made planting pot (see Macaulay, ). The cylindrical-ovoid amphora is almost certainly a planting pot in reuse, a practice commonly seen at Pompeii, Hadrian’s Villa, and other Roman garden sites in Italy. The breaking of pots prior to planting is suggested in the ancient literature. The plant remains found in the fill of these pots are characteristic of the fertilizer rather than plants that may have grown in the pots themselves. The other feature in the line is a planting pit without a pot.

Activity 7: This represents a series of activities indicated by small dark pits and dark circular holes surrounded by cemented stone (fig. 23). It is not possible to determine a sequence or chronology of these features in relationship with those described in activity 6. The features were not excavated, so there are no associated SU numbers. No datable material was recovered.

On the north side of the garden, two small holes (ca. 7 cm in diameter) of brown, organic soil surrounded with mortar and small stones (20 cm in diameter) were found, one on either side of the north-south axis, though not paired (“stake holes,” in fig. 17).

Round postholes, 5-7 cm in diameter, filled with brown soil, were noted during the excavation of the yellow fill of activity 6 in a number of locations. These are noted on the plan and tentatively appear to conform to a pattern symmetrical to that of the west side of the garden’s central axis.

On the east side of the central axis, where the overlying cultivated soil becomes very thin (7-9 cm) in the areas where it overlies higher bedrock, excavation of the fill below revealed various pits, up to 30 cm in diameter (fig. 24). Further exploration is required to determine if the shallowness of the soil is due to erosion or to some feature of the garden requiring little depth to the soil.

The small holes surrounded by mortared stone may be stake holes for a light reed or wooden garden feature, as the diameter of the hole is appropriate. The support provided by the stones and mortar is slight (we did not excavate the features), and the purpose may have been to protect the base of the feature from rot.

The 5- to 7-cm postholes without supports are more difficult to analyze at this point. They are almost perfectly round, which suggests the interpretation of a posthole rather than a plant hole. These posts have been pressed or hammered directly into the ground; there is no evidence of pits dug first then backfilled around the posts. As the posts are found in the layer below the cultivated soil, they appear to be sturdy supports, perhaps for a lightweight fence, such as the type of reed fence seen in garden paintings. They would not provide sufficient support for an architectural feature, such as a trellis.

The larger irregular pits seen in the yellow fill below the garden level are most easily interpreted as pits for small shrubs or plants. They are not carefully made, and it is not possible to ascertain from the pits themselves if the plants grew intentionally or wild; however, preliminary studies of their location suggest a place in a coherent design pattern.

Activity 8: SU 7005. Traces of a plaster surface were identified along the northernmost edge of the trench. It consisted of a layer of lime mortar only partially revealed and not removed. No datable ceramics or artifacts were recovered. This activity represents the poorly preserved remains of a walk, landing or other feature, probably associated with the lowest unrestored step on the central axis (activity 9). Activity 9: SU 7006. This activity represents a constructed ledge of mortar, corresponding with the dimensions of the reconstructed stairs above, although this step has a higher riser than the restored ones above in the series (fig. 25). Disturbed soil (SU 7002) above this step makes it unclear whether or not the step had been discovered by Pasqui and deliberately left unrestored, or if it was first discovered in 1998. In the center of this feature is an opening framed with brick tile (SU 7007). No significant pottery was identified from the stratigraphic units in this activity.

The most obvious interpretation is that this one step represents the construction of central steps into the garden, more or less as seen restored. It is unclear, however, why the steps were furnished with a channel (SU 7007) set above the lowest step. This feature may be contemporary with or earlier than the activities above, as it is not possible to discern the phasing from the evidence exposed within the sounding. Considering the presence of this drain and the absence of any paving or definitively compacted surface at the base of the stairs, we should keep open the possibility that the reconstructed stairs were incorrectly interpreted. It is worth considering whether the steps were intended to provide access, or were part of a stepped water fountain, popular in the first and second centuries A.D. Further excavation is needed to explore this feature; as noted above, however, after our 1998 season, electricians dug a narrow trench and laid in wiring for lighting at each side of the central stair.

Period III: Decline of the Garden

Period III (Activity 10) represents the decline of the garden, and probably of the adjacent quadriporticus.

We have poor stratigraphy for this Period, which is broadly dated by coins and artifacts from the third to the fifth century A.D. Any phasing information that might have come from the association with the surrounding architecture was destroyed when Pasqui’s team severed the relationship with the architecture in pursuit of the line of the walls. Activity 10: SU 7020, 7037, 7038, 7039. The mixed fills of material from this period seal portions of the cultivated layer discussed above. Lenses of wall painting fragments (primarily red, yellow, and white, most without decoration), eroded plaster, patches of debris, irregular surfaces, and finely eroded materials cover the cultivated surfaces between 10-23 cm in the northern part of the trench, closest to the building. In the southern part of the trench, as one moves out into the courtyard, there is hardly any distinction between the Flavian cultivated surface and those above. Plant remains from SU 7038 consisted of wild grasses.

Ceramics from this stratum range in date from residual material of the mid-first century A.D. to more significant material dating from the third century, with types in use until the fifth. Coin evidence also offers a terminus post quem of the fifth century A.D., although the types found were in use from the late third through the fifth (see Buttrey, ).

The lowest fills over the cultivated surface of the garden can be interpreted as the decay, collapse and erosion of the plasters of the surrounding porticos onto the garden surface immediately nearby (fig. 18). This process did not apparently extend far out into the courtyard, and if it did, later agricultural processes obliterated the traces. The layer is best preserved where protected by later fills.

Within the limited area of the trench, we did not identify any deliberate destruction in this phase. We need to see more of this layer to interpret the role of the artifacts. If they are residual, this suggests activity in the third to the fifth centuries prior to the deposition of this layer, or activity occurring at the villa during a period of neglect of the stucco decoration of the building exterior around the courtyard.

The preservation of carbonized seeds of wild grasses may be consistent with some burning off of wild grasses growing over the site, or they could have been blown in from burning in the greater vicinity.

Period IV: Medieval Occupation

Period IV (Activities 11-12) is characterized by the intentional dumping of debris, probably indicating the medieval occupation of the adjacent building. Some three meters from the building wall into the garden, the debris piles taper off and the ground is marked by cultivated soils that merge with cultivated levels above and below (fig. 18).

This Period is distinguished by deliberate human activities taking place after the main occupation phases of the first century B.C. to the second century A.D. Judging from the garden trenches alone, the activities appear to be a clearing of debris from the villa, and probably cultivation of the courtyard beyond the zone of dumping. The Pasqui excavations severed the stratigraphy of these later activities from the stratigraphy of the architecture by trenching along the outer wall of the building, so it is difficult to make specific associations between activities in the building and those seen in the garden. One has the impression from the nature of the materials that the building was cleared out for reoccupation and usable pieces of construction material were removed. Activity 11: SU 7004, 7010, 7011, 7016, 7025, 7035, 7036, 7052, 7055, 7058, 7059, 7062, 7064, 7074, 7079, 7082. A clear margin distinguished the soils of activity 10, which had little evidence of human effort in the deposition of the fill, from a complex area of apparently intentional deposition above it. This level consists of adjacent deposits of debris, too numerous to define and number individually during the feasibility study (fig. 26). These deposits contained construction material and artifacts from the building: fragments of slate, tegulae, imbrices, bricks, triangular column bricks, marble, painted plaster from the columns and walls; rough volcanic rocks (of the type often employed for grotto effects); blue glass tesserae, black, white, and red tesserae of varying sizes; and a range of artifacts, shells, molluscs, and carbonized plant remains. Among the artifacts were pottery, coins, various metal fragments, and extraordinarily well-preserved fragments of window glass and other glass artifacts. Plant remains retrieved from two separate dumps indicate varied uses: sample 13, for example, contained cultivated wheat, barley, and olive, while sample 16 simply contained wild grasses.

Ceramic remains are largely residual fragments from the late Republican period through the imperial age, with types dating from the fourth to the sixth century A.D. Two coins are datable to the fourth century. The terminus post quem is thus the sixth century A.D., although an earlier date may be more likely.

The size of each rubble unit suggests individual dumping episodes. Two impressions stand out when assessing the artifacts recovered from the deposits. First, among the roof-tiles and bricks, nothing was whole. The composition of the piles strongly indicates that the debris was first sorted, then dumped in the courtyard. Similarly, with the artifact assemblage, we have an impression of sweepings that contained things either not noticeable or not worth collecting, as opposed to a habitation dump or artifacts lost accidentally in an inhabited area. Most of the pottery, glass plates and glasses, metal objects, even coins, are of Roman date, but the occasional fourth to sixth century potsherds provide a terminus post quem.

Overall, we interpret activity 11 as an effort to clear out the building for reuse, dumping the material into the garden. We cannot rule out the possibility that the dumps only represent a kind of sorting through or pillaging of the ruins for building materials, although if so, the pillagers were doing some fine sweeping as they went along. The lay of the dumps, sloping down towards the courtyard, suggests that the debris may have been dumped “out” of a relatively intact building into the courtyard. The debris apron appears primarily on the north and west sides of the courtyard where substantial buildings remained behind the porticoes. It does not appear that anyone made the effort to dump more than 2-3 meters into the courtyard. Beyond the edge of the dump area is cultivated soil, and it is possible that the process of later cultivation disturbed earlier debris, at least at the edges. Pasqui’s excavators cut a trench along the building wall, which prevents us from determining if the material was simply dumped off the retaining wall at the north. The piles taper off on the south side, and there is some directionality to the lay of the fragments in some piles, but not others (fig. 18).

Activity 12: SU 7033, 7034, 7051, 7053, 7054, 7056, 7073. South of the debris piles, the soil is disturbed by cultivation from the surface to levels lower than we were able to reach. Units merged and were assigned new numbers based on the most subtle of features. Essentially, where there were no rubble piles from late antiquity, no clear traces of activities 6 to 8 could be found. From the surface downwards excavators encountered a series of merging layers, all characteristic of cultivated soils, whose distinctions in content were more evident in the sections than in the horizontal surfaces of the trench. In general, we can observe different ceramic content in different broad stratigraphic zones, and the soil is more calcareous in the more recently cultivated soils than in the clayey shale-derived ancient ones. There are no structures or even particularly evident soil features, such as tree pits, although in the upper levels we searched for some indication of the trees evident in the 1911 and 1930 photographs of the site (fig. 27).

Ceramic evidence and soil structure suggest that cultivation took place regularly, if not continuously, from the early through the late medieval period. The level of the cultivated surfaces has built up nearly two meters since the earliest detected cultivated soil of the late Republican period. This can be attributed to the combination of material brought in for soil amendment and alluvial deposition of soils from cultivated slopes east and west of the site.

It is notable, however, that the ceramic content in the cultivated areas of Sector VII is much greater than that seen in Sector VI. This may be due to the proximity of Sector VII to the building complex. Moreover, Sector VII lies in historically different property (parcel 1215, owned by Foschi) than Sector VI.1 (parcel 1214, owned by Angeletti) or VI.2 (parcel 1213, owned by Caponetti). Depending upon the longevity of these holdings, agricultural practices may have varied between properties. It would make an interesting contribution to the cultural landscape history of the area to examine this possibility in the baulks as excavations continue.

It is not possible to judge from these excavations, therefore, at what point any medieval inhabitation of the site ended. A July entry in the Regesto Farfense from 1011 points to local initiatives towards incastellamento. This may be a plausible terminus post quem, as the cultivated levels directly above this rubble layer contained pottery later than the thirteenth century reoccupation of the site. The presence of potsherds in the soil is not in itself an indication of inhabitation on the property, as it remained a customary practice to fertilize soils with kitchen debris, including broken pottery from various periods.

Period V: Surface Levels of the Courtyard

Period V (Activity 13) is a continuation of the agricultural phases of Period IV. We identify this level as the agricultural activities that can be seen in the photographs of the Pasqui excavations—orchards and gardens still under cultivation as excavations progressed. The absence of well-formed surfaces in the upper two meters suggests that continuous cultivation begins between the thirteenth and fifteenth centuries. Activity 13: SU 7001. Agriculture appears to have been fairly continuous, with pottery gently gradating from the twentieth century to medieval times as one progresses down through the mixed levels. Horizons are merged, as one might expect in plowed soils. No distinct soil features were seen when these surface levels were manually excavated in 1998 and 2000. In the expansion of the trench, these levels were removed mechanically.

Period VI: The Twentieth Century

Period VI (Activities 14-16) is the pre-excavation surface level. It represents the surface seen in the 1911 and 1931 photographs of the Pasqui and Lugli/Price excavations. Cultivation is active in photographs of the Pasqui excavations. During the Lugli/Price excavation, the orchards remain, but without evidence of activity. Rather, the site begins to be groomed for presentation as an archaeological park. The excavators’ trenches within the garden date to this Period as well. This Period represents the excavation, restoration, and planting that created the site as it was encountered at the beginning of the current project. Activity 14: SU 7002, 7012, 7013, 7017. A trench, ca. 3 m wide, cut along the building wall on the north side of the garden through the ancient levels.

Photographs from the Lugli/Price excavation of the 1930s show the garden side of the trench to be quite irregular (fig. 28), although without comparative photographs from the earlier excavations, it is difficult to tell if this was the weathering of two decades or if vertical baulks were simply never created by the excavators. During the course of Pasqui’s work, cultivation continued above the trench to the south. The cut of the excavation trench was readily identified, as were the later fills of it.

In 1998, we were initially unable to date the garden level and the rubble piles, as both contained small bits of a grey rock that at first sight looked like the bits of Portland Cement used in the restoration of the steps at the time of the Pasqui excavation, but it was soon identified by Foss (see Foss et al., ) as a type of local tufa. For the most part, Pasqui’s team had excavated down to the level of activity 3 and not lower. A half-meter of erosion and fill from the later landscaping had added 25 cm to the bottom of the original excavator’s trench, but it was never completely backfilled (fig. 18).

Activity 15: SU 7009, 7032, 7050, 7069. Sloping fills extending from the unexcavated upper surface at the south to the area of the lowest step of the restored stairs.

During the Lugli/Price excavation, the site was altered, and a small archaeological park was created using the excavated earth to groom the slopes. The baulks left standing in the area of the quadriporticus were graded to produce sloping banks leading from the unexcavated surfaces down to the ancient level. Stratigraphically, this appears to have been primarily a process of cut and fill; some material was taken off the top of the baulk and deposited at the base, creating a slope. This would have happened through erosion to a certain degree, but the photographs and Price’s model indicate that the slopes were intentionally regraded (fig. 29). Price shows the sloping banks in his model, which he created either in the winter or late spring of 1932.

Activity 16: Surface activity—no SU were assigned. The surface level of the site reflects the last phase of cultivation, seen today as grass, spirea, rose, and rosemary, removed for the excavation, and cypress and other trees left in place.

The area surrounded by the quadriporticus was also planted with ornamental trees and shrubs, perhaps as the old orchard trees died. The cypress trees were planted around the perimeter, and other trees appear to have been selected in accordance with Lugli and Price’s interpretation of the area as a combination of ornamental and fruit-producing trees and shrubs.

The earliest construction that we discovered here occurred between the second half of the second century B.C. and the first half of the first century B.C. The area was transformed by the creation of a large quadrangular space (10.40 m x 10.40 m, corresponding to 35 Roman feet), in the center of which was found a square impluvium surrounded by four columns (figs. 7-10).

The only stratigraphic sequence pertaining to this period came to light during the excavation of the interior of room 38. Here the natural clay had been cut by the insertion of a drain or channel (activity 3) that had already been identified during the excavations of Pasqui. The drain was made of two small walls in stone, a bed formed of tegulae and a pitched cover of roof-tiles. The structure was sealed by stones bonded by a yellowish-gray mortar. A very compact stratum of clay rich in carbon covered the drain. Then the walls of the room were built in opus incertum with the first course composed of uniformly shaped cubilia (activity 2). The floor of the room consisted of stones mixed with earth (activity 4), on which was created a square impluvium connected to the sewer below by means of a conduit covered by mortar. Four piers, which probably carried the compluvium roof, were raised directly on the stone surface, at the four corners of the impluvium. The complete absence in the room of a proper floor, as well as the absence of an internal revetment of the impluvium, leads to the conclusion that the room was never completed and that its structures were completely obliterated and reused as foundations in a later building phase. Given the poor state of preservation of the room it can be very cautiously hypothesized that it was an atrium (Corinthian?) perhaps associated with the opus incertum wall that would later become part of the quadriporticus (see De Simone, ).

Activity 1: Deposit of late Republican strata inside room 38 (SU 874, 875, 877)

In the zone excavated inside room 38—and probably also in the other areas not investigated to this depth in the location of the future frigidarium—the space was filled by three firm strata of clayey earth that raised the original level of the terrain in order to fill in the difference in level existing between the plateau of the villa and the hill above. The composition of the various strata (rocks, bricks, and charcoal), mixed with a few potsherds, is analogous, and their temporal relation is very close: the division into three separate stratigraphic units is only intended to highlight the different levels of the fill.

Activity 2: Construction of the square room (SU 291, 292, 310, 311, 420, 464, 465, 612, 614, 620, 637, 644, 652, 653, 802, 810, 833, 837, 838, 845, 871, 876, 879, 880, 1001, 1014)

In the western zone of the villa a large square room was built. The perimeter walls of the space have foundations set against earth (controterra) and composed of chips of limestone mixed with earth; the walls show a first course of cubilia laid horizontally and an elevation in opus incertum. At the center of the room four plinths or piers were built, composed of stones and tile fragments laid in horizontal courses.

The eastern pier of room 40 (MSU 653) was covered by a squared stone of good workmanship (MSU 612), on which perhaps was supported the base of a column. The western pier (MSU 614) was preserved only to a low level, as was also the case for the pier inside room 39 (MSU 1001), of which only two courses remain.

In contrast, the pier now inside room 38 (MSU 810) is much better preserved. In it the building technique is clearly visible, characterized by a first horizontal course in bricks on which are laid courses of small blocks of the local limestone, known as cardellino (fig. 11). The core of the wall was comprised of brick fragments and small stones. The very impressive foundation of this pier (MSU 876, related trench SU 879) is made of chips of stone arranged without pattern and mortared together.

Figure 11 Pier now inside room 38 (MSU 810=10042).

In the space between the piers was built the impluvium, of which remains only a floor paved in stones laid horizontally. Only in the later room 39 does the floor tend to rise in proximity to the pier. A central opening led directly to the drain below (cf. activity 3).

Activity 3: Construction of a drain (SU 808, 809, 817, 824, 825, 826, 827, 872, 1017)

At the same time that the square room was built, a drain with an east-west orientation was constructed along the main axis of the room. The drain was covered by roof-tiles (78 x 42 cm), pitched and resting on two simple walls built against earth, with a thin layer of mortar on the lower part. Two meters below the ancient surface was the bottom of the drain, on which the waste water ran. It was made of tiles laid out flat and set close together. Only a small section of the original channel remains, so it is impossible to determine where the drain discharged its water. It cannot be excluded that it had a relationship to the principal sewer of the villa, which runs parallel to the quadriporticus.

Activity 4: Installation of the surface level of the atrium (SU 632, 651, 869, 873, 1015)

Over the entire surface of the new room a uniform stratum was laid down, consisting of stones of small and medium size mixed with a little earth (SU 651, 873, 1015). Above this was placed a further layer of earth with a sandy and very compact matrix, evidently laid down to even out the irregularities of the surface below, which was characterized by many dips and subsidences. No traces of a proper pavement were found. It seems quite probable that, until the reconstruction of Period II (see the following discussion), this earthen floor served as the surface level for the entire room.

As a result of the construction of the bath complex, which significantly modified the aspect of the western part of the villa, the entire area underwent a dramatic transformation in a period that can be dated to ca. 80-150 A.D.

Stratigraphic data concerning the construction of these rooms come from all the areas investigated and have permitted us to clarify the chronology and the building sequence of the new structures.

The floor of the square room (Period I) was raised by dumping earth (activity 5) containing very little pottery but a great quantity of fresco fragments datable to the second half of the first century A.D. (see Mols, ). This intervention saw the destruction of the four piers of the atrium and the obliteration of the impluvium. On these strata a uniform surface of rocks was later laid, which functioned as a loose stone foundation, and then a concrete layer was added. Atop this was placed a very simple black ground mosaic with a double frame along the sides and decorated internally, at least in the part that is preserved, by triangular crustae of marble (activity 6; see Werner, ).

To the west of this room was built a new space with a semicircular shape (room 37; activity 9). It was designed with a central niche having three steps, and two other niches, smaller in size, which allowed access to the square room (fig. 12). The floor, set about one meter lower than that of the adjacent room, was covered with a mosaic made of white tesserae in the parts preserved (see Werner, ). The walls were reveted with slabs of white marble set on a thick layer of cocciopesto. At the same time, a narrow trench was dug in the western zone of room 37 on line with the central niche. It was cut out of the natural clay and was intended as the bedding for a lead fistula to bring water into the space (activity 11). The presence of the layer of cocciopesto, of the fistula, and of a bench covered with marble along the perimeter of the room allows us to identify the use to which the space was put. It was a plunge pool connected directly to what appears to have been the great frigidarium of villa’s new thermal complex.

To the south of the pool a series of strata (activity 7) that is related to the construction activities in the area was intentionally laid down. Above these were superimposed, at a slightly later time, some strata of dumping (activity 8), which raised the level of the surface by about 50 cm. The terminus post quem for this action is set by several pottery fragments assignable to the Flavian period. A lead fistula was set into a cut made into this fill (activity 12). The pipe, connected to that of the frigidarium, carried water into room 43. The presence of the stamp of the manufacturer, C. Iulius Priscus, does not help to establish a date, since the name is quite common (see Bruun, ). Above the pipe was built a new room that was rectangular in shape (room 50; activity 13) abutting the frigidarium, with walls made of small blocks of cardellino with soil used as a binder. The room had a beaten earth floor—very compact and uniform—consisting of irregularly shaped small stones and tile fragments (activity 15). It is difficult to interpret the purpose of this space, but the presence of an earthen floor makes it likely that it was a service area.

In this phase, on the upper plateau of the villa, a series of very compact clayey strata (activity 16) was laid down, on which was created a floor in beaten rubble. On this surface a colonnade in limestone was erected, of which only one column remains. The colonnade probably ran along the western side of the villa in the area of the bath complex (activity 17).

To the south of room 50 a new space (room 51) was built, with two shallow apses opposite each other, whose walls were made of cardellino (fig. 13). Partially excavated, the room may be topographically related to the southern zone of the baths and may have functioned in tandem with Areas 43-49.

At the same time, to the north of the frigidarium, seven columns were erected with a north-south alignment. They were probably elements of a colonnade overlooking a garden (Area 35). In the same zone a new room with a rectangular shape was built in the bath complex (room 34); it was completely explored and heavily restored by Pasqui. The recent excavations brought to light the foundations of the western wall of this room, which abutted the northern wall of the late Republican square room (atrium). Below the floor level of the colonnade ran a channel, oriented east-west, which carried water into room 34.

Activity 5: Fill strata of the rectangular room (SU 613, 615, 629, 630, 631, 645, 819, 854, 858, 860, 865, 868, 870, 1006, 1013)

In this period, the surface level of the square room was raised quite substantially (on average 50 cm) through the accumulation of a series of strata of clayey earth that definitely eliminated the loose stone floor, the impluvium, and, probably, the four piers. Some of these strata are characterized by the presence of numerous fragments of plaster, datable to the second half of the first century A.D.

Activity 6: Creation of the mosaic of the frigidarium (SU 602, 603, 604, 621, 624, 626, 813, 823, 830, 840, 857, 1002)

The complete remodelling of the northern area of the baths is attested by the building of a new mosaic pavement in the zone previously occupied by the atrium and impluvium.

A uniform layer of stones (SU 604, 830, 1002), ca. 30 cm thick, was laid out directly upon the strata of clay accumulated above the Republican floor (activity 5). Atop this was set a nicely leveled layer of conglomerate made of lime and tile fragments, ca. 3 cm thick (SU 603, 621, 823, 840, 857). Above this a top bed of mortar was laid to support a pavement mosaic of black tesserae framed by two white bands and decorated with triangular marble inserts (SU 602, 624). The decoration of the room was completed with the revetment of the walls with white marble slabs (SU 626).

The function of the room is not completely certain, although in view of the plan of the baths and the room’s connection to the adjacent pool for cold plunges (cf. activity 9), one may speculate that it served as a frigidarium.

Activity 7: Construction activity inside room 50 (SU 339, 340, 341, 342)

Evidence of construction work in room 50 connected to the building of the frigidarium consists of two rectangular trenches oriented northwest-southeast (SU 339 and 341). The respective fills (SU 340 and 342) are composed of clayey earth of compact consistency with an admixture of charcoal and small tile fragments.

Activity 8: Raising of the surface level of room 50 (SU 288, 316, 319, 326, 327, 328, 334, 336, 337)

This activity relates to a series of strata that indicate a raising of the entire area occupied by room 50. It is of a piece with what is observed elsewhere in the baths at the same time (activity 21). The rise in quota of the surface level of the room was achieved by spreading a layer of stones (SU 334, 336), which served as a loose foundation and lifted the level of the floor by ca. 50 cm above the previous floor. The analysis of the pottery in these strata indicates a date late in the Flavian period or just afterward, a dating confirmed by the analysis of the stratigraphic sequence.

Activity 9: Construction of the pool (SU 248, 280, 296, 343, 344, 409, 423, 432, 437, 438, 439, 441, 445, 446, 456, 459, 460, 462, 463, 803, 849)

A new room (37) was built behind the western walls of the late Republican atrium. It consisted of a semicircular pool with three niches (a central niche on the axis of the room and two side niches), furnished with a bench (50 cm wide) along the walls. The central niche, 1.80 m wide with three steps at the base, was used for the flow of water carried by a lead fistula (activity 16) that was identified behind the west wall of the pool. The lateral stairs, which also had three steps that were smaller in size, provided a passage from the pool to the frigidarium (fig. 14). The pool abutted the frigidarium on its western wall (MSU 420=802). The elevation of the inner core of the perimeter walls was made of brick, while the exterior face was opus mixtum with regular bands of cubilia alternating with cardellino blocks. On the basis of the existing remains and design, it is possible to hypothesize that this room was covered by a groin vault open toward the east.

Figure 14 Room 37, northern lateral steps.

Activity 10: Construction of the apse of room 51 (SU 214, 215, 216, 281, 299, 300, 301, 302, 308, 312, 313)

In the course of this activity room 51 was built. During the excavation of room 50, however, it was possible to reveal only the northern apsidal wall of room 51. This structure (MSU 214=299, 215=300, 216=301, 302=MSU 10009) consists of horizontal bands of cardellino alternating with brick courses. Room 51 was excavated only superficially, but on the basis of its plan and owing to the presence of a praefurnium located in the eastern wall of the room, it was probably one of the heated rooms in the bath complex. The foundation of the apsidal wall (MSU 308=281) is composed of small blocks of cardellino and is characterized by a change in level from west to east following the natural surface, which in this part of the villa appears to have a slight slope. With the foundation trench of the wall (SU 312), the fill layer (SU 313) contains ceramic material dating to the Flavian period, which permits us to date the construction of the wall and hence the entire room 51 to a period stretching from ca. 80 A.D. to the middle of the second century. Regarding the function of room 50, we may deduce that it was a service area connected to the use of thermal rooms 51 and 37 located to its sides. Access was most probably from the east, although the complete restoration of wall MSU 298, undertaken in modern times, prevents verification of this hypothesis. It also seems barely possible that there was an entrance on the south side; moreover, one cannot completely reject the hypothesis that there was also access available from the terrace above, situated to the west of room 50.

Activity 11: Installation of a fistula inside room 37 (SU 410, 411, 457, 458)

A channel (SU 410) was dug immediately behind the western wall of room 37. It runs in a north-south direction and is the bed for a lead fistula (SU 457). The pipe was almost certainly intended to carry water into the pool of the frigidarium. A preparatory layer (SU 458), related to the housing of the fistula, was laid on the bottom of the ditch. It was composed of mortar, marble chips, and fragments of roof-tiles and bricks.

The poor state of conservation of the fistula, of which only a short section (ca. 30 cm long) remains, does not allow us to understand its relationship with the ridge of the hill immediately behind. It is nevertheless likely that this pipe was connected to the fistula found in room 50 (cf. activity 12; on the hydraulic system in this part of the villa, see De Simone, ).

Activity 12: Placement of a lead fistula in room 50 (SU 232, 240, 243, 317, 318, 331, 333, 335)

In the center of room 50 a cut (SU 232 and 317) was made in order to set a lead pipe (SU 331) running northwest to southeast on a diagonal through the room (fig. 15). The pipe was found covered by a stratum of earth (SU 243 and 318), which also filled in the empty areas of the cut made for the fistula after it had been installed. The water pipe—which probably belongs to the same system of plumbing as the pipe fragment found in the western area of the excavation (SU 240)—was protected by two side walls made of brick and stone (SU 333), found beneath wall MSU 266. To facilitate the passage of the pipe through the wall toward the square room 43, situated to the southwest of room 50, a small square opening was created in wall MSU 302. The pipe was inscribed with the name of the manufacturer, C. Iulius Priscus (see Bruun, ).

Figure 15 Lead pipe 331 in Area 50, from the north-west.

The pottery found in the fill of the channel dates this activity to the second half of the first century A.D. Thus, although the name of C. Iulius Priscus is too common to serve as a dating element, the activity can be dated to the end of the first century A.D. or somewhat later on the basis of the ceramic finds.

Activity 13: Construction of the walls of cardellino in room 50 (SU 245, 246, 266, 307, 314, 315)

The walls MSU 246=245, 266, and 307 were built during this activity. They delimit a room with an irregular shape abutting the southern wall of room 37. The walls were made of small blocks of cardellino disposed in horizontal courses and mortared with earth. At the time of their discovery, the walls were in a very poor state of conservation. The cut (SU 314) made for the foundation of these walls was filled with a layer of soil that did not produce any ceramic material. Nevertheless, the dating of these structures can be hypothesized on the basis of stratigraphic data alone, which appear to indicate a relative chronology in the late Flavian period or afterward; this can be deduced from the close relationship of these walls to the fistula described in the preceding activity. The space delimited by the construction of these walls can be interpreted as a service area connected to the use of the pool (room 37).

Activity 14: Laying of the mosaic and revetment of the walls of the pool of the frigidarium (SU 412, 413, 419, 421, 442, 443, 453, 454, 455, 814, 815, 816, 828, 834, 835)

The whole space used as the pool of the frigidarium was waterproofed by means of a thick layer of cocciopesto (SU 413, 419, 421, 814), which was covered by a marble revetment with slabs of regular dimensions (SU 412, 455, 828). These plaques were attached to the walls by means of “L”-shaped bronze cramps (fig. 16; on the cramps, see Martin, ). The cramps had square housings from 2 to 4 cm on a side and were secured by means of mortar and small marble shims. In the hollow space between the slabs and the wall, a mortar composed of a pozzolana base mixed with brick chips (SU 442, 443) was spread to bind the entire surface of the plaques to the wall. The pavement of the pool was covered with a mosaic of white tesserae (SU 453, 835) set on a very thick layer of sandy mortar mixed with rubble and brick fragments (SU 454, 823).

Figure 16 Detail of the marble revetment in room 37: slab and bronze clip.

Activity 15: Paving the surface of room 50 (SU 261, 274, 276, 321, 322)

After being filled with two preparatory layers of soil (SU 276, 321), the room was completely covered with a pavement of rather second-rate cocciopesto (SU 261, 274, 322), which is not very well preserved. The pottery remains date this pavement to sometime between ca. 80-150 A.D., that is, contemporaneous with the walls described in activities 12 and 13. The find of a coin of Hadrian in the layer SU 261 (on the coin, see Buttrey, ) may serve to narrow the dating parameters, but its context needs further discussion.

Activity 16: Rise in the level of room 50 (SU 211, 229)

The surface level of the area situated in the plateau to the west of room 50 was raised by the dumping of clayey strata mixed with brick and stone fragments. Such characteristics permit us to relate them to the strata described in activity 8, where the chronology has already been discussed.

Activity 17: Pavement levels atop the upper surface of room 50 (SU 218, 221, 230)

Soon after activity 16, a series of strata of very compact soil was laid down in room 50 to serve as floor levels in this area. This intervention, contemporaneous with the remodelling of room 50 (activity 15), also included the installation of a limestone column with a diameter of 34 cm (SU 230), of which a fragment 50 cm long is preserved.

Activity 18: Construction floors in the northern zone of the baths (SU 1219, 1227, 1229, 1230, 1232, 1239, 1242)

In the northern zone of the bath complex, behind the large square space of the late Republican period, a series of strata of various compositions and matrices was laid in order to raise the surface level. The intervention, contemporaneous with the activities of levelling identified in the other rooms of the bath complex (activities 5, 7, 8, 16) is datable, on the basis of pottery finds, to a period after the last quarter of the first century A.D. Noteworthy is the presence within SU 1229, 1230, 1239, and 1242 of numerous marble chips and mosaic tesserae, some of which were not completely worked. It is likely that this material results from the working in situ of the numerous marble slabs that decorated the walls and, perhaps, some of the pavements of the thermal rooms.

In addition to this material, the excavation brought to light some fragments of terracotta architectonic plaques that probably belonged to the first decorative phase of the villa (see Strazzulla, ).

Activity 19: Construction of a channel in the central zone of Area 35 (SU 1236, 1237, 1238, 1250)

A channel for water drainage, oriented east-west and connected to room 34, was built directly atop the strata of soil laid down to level the surface of Area 35 (activity 18; fig. 17). The side walls of the drainage channel, built against earth, were made of tufa cubilia, probably reused from an earlier structure, and of roughly worked blocks of small and medium size. The bottom of the channel, sloping downward toward the east, consisted of roof-tiles laid horizontally and close together. The use of this channel can be related to the lead pipes found inside rooms 37 and 50. The drainage channel was probably connected at its western end to a fistula that was part of the general hydraulic system of the villa (cf. activities 11 and 12).

Figure 17 Channel for the drainage of water in Area 35, from the east.

Activity 20: Construction of the perimeter wall of rooms 32 and 34 (SU 1226, 1243, 1246, 1248, 1249)

In order to construct rooms 32, 33, and 34, a wall running north-south along the eastern side of Area 35 was built. The foundation was made of irregularly shaped stones and pebbles of small and medium size, not worked, which were arranged in fairly regular courses and bound with a sandy, rough mortar. The elevation of the structure, preserved in part only in the southeast corner, was built in opus reticulatum with cubilia of tufa and limestone. Only the relative chronology of the structure can be determined; it is later than the walls of the square room, which gives us a terminus post quem.

Activity 21: Construction of a colonnade inside Area 35 (SU 1233, 1235, 1244, 1245, 1251, 1252)

Seven columns were erected at regular intervals in the central space of Area 35. These were aligned north-south and perhaps formed part of a porticus outside the bath complex. The colonnade, which connected the residential part of the villa with the baths, was probably delimited by a perimeter wall situated in the part of the western hill still not excavated, while the entrance was perhaps situated on the eastern side. The columns were made of bricks that were cut to have the shape of a quarter circle (see below, activity 28).

In contrast to the piers belonging to the atrium and impluvium (Period I, activity 2), these were made entirely of bricks of regular dimensions, bonded by a friable mortar, rather similar to the mortar used in the construction of room 37.

In the bath complex in this period are attested the first activities of destruction and spoliation or, at any rate, some significant transformations in use. The mosaic floor of the square room was partly dismantled and replaced by a simple floor of beaten earth. Inside room 50 several strata of debris accumulated, partly resulting from the collapse of the perimeter walls (activities 25 and 26). An analogous situation is encountered within the pool (room 37), where strata consisting mainly of building materials were found near the walls. At the same time, in the central zone of room 40, a trench tomb was dug (figs. 18-19). It was oriented northwest-southeast and had a pitched covering of roof-tiles (a cappuccina). On the basis of C14 dating of the bones, this burial can be dated to the fourth century A.D. In the same period, but perhaps somewhat later than the previously described activities, the plan of this sector began to be partially modified through the building of some walls in the large square room. These walls defined a new space within the old frigidarium (activity 23).

The northern zone of the baths was affected by a demolition activity, which is attested by a series of rubble strata originating from the destruction and/or spoliation of the porticus (activity 28). To this activity can also be added the excavation of a large trench, oriented east-west, which was filled up with building material (activity 27).

Activity 22: Spoliation of the wall structures and pavement of the frigidarium (SU 622, 820, 861)

The partial abandonment of this part of the villa is stratigraphically attested by the destruction and cutting back, practically to the foundations, of some wall structures. The northern end of the western perimeter wall of the square room was completely destroyed (SU 820), probably to recover building material. A similar situation was found inside the pool, where the walls were partially demolished and where all the marbles, the pavement, and also the statues decorating it (if there were any) were completely removed. To be added to these actions of stripping is the systematic removal of the mosaic of the square room, which for this reason is preserved only in small areas along the southern wall of the space.

Unfortunately, there are no dating elements that can provide an absolute chronology for these activities, but it is quite plausible that they occurred shortly before the excavation of a trench tomb identified in the center of room 40 and assignable to the fourth century A.D. (see below, activity 24).

Activity 23: Construction of a series of walls inside room 40 (SU 634, 642, 643, 647, 648)

Three walls were built inside this room. Their poor state of preservation does not permit us to understand their function nor to define the spaces that they delimited. The foundations of these structures, which have heavily cut into the earlier pavement levels, were made of medium-size stones bonded with clay and sand.

Activity 24: Tomb a cappuccina in room 40 (SU 635, 636, 649, 650)

Near the northwest corner of room 40 a rectangular ditch (SU 635) with a northwest-southeast orientation was dug; its purpose was to house a burial a cappuccina. This presents the characteristic pitched cover made of roof-tiles with cover-tiles arranged above the upturned rims of the roof-tiles (SU 649). The top of the cover consists of imbrices. Two roof-tiles set on their sides and abutting the sloping tiles were arranged along the short sides of the burial (on the east and west); this arrangement constituted the closure of the short sides of the tomb. On the western side, a reused marble slab (25 cm x 55 cm) was placed in contact with the roof-tile of closure and in contact with the cranium of the skeleton. Inside the trench the inhumed corpse (SU 650) was deposited, orientated east-west, with the cranium toward the east. The bones of the upper part of the skeleton, down to the pelvis, were in complete disarray, owing to the collapse of the tomb cover (SU 649). In contrast, the lower half (legs and feet) was better preserved, especially the feet, thanks to the resistence of the tomb cover at this point. The tomb itself was in good condition, especially on the west, but the eastern section had collapsed inward.

The radiocarbon dating of the bones of the skeleton (performed by the NSF-Arizona AMS Laboratory) permits a dating of this burial to 1,682 years BP (i.e., ca. 318 A.D.) with a standard deviation of +/- 58 years.

Activity 25: Strata of rubble inside room 50 (SU 255, 262, 271, 273, 275, 283, 285)

The numerous strata identified inside room 50 relate to a single intervention. As already noted, this event was the dumping of soils, pottery, and building and organic materials in an area long since abandoned. The distinct contexts noted during excavation did not correspond to any chronological sequence in terms of the different deposits, which produced pottery that was largely residual (mostly dating to the mid-imperial period). There was a small amount of pottery found that dated to the fifth century. This provides a terminus post quem for dating this activity.

Activity 26: Abandonment of the upper surface of Area 50 (SU 209, 229)

The entire upper surface of Area 50, occupied in the imperial period by a porticus that faced the baths, was raised in level by the accumulation of several strata rich in building materials, in large part originating from the collapse of adjacent structures. The deposit of these strata signified the definitive obliteration of the porticus and of the floor belonging to it at a time subsequent to its destruction (see below, activity 29).

Activity 27: Excavation and fill of a ditch in the central zone of Area 35 (SU 1228, 1231)

The various dumps of soil presumably laid during the course of the first half of the second century (Period II) to raise the surface level of the northern colonnade of the baths were cut by a ditch (SU 1228), roughly rectangular in shape (1.70 m x 0.70 m), oriented east-west.

The fill (SU 1231) consists primarily of soil mixed with fragments of marble and pottery, as well as building materials. Nothing in the fill allows us to date the intervention with precision or to shed light on the purpose of the cut.

Activity 28: Abandonment of Area 35 (SU 1222)

The center part of the southern zone of Area 35 was partially occupied by a stratum rich in building materials that mainly originated from the collapse of adjacent structures. In particular, conspicuous among the elements making up the stratum were regularly shaped bricks still adhering to the core of the wall, as well as bricks in the shape of a quarter circle. The latter were elements of columns and can be assigned with certainty to the demolished colonnade.

Activity 29: Construction of late-antique walls in room 50 (SU 237, 250, 251, 289, 290, 303, 305)

In a phase shortly after the first abandonment of the bath complex and after the formation of rubble strata in all its rooms, several small walls with uncertain function were built within room 50. Wall 289 was built in the central part of the space, completely closing the room on its west side. This wall, whose construction necessitated the partial destruction of the northern wall of room 51 (MSU 248) was crudely constructed of small and medium sized stones arranged in an irregular way. The foundation of the wall (SU 237, 290, 303) was also made of irregularly placed rough-hewn stones. At the point at which the wall abuts MSU 246-266, a cornice fragment (SU 250) was added, reused as a buttress to support the corner of the wall. It was not possible to establish the construction date of the wall with any precision, but on the basis of its building technique and an analysis of the stratigraphic sequence, we can hypothesize a late-antique date.

In the northern zone of the baths, a new flurry of building activity can be detected in the early medieval period. The resumed activity was certainly due to the visible presence of a considerable portion of the ancient structures. The buildings must have been preserved to some height, at least in the western and northern zones, and it was indeed precisely in this area that three new rooms were constructed. To the south, a large room with a rectangular shape (room 40) was created when a wall was built in an east-west direction. The pre-existing opus incertum walls of the frigidarium to the south, east, and west were reused in order to form the other three sides of the room. In the northwest part, a new space was made (room 38) whose west wall was built on the pavement of the pool of the frigidarium; this necessitated the partial destruction of the earlier imperial structure. To the northeast, another room was built, smaller than the others, which reused the Republican walls still in existence to the east and north (room 39).

The southwestern part was not affected by substantial modifications, if one excepts the reuse of a lead pipe, which brought water into the new room 40 (activity 34).

Activity 30: Excavation and fill of a ditch in room 39 (SU 1004, 1005)

A trench oriented north-south was dug just behind wall SU 843. The fill of the ditch, added soon after, consisted of earth with a clayey matrix, blackish in color and including building material relating to imperial structures. It also contained numerous pottery fragments that permitted a dating of the activity to a time later than the seventh to eighth century A.D.

Activity 31: Spoliation ditch in the center of Area 35 (SU 1220, 1221)

Another deep cut, oriented north-south, was made in the central zone of Area 35. The pit was located behind the second column and the fill consists almost exclusively of building material (stones, bricks, and blocks of cardellino) that probably derived from the imperial colonnade. Noteworthy among the material found are bricks cut in a quarter circle; these are associated with the demolished colonnade (cf. activity 28). Similar bricks were found in the garden (see Gleason, , activity 11).

The function of the pit could not be determined, although it is probably related to the stripping of the bath complex. Thanks to the presence of numerous pottery fragments found in the fill, the date can clearly be assigned to the eighth and ninth centuries A.D.

Activity 32: Circular cut in Area 35 (SU 1240, 1241)

Beside the perimeter wall of rooms 32, 33, and 34 (activity 20) a small circular pit was dug, the walls of which were covered with stones and brick fragments. The interpretation of this intervention is uncertain; it may be connected to the construction activities of this period.

Activity 33: Construction of rooms 38, 39, and 40 (SU 422, 435, 461, 466, 467, 633, 804, 807, 829, 831, 832, 843, 846, 847, 850, 864, 1003, 1016)

The remodeling of this zone in the early medieval period resulted in the laying out of three new rooms inside the old frigidarium of the imperial age. The walls of late date that had invaded the ancient area (activity 29) and the wall in opus incertum 420=802 were razed to their foundations; three new walls were built in their place. Stones, brick fragments and earth composed the inner core. The facings were made of rough-hewn stones of various sizes, which were laid in horizontal courses with staggered joints. In the elevation of the walls, limestone, blocks of cardellino, and brick fragments were used indiscriminately. One wall (SU 435-846-847-850), which runs in an east-west direction, is partly seated on one of the previous structures (MSU 634) and partly on the preparatory layer (statumen) of the mosaic, destroying part of the wall in opus incertum that delimited the earlier room on its west side. A second wall (461), oriented north-south and bonded to the preceding wall, was built inside the pool of the frigidarium. Its deep foundation partially destroyed the mosaic and the northern niche, implying thereby that this room was no longer in use as part of a bath complex. A third wall (843), of small dimensions, was erected approximately in the middle of the square room in order to divide it into two separate rooms (38 and 39).

Activity 34: Reuse and modification of the ancient drainage system (SU 293, 294, 295, 297)

The creation of three new rooms was accompanied by equally significant changes to the hydraulic system of the area, which underwent radical modification. A cut (SU 293) was made under the central area of the opus incertum wall 291 in order to allow the passage of a waterpipe (SU 241; the waterpipe had been uncovered during earlier excavations, see activity 48) situated along the northern wall of room 50. A stratum of mortar (SU 294) was laid down inside the opening of the passageway for securing the pipe, while the embankment that supported the channel was created by reusing a fragment of pavement in cocciopesto (SU 295). The recent archaeological interventions and restorations by the Superintendency made it impossible to completely understand the development and course of the new channel. Nevertheless, it is worth noting that the ancient plumbing system, with repairs and modifications, was still operational in the early Middle Ages.

Activity 35: Collapse of the walls of the pool of the frigidarium (SU 403, 433, 444)

During the early medieval period several perimeter walls of the frigidarium collapsed, causing the accumulation of a substantial quantity of rubble (bricks, mortar, and marble chips) near the western wall of the basin (SU 403), on the central step (SU 444), and above the southern step (SU 433).

Activity 36: Abandonment layer in Area 35 (SU 1213)

While the central part of the bath complex was reused as a habitation, the entire surface of room 35 was covered with a stratum of uniform soil that definitively obliterated the remains of the imperial colonnade. The layer, composed of a very compact soil mixed with fragments of building material, has yielded a notable quantity of residual material from the early and middle imperial ages, while later periods are barely represented. Noteworthy, among the latter, is a fragment of white majolica datable to the seventeenth or eighteenth century, probably the result of agricultural activity in modern times.

Activity 37: Interventions of spoliation and destruction (SU 258, 259, 265, 268, 277, 278, 279, 286, 287, 309, 414, 434, 447, 448, 449, 450, 451, 452, 821, 822)

Toward the end of the early medieval period the entire northern area of the imperial baths was affected by a series of interventions of destruction and spoliation. In room 50, a series of irregularly shaped holes were dug (SU 258, 259, 265, 268, 277, 278, 279, 286, 287, 309), which were later filled with rubble derived from the demolition of the walls and mixed with clayey soil. The material contained in the trenches was of modest quantity and mainly represents residual pottery fragments of the classical and late antique periods.

Inside the pool of the frigidarium a large circular pit (452) was dug. The excavation work probably lasted for a relatively long period of time, as is attested by the superimposition of cuts, perhaps due to later adjustments and levelings. The pit was most likely created to search for reusable building material. During this activity, part of the mosaic on the floor of the pool and all the marble slabs from the walls of the pool were stripped away. The recent excavation revealed that the filling of the ditch occurred in distinct phases, each linkable to a specific stratum (SU 414, 434, 447, 448, 450, 451, 822) but as part of a single general dumping activity. The fill contained a great deal of residual material, mainly of imperial date. Noteworthy in this fill were three fragmentary marble statuettes (see Lattimore, ). Also found was material contemporaneous with this activity, including a great deal of unpainted (achromatic) pottery with a wave decoration, pottery with heavy glaze, and African red slip “D” ware (see Angelelli, ). Analysis of this ceramic material suggests a date of ca. 700-800 A.D., or perhaps a little later, for this activity.

Activity 38: Burial west of room 50 (SU 223, 224, 225, 226, 227, 228)

A rectangular ditch was dug on the plateau west of room 50, ca. 60 cm deep and oriented east-west. Its purpose was to receive an inhumation burial (fig. 20). The ditch (SU 223) was reinforced on its sides by two banked walls (226, 227) made of limestone and brick fragments bonded with soil. Because the burial was discovered during a cleaning operation on a steep slope just below the edge of the plateau, only the eastern end of the tomb was explored. The bones of the legs (femurs, tibias, and fibulas, SU 225) were found in their original position. On the basis of stratigraphic data, this burial can be related to the two burials found in Area 35, which were given radiocarbon dates of ca. 1,194 and 1,118 years BP (see below, activities 39 and 40).

Figure 20 Tomb on plateau west of room 5.

Activity 39: Burial in the western zone of Area 35 (SU 1214, 1215, 1216)

To the east of the pier SU 1207, a ditch (1214), more or less rectangular and oriented east-west, was dug for an inhumation burial. Inside the ditch, the skeleton was buried with the cranium turned to the west and resting on a sort of raised earthen level. This factor determined the position of the shoulders as well as the head, which was found angled forward and resting on the right shoulder. The arms were arranged with the elbows resting on the sides of the ditch, while the hands were placed over the pelvis. The spinal column was perfectly straight, and the legs were extended. The right foot abutted a stone placed at the bottom of the ditch. The burial had no grave gift, so an absolute date was sought by means of C14 analysis of a bone sample. The results indicated a date of 1,118 BP (i.e., ca. 882 A.D.) with a standard deviation of +/- 40 years.

Activity 40: Tomb in the western zone of Area 35 (SU 1201, 1202, 1203)

In the central western zone of Area 35, near the burial just described, a second ditch (1201) was dug for an inhumation burial (fig. 21). It, too, had an east-west orientation. All of the upper part of the ditch had been destroyed by modern interventions in the area; only the lower part remained. The skeleton was buried with its cranium turned toward the west. Unfortunately, the bones were found in a poor state of preservation. The cranium was caved in, the right arm was folded above the thoracic cavity, and the wrist was placed near the right scapula. No grave goods were found, so once again an absolute date was sought through radiocarbon dating. The results indicated a date that is 1,194 BP (i.e., ca. 806 A.D.) with a standard deviation of +/- 40 years.

Figure 21 Tomb in the western zone of Area 35.

In this period, the structures of the bath complex of “Horace’s Villa” were demolished and definitively despoiled. In the various areas investigated we found that strata of earth had accumulated, which obliterated whatever still remained of the imperial and early medieval structures. Only in the southern zone of the upper plateau of the villa was a new structure erected. It was composed of a large wall running north-south, plastered along its western face. The purpose of this structure remains uncertain.

In the southern part of room 40 a small circular ditch was dug (activity 41) for the burial of two skeletons found in a secondary deposit and perhaps to be connected to the two burials of the preceding period (see activities 39 and 40).

Activity 41: Pit for the deposit of two skeletons in room 40 (SU 611, 639=610, 640, 641)

In the southern part of room 40, beside the wall in opus incertum (SU 620) a pit (639=610) was dug, roughly circular in shape, to a depth of ca. 30 cm. Into this pit numerous fragments of human bones belonging to two individuals were placed; this was clearly a secondary deposit (fig. 22). The burials are probably to be linked to the small late-antique cemetery attested by the tomb a cappuccina identified in room 40. It is probable that the skeletons were removed from their original place of burial as a result of later activities of building or spoliation.

Figure 22 Room 40: circular pit (SU 639=610) with human bones belonging to two individuals.

Activity 42: Strata of the rubble of Roman and early medieval structures (SU 428, 440, 818)

The abandonment of the early medieval rooms as well as those dating from the Roman period resulted in the creation of a series of small strata consisting of stones, bricks, and mortar, located within the pool of the frigidarium and in room 38. The presence in these strata of marble fragments—including some of large size—leads us to conclude that part of the rubble relates to Roman imperial structures reused down to the abandonment of the site.

Activity 43: Medieval wall in room 50 (SU 212, 217, 238)

In the southwest zone of room 50 a large wall (SU 212) was built, which encroached on part of room 51. The wall was only partially investigated and only its western face was exposed. It was built of irregularly shaped stones and reused building materials (including various cubilia) and was covered with a layer of plaster (SU 238). The structure SU 217 abutted the wall. It was made with the same building technique and served as a buttress of the wall.

Activity 44: Strata from the collapse of the medieval structure in room 50 (SU 244, 247, 269)

At an undefined time, the medieval structure just described (activity 43) collapsed. This was probably connected to the abandonment of the entire thermal zone. The strata formed by this activity were entirely composed of stones and fragments of brick and mortar. There was a complete absence of post-classical ceramic material.

Activity 45: Ditch on the upper plateau of Area 50 (SU 235, 236)

A deep ditch (SU 235) was dug beside the medieval wall built along the southern limit of Area 50. Oriented east-west, it was about 1 meter deep. The fill in the trench yielded a small amount of finds, all of which relate to the classical period. Thus it is impossible to establish an absolute chronology for this intervention, which clearly must have happened after the definitive collapse of the large medieval wall.

Activity 46: Strata of abandonment and collapse on the plateau of Area 50 (SU 202, 206, 213, 219)

On the plateau west of Area 50, the complete abandonment of the site led to the accumulation of a series of strata attributable in part to the collapse of the medieval structure (activity 43) and in part to the natural erosion of the hillside above. The materials found indicate a chronological time frame extending from the medieval to the modern period.

After the abandonment of the buildings, most of the remains were completely covered by earth; only bits of them were still partially preserved above ground. The modern interventions to the site made from 1911 to 1914 by Angelo Pasqui and later in the twentieth century by the Archaeological Superintendency for Lazio represent the last activities that we identified. Our excavation allowed us to bring to light numerous excavation trenches, particularly along the walls of the bath complex, the purpose of which was to permit the earlier excavators to reconstruct the plan of the complex in order to understand the function of its various parts. In addition to these activities, there were numerous interventions of conservation and restoration (in whole or in part) of the walls and pavements beginning in the time of Pasqui and extending throughout the twentieth century.

Activity 47: Modern restorations (SU 203, 204, 401, 402, 404, 405, 406, 601, 605, 623, 625, 627, 811, 812)

This activity relates to a series of restorations that were begun on the site with Pasqui’s excavations in 1911-1914. This activity affected the tops of the visible walls, which were protected by a layer of cement. The restorations also affected the external faces of the walls, sometimes making it difficult to distinguish the ancient from the modern (see De Simone, ). Other conservation interventions concerned one of the two mosaic fragments located in room 40, which appears to have been removed, restored, and reinstalled in its original position.

Activity 48: Modern archaeological interventions (SU 205, 207, 208, 210, 231, 233, 234, 241, 249, 252, 254, 256, 257, 263, 264, 267, 270, 282, 320, 323, 324, 415, 416, 417, 418, 424, 425, 426, 427, 429, 430, 431, 436, 605, 606, 607, 608, 609, 616, 617, 618, 619, 638, 805, 806, 836, 841, 842, 848, 852, 853, 855, 856, 862, 863, 866, 867, 1007, 1008, 1009, 1010, 1011, 1012, 1218)

The excavation interventions executed during the last century affected a great part of the bath complex, causing serious damage to the ancient stratigraphy and significantly compromising our ability to understand the relationships between the various structures and the strata around them.

As can be reconstructed from the course of the trenches dug prior to our campaigns of 1997-1999, the principal objective of the earlier excavations was to identify the plan of the site, including the course of the walls and the location of related floors. The principal method used was wall chasing. Particularly noteworthy are the deep ditch (SU 256), which was dug along the eastern wall of room 40 in order to restore the wall and the pavement of room 42, and the deep trench (SU 415 and 417) that follows exactly the apsidal structure of room 35, onto the bottom of which was set a crude channel composed of bricks and used to carry the runoff of water from the hillside toward room 34.

Activity 49: Formation of strata of humus (SU 200, 254, 400, 600, 800, 1000, 1200)

A rather thick layer of humus with dense vegetation formed atop the structures identified as a result of the extensive excavations and restorations undertaken in the twentieth century.

The first step was to distinguish the restorations from the original structures, through non-destructive direct observation and by comparison with archival photographs illustrating the various phases of excavation and other interventions (cleaning, restoration, etc.), beginning with those of 1911-1914 and including those in 1930-1931 and in the 1970s.Besides Lugli’s fundamental publication, we have also had at our disposal some of Pasqui’s documents, which mainly consist of an inventory of finds, as well as the excavation diaries written by De Rossi, the site foreman. In addition, Bernard Frischer also found the correspondence between De Rossi and Pasqui, along with some letters from the restorer, Verduchi, to Pasqui, which were intended to keep Pasqui apprised of work at the villa. These letters have also been taken into account in what follows, and have furnished a great deal of surprising information.

In general, the letters are administrative in nature (numbers of workers in attendance, weather, storage arrangements for the finds, etc.), but information pertinent to our purposes can be found regarding management of the work on the site, purchases made and procedures followed (fig. 1). The passages below have been selected because they refer to the masonry structures. They document the phases of excavation, including some structures that are no longer visible, and the phases of restoration. These passages, which have been translated into English (for the original Italian version see Frischer, ), are quoted in extenso, except in those places where italics indicate that a summary is given. In several instances, the telegraphic style has been supplemented to facilitate the reader’s understanding, with the additions put inside square brackets ([ ]). A brief commentary follows in italics, explaining the importance of a passage, or group of related passages, for the present investigation.

1. Nicola De Rossi to Pasqui, 28 March 1912

In the cryptoporticus, work on the right-hand side has brought to light about two meters of plaster to a height of about 25 cm, red in color like that found before.

2. Nicola De Rossi to Pasqui, 30 March 1912

The excavation proceeds well. The cryptoporticus will be completed by Monday evening up to the boundary with Rocco Foschi[’s property = parcel 1215; see Frischer and cadastral map, fig. 17]. The reticulate walls follow on both sides, where plaster painted red was found to a height of about 25 cm.

It seems correct to interpret the right-hand side as the western wing. The plaster must necessarily have covered a still intact wall surface. Unfortunately, no unequivocal indication is given about the construction technique of the wall in question (opus reticulatum or opus incertum?). A section of red plaster is still preserved, a facing on the wall (in opus incertum) next to the stairs that lead to the residential area (see De Simone et al., ). The boundary with Rocco Foschi’s property, in fact, ran through the flight of stairs (see Frischer, , fig. 4, where the modern plan is superimposed over the property lines, and , table 2). 3. Nicola De Rossi to Pasqui, 27 April 1912

The excavation on Foschi’s land is almost entirely completed; only foundations were found.

The area in question is located in the southern part of the residential area. The observation that only foundations had been discovered is interesting, since the reconstructed walls were apparently built on these. 4. Giuseppe Verduchi to Pasqui, 29 May 1912

The reticulate wall adjacent to the stairs has been completed for a length of about 7 m. The work continues well, both the excavation work and the wall building.

From this passage it is not clear if the completion refers to excavation or restoration work; since the person in charge of the restoration/reconstruction of the masonry (“lavoro di muratura”) was in fact Signor Verduchi, it is likely that he means the latter. 5. Nicola De Rossi to Pasqui, 1 June 1912

Transport of bricks and 1860 kg of lime from Roccagiovine to Licenza, because the limekiln owner has no more of lime to sell. On Monday we will get another shipment of lime.

6. Nicola De Rossi to Pasqui, 4 June 1912

Expense note for lime: 20 quintals. The masonry work continues as planned since Signor Verduchi is not worried by the sun. The investigation ordered by Your Most Illustrious Lordship in the center of the pool has been carried out by Signor Verduchi, but without any sign of a source being found, only a lining of stones and pozzolana. […] [The small room] shows the end [top?] of the wall facing, because the wall is terminated and is made of reticulate with edges of roof-tiles. […] transport of gesso [plaster?].

The masonry activity, as is evident from the expense note, required a constant and abundant supply of building materials, which were immediately put to use by the industrious Verduchi. The “pool” he is referring to is, in reality, room 33 (interpreted, evidently together with room 32, first by Pasqui and then by Lugli, as a swimming pool). However, it is impossible to identify the precise location of the test-pit. For the south wall of room 33, the northern face of which is of terracotta opus reticulatum , see the section on construction typologies (appendix I, , type 5.3.3). 7. Nicola De Rossi to Pasqui, 8 June 1912

Expense note for sand and 1000 bricks. The masonry work continues on the reticulate wall adjacent to the stairs; more than half is now done. In the digging, new walls are being continuously discovered. Just today a reticulate wall was found with fine red-colored cocciopesto plaster, 3.2 m in length, with two stairs at the top and two at the foot (of guide also covered by plaster), [which have] a length of 1.00 m and a height of 30 cm.

In this case the (reconstructive) restoration can be identified with certainty, even if the precise location of the work remains to be determined. The new area mentioned is the northern part of room 34. Two points are of interest here: the indications of the waterproof coating of cocciopesto and the note about the stairs “of guide also covered by plaster.” I will return to the meaning of the term “guide” below (cf. the letter from Nicola De Rossi to Pasqui, 27 December 1912 = no. 20). The reference to the “two stairs at the top and two at the foot” actually indicates the two steps discovered at that time for each of the two staircases; consequently, the height must refer to a single step. 8. Nicola De Rossi to Pasqui, 15 June 1912

Purchase of pozzolana and sand. The basin is completely cleaned and glued, even on the bottom. It has a drainage hole that must lead into the main sewer.

The room in question is number 34. By “glued” (Italian: incollata) De Rossi doubtless means “coated throughout with cocciopesto.” 9. Nicola De Rossi to Pasqui, 21 June 1912

Investigating the perimeter wall on the Onorati’s land, it was found to be truncated by a lime pit.

10. Nicola De Rossi to Pasqui, 29 June 1912

On the land of the Onorati heirs, work proceeds with some laborers to expose the perimeter wall. The wall is cut off by a lime pit, but it comes up close to a large foundation that continues toward the waterfall.

The area in question is north of the gravel road that crosses the remains of the villa (see Frischer, , fig. 17), i.e. the area north and northeast of the excavated residential section. The perimeter wall is the northern wall, interpreted then as the limit of the villa. The foundation structure noted as going toward the waterfall (i.e. toward the west) is probably outside the rectangle of the residence block visible today. This hypothesis is supported by evidence gathered from the testimony of various custodians at the site. The lime pit indirectly attests to a later spoliation and reuse of building materials. Bernard Frischer, in a personal communication, notes that such activity is definitely known in connection with the construction of the local church in the 1840s and 1850s (see Frischer ). 11. Giuseppe Verduchi to Pasqui, undated

The masonry work in the cryptoporticus is almost finished. On Monday I will begin work on the masonry of the stairs in the garden.

12. Nicola De Rossi to Pasqui, 10 August 1912 Expense note for the stonemason’s working days (De Rossi says “for eight linear meters”) and for the transport of sand. 13. Nicola De Rossi to Pasqui, 24 August 1912 Expense note for pozzolana; the masonry work continues. 14. Nicola De Rossi to Pasqui, 28 August 1912

The stones that were in the vicinity of the cryptoporticus have been transported away from the excavation. The others serve for the construction of the restored walls. A good part of the walls has been aligned; if he wishes, Signor Gatti may come.

The substantial masonry reconstructions are clearly attested by this letter. It is obvious that these are not only repairs, but also actual constructions. The draftsman, Signor Gatti, was invited to carry out the survey for the plan after the walls had been “aligned.” 15. Nicola De Rossi to Pasqui, 31 August 1912

New order for sand. A [mosaic] 2 m x 90 [cm], nearby, fragments of another one without any surrounding wall [have been found]. Suspension of excavation and De Rossi asks if the masonry work will continue.

The restoration work continued non-stop. Sometimes parts of mosaics were found in situ, but the walls of the rooms in which they were located had been reduced to nothing. The note here concerns room 11, as may be deduced from the schematic drawing of the mosaic that De Rossi attached to the letter. It is not possible to identify the second room. 16. Giuseppe Verduchi to Pasqui, 16 September 1912

The masonry work continues very well; about the work on the basin of the bath, the arch and the wall facing, what was left from last year was finished. The cryptoporticus was finished equally on both sides, and now I am working on the heating system on Ricciotti’s land, to restore the small pillars […]; small repairs with cement around the mosaics on the land of the Onorati.

17. Nicola De Rossi to Pasqui, 28 September 1912

There are 12 quintals of lime still to slake […]. Various materials are still to be removed from the cemetery corridor of the bath basin, precisely where the masonry has been done and the hole has been closed, which leads to the sewer. […] The old stock of lime is quite diminished; it would be useful if Your Most Illustrious Lordship could make a larger acquisition of that, because it is lime of good quality and well baked, and not too far to transport, because there are still many walls to restore and they take a lot of material, especially the wall of the cryptoporticus near the gate, which must be urgently done. The mosaics have had all the broken parts repaired (with cement) by Signor Verduchi. The masonry work has proceeded in good order, so that when you arrive on site you will be satisfied.

It is impossible to identify with certainty the room that is referred to as the “bath.” It could refer either to the large apsidal hall (room 33) or to the so-called vivarium (room 53). The fact that De Rossi, in the letter of 17 September 1912, mentions a single arch and refers to the remainder of work from the year before points to the latter hypothesis. The excavation seems to have started from the building that had the most substantial remains left above ground. The reference to a “cemetery corridor of the bath” could create a misunderstanding. We know with certainty that the so-called vivarium was used at some point as a burial place, but one cannot exclude an analogous use for the apsidal hall 33, especially since the 1997-2001 excavations have found burials in nearby Areas 40 and 35. The reference to “the hole which leads to the sewer” is the decisive point. Lugli records at least two holes (one for a water jet and one for an outflow) in the floor (in reality, the sub-floor) of the so-called vivarium. These holes are no longer visible.

The use of cement, in place of the usual mortar, is also remarkable in a period in which there was still a cautious attitude toward the use of this material. This detail, however, testifies to a differentiation in the use of the various materials: cement only for the mosaics, mortar and traditional inert materials, necessary in large amounts, for the walls.

18. Giuseppe Verduchi to Pasqui, 10 November 1912

Regarding the work that I oversee, I have taken the greatest care to find the first course of the doors [“spiccato,” i.e., the level from which the wall starts atop the foundations] and, indeed, in the rooms of the mosaics, excavating below the level of the mosaics, I have found the spiccati for three doors, one in the direction of the stairs of the cryptoporticus, and two passageways, one from the first room and another from the second room (that is, the [room with the] mosaic that has always been known). […] Signor Gatti can come at his convenience, for we will attempt, as far as possible, to have it all organized for him.

The rooms discussed seem to be 1, 2 and 4. It is interesting to note that for the first time, after almost two years of excavations, there is sudden talk about finding doorways; no mention of them had been made in the previous series of letters. This is probably in response to an explicit request by Pasqui, who, while not present during most of the excavations, had perhaps just visited the villa and noted this illogical absence from the record (cf. the next letter). Of course, it is hard to find a doorway when the wall is not standing at all. At best, De Rossi could be on the lookout for an area in which a doorway-sized opening of the bare foundation was flanked by at least one course of elevation of the wall (i.e., spiccato [singular], spiccati [plural]), which could be interpreted as the remains of the doorjambs delimiting a threshold. 19. Giuseppe Verduchi to Pasqui, 23 November 1912

The masonry work continues to proceed well with further spiccati for doorways. While working last Monday, De Rossi found spiccati of two other doorways near the mosaic on the land of the Onorati heirs. He is now carrying out the work on the aforementioned spiccati and on the extension of the reticulate walls. It is an important job. At the moment there seem to be six doors. I will do my utmost to see to it that when Your Most Illustrious Lordship visits the excavations, you will be pleased.

See the comment on the preceding letter (also for the meaning of spiccati). 20. Nicola De Rossi to Pasqui, 27 December 1912

I will do as Your Most Illustrious Lordship has ordered, that is, to put the chiodi and guide around the Ricciotti mosaic. Or rather, I propose that since there are not any chiodi and guide in the area of the villa, if Signor Verduchi is late, I can choose from the spoil heap near the bath basin; those [chiodi and guide] are mixed, good and bad, and stones. Then I can move them to the area of the villa and keep them ready for when the restorations are being done. The other extremely important work is that of covering both the restored walls and those still to be restored with earth, especially in that part of the cryptoporticus near the gate that was cleaned for restoration in September, but which was not done. The same should be done for the most recently restored walls, which are still covered with planks, since every clear night brings a lighter or heavier frost, and the walls are being damaged.

The Ricciotti mosaic is probably to be identified with the floor in room 16. Most interesting here is the meaning that can be attributed to the words guide (cf. the letter from Nicola De Rossi to Pasqui, 8 June 1912 = no. 7) and chiodi (cf. the letter from Giuseppe Verduchi to Pasqui, 17 November 1913 = no. 24). We may deduce from these passages that the discussion concerns construction materials. Chiodi, in my opinion, are the opus reticulatum building blocks, which may be informally defined that way because of their long and narrow form, reminiscent of a nail (even if the shape more closely resembles a truncated pyramid). Guide, on the other hand, may be the little rectangular parallelepiped blocks that are used for the edging of walls in opus reticulatum. The stairs of room 34 were, in fact, made of guide and chiodi, and the guide to be reused must have been chosen from the heap of stones near the “bath basin.” As we have seen, this must be identified with the so-called vivarium (room 53; on the characteristics and actual function of this building see ). Building materials must have been present in great abundance, although we do not know the percentage of chiodi with respect to “stones that were not good” (evidently the stones might well have come from opus incertum structures). This confirms once again that falsifications could easily have occurred when walls were restored in opus reticulatum when only foundations were preserved. 21. Giuseppe Verduchi to Pasqui, 12 December 1912

The masonry work continues in the vicinity of the mosaics.

22. Giuseppe Verduchi to Pasqui, 27 August 1913

The two olive trees, one on the Onorati land and the other on the Foschi land, have been felled. The latter was the tree that is close to the stairs that are to be restored. The sand is being extracted and tomorrow the transportation of it will begin. […] The stonemason has started the arrangement of the corners that will be used for the small piers of the stairs. […] The masonry work continues very well, mainly concentrating on the doors and the perimeter wall.

23. Giuseppe Verduchi to Pasqui, 25 September 1913

As soon as I arrived at Licenza I went to the excavations at the Villa of Horace, where I found the masonry work proceeding regularly […]. Meanwhile, in my absence, I will leave work for the coming week; this is work on the foundations that can be done easily. In addition there is the work of shaping the bricks.

24. Nicola De Rossi to Pasqui, 17 November 1913

The shaping of the small bricks, 500 in number, was finished by the tenth of this month. Also, a stretch of reticulate wall has been completed that runs from the room of the rough mosaic and joins the boundary wall of the villa [...]. This section being finished, work has started on another small stretch of the perimeter wall that joins the remaining corner that is still to be done. Today, work was also carried out on the above-mentioned wall because the lime was ready; tomorrow we will begin to gather the chiodi scattered throughout the excavation and to reduce them to the size of those already put back into use, as Your Most Illustrious Lordship has ordered me. I would like to know, please, if the chiodi now at the end of the Caponnetti cryptoporticus, that is, at the entrance to the garden of the villa, are to be taken back and reduced as well.

The passages in question testify to the intense activity of reconstruction, even at the level of the foundations. The building materials, whether modern (bricks) or ancient (chiodi = building blocks of reticulate facing; cf. comment on no. 20) were shaped (“reduced”) as needed.

A database of photographs was assembled from items in the photographic archive of the Archaeological Superintendency of Lazio, the Fototeca Unione housed in the American Academy in Rome, and the personal archive of Thomas D. Price, now in the possession of his sister. The photos were mainly found by Bernard Frischer, while Kathryn Gleason found the Price archive. The photographs of the Superintendency and of Price lack captions, dates and other details; like the other photos, they are casual and panoramic in nature. But even if they were not taken with the intent to provide scientific documentation, they nonetheless testify to the various phases of work and allow us a glimpse of the conditions prior to restoration interventions. They have been taken into account in compilation of the individual masonry stratigraphic units (MSU, for which see ); the information on these gathered from the photographs has been annotated there. A selection of the most representative photographs is given here, listed in approximate chronological order and accompanied by brief comments.

1. SAL E 661 – SAL E 662 (figs. 2 and 3)

Taken in 1911-1914. Room 53. The so-called vivarium is almost unrecognizable (for the function of this room, cf. ). The photograph shows a series of walls, no longer visible today. These walls, which had been made of reused materials, were demolished at the time of the excavation because they were ascribed to the medieval period. The demolition, besides having abolished a phase in the life of the building (thus making it impossible to formulate hypotheses for its later use), has also undoubtedly altered the evidence for its original functions. In this case, the reconstruction work was particularly invasive, as may be inferred from the state of the monument at the time of the excavation, as well as by direct analysis of the remains on the site. Extremely interesting is the relief of a flower with a central button that was inserted into the later masonry. The demolished dividing walls are only indicated by broken lines on the plan published by Lugli. Neither Lugli’s plan nor his text, however, indicates the presence of stairs (visible in the photograph), nor do they clearly describe the structure in its last phase (i.e., the actual state of the hypogeum space, which exploited the vaulted passage).

2. SAL E 729 (fig. 4)

Taken in 1911-1914. Views of many rooms, including 34 (foreground) and 21 (mid-ground), from the south. The condition of the site at the time of discovery, also affirmed by the letters addressed to Pasqui and by Lugli’s publication, is easily recognizable. Room 34 is entirely covered by a layer of cocciopesto, which is now preserved only on the floor, with a cordolo (i.e., a rounded base molding). Traces of the cocciopesto can still be observed on the reticulate stones (but never on the mortar, which is completely modern). The visible section of opus reticulatum (wall MSU 10001) is a modern restoration, except for four or five blocks near the second step on the east. The iron nails mentioned by Lugli are not identifiable in the photograph, but are still present on site, in wall 10048. This wall has, however, undergone substantial restoration. The photograph also shows evidence of an opening in the (reconstructed) southern wall of room 21, which was later restored as a continuous stretch, as indicated by Lugli on the plan and still seen today.

3. SAL E 730 (fig. 5)

Taken in 1911-1914. We see the residential block, from the west, with workers during the construction of the walls. Clearly visible are the planks keeping the strings taut. Many of the rooms have already been restored. In the foreground, the ancient section of wall 10056 in opus reticulatum and apse 10055 can be seen; the exterior of apse 10054 appears almost completely without facing. In room 21 the small piers of the suspensurae are evident. Some stretches of walls that are present today do not appear at all. Judging from the darker color of the surface of the edging (the mortar not yet being dry), the passage between rooms 14 and 15 seems to be a “correction” in progress, as is the unhappy compromise for the door of room 20. In the background the small sections of opus reticulatum in the eastern perimeter wall are recognizable, as are the huge piles of building material to be reused in the restoration.

4. SAL F 368 (fig. 6)

Taken in 1911-1914. The residential block is seen from the east, showing phases of excavation and restoration. The absence of currently existing walls is rather clear from the picture, as is the regularity of the restorations already completed at the time of the photograph. Furthermore, two mounds—probably divided according to the shape of the stones—may be distinguished at the edge of the excavation area for use in the restorations (cf. the letter of 27 December 1912 = no. 20). For the situation of room 12, cf. De Simone, .

5. SAL F 372 (fig. 7)

Taken in 1911-1914. Western rooms, from the south. The presence of masonry remains in opus reticulatum in Area 36 (west of the quadriporticus) is noteworthy. The antiquity of this material cannot be ascertained, and it is difficult to imagine it as the collapse of a section of the wall of the quadriporticus, but perhaps it may be identified as the wall indicated in Lugli’s plan as opus vittatum mixtum. In any case, the presence of this wall, which is no longer visible, further complicates matters. In the quadriporticus the internal wall still appears continuous in the photograph; later, as a result of Price’s excavation, it was re-restored with windows and doors. This detail clearly illustrates the scanty remains from which the walls were reconstructed. We can also see the relationship of the modern level to the ancient, and how the trees planted have disturbed the ancient strata and structures. Moreover, the deep trench that has been dug into the garden to a width of about 2 to 3 meters along the eastern side of the western arm of the quadriporticus has completely destroyed the ancient stratigraphy.

6. SAL A 96 – 1070 (fig. 8)

Taken in 1911-1914. Rooms 36, 52 and 53, from the east. In this photograph we can easily identify the first restoration work on structure 53, which was delineated by the removal of later walls. Most of the brick facing is shown to be an addition; this is verifiable on the monument itself. Successive work further changed the profile of the structure to its appearance today. The view of the two projecting walls (MSU 10035 and 10038) is equally interesting. They are located near the southeast corner of the western rooms, between these and the external wall of the quadriporticus (Area 36). From the photograph, it appears that one was empty, while the other could have housed stairs (the profile suggests two steps). One cannot but note how substantial the reconstructions are for all the visible structures and not just the wall facings.

7. Price Family Archive (fig. 9)

Taken in 1931. The eastern wing of the quadriporticus, Area 55, during the excavation. Three niches, two oval and one rectangular, are perfectly visible. These were made in antiquity by cutting back the eastern perimeter wall of the quadriporticus. Now almost completely obliterated by restoration, these suggest the presence of a fountain structure, as has been demonstrated by recent excavation (cf. De Simone et al., ).

8. AAR 2721 (fig. 10)

Taken in 1955. Room 33 from the southwest. An arched lintel inside wall MSU 10056 can be seen, which is no longer visible. It cannot be established if wall 10002, which abuts wall 10056, had a similar passage at that point, or if the construction of the wall obliterated it. Depending on the hypothesis, it can be connected either to a hypocaust system that will have existed between rooms 21 and 33 or (and this is more probable) to the passage of the sewer, which will later have been redirected with the construction of room 33). The arched lintel seems to be constructed in brick or roof-tiles. Moreover, we cannot tell from the photograph whether it was made at the same time as wall 10056 or by cutting through the wall during a later phase. Above and on both sides of the arched lintel the original stretch of wall in opus reticulatum is easily identifiable. The eastern side of the arch is visible in photograph SAL L2 30659, but the walls seem to have undergone considerable restoration, a fact that prevents further speculations on this subject.

9. SAL L 20650 (fig. 11)

Taken in the 1960s or 1970s. The state of decay of the walls in room 34 is evident from the reticulate facing that has fallen. The surviving section is almost certainly attributable to the restorations of 1911-1914. We are therefore faced with the paradoxical situation that the wall had been restored twice, once in the period 1911-1914 and again in the 1960s or 1970s. The wall facing of the second intervention is offset with respect to the first restoration, as if it were the ancient original.

10. SAL I 1625 (fig. 12)

Taken in the 1960s or 1970s. Restoration work in Area 36, from the north. The previously reconstructed sections of masonry are easily distinguishable. First and foremost, it is important to note the remnants of structure MSU 10046. These include a tract of wall made of complete bricks for the first two courses and fragments of bricks (or roof-tiles?) for the remainder of the section. Despite the poor state of conservation, no limestone blocks can be distinguished, but these are seen today and were evidently inserted during subsequent restorations. However, the most noteworthy fact is that this tract of wall rests on another, which is now no longer visible on the site; it has been covered by the stone floor being laid at the time the photograph was taken. The demolished structure may be associated with the northern side of the drainage conduit leading into the main drainage channel. While it is impossible to determine whether the facing of this wall was opus incertum or reticulatum, the fact that it had a facing, according to this photograph, is noteworthy considering its level. Wall 10017, which delimits the west rooms, has a facing, albeit ruined, of opus incertum.

The small buttresses that abut the outside of the western wall of the quadriporticus (MSU 10056/10068), which are still visible at the site, have never been preserved above the level of the opus incertum section (10068). Furthermore, their original profile suggests that they are to be associated with the cover of the main drainage conduit, subsequently ruined and collapsed. Such indications may support the hypothesis that the buttresses were intended to be buried.

The extent of the reconstruction work carried out on the site from 1911 to 1914 and later in the twentieth century is now well documented. To some extent these early efforts were forgotten or were accepted at face value as a reconstruction that was faithful to the original.

These restorations aimed not only at reconstructing some walls, but also at the demolition or obliteration of others, particularly those thought to be of the medieval period. Because the work was carried out without first having come to an adequate understanding of the structures and without having made a thorough documentation of the original remains, it has caused irreversible damage. The sudden interruption of work in 1914 (see Frischer, ) created a further unevenness in the complex. In Area 17, for example, the structures partially preserved at the level of the foundations were not rebuilt, as they were in other areas, and, as a result, we obtain an image of the monument that has been falsified yet again—this time precisely by a reconstruction that is inconsistent and hence (in the total absence of a sign on the site or at least a scientific publication explaining what was done, not done, and why) confusing and misleading. The most recent restorations, sometimes employing modern materials, have in turn created further ambiguity, because the vertical line of the facing was set back in relation to the 1911-1914 restoration, carried out primarily using ancient materials. The unfortunate result is that, at first glance, the 1911-1914 restorations appear to be original walls.

The various superimpositions and demolitions of stratigraphic relationships make the interpretation of very partially preserved structures even more complicated and have consequently given rise to a series of new questions that have not all been resolved.

Parallel to the collection and analysis of archival documents and photographs, the arduous task of analyzing the wall structures on site was undertaken. This work included measured drawings, creation of a database of wall features, and a sampling of mortars (cf. Appendix II, ). The collection and collation of data was not enough on its own, but it was a convenient way to approach the study, and it provided the starting-point for the reflections that follow.

To document the individual structures, a Masonry Stratigraphic Unit (MSU; see also Frischer et al. ) catalogue form was used. It was modeled on the MSU form recommended by the University of Siena.

Granted that it is impossible to create a form that could contain all the terms and fields useful for the study of the monument and that a database alone is not sufficient for understanding it, the creation of a catalogue is still a useful procedure, since it helps to keep our critical faculties constantly alert as we confront the monument. The study must therefore be based upon direct knowledge of the structure, whose data can be only partially captured on a catalogue form, which for this reason is only a means, not an end. These kinds of documentation are not exhaustive by themselves and a greater effort needs to be focused on precisely the personal choice of the data that ought to be entered. But the documentation phase has its undeniable importance, especially with respect to the lack of understanding that can arise from excessively fanciful restorations. Of course, every intervention on an ancient monument ought to be rigorously documented, within the limits imposed by subjectivity, in order to avoid misunderstanding or loss of information. This is a policy to which assent is very readily given and that can also be considered completely obvious, yet it has often not been applied and—what is worse—is even today not always observed.

As noted, some of the data we have gathered were drawn from materials not originally intended to furnish exact documentation, such as letters, notes about expenses, and casual photographs. In the present case, we have attempted to proceed in the opposite direction, reconstructing the modern interventions that have affected the walls, isolating the parts that are clearly original and seeking, insofar as is still possible, to identify remains that are helpful to the interpretation of the phases and function of the various rooms and spaces of the villa.

In the residence (rooms 1-21), the walls that have been restored or entirely reconstructed have been examined, but only a sample of them has been catalogued, whereas in the rest of the monument we have documented not only the walls identified in the new excavations but also the walls that were excavated earlier and subsequently restored.

Another limitation of this study should be mentioned. We have restricted our investigations, in the areas previously excavated, to the features presently visible on the surface. We have not been able to undertake new excavations or cleaning beneath the currently visible remains. It is thus probable that portions of a wall indicated in the catalogue as “completely restored” still preserve an original course below the surface. On the other hand, we also know that some structures have certainly been completely reconstructed atop the foundations or even from the foundations up.

Because it was merely the means of the study (not the end), the catalogue form has been used with a certain flexibility, which admittedly runs the risk of subjectivity. Where it has been impossible to decide if we are confronted with one or more masonry stratigraphic units, a single form has been filled out treating the wall as it currently appears (e.g., MSU 10004) and putting the various issues into a note. Where there are various ancient wall segments that can be clearly identified as belonging to a single structure, only one catalogue form has been used (e.g., MSU 10061, fountain). For the building with an internal oval plan furnished with niches (building 53) no catalogue entry was made since it would have been impossible to do justice to its complexity in a short entry. The structure has been treated in a separate paragraph (). This activity of “wall census” has been identified by its own operational sector (Sector X) and the pertinent MSU belong to group 10000 (see fig. 13 for MSU numbers). Given the fact that so many stratigraphic relationships are no longer legible, the construction of a stratigraphic matrix for the wall structures unfortunately has not been possible.

The direct analysis of the monument, corroborated by archival documentation, enables us to establish some fixed points from which a new reading of the complex could proceed. On the basis of various observations, we have therefore created a plan showing those features, currently visible on the site, that have at least one genuinely ancient course of elevation. A two-dimensional plan could not, of course, handle the situation in which walls are preserved one atop the other, when the upper wall is entirely or partially restored.

It appears clear that, at the Licenza villa, the portions of ancient masonry that can be analyzed are quite small, sometimes limited to just one course. As noted, however, there may be additional ancient masonry preserved under the surface (including, as the 1997-2001 excavations showed, entirely new building phases hidden beneath the levels where the previous excavations stopped and where restorations were consequently made). It should not be forgotten, furthermore, that the medieval period is under-represented because the excavations of 1911-1914 intentionally removed much of what remained from this period in order to reach the classical levels. Sometimes what seemed to the previous excavators to be medieval could also have belonged to other phases, thereby further complicating any precise reading of the site.

The information derived from the plan should be combined with the MSU catalogue entries. The numeration of the rooms or areas is, of course, purely arbitrary and was made for the sake of convenience. It was often not possible to take account of distinct units nor, obviously, of distinct phases in the life of the villa.

It is evident that the walls in opus reticulatum in the residential part of the villa (rooms 1-21) were heavily restored and sometimes even completely rebuilt. A country road passed through this area (cf. Frischer, , fig. 17) and the structures were preserved for only a few courses or simply at the foundation level. Moreover, some walls were re-erected that probably had already been razed to the ground in antiquity. The reconstruction of many of the structures down to the foundation level reduces everything to a single phase, while it is quite possible that in several places earlier structures affected, or were in part exploited for, the terracing of the residence. These pre-existing structures have been verified in at least one case (see De Simone, ).

Some of the problems encountered concern the whole area in general, while others refer to individual loci. Among the general problems the following may be noted: the masonry work; the true dimensions of the complex; the doorways and the complete absence of evidence that the doorways were walled up; the floor levels; the circulation routes; and the main entrance. Among the specific problems one can list: the organization of Areas 8, 12, and 17; the opening in the foundation between rooms 24 and 33; and the access to room 33. Many of the definitely original tracts of walls in opus reticulatum present tufa facing-blocks, while the reconstruction exclusively used limestone. This situation does not necessarily offer a dating criterion or an indication of phase, but can simply reflect the use of a mixture of materials readily available in the area.

The pars urbana must undoubtedly have been more extensive. Evidence of this, at least with reference to later phases, is the strange situation of Area 8 and the relationship between the structure identifiable as a fountain and the northern wall, which has always been interpreted as the limit of the building. In particular, the northern wall, of which only the foundation level was found for nearly its entire length, now appears as continuous, but it probably had openings that connected it to the zone farther north. In Area 8, it is likely that part of that wall was cut and razed to the ground when the fountain was built in order to permit the fountain to be used.

The square structure of the brick fountain was inserted into an area that was presumably not covered at the time, but which must have had a completely different arrangement. The fact that the intervention was later has not been proved by a difference in wall type, but is based on other considerations. Traces of a wall in opus reticulatum, certainly razed in antiquity, are still visible in the southwest corner of Area 8. The wall, which cannot be considered a foundation, is disposed at a 90-degree angle with respect to the south wall, and is interrupted at the point where there is a manhole giving access to the drain below. In a first phase the wall might have delimited a small room—perhaps a cubiculum—and there might have been an analogous wall on the north side (now occupied by the fountain), of which no visible trace remains. Later, the open area—perhaps originally furnished with a small peristyle—had been enlarged, by way of leveling off the structures of the west side. It was paved with limestone slabs. Traces of this arrangement survive in the southwest corner, where we also find, cut into a big slab, a small drain that leads to the sewer below. These changes were probably undertaken to make space for the insertion of the monumental fountain. Its present arrangement seems somewhat anomalous because it does not occupy the geometric center of the area, and it is difficult to recognize any axial relationship that may have existed within the space as restored. Moreover, to the north it pierces through the perimeter wall, thereby being surrounded on its other three sides by an irregular U-shaped space.

There are hardly any legible remains of the structure of the fountain, which has also been greatly affected by modern restorations, and no hypothesis can be made about the revetment that protected the walls. Almost square in shape, the fountain is equipped with an outer channel as well as a smaller inner one, which is essentially a hollow space (intercapedo). The latter communicates with four semicircular niches, each set in the center of one side. In creating the two channels, the builders of the fountain dug down to a level lower than that of the pavement, except for the center of the fountain, which was left at the original height. The exterior side of the outer wall of the fountain was built directly against the baulk of the construction trench. One side of the interior wall with its niches lined the center of the fountain, which means that it was built against the opposite side of the construction trench. The center of the fountain does not present evidence of basins, and, in the absence of a small pond, we have to imagine that it was kept as a garden, perhaps with one or two ornamental statues. Water gushed from the niches, probably emerging from protomes. The water flowed within the narrow intercapedo, thereby forming waterworks of the “cascading” type, flowing finally into the external channel.

It has already been noted that the addition of the fountain changed the original layout of the first phase of this zone. The most striking aspect of this modification was precisely the demolition of the north wall, a feature that presupposes the opening on that side. The form of the structure, furnished with niches on all four sides, identifies it precisely as a fountain to be centered (da centro), and this brings up yet again the problem of how it could have been enjoyed from the north. In this connection one should recall what Lavagne said about analogous structures: “The crucial phase in which the fountain begins to be treated as an autonomous subject, to be placed in the middle of a room or a courtyard, and no longer as a decorative motif within a niche, seems to be in the first years of the second century A.D.” It should be emphasized that in reality there are already precursors of this trend attested in the late first century A.D. (cf., for example, the “pelta-shield” fountain on the Palatine from the late Flavian period).

The wall that functions as a partition and comprises the north side of corridor 5 ought, most likely, to have had an opening that permitted Area 8, where the fountain was inserted, to communicate with it and thus with room 12 as well. Because of these walls, as reinterpreted by restoration, we lose the ability to see the circulation routes and the various phases. In discussing the documents, we have already noted how the doorways have been sometimes ignored, rethought, or arbitrarily reconstructed, not to mention the fact that there is no recorded evidence that doorways were ever walled up, generally a fairly frequent occurrence in any building complex. In several cases the openings, in their present condition, do not show any original features. One will have to take into account these observations before hazarding a formal appraisal of the residence that emphasizes its symmetry, functionality, and the axiality of circulation routes visible today.

It has been practically impossible to identify the ancient pavement levels. For the most part, the foundations were rebuilt to a uniform level, as were the first courses of a large part of the reconstructed walls. After the removal of the mosaics, carried out in the 1970s, the pavements were reset at a different level. Moreover, no data are known about the subfloors of the mosaics, which could also belong to a different phase with respect to the surrounding walls. Some walls appear arbitrarily aligned at the edge of one of the two sides of the foundation, while others are set (as is usually the case in Roman architecture) on the middle of the foundations; consequently, the reconstructions falsify the original wall thickness.

The arrangement of Areas 12 and 17, for different reasons, betrays the subjectivity of the first excavators. Area 12, generally identified as an atrium, today consists of a rectangular space ca. 9.75 m x 8.20 m, with a small square brick basin, which, however, is not located in the geometric center of the room. The excavation has shown that not all the foundation remains were taken into account equally at the time of the restorations, and other structures were identified that attest to different phases and different arrangements of the space (see De Simone, ). Accepting for the sake of argument that the space was used as an atrium, at least at a specific moment in the life of the building, then such a space, the purpose of which is so intimately connected to the circulation of people, light, and air, ought to have been equipped with further openings, but the current state of the evidence shows no trace of them.

In Area 17, tracts of walls also have been identified that were not taken into account at the time of the modern restorations. Together with the neighboring rooms 18, 26, and 27, for which no ancient segments of wall have been recognized, it constitutes an area that is difficult to understand. Some walls now appear to be made of small blocks of limestone and cardellino (cf. D.1.4 for the definition of cardellino) and respect the surviving mosaic flooring. The latter disrupts the continuity of the wall in opus reticulatum traditionally considered as the west limit of the villa in its original design.

Although in the case of rooms 19-21 the portions of ancient masonry are also very scarce, it is clear that this zone underwent an ancient intervention intended to change its function. The insertion of the small piers of a suspensurae system left the foundation of the pre-existing wall uncovered. The new use necessitated a different arrangement of the spaces, with the creation of two walls to bound room 20, which was also furnished with an apse. The latter feature seems to be in phase with room 33 behind, while the two partition walls do not provide dating elements, except inasmuch as they are later than the so-called perimeter wall (west side). The partial closing of the apse—this also restored—may be associated with the division into rooms at a later date. What is most interesting in this area, however, is the existence of an arch in that part of the foundation of the so-called perimeter wall which corresponds to room 21, an arch that is now no longer visible (cf. the commentary on photograph AAR 2721, , no. 8, and see also the discussion of the sewer system, ).

It seems plausible that at the time room 33 was built there could have been access to it on the east side, even though there is no trace of any opening, in that the so-called perimeter wall in opus reticulatum has been completely restored and follows the course of the foundation without a break.

At present, most of the archaeological site is occupied by the area of the garden and the quadriporticus that surrounds it (ca. 42 m x 85 m). Since it is at the same level as the residence, the northern side of the quadriporticus serves as a joining structure with it and produces a sort of veranda. This, in turn, gives onto the garden and must have had windows to provide the rooms with light and air. The connection between the different levels was provided by three ramps, two lateral ones for the long porticoed wings and a central one for the uncovered area. Except for the veranda, which rises on a small terrace, the other sides of the quadriporticus were constructed following the natural slope.

As we have seen, in this zone the massive restorations have also resulted in situations that are difficult to understand. The masonry remains must have been minimal. In fact, at first, an almost complete reconstruction of the raised part of the walls in opus reticulatum (at least in the northern half of the quadriporticus) was proposed by the restorers, thus creating a continuous internal wall. The subsequent excavation of Price, however, showed that at least at one point in the internal wall there were openings that communicated with the garden. Therefore, a new restoration was undertaken, which fixed the arrangement visible today.

If such a question has found a reasonable solution, the same cannot be said of the problems posed by the external walls, particularly on the west wing, nor of the structures that impinge on the east wing.

On the west wing, the structure in opus incertum faced with plaster, brought to light in the recent excavations, attests to a distinct phase; at present, this structure is visible for the whole of the western face of the perimeter wall and in part on the outside of the south wing (fig. 14). It was previously interpreted as being a simple foundation offset for the support of the plastering on the opus reticulatum wall. In the documents, there is never any mention of a structure in opus incertum, although stretches of plastering are noted (cf. , nos. 1 and 2). The wall in opus reticulatum therefore belongs to a later phase, when the quadriporticus was constructed, after the ground level was raised. The erection of structure 4203, which lines the portion of the wall in opus incertum on its eastern face, must be connected to this work. The correct interpretation of this activity cannot be separated from the question of the modern reconstructive restorations. The external perimeter wall of the quadriporticus (MSU 10056=4211=4005) has on its west side a long stretch of a structure in opus reticulatum that is undoubtedly original. There remains reasonable doubt concerning the antiquity of the uppermost part, however, because it seems that the first excavations reached the very level indicated by that portion of opus reticulatum that is certainly original. Did they dig until they uncovered the top of the remaining structure, stop excavating at that level, and then reconstruct the wall to its present height? Or had they already found the wall at the current height and therefore it is only the upper part that is sealed with modern mortar set onto the surface of the facing? In the 1970s the excavators went deeper, bringing to light the portion of opus reticulatum that was certainly original. They did not reconstruct it at all for the rest of its length and instead uncovered the wall in opus incertum beneath it.

We have three elements, of varying importance, which lead us to hypothesize the almost complete reconstruction of the upper part. First, in Sector IV.1, the original wall in opus reticulatum juts out in relation to the line of the southward continuation, which is set back and evidently restored. Second, the external buttresses survive to a height that never reaches the level of the structure in opus reticulatum that we believe was completely reconstructed in the process of restoration. Third, the internal wall of the quadriporticus is also poorly preserved, and after the first reconstruction there was a subsequent restoration, which altered the arrangement of the full and empty spaces. This modification would not have been possible if it had been necessary to demolish the original wall. The practice followed by the excavators of 1911-1914 seems to have been consistent: identification of the top of the wall and the reconstruction of it, set back in relation to the line of the facing. It is also possible that the excavators identified the structure simply as a more or less coherent nucleus, which was preserved up to a greater height than that hypothesized here as being original. They therefore unified it by means of refacing, as seems to have happened in other parts of the villa. It is thus shown that the east side of the external wall is wholly the result of reconstruction in the early 1900s, except for the undoubtedly original portion of Sector IV.1. But what was the purpose of structure 4203? I maintain that, together with the structure in opus incertum, it constituted the foundation of the wall in opus reticulatum, which was almost twice as thick as that proposed in the restoration (ca. 45 cm). If we accept, on the other hand, the restored wall of 45 cm, it is necessary to hypothesize a structure, now completely lost, which was founded on 4203, evidently with the purpose of reinforcing the perimeter wall. An analogous situation presents itself on the east wing, which has either two parallel walls built up next to each other (according to the restorations), or conversely a single thick wall; it is impossible to identify, however, those parts that are certainly original.

The perimeter wall of the quadriporticus, on the south and west sides, is furnished with a series of limestone opus vittatum buttresses set at right angles (a pettine) and backing onto the wall. The structural reasons for this intervention are multiple. First, the buttresses countered the pressure of the soil caused by the difference in ground level. Second, they served to evenly distribute the forces pushing against the continuous wall, which, because it is so long, is exposed both to a high risk of rotating, due to its own weight and that of the roof, and to cracks caused by thermal fluctuations. In addition, the conduit of the principal drain runs parallel to the outside face of the wall of the quadriporticus; this further indicates that the lowest point of the original ground level must have been precisely in this area, that is, the meeting-point (compluvium) of two slopes on the east-west axis, which created a natural channel. This originally gathered and drained away the surface water naturally, and then later was reorganized, regulated and used as the sewer system. We must still consider the original height of the buttresses, which could not have been very high, even at the moment of construction, given their limited length. The height of the wall of the quadriporticus can be partially worked out, albeit indirectly, from the spacing of the buttresses on the west side; in fact, this type of buttressing performs its static function only if the distance between each element is less than the height of the wall. Since the distance between the buttresses is ca. 2.96 m (= 10 Roman feet), we can deduce that the wall supported by them was certainly higher than that. On the outside of the south wing, the buttresses found at the level of the foundations are spaced more closely together, evidently on account of the natural slope of the site, which slants from north to south.

The east wing, which bends slightly to the southwest, includes a series of structures that modify its original appearance and function (cf. De Simone et al., and ). These works partly eliminated the wall in opus reticulatum and modified both the circulation routes and the visual axes. It is possible that these were necessitated by a partial structural failure of the perimeter wall; caissons would then have been built, filled with earth and stones, to counter the pressure from uphill. Perhaps in order to mask these structures and to render the rest of the portico usable, a construction with oval and rectangular niches was added, presumably a fountain; since the portico at this point was no longer usable, it is possible that a sort of detour was created through the garden in order to bypass the obstacle (cf. Gleason, , fig. 13). In this part of the quadriporticus, then, they had to destroy the original roof, which was almost certainly one with an inward slope. Neither a flat roof nor one sloping outward is possible here. A flat roof, which is not particularly common in this type of structure, would have been avoided in this geographical area due to the amount of snow and heavy rainfall. An outward-sloping roof was not normally used for covering a quadriporticus, because it makes the joining of the slopes of the roof for the compluvium difficult. Furthermore, the roof of the veranda certainly sloped outward toward the garden. In order to cover a span of only 3.00 to 3.20 m, there was no need for a pitched roof, which in this case would have created problems for the watershed.

A problem unresolved on account of the scarcity of remains concerns the openings that must have linked the quadriporticus with the area to the west. Nor can we be certain, although it is plausible, that there was access from the outside to the south wing, as Lugli hypothesized (col. 541), but it certainly cannot have been the only entrance to the villa, nor even the main one.

The area of the garden has only partially been examined and the sole structure found is a large pool in limestone opus signinum. This should be considered as an open-air water reservoir, which occupies a large space (24.50 m x 12.90 m), almost at the center of the garden. The fact that the villa received water from a small aqueduct coming from the west and that no traces of any piping leading from the reservoir have been discovered, lead us to believe that this is not a real cistern. The only link with the water system of the villa seems to have been the outflow conduit, which starts from about the middle of the west side and flows into the main drain; it follows an oblique course, in order to take advantage of the natural slope. The outflow is regulated by a small sump with footholds consisting of two cover-tiles (imbrices). The sump abuts the west wall of the pool, and serves as a manhole for regulating the level of the water. The excavations carried out in the first half of the twentieth century followed the line of the walls and removed the earth only from the inside of the tank, creating a situation similar to that of antiquity, but compromising the static equilibrium of the structure; in fact, with no water on the inside, there is no counterthrust for the earth, which puts greater pressure on the long sides. The same problem existed in antiquity, when the structure no longer held water and began to be buried. In fact, the north side has a tranche, which is broken and has slipped southward, while the south side is missing in the middle, and on the west has been forced out of plane, with damage at the corner. The structure has two external bodies on each of the long sides. A calculation of a depth of about 2 m gives a volume of more than 600 cubic meters (more than 600 tons of water). To understand the forces working on the masonry, it is necessary on one hand to calculate the weight of the water and the incidental load (due, for example, to wind) and on the other hand, the weight of the earth; in addition, the difference in level between north and south as visible from the longitudinal section of the whole villa should also be considered.

Price had already questioned whether the four avant-corps served a static function as buttresses, or as foundation-bases for other superstructures, perhaps of an ornamental nature (such as statuary groups). In considering this issue, we must remember that the two pairs of buttresses are of different sizes. Those to the south are built to counter the lack of static equilibrium of the pool when it is full; they are bigger, because the slope necessitates more support on this side. Furthermore, it is probable that this area was partially levelled with less compacted soil brought from elsewhere and thus requiring more powerful and deeply sunk structures in order to achieve the maximum anchorage possible. In order to understand correctly the function of the structure, we must remember that the southwest buttress, the only one today not completely buried, has a cavity opening towards the inside of the piscina. This is not enough to support a hypothesis that the whole buttress was hollow; if it had been, it could not have performed its static purpose. It is possible that such a recess may have been a refuge for fish; although the documentation provides no information for the other three buttresses, we cannot exclude the same function for them as well. We can, therefore, visualize the piscina as a fish-pond. Unfortunately, this structure, which is entirely attributed to the same building phase as the quadriporticus, does not have the elements for a certain and absolute dating.

The constructions on the west side of the site consist of a series of rooms that can be generally attributed as the baths of the villa. The excavations of 1911-1914 did not completely uncover these buildings, but they did compromise any interpretation, in that, in addition to the restorations previously discussed, some walls were demolished (cf. figs. 2-8), while others were reconstructed on minimal evidence. The last intervention, in the 1980s, affected the upper strata of rooms 37-40 and 50-51. Based on the documentation and what is still visible on the site, it is possible to deduce the primary characteristics of the complex.

In this area, too, which was previously occupied by other structures, work was carried out to counter the incline of the natural slope (at this point slanting from west to east). Thus, the floor levels are more or less at the same level as those of the dwelling area, thereby creating a kind of terrace in relation to the quadriporticus.

The space occupied by the thermal rooms gives the impression of being excessively large in relation to the size of the residential block. This disproportion diminishes if we remember that the living quarters must undoubtedly have been more extensive, stretching to the north and probably to the northwest. The thermal buildings are the result of two distinct construction phases, the second of which affects the southern part, with the addition of a series of heated rooms and the construction of a building of considerable architectural importance, the laconicum (room 53). The two different construction phases can be deduced not only from the different building techniques, but also from an apparent irrationality in the layout of the service rooms, which denotes two completely distinct heated areas. Having ascertained the difference in the building phases, we cannot exclude the possibility that all the rooms functioned at the same time. Another equally plausible hypothesis, however, is that the earlier buildings were modified wholly or partially and adapted to different uses.

The first thermal rooms to be built were undoubtedly rooms 32-34, which abut both the west perimeter wall of the villa and the north wall of the Republican atrium (a space subsequently subdivided into rooms 38, 39, 40) and which were modified through the incorporation of later structures. Lugli saw in these constructions a variety of phases, which cannot be detected in the masonry; his reading is not convincing from the point of view of their presumed functions. We have in fact a homogeneous group of rooms, composed of a large hall with an apse (33), flanked by a service corridor (32) and linked to a single room that contained a stairway (34) (figs. 15 and 16). In order to construct this complex it must have been necessary to go far below the floor level of the residential block in order to have a height sufficient for the functioning of the hypocaust system. Room 32 consists of a long, narrow corridor (ca. 13 m x 1.30 m) in opus reticulatum and coigning in limestone blocks, at present accessible from room 31 by way of a series of steps which are most likely products of the restoration. The presumed floor level is at ca. 2.80 m below ground level. The west wall (MSU 10053) and the south wall (MSU 10062) are wholly constructed against the regularized slope, with only the exposed side provided with facing; about halfway along, the west wall turns into a very shallow apse, with a span of ca. 4.40 m. In the east wall (MSU 10057) there are two small arches for the praefurnia, with similar characteristics (height at the keystone 1.70 m, width 0.80 m, radius of the arch 0.35 m), at the level of the apse. The praefurnia connect the corridor with room 33, which is a large rectangular hall, with an apse on the north side (ca. 12 m x 5 m, excluding the apse; the span of the apse is 4.36 m, the radius 2.10 m). The hall, which survives only at the subfloor level, has internal wall facings in cotto (cf. MSU 10001, 10002, 10051, 10052, 10053, 10054). Rather than bricks, broken and trimmed roof-tiles are used, which sometimes are put in courses with the unbroken rim showing (creating a string-course, e.g., for the installation of the suspended floor, cf. , types 4.1, 5.3.1, and 5.3.2). The partially preserved subfloor level consists of roof-tiles and is raised in relation to the service corridor. No trace remains of the suspended floor, but its level is indirectly discernable from the line of the string-course roof-tiles and the height of the arches of the praefurnia. An offset along the east wall most probably served to support a fistula (cf. ). The south end, unique for its construction technique (MSU 10001; cf. , type 5.3.3), is articulated by two quadrangular brick avant-corps (MSU 10051 and 10052), which form a small recessed area. The passages to room 34 should be identified as corresponding with the two avant-corps. This room (ca. 3.80 m x 6 m) is not an ordinary space. Against the north wall is built a narrow platform-like structure, approached by two small lateral ramps next to the avant-corps of room 33; the floor level of the room, which is accessed by way of another ramp situated against the east wall, is essentially at the same level as that of corridor 32, while the top of the platform is about 1.50 m higher. A small rectangular avant-corps is located between the stairway and the south wall. At the time of Pasqui’s excavations, the walls in opus reticulatum and the steps in limestone blocks were found to be completely coated with a layer of cocciopesto, of which only some small patches, belonging to the watertight edging of the floor, are now visible. A hole with a lead outflow fistula is visible. The west wall is constructed against the regularized slope (controterra) only up to ground level, and has a small specus linked to a small channel; part of the south wall partially abuts the north wall (MSU 10045) of the Republican atrium. The room has an outflow hole, with the related fistula and lead protection sheet. Lugli also mentions another outflow (or overflow?) at the level of the platform (col. 537), of which no trace has been found.

These rooms formed the first thermal nucleus of the villa. Room 33 is certainly to be identified as a calidarium, but it is difficult to locate where its tanks may have been, in that there are no traces of furnaces or water pipes. Due to lack of space, it would have been nearly impossible to place the tanks close to the praefurnia, as usual practice would dictate. We could hypothesize the existence of a basin in the rectangular niche of Room 33, although this would have resulted in considerable heat loss. Lugli mentions a fountain, of which there is no trace, in the apse of the same room. It is probable, however, that within the apse there was a labrum attached to a water supply. The calidarium, which was certainly of notable proportions for a private complex, was accessed from the east, probably from room 21, which served as a changing room. It is impossible, however, to specify the chronological sequence of rooms 19-21, which occupy the southwest corner of the rectangle of the living quarters; this part was evidently remodeled in order to be used as thermae. It is fairly clear that the apse of room 20 (MSU 10055), which was created by cutting an opening in the perimeter wall of the villa, is perfectly in phase with the apse of the calidarium lying behind. Little can be said, however, about the internal arrangement and the purpose of rooms 19-21 at that time. Inasmuch as it is visible today, room 20 was subsequently turned into a heated room; in the course of this work the foundation of the perimeter wall was uncovered to give enough space for the hypocaust. This was probably heated independently of calidarium 33, by way of a furnace in room 19. It is probable, however, that this last room was in direct communication with service corridor 32, if we hypothesize a continuation of the oblique wall.

Room 33 also allowed passage to the south, into room 34, and to the west directly to the porticoed area (35). Corridor 32 in fact was completely hypogean, and the portico was constructed with a series of masonry columns, aligned parallel to wall 10003/10057 (the west side of rooms 33 and 34), upon which rested the ridge of the roof slope. The covering of the corridor may have consisted of simple planking or of a proper concrete ceiling laid on a wooden framework. No trace remains of these structures. The shallow apse that characterizes the corridor certainly did not have an aesthetic purpose. Rather, it served a structural need, countering the thrust of the earth behind, and at the same time it fulfilled a functional requirement by providing as much space as possible for working at the mouths of the praefurnia. It is not possible to verify the existence of supports for hot-water tanks on the inside of the corridor, which appears too narrow, but we may hypothesize that the blind end of the corridor, south of the furnaces, was used to store wood.

The interpretation of room 34 is more difficult. Evidently, this room was intended to hold water, but it is not possible that its level reached the top of the platform, which must have been used as a narrow passage, while the central stairs led down to the immersion pool. The south avant-corps, which skirts the stairs, must have served as a passage to the area even further south, previously occupied by the quadrangular room of the so-called Republican atrium (rooms 38-40). Both for functional reasons and because of the difference in level, I maintain that there was no communication between the quadriporticus and room 34. This room may be identified as a small frigidarium/natatio, rather unusual for its layout as well as for its size. This small thermal complex (calidarium 33, frigidarium/natatio 34, service corridor 32, group of rooms 19-21) was protected on the west side by a portico (Area 35). The remains of this portico consist of the foundations and, sometimes, the first course of six brick columns and one brick pilaster. As we have noted, the portico was partly placed over the hypogean corridor (32) and made use of the west wall of rooms 33 and 34; it was therefore ca. 3.80 m wide (ca. 13 Roman feet, calculating from the center of the column). The interaxis between the columns, approximated by reconstruction, is about 10 Roman feet, except for the last one on the south, between the column and the pilaster (ca. 11 Roman feet). It is difficult to establish when this work was carried out. The fact is that the portico is perfectly aligned with the two west piers of the Republican atrium; this can hardly be by chance. It is possible, therefore, that the colonnade was constructed with the intention of harmonizing the new structures with the earlier ones, by providing an element of continuity. Nothing is known about the arrangement of the area to the west of the portico, which may have been used as a palaestra.

The incorporation of these structures may have partly altered the disposition of the roofs of the residential block. The sloped roof of the residence must have been at a higher level than the roof of the small thermal complex. The conjectured covering for room 33 is a barrel vault, rectangular in plan, protected by a single-sloped roof, inclined to the west, and overhanging the slope of the roof of the portico.

Immediately to the south of these first thermal buildings an extension was built. An apsidal basin (room 37) was constructed, with a platform and lateral steps, and arranged with rectangular niches (two lateral ones, for access to the basin, and a wider, central one). The structure was built next to the west wall (MSU 10016) of rooms 38-40, and, in order to reach the depth necessary for the basin, the foundations of the earlier wall 10016 were partially exposed. In this way the Republican atrium was adapted into a frigidarium. The floor level of 38-40 was raised and a mosaic floor with triangular marble inlay was laid. In order that the apsidal basin (MSU 10010) might be used, the west wall of the old atrium (MSU 10016) was demolished; it is possible, however, that the upper part was retained, or rather remodeled, in such a way as to create an arch over the entrance opening toward the basin. In this case, if the four central piers continued to be in use, it would not have been necessary to alter the layout of the roof, and the compluvium structure could have been retained. Obviously, the basin must have had its own cover, with an apsidal semi-dome, itself protected by a semicircular roof.

Unfortunately, the excavation data are not sufficient to resolve this question, both on account of the earlier work and because of the poor state of preservation of the remains, which were subsequently transformed over the centuries. In fact, we do not know with certainty whether the piers were in use at the same time as the mosaic floor. Indirect support of such a hypothesis, however, can be detected by the intentional alignment of portico 35, which as noted harmonized the pre-existing structures in view of their renewed use. Further indication for the process of homogenization is provided by the very dimensions of the basin, whose apsidal platform has nearly the same radius as the apse of calidarium 33.

Access to the frigidarium could have been from the east, perhaps directly from the quadriporticus, by way of several steps that linked the different levels.Traces of a small ramp may perhaps be seen in the two masonry fragments that are equidistant from the median axis of the frigidarium and which might indicate a possible access to the room. South of the stairs, at a level slightly lower than the terracing on which the frigidarium lies, the latrine would have been built in an unused space. Its walls have in part disappeared but are recorded in Lugli’s plan and in an archival photograph (SAL F 372, fig. 7). At the moment, the typically shaped slabs have been repositioned, perhaps at an arbitrary level, but the drain conduits underneath guarantee their planimetric layout.

If we analyze the rooms to the south of the frigidarium 37-40, it is immediately evident how a new thermal complex, which was completely distinct from the original nucleus 32-34, was created. It is not possible to establish whether this work was carried out simultaneously with the new frigidarium or later. The building technique provides no clues, as it is linked in this case to purely functional requirements. The south buildings (43-49, 51, 53) are all heated and separated from the frigidarium 37-40 by a screen of rooms (41-42). Room 41, from its narrow and elongated shape, seems to have been part of the circulation plan of the thermae, and/or a changing room separating the heated areas from the others. Its connection with the small rectangular basin 42 is not entirely clear, as there are some anomalous steps set at its corner. Moreover, for this phase, we do not know the arrangement of the area immediately east of room 41. The floor levels of the heated rooms are similar to those of the frigidarium 37-40. Indirect evidence for reconstructing floor levels is provided by the identification of the door that gives access between room 49 and building 53.

The corridor that served these rooms is to be found to the east of room 47, which, on account of its proximity to the furnace, should be interpreted as a calidarium. To the west is a series of spaces, heated indirectly, whose arrangement into distinct rooms is difficult to determine, with the sole exception of room 51. This room, even though it is not completely visible today, presents a rectangular plan, with very shallow apses on the short sides; it is equipped with an arch for the passage of hot air. The disposition of the rooms and the actual levels that can be seen imply that the domestic staff would not have had direct access to this praefurnium. It is unlikely, too, that a system of praefurnia existed to the west, since on that side the slope rises steeply. It is possible to hypothesize an extension of the complex to the west, but only at a higher level.

Room 52 may possibly have been a staircase, but there is no data enabling us to relate its construction directly to the use of the thermal buildings. The scant remains available do not allow us to hazard hypotheses about the roofing in this area.

Directly linked to this group of heated rooms, but with an extremely unusual appearance, is the most famous building of the site, which Lugli had identified as an ornamental fishpond.

Building 53 is without doubt the most attractive and monumental structure of the entire archaeological site; this is partly due to its relatively well-preserved state, at least when compared to the other surviving masonry structures, and partly to Lugli’s fascinating interpretation of it, in which he visualized a nymphaeum that was subsequently converted into a vivarium (col. 559).

The building is elliptical in plan, incorporated into a trapezoidal form (fig. 17). The ellipse is further arranged into four semicircular niches that correspond to the four corners of the trapezoid. Apart from the alterations, which reflect the addition of masonry now partly demolished, and the construction of a longitudinal trench at the bottom, three different internal levels are visible today. These can be described as follows, from top to bottom: a) corresponding to the elliptical ring that is joined to the niches—residues of brick flooring; b) level of the floor bed that marks out the smaller ellipse––residues of brick flooring; and c) floor level of the hypogean corridor—brick flooring. The structure itself is entirely built of brick, with the exception of some peripheral parts, which are faced in cardellino blocks. In the wall that marks the change in level between the present upper and lower ellipses, there are some small rectangular niches (51 cm x 49 cm, ca. 79 cm in height), with a slope at the bottom and covered a cappuccina (fig. 18). Each of these communicates with the floor of the ellipse above by way of a chimney-like feature. There are four niches on the east side, into which the arch of the hypogean corridor opens, and five on the west side. The distance between them is not consistent. In addition, next to the elliptical internal wall to the north and south, there are two solid avant-corps that jut out toward the center of the building. In the northern one, it is still possible to see a hint of a curved wall. In the southern one there is a small conduit, triangular in cross-section and connected to the outflow system. The building still has two openings, which are at different levels; the higher one is in the northwest niche, and the lower one is in the curved section between the two northern niches. In addition to these openings, there is, as previously noted, a hypogean corridor that goes off at an angle, penetrating the building from the east through an arch faced with bricks sesquipedales. Just by the arch, at the level of the lower ellipse (level b), there are two small walls in brick, which extend the lines of this corridor into the building.

Upon analyzing the various data, we see that there is no shortage of evidence that contradicts Lugli’s hypothesis, and which can be summarized as follows: location, orientation and functional characteristics.

The first is the simplest and may also be the least binding of the three arguments; nonetheless its implications must be carefully considered. The building, which belongs to the same construction phase as those immediately to the north, is an essential component of them. Building 53 was accessed by means of a door communicating with room 49, which was probably a room for passage. This door has been interpreted by some observers as a window because of its current level, which is effectively above the present floor. This misunderstanding has been caused by the fact that in all the heated rooms the suspended floors no longer survive, as mentioned earlier. Therefore, the present floor level corresponds to the subfloor upon which stood the small piers that supported the use-level floor. The users of the thermae would walk on this level, which coincides with the floor of the north opening of building 53 and is therefore undoubtedly a door. In this way, the first of Lugli’s arguments is refuted, as he did not recognize any access other than through the hypogean corridor.

Furthermore, we should consider why the builder would have created such an intentional difficulty in the construction of this building by rotating it slightly and thereby putting it out of alignment with the rest of the complex. If he did not do it by mistake (which is the only logical conclusion), the orientation must have been of critical importance. This is comprehensible only in connection with the purpose of the building. We can immediately deduce from the plan that the axis of the building is aligned to the north and that it was constructed in relation to the cardinal points. It is protected to the north (the coldest side) by the presence of other buildings, but completely open on the other three sides, in order to exploit the maximum exposure to the sun. To the fish the orientation of the building certainly would have been inconsequential, but to an architect intent on obtaining the optimal results from a special construction, this aspect would have been paramount.

The functional element is the most evident, however, and it is truly surprising that it was not recognized earlier. First of all, the hypogean corridor, which goes directly into the center of the room, is nothing other than the service area for heating the building itself, in accordance with the usual hypocaust system. The extant levels indicated by a and b (fig. 19) correspond respectively to the subfloors, paved in brick, on which stood the small piers for the suspended floors. These are at two separate levels because inside the building there was a high step of the type commonly found in both laconica and frigidaria of circular plan. The internal elliptical wall supported the upper floor, positioned at the same level as the north door. The rectangular niches, with their sloping floors and their flues, served to circulate the hot air at the upper level, and to transmit it into the walls, which were probably originally equipped with a hollow space (intercapedo). The niches were the unifying part of the system, which otherwise could not have functioned. The two projecting avant-corps, placed to the north and south, supported the steps that gave easy access to the lower level; at the same time, the southern avant-corps contained conduits, as it is likely for the northern one, although no decisive evidence is visible at present (see ). The two small walls flanking the portion of the corridor that entered the building had the double function of supporting the floor more solidly than the small piers could have done at that point, and of preventing the loss of heat. The only remaining question is why an oblique corridor was created. It is possible that this, too, was a device designed to maximize heat retention as much as possible, since the temperature within the building had to be maintained at a constant level, avoiding unexpected drafts and sharp changes in temperature.

It is now clear why such a building required a completely autonomous service system, although this does not mean that we must postulate different phases in relation to the rooms immediately to the north. The semicircular niches, of which the northwest one certainly had a window, were covered by an apsidal semi-dome. This semi-dome did not form an extrados, but was buried inside the masonry of the corner. The geometrical unity of the original ellipse, on which an elliptical vault rested, was thus reconstituted at the upper level. The vault, in turn, could have supported a false dome, which allowed the hot air to rise to the top. The vault itself may have been sunk inside a compact block, protected by a pitched roof, but a lighter and cleaner solution would have been a vault with extrados, reinforced up to a certain point with a stepped feature and possibly equipped with a central oculus. This second hypothesis would affect the dating, however, putting it back to at least the time of Hadrian.

The room was undoubtedly very well heated and possibly supplied with water, although not furnished with a true alveus. The lower level may have constituted a pool, presumably lined with marble panels, as was the upper step. In fact, an outflow hole has been identified in the south avant-corps, but since no fragment of this floor has survived, it is absolutely impossible to make a definitive statement. It is probable that the building’s primary purpose was for sudationes, but it is also possible that the lower floor was flooded occasionally to create a heated swimming pool. In this case, however, we have no traces of the possible methods used for direct heating of the water. Usually, in fact, these intensely heated buildings were supplied with cold water in very small quantities, as it was only used to produce steam.

Although such buildings are fairly common in the Roman architecture of the imperial period, the originality of the planning in the Licenza example is extraordinary. The only comparanda, based on geometrical shapes similar to the ellipse or the oval, are the laconicum in the Forum Baths at Ostia, also sited for the best possible exposure to the sun’s rays, but without the corner niches, and the irregular room, almost octagonal and also supplied with a high-temperature heating system, in the so-called Heliocaminus Baths at Hadrian’s Villa. On quite a different scale, but vaguely similar in their basic schemes, are the twin rooms that have also been identified as laconica in the Baths of Caracalla and the Baths of Diocletian in Rome. One might object that among these three are public buildings and one of them is part of the villa of the emperor Hadrian, who is known for his experimental architecture. Furthermore, the earliest of the four examples cited dates to the period of Hadrian. Although it may seem anomalous, the building at Licenza has every right to be included in this innovative series, and its inclusion could have an impact on the question of the identity of the owner of this villa at the time when building 53 was constructed.

In the complete absence of accurate dating, it must be remembered that circular buildings of this type were quite common from the first century A.D. on, and that they continued to be widely used, with ever-changing variations, through the third and fourth centuries, although they were different in size and purpose. We would not be far from the truth in proposing a date between the end of the first century and the middle of the second century A.D., with a preference, however, for the later limit, that is to say, the age of Hadrian and Antoninus Pius, for the reasons previously stated.

The building was subsequently reused, with the creation of walls at various levels; a hypogean room that was accessible from the former service corridor was also built, in which were found many human bones.

The whole complex must have received water from a small aqueduct coming from the west, of which no trace has been found.

During the recent excavation, it was discovered that several lead conduits reached the building near Area 50. Together with those still in situ and with others that were found and removed during the 1911-1914 excavations, it is possible to reconstruct the principal outline of the water system. The drainage system is better preserved, for reasons linked chiefly to questions of level. In figure 20, the two systems are shown in different colors. When there is uncertainty about to which system a conduit belongs, it is marked in a third color, while a dotted line indicates a conjecture.

The natural topography of the site conditioned the positioning of the original drain. Its principal axis (i) runs north-south on the outside of the west side of the quadriporticus, precisely at the point where the two slopes naturally meet. Clearly, the existing topography was exploited and regularized, imitating the path of the surface drainage. The conduit, at present open to the air for a considerable length, averages 60-70 cm in width and must have had a covering that was ogival in cross-section, as is visible further south in the unexcavated portion. There, two channels join from the east. The first (d-f-h), which links the zone of the impluvium and that of the fountain, flows at an angle into the principal system near the southeast external corner of room 34, clearly respecting this construction in the course of its alteration. The second channel (o), following an oblique line in order to exploit better the natural slope of the terrain, starts at the sump of the pool (25) in the center of the garden and, going in a southwest direction, joins the large drain about halfway along its length. Channel d-f-h, which crosses the western part of the residential block obliquely, and then aligns itself on the inside of the perimeter wall, is in fact the product of several distinct building phases. Before the incorporation of the group of rooms 32-34, the conduit left the outside of the buildings at exactly the point (g) where it shows a change of direction, which was necessitated by the construction of subsequent buildings (cf. fig. 10 and comment on photograph AAR 2721, no. 8). The drain was thus forced to follow the inside of the perimeter wall closely for the stretch (h) to the east of room 34. Since the covering of the conduit at this point obliterates the portion in opus incertum of the west perimeter wall, it is clear that at this time, or perhaps earlier, the level of the quadriporticus and of the garden was raised. On the other hand, the segments h-f seem to have been restored, since they are rather different in shape from the length i. A water drainage system for the residence must have been planned from the beginning of the construction, but the channel visible today is different from conduit i, which we know with certainty was the principal one and the most ancient. Drain f-h, in fact, has a wider cavity and a vaulted ceiling, ribbed with a series of brick arches, and part of its lining is preserved. Also, on account of the direction of this conduit, which was linked to the construction of the fountain, we can suggest that it was built in the same period, or rather adapted from the pre-existing system and restructured. The sewer f-h received water from at least three drains: conduit d, of which a length of lead fistula was found in the 1911-1914 excavations (Lugli 1926, col. 582); the drain of the fountain in Area 8, through outflow hole k; and probably rainwater also, through the outflow j carved in a limestone paving block. The channels from the west (l and m with covering a cappuccina), which were linked to the thermal buildings and inexplicably considered to be water pipes by Lugli (col. 549), and the outflow for the latrine (n), also joined the principal outflow conduit (i). In l was found a large lead fragment of a tank (a). By l, near the southwest corner of room 41, there is an outflow hole in a marble paver (w) and, connected to the conduit, is also a rectangular manhole. Unfortunately, it is impossible to understand how much of this structure is due to restoration.

We can also hypothesize that conduit q, which in its turn took the sewage of p1 and p2, joined the main system further south. We do not know the route taken by r.

Exactly at the point where the last stretch (h) of the conduit of the residence flows into the main system (i), the drainage channels relating to the hydraulic equipment of rooms 33 and 34 (x and α) must have also joined. On the southeast corner of 34, there is still in situ a length of lead fistula (α), at the point where there is also preserved a lead facing panel, inserted between the wall and the cocciopesto coating.

The arrangements for supplying water are less well preserved; there are some lead fistulae, of which at least four series are known, with inscriptions. In one case, it appears that the water pipe simply consisted of a small channel, with a brick bottom (s), which flowed from the west into room 34. We do not have sufficient traces, however, to reconstruct the path of the water required for calidarium 33. An offset along the east wall (MSU 10002) of this room joins two lodging holes created respectively in the lateral shoulder of the apse (MSU 10054) on the north and in the avant-corps (MSU 10051) to the south. It can be hypothesized that the offset partly supported a fistula, most likely belonging to the drainage system.

Water must have arrived at the fountain of Area 8 by way of the lead conduit (y) still in situ under the mosaic of room 16. Fragments of fistulae are recorded by Lugli; one of these (e), probably not in situ, was found at the west end of the so-called veranda 13.

There is more documentation on the thermal areas further to the south. The recent excavation, in fact, brought to light three distinct conduits (t, u, v); the first (t) served the frigidarium, while the other two led to the heated rooms, crossing the structures by way of specific recesses (as in MSU 10058). It is very likely that conduit u was linked to fragment z, which was set into a limestone threshold slab.

For the laconicum, there is no evidence for the water supply, if we exclude the fistula (c) that Lugli mentions was incorporated into the curved wall (attributed by him to the drainage system), and hole (p3), no longer visible, in front of the north avant-corps.

A fistula was found on the site by the Baron de Saint’Odile and another one by some peasants, in the period when de Chaupy took interest in this area. Unfortunately there are not sufficient data to locate exactly their find-spots. The first carried the inscription M BURRUS and the other TI CLAUDI BURRI. It is probable that the first of the two, however, if not both, came from the area of the laconicum, which at that time was the only building partially visible. Baron de Saint’Odile had also discovered underground chambers, which may well be identified as the service corridor of the laconicum.

No hydraulic system is known for Area 55, where there may have been a fountain.

To help with consultation and to aid in understanding the plan, a list of the individual elements marked on the plan follows: a) Wall of furnace or other type of reservoir, found (not in situ) during the excavations of 1911-1914 near the drain l; inscription CAESERNIUS–LUCERNIO FEC (Lugli, col. 581). b) Length of lead fistula (1.70 m long) consisting of two tubes joined together, which Lugli reports was the drain for room 34, emptying into i; one tube carries the inscription P OSTILI FIRMINI (Lugli, col. 581); at the Museum of Licenza two fragments are preserved (inv. SAL 00403242 and 00403243), both of which also carry the same inscription (cf. Bruun, ). Frischer, in a personal communcation, notes that Lugli is in error (see Pasqui’s Catalogue G, no. 2 and no. 4, in Frischer ) and that this pipe was found somewhere in rooms 44, 47, 48, 49, or 51. For the drain of room 34 see α. c) Three fragments of the same fistula, one of which has the inscription C IULIUS PRISCUS FEC, found inside the east wall of building 53, connected to the drain p2 (Lugli, cols. 581-582; inv. SAL 00403241). d) Fragment of small fistula, which formed the last part of the drainage conduit of the so-called impluvium (Lugli, col. 582). e) Fragment of fistula (diameter 7.5 cm), at the end of which is fastened another fistula (diameter 3 cm), found probably not in situ (Lugli, col. 582). f) Drain conduit, in masonry of unworked rough limestone; the covering is ribbed with a series of brick arches; most likely it was rebuilt during the life of the villa. g) Arch on the inside of the wall; corresponds to the exit point of conduit f in phase I (cf. fig. 10 and , no. 8). h) Drain conduit in masonry; constitutes the connecting stretch between f and i, constructed when f could no longer issue forth via g. i) Drain conduit in masonry; constitutes the main axis on the line of the natural compluvium, on the outside of the so-called perimeter wall. In the first phase, it probably extended northward and joined tract f at the point g. j) Limestone paver with outflow hole for draining water into drain f; perhaps gathered the discharge from a downpipe. k) Outflow hole of the brick fountain of Area 8, for discharging into the sewer f. l) Drain conduit in masonry, covered a cappuccina; constitutes the outflow for basin 42; probably received also the discharge from room 41 (of a labrum?, cf. w) and also perhaps of those rooms directly to the south, to then discharge itself into the main conduit i. m) Drain conduit in masonry, covered a cappuccina; used as outflow pipe for tank 37; it originally served the so-called Republican atrium (see Camaiani et al., , activity 3). n) L-shaped drain conduit in masonry, relating to the outflow system of the latrine. Heavily restored. o) Drain conduit in masonry; constitutes the outflow for the pool 25, with which it is connected by way of a drainage sump. It follows the natural slope of the terrain, thereby reaching the main conduit i obliquely, into which it flows. p) 1. Small conduit, triangular in cross-section, incorporated into the south avant-corps of building 53. It is possible that it held a small fistula. It presumably emptied into drain q. 2. Lugli refers to a rectangular hole for the outflow, positioned lower down than p1, to the east in the corner between the south avant-corps and the elliptical wall. Fistula c must have flowed into p2 (Lugli, cols. 553 and 582). No trace of it remains. 3. Lugli also mentions a hole (for clean water?) in front of the north avant-corps (Lugli, col. 553). No trace of it remains. q) Drain conduit in masonry; took the outflow from building 53 and presumably continued until it joined up with the main conduit i. Lugli remarks that this conduit was covered a cappuccina (Lugli, col. 549, n1, letter p of his attached plan. There are, however, errors in the attribution of the letters). r) Recess for a conduit, presumably a fistula, which runs on the outside of the north side of building 53. It is difficult to establish whether it is a water-supply pipe or a drainpipe. s) Small channel bringing water for room 34 (see Camaiani et al., , activity 19). t) Fragment of fistula relating to the system for bringing water to basin 37 (see Camaiani et al., , activity 11, SU 457). Perhaps connected to the fragment further west (SU 347; but cf. also u). u) Stretch of lead fistula (VH 085, 086, 087; inv. SAL 114587, 114588, 114589) composed of three separate elements soldered together for a total length of ca. 5.20 m. One of the tubes has a joint for another emission channel (see Camaiani et al., , activity 34, SU 241). We cannot exclude the possibility that the fragment of SU 347 should be attributed to this conduit, and not to t. v) Stretch of lead fistula (VH 121, 122, 207, 208, 209, 210; inv. SAL 114585, 114590, 114591, 114546, 114547, 114548) composed of five parts soldered together, plus fragments, for a total length of more than 6 m (see Camaiani et al., , activity 12, SU 331 and 240, and , SU 355). There are on all five identical inscriptions (C - IULIUS - PRISCUS - F), after one of which is the indication III. w) Marble paver, with central outflow hole in the form of stylized petals. x) Hypothetical conduit, which crosses the east avant-corps of room 33 and joins the main drain i. Perhaps connected with the passage of a fistula along the east side of room 33, attested by the offset along the east wall, and by a protected passage in the east embrasure of the apse. y) Stretch of lead fistula, in situ, under the mosaic floor of room 16. z) Fragment of lead fistula, in situ, preserved under the wall MSU 10019 and set into a stone block, probably a threshold slab. Perhaps to be attributed to conduit u coming from the north. α) Fragment of lead, in situ, partially covered by cocciopesto, beside a drain conduit; southeast corner of room 34, flowing into i.

Like any architectural complex, the so-called Horace’s Villa underwent a series of transformations through the ages, with additions and rebuilding of what already existed. The scant masonry remains and the modern reconstructive restorations severely limit our ability to reach an interpretation that explains in detail the sequence of these changes. In spite of that, we have attempted to develop a plan that shows the principal phases of activity on the site, even if it is not always possible to relate all the diverse elements to one another. The following observations are based on the stratigraphical data, where available, as well as on structural and functional considerations. We also refer to all earlier analyses, both of the documentary material and of the physical remains, in an effort to offer a synthesis of the conclusions reached.

The occupation of the site in antiquity is attested from the fourth and third centuries B.C., as some residual material from the lower strata of the excavation in Sector IV.2 demonstrates. It is possible that these finds were washed into this particular zone. The site rises on the saddle of a hill and is characterized by various differences in level: a predominant and constant slope that descends from north to south; a slope that slants gradually from east to west; and a more accentuated slope from west to east. The point at which these last two join, on an east-west axis, constitutes the natural meeting point for the draining of surface water and coincides almost exactly with the west side of the quadriporticus. This difference in level diminishes to the south.

Period I

At the site, the first constructions that we have any knowledge of can be classified in three distinct groups, which, in their present state, appear to have no relationship to each other. It is difficult to give an overall picture of the first phase of the life of the building, which cannot even be described as such, and therefore the structures in question are here indicated generally as “pre-existences” (fig. 21). We lack further deep excavation data, especially under the residential block as it is visible today, which could supply valuable clues as to the design of the first building. We can recognize, however, that the elements identified point to a commitment to build a structure of substance. Referring to the plan we can see these features marked and can address some of them in detail. • The long wall in opus incertum (MSU 10068 and 10080), of which only a few centimeters survive, later was reused and in part overbuilt with an analogous wall in opus reticulatum. In the first phase of settlement, it could only have served as an enclosing wall, as there is no data to enable us to relate it to porticoed structures. • The remains found underneath room 12 and belonging to a basin that was adapted to the natural slope, as the terracing at that point had not yet been created. • The group of structures to the west that were subsequently incorporated into the thermal buildings; in addition to the Republican atrium, we must note other structures in opus incertum (MSU 10004 and MSU 10006, cf. figs. 56-58 and 42-43), which can hardly be attributed to the later interventions. Also to be considered as “pre-existences” are the fragment of flooring in opus spicatum to the east of MSU 10004 and the threshold slab subsequently obliterated by the later MSU 10019.

For all of these structures, which were not necessarily built at the same time, an indicative chronological span, which runs from the end of the second century B.C. through the entire first century B.C., can be proposed.

Period II

The phase that received most attention during the excavations of 1911-1914 was undoubtedly the one regarded as being Horatian, a simplification due to the presence of masonry in opus reticulatum. Because of the modern demolitions and reconstructions, however, the first feature identifiable in its entirety is the large elongated rectangle created by the construction of the residential block and of the quadriporticus that surrounds the garden (fig. 22). It should be recalled that its image, too, is partly altered in view of the now certain extension of the masonry walls northward. There was considerable building activity at this time, correcting the slope on the side to the north and thereby creating the terrace of the living quarters. In the garden and in the quadriporticus, levels were raised as well. In the garden the large pool 25 was created, which, while not datable, can plausibly be included in this phase.

At this time, it is probable that Area 8 of the residential zone had a peristyle, with an arrangement of the spaces different from what we see today (fig. 23). The living quarters must have already had a drain, which was linked to the main system, emerging on the outside of the perimeter near room 21. It is probable that the residence had an upper floor. The location of the entrance is not known, but it could have been either on the north side, of which we know nothing, or more likely, near the present northeast corner, by rooms 26 and 27. In this case, there would be two visual axes and circulation routes set at right angles, a feature that is well-attested to in other complexes (fig. 24). The fact that these rooms (26-27), which presumably served as an entrance, jut out in relation to the compact and regular rectangle, is not an obstacle to such an interpretation, especially if we take into account the fact that there may have been a modification during the construction. More problematic is the fact that we know very little about these original masonry structures; even if we accept the indications of the restoration and of the mosaic floors in situ, we still must use great caution in formulating such a hypothesis about the entrance in this phase. The dating of this work can only be tentative, as there is absolutely no stratigraphic data, and can therefore only be guardedly attributed to a chronological span that extends from the Augustan age through the end first century A.D.

Period III

The intervention that immediately followed consists of the addition of the first thermal complex with rooms 32-34 (fig. 25) and the consequent adaptations to the inside of the residential rectangle (definition of the rooms 19-21?; deviation of the drainage conduit; creation of portico 35). The later date of the structures 32-34 in relation to the living quarters is not proven by the building techniques used, but by the stratigraphic relationships, the planimetric arrangement, and the hypothetical presence of other features from the same phase as the residential block. The use of brick by itself is not a sufficient indicator of a distinct phase, whereas the determining factor is the stratigraphic relationship between walls. In fact, the east wall of the calidarium 33 (MSU 10002, in trimmed roof-tile brickwork) was built right up against the west perimeter wall of the villa. This additional wall covers not only the facing, but also the foundation of the perimeter wall, because in order to build room 33, the level had to be lowered to accommodate the hypocaust system. Exposed foundations, in fact, very often indicate a different phase or a change in the planned project post operam.

Furthermore, we cannot ignore the very position of the group of buildings that undoubtedly project forward from the compact block created by the quadriporticus and residence. Prime consideration must also be given to the arch that opened in the foundation of the west perimeter wall (cf. fig. 10 and , no. 8). This was the point where the drain conduit first exited the quadriporticus-residence rectangle, running alongside the wall for its whole length. The incorporation of rooms 32-34, with the required lowering of the level, made it necessary to re-route the drain to the inside for some distance. It then exited immediately to the south of room 34. The constant and almost exclusive use of roof-tiles in the brick facings of this period is indicative of a dating no later than the middle of the first century A.D.

Perhaps the great frigidarium (37-40) was built only shortly afterwards, as an addition to the first thermal complex, exploiting the pre-existing structure of the Republican atrium, with a considerable raising of the level. It is impossible to say whether the other important works were carried out at the villa at the same time; in the absence of reciprocal relationships, they can only be considered as belonging to one period, without division into sub-phases, that extends from the middle of the first century A.D. to the middle of the second century A.D.

There was then (phase IIIB?, fig. 26) a further extension of the thermal complex to the south, with an arrangement of heated chambers, architecturally notable (cf. the laconicum). At this point, there were two complete thermal complexes, which could have functioned simultaneously and independently. The higher level of the suspended floors, required by the additions to the bath complex, produced another terrace on the west, more or less at the same level as the residential block.

At this time, the residence was altered by the addition, in Area 8, of a monumental fountain da centro, which completely changed the appearance of this zone. The features built in the east wing of the quadiporticus are also to be attributed to this period, probably for structural reasons and not just aesthetic ones, with the addition of a fountain (or some scenographic setting), whose center is perfectly aligned with the axis of the piscina of the garden.

Period IV

The last phases of the life of the villa cannot be analyzed in detail, except where the stratigraphy is preserved (see Camaiani et al., , their Periods III-V). New construction becomes less frequent (fig. 27), and there is almost always an attempt to readapt an existing building according to the needs of the moment. Materials from the earlier structures are reused, the thermal complexes fall into disuse, and the suspended floors are wholly or partly dismantled, resulting in the creation of new rooms on new foundations. The modern partial demolitions, undertaken in order to uncover the presumed Horatian phase, render the layout and the stratigraphic sequence of these new phases of the life of the villa almost unidentifiable. As it is now, the residential area does not offer any elements attributable to these later stages, when some parts were even used for burials (from the fourth century onward). The laconicum, which after a period of abandonment was perhaps the most serviceable building remaining, was probably reused as a small church, in which was created a crypt, designed for burials and reached by way of the old service corridor of the hypocaust.

Conclusions

In spite of the minuscule quantity of the remains it is still possible to attempt to sketch some outlines of the plan, at least for the first phase of an architectural complex that can be evaluated as a single unit. If we include in our calculations its north side, that of the so-called veranda, the quadriporticus forms a rectangle of ca. 290 x 145 Roman feet; the proportion of 2 to 1, between the length and the width of the porticoed rectangle around the garden, is evident and the garden is therefore formed of two squares. In reality, the proportions of the quadriporticus were slightly modified during the carrying out of the project; the south wing is 3 meters shorter than its northern counterpart. This situation was evidently created by the deviation of the east wing, which was almost certainly caused by the terrain that required the modification of the plan in the course of building.

The module of the “two squares” also constitutes a guideline for the incorporation of the large pool 25. The south wall of the piscina is set along the median latitudinal axis of the quadriporticus. A visual axis is also identifiable on the east-west median of the piscina, which coincides perfectly with that of the fountain incorporated into the east wing of the quadriporticus. It is more difficult to establish the internal relationships for the living quarters, because, as we have seen, we do not know for certain the extent of the space it occupied. We could perhaps hypothesize that it occupied the module of one square (namely, ca. 145 x 145 Roman feet).

Activity 1: Republican strata in room 38

The stratigraphy associated with this phase provides little ceramic material; it is chronologically homogenous and attributable to the late Republican age (second to first centuries B.C.).

SU 874: There are fragments of a black-glazed pottery base, Morel 300 type, belonging to an unidentified open shape, generically datable between the third and second centuries B.C., and an example of Sutri A60 olla, dating to the first century B.C.

SU 875: The only recognizable fragment is that of an olla with almond-shaped rim, typical of the late Republican period.

Activity 2: Construction of the quadrangle (atrium with impluvium)

Little significant ceramic material emerged from the strata pertaining to this activity, just a very few fragments of common pottery sides, which are not attributable to specific shapes or types. We have more evidence as far as the stratigraphy of the impluvium construction is concerned, in particular from SU 644, which has produced a fragment of cup foot of Italic terra sigillata, of unidentified type.

Activity 3: The drain construction

The SU connected with this work did not produce ceramic material.

Activity 4 : Preparation of the floor surface of the atrium

The extremely sparse ceramic material produced from the stratigraphy here is largely unidentifiable. Only SU 873 yielded some thin-walled ceramic fragments, unfortunately not attributable to specific shapes or types.

Activity 5: Strata of material raising the level of the quadrangle

The strata connected with this activity yielded a certain amount of ceramic material, datable mainly between the Augustan and Flavian periods, with a small nucleus of residual material (consisting mainly of black-glazed pottery), dating to the late Republican period.

SU 629: The pottery material is both very scarce and in a very poor state of preservation. The surfaces have been washed away and therefore it is not identifiable.

SU 646: The only identifiable fragment is an example of an Albintimilium 109-type pan with rim a tesa, dating from the Augustan period, together with a fragment of a thin-walled cup, which has no decoration and is therefore chronologically unidentifiable. In the stratum, among the residues, there is also a fragment of a black-glazed ceramic cup rim, Morel 2615 type.

SU 631: No important material was found (only pot walls). The single recognizable fragment belongs to a thin-walled ceramic cup.

SU 860: One of the few identifiable fragments is the rim of a common kitchenware olla to which can be added a rim and wall belonging to a thin-walled cup, decorated à la barbotine, of the Atlante 1/39 type, dating from the Augustan Age. In this level, as in the preceding one, the residual materials include some black-glazed pottery fragments, among which is the rim of a Morel 2900-type cup.

SU 865: The material is generally attributable to the early imperial period (Italic terra sigillata, thin-walled ware, etc.). In particular, some fragments of everyday kitchenware, such as the Curia 101 pot and the Curia 189 olla, can be attributed to the Flavian period.This chronology was also suggested by a fragment of the wall of a vessel of African red slip “A” ware, from a piece of unfortunately undetermined shape.

SU 870: There is one fragment of a plate of Atlante XIX-type Italic terra sigillata, datable to the first decades of the first century A.D., and a fragment of a Curia 101-type pot rim, found also in SU 865 (see above).

SU 1006: No identifiable material was found, just some common pottery wall fragments and fragments of common kitchenware.

The presence of joins between fragments from different strata and their typological and chronological homogeneity confirms the uniformity of the intervention and the rapid superimposing of the various levels of earthing-over (necessary for the creation of the mosaic floor of the frigidarium; cf. activity 6) in a fairly short span of time, perhaps to be placed in the first half of the second century A.D.

Activity 6: Creation of the frigidarium mosaic

The only stratum with identifiable ceramic material (SU 604) yielded just one important fragment, belonging to a Curia 72 olla of the Flavian period, which is therefore to be considered the terminus post or ad quem for the dating of the activity. The other pottery and non-pottery material does not give us any further chronological indications.

Activity 7: The SU connected with this intervention did not produce ceramic material. Activity 8: Material raising the floor level of room 50

SU 288: The most recent material is a cup of African red slip “A” ware, together with fragments of amphorae of the Dressel 2-4 and 9 types. Fragments of Italic terra sigillata were also found, including the rim of a Conspectus 34.1-type cup, dating to between the Tiberian and Flavian periods, and an unidentifiable foot fragment, with a seal in planta pedis: MA [---] (VH 076, fig. 23).

SU 316: This layer contains common table- and kitchenware from the second half of the first century A.D., presumably from the Flavian period. Among these are a Curia 63-type bottle, cups of the Albintimilium 303, 315, 32-type, and Curia 105 and 148-type ollae. The residual material is scarce, particularly the thin-walled and Italic terra sigillata pottery, of which one fragment (VH 079) bears the seal C […].

SU 319: This stratum unfortunately does not contain diagnostic material (only fragments of pot sides). A terminus post quem is established, however, by the presence of some fragments of plastering dating from the Augustan age (see Mols, ).

SU 327: It contains three fragments of rims and a wall belonging to a single example of Iberian amphora, Beltran II A type (cf. SU 333, activity 12), together with one fragment of Italic terra sigillata wall. Among the marble fragments from this level, the presence of several relating to some small fragments of bardiglio is probably datable to the first century A.D. This marble is generally not used after this period, although it is occasionally recycled.

SU 337: There are two fragments of Italic terra sigillata wall, which because of their fabric and glazing are perhaps datable to the period of Augustus or Tiberius. The same chronology seems to be confirmed by the other pottery elements found in this level, in particular some fragments of Albintimilium 37 and 41-type ollae attributable to the late Republican period.

Activity 9: The construction of the basin (37)

The SU relating to this work did not produce ceramic material.

Activity 10: The construction of the apse in room 51

Various ceramic fragments were recovered from the excavation of the fill (SU 313) of the trench in the foundations of the apse (SU 312). The material dates to the period between the third quarter of the first century and the first decades of the second century A.D. Fairly characteristic of this period are the Curia 105-type saucepan, the Albintimilium 175-type lid, and the olla, probably of local production, of the type VH-CM/14-15 (fig. 14); for the dating, cf. activity 15). The stratum also contains some pieces datable to late antiquity and the early Middle Ages, such as the Albintimilium 80-82-type olla and the Albintimilium 305-type small cup.These last are certainly adulterating elements, having ended up in the soil through the upheavals of the numerous excavations in the twentieth century.

Activity 11: The installation of a lead fistula in room 37

The SU relating to this work did not produce ceramic material.

Activity 12: The installation of a lead fistula in room 50

SU 318: Among the few pottery elements were found a fragment of Italic terra sigillata cup rim of the Atlante XXV, 7 type, datable to the Augustan/Tiberian period; three fragments of thin-walled pottery cup, of an unidentified type; and one fragment of Italic production amphora neck (perhaps to be attributed to the Tyrrhenian area, because of the characteristics of the clay).

SU 333: There are two joining fragments of Beltran II A-type amphora rim (late Augustan Age – second century A.D., cf. SU 337), and one fragment of an amphora handle of African production, which cannot date from before the early imperial age on account of its morphological characteristics.

Activity 13: Construction of the walls of room 50

The one stratum containing pottery material (SU 266) offers only walls of amphorae of medium-large dimensions, probably produced in Africa and generically attributable to the early imperial age.

Activity 14: Construction of the mosaic and of the plastering of the frigidarium wall.

The only chronological evidence comes from the excavation of SU 834. Here was found a fragment of amphora, which, because of the characteristics of its fabric, is classifiable as Iberian production of the second half of the first-second century. A terminus post quem, which at first sight would seem to confirm this dating, is the presence in the stratum of some elements of a sectile pavement, triangles in slate and white marble, probably coming from the demolition of a floor of the early imperial period (see below).

Activity 15: Construction of the floor level in room 50

SU 261: Notable among the ceramic material are two fragments of Beltran II A-type amphora rim (of which one joins another fragment from SU 327: see above); one fragment of Curia 97-type olla; and one fragment of Albintimilium 175-type lid.

SU 274: Among the important fragments are the rim of an olla with out-turned and slightly thickened rim, short neck and shoulder marked by two parallel grooves; it is the VH-CC/1 type (fig. 2), datable from analogous pieces to the first century A.D. There are also two examples of rims of ollae of common pottery of the VH-CM/14-15 type (fig. 14). This shape, characterized by a rounded rim, tapering on the upper side, short neck, shoulder marked by a small listel, and cylindrical body, has been attributed to late antiquity on the basis of very few comparisons. Its presence in strata linked to activities datable to the first decades of the second century allows us to put back considerably the chronology thus far accepted.

SU 321: Among the everyday kitchen-ware are some forms characteristic of the early imperial period, in particular one fragment of a saucepan, with rim a tesa, of the Albimintilium 109 type, from a stratum dating to the early Augustan Age, and one fragment of Albimintilium 175-type lid, a shape somewhat characteristic of the Flavian period. The residual material includes one fragment of black-glazed pottery and several fragments of Graeco-Italic amphora. Among the marble finds, note the prevalence of types typical of the early imperial period, such as white Luni marble and bardiglio.

The presence of joins with fragments from SU 327 (cf. activity 8, above) provides the archaeological proof that the various activities belong to a single building phase.

Activity 16: Strata of material used to raise the level of room 50

SU 211: This has yielded one single example of pottery, a cooking-bowl rim attributable to the early imperial period.

SU 229: Among the few identifiable pottery fragments are two rims of pottery cooking lids, with raised edges (VH-CC/51 and CC/56 type: fig. 11), characterized by a shape typical of the early imperial age, and particularly of the Flavian period. The rim of an Italic terra sigillata cup of the Atlante XXV type, datable to between the first decades of the first century B.C. and the Tiberian period, points to the same chronology, as does a fragment of a thin-walled cup of an unidentified type, decorated à la barbotine.

Activities 17-21: No significant pottery elements were found.

Activity 22: No significant pottery elements were found. Activity 23: Construction of a series of walls inside room 40.

The pottery material was all recovered in the excavation carried out in the foundation trench SU 647. Among the identifiable elements are an example of a Dressel 2-4-type amphora; two joining rims of Dressel 1-type amphorae; several fragments of Italic terra sigillata; and a handle and two pot walls belonging to a single example of a thin-walled cup of unidentified type. The early imperial chronology of this group of elements leads us to hypothesize that the foundation trench was filled with the excavated earth itself.

Activity 24: No significant pottery elements were found. Activity 25: Strata of rubble inside room 50.

The SU levels relating to this work are strata of material used to raise the level of the floor. These consist of soil, probably taken from the vicinity of the villa, unloaded in various heaps close to one another, as is shown by the various joins found between ceramic fragments coming from stratigraphically distinct units.

SU 211: The only identifiable fragment is a sample of bowl rim, attributable to the early imperial age.

SU 255: This level contains a fragment of the rim of a pottery cooking-vessel (VH-AC/2: fig. 21), whose shape is not exactly analogous to the common categories of local production but is quite near to that of the examples of African red slip “D” ware. Note also the VH-AC/3 type (fig. 21), whose shape recalls that of the Ostia 216 type and is similar to an example from Lugnano. There are also examples of African production among the kitchenware, with one fragment of a Hayes 196-type saucepan (end of the second to the fifth century). Together with this body of later material, there is also a nucleus of residual elements, including thin-walled pottery (some with decoration à la barbotine) and numerous fragments of kitchenware. Significant among these are some rims of pans a tesa, typical of the early imperial age; the VH-CC/41-type vessel (fig. 8); a type of olla with rim thickened and slightly out-turning (VH-CM/10: fig. 13); and another, with its rim thickened and folded outward, flattened on its upper surface and rounded on its exterior, with cylindrical body (VH-CM/13: fig. 13).

SU 262: This stratum contains no typologically important material (a few fragments of amphora walls and handles, of unidentified production, and some table- and kitchenware walls).

SU 271: The material is exclusively residual. There are two exemplars of African red slip ware cup rims (one of which joins a fragment from SU 272: see below), of the Atlante XXV and XXVII types, datable to between the first decades of the first century B.C. and the Tiberian period. Among the common pottery, there are several examples of bowls, all characterized by the rim, thickened and rounded on the outside, and by the deep body (types VH-CM/18-20: figs. 15-16). This shape has precise analogies at Luni and at Albintimilium. Also note the fragment of a pot rim, type VH-CC/41 (joining a fragment from SU 255: fig. 8). Most of the ceramic material, therefore, can be ascribed to the period between the end of the first century and the first decades of the second century A.D.

SU 272: In this stratum only residual elements were found, dated to the first and second centuries. Important are one fragment of Curia 145-type pan rim; one fragment of African kitchenware rim, Hayes 196 type; and one fragment of Italic terra sigillata cup rim, Atlante XXVII type, which joins a fragment from SU 271 and is datable to the period between the Augustan Age and the second century A.D.

SU 275: This stratum contains a high degree of residual elements. It yielded one fragment of Italic terra sigillata cup, Atlante XXV type (VH-TS/1: fig. 20), datable between the first decades of the first century B.C. and the Tiberian period, as well as numerous fragments of thin-walled pottery (perhaps from the Augustan period) and two amphorae, one of Graeco-Italic production (VH-A/2: fig. 1) and one Dressel 8 (VH-A/6: fig. 1). These are to be placed in a rather broad chronological span, between the second century B.C. and the first century A.D., along with a large quantity of walls, some of which probably belong to examples of Dressel 1 and 2/4. Among the common pottery, several fragments are attributable to the early imperial age, such as the Lugnano 83-type olla and the Curia 124-type lid. The local productions include the VH-CC/4 type (fig. 2), comparable with examples from Lugnano and Luni datable to the first century A.D. On the other hand, two fragments with hollowed rims, which are comparable to the Lugnano 200 type,are probably datable to late antiquity.

SU 283: The class mostly represented is common tableware, among which figures the VH-CM/12-type cup (fig. 13), characterized by a rim of four-cornered cross section and deep central grooving, flattened on top and slightly hollowed on the inside. This has analogies in Campania, at Luni, and at Albintimilium, all datable within the first century A.D. There is also a fragment of VH-CM/14-15-type olla (fig. 14; for the chronology, cf. activity 15). Among the kitchenware, note the Curia 106-type pan and the VH-CC/28 type (fig. 6), similar to the previous one, which has analogies with examples from Cosa, Amelia, and Lugnano. There is also a VH-CC/26 olla (fig. 6), which is comparable to a similar type from Lugnano. Furthermore, the African red slip ware cup rim of Hayes 8 type is also from the same time period, datable to between the end of the first and the middle of the second century.

SU 285: From this stratum we have only material from the early imperial period, clearly residual. Among these finds are a fragment of Italic terra sigillata, unidentifiable thin walls and common pottery, and numerous fragments of amphorae, whose provenance unfortunately cannot be established. The only significant fragment is an example of Albintimilium 109 pan from the early Augustan period.

Activity 26: Abandonment of the upper flat level of room 50

SU 209: Among the few pottery fragments, there is an amphora (VH-A/7: fig. 1), a variant of the Dressel 6 A type, datable to between the second half of the first century B.C. and the first century A.D. From the same chronological range are two fragments of basin rim, of which one is similar to the Hartley 2 type and one with thickened and molded lip, as the VH-CM/24 type (fig. 17). Dating is established by a bowl rim of common kitchen pottery, VH-CC/45 type (fig. 9), with deep belly, rounded rim and shoulder marked with grooving. This is similar to the Curia 387 type and related to the Hayes 63 type, produced in African red slip ware between the middle of the fourth century and the middle of the fifth century.

SU 220: This stratum contains several fragments of African kitchenware, of which the only identifiable piece is a Hayes 196-type bowl lid rim, dating to anywhere in a very long time span. There are two lid rim fragments of kitchenware, which are unfortunately not attributable to any particular type, although the morphology, with thickened and raised rim, would indicate the early imperial period.

Activities 27-29: No significant pottery elements were found.

Activity 30: Excavation and refilling of a trench in room 39

SU 1005: This SU contains a considerable quantity of pottery material, presumably of primary deposition, attributable overall to late antiquity and the early Middle Ages. In addition to two fragments of spathia, fragments of ollae were found of the following types: Albintimilium 71-74 (fourth to sixth century), 202 (beginning of the fifth to the seventh century), and 204 (end of the sixth to the seventh century). The common pottery, probably of local production, seems always to be of a type of fabric characterized by an extremely high level of calcareous ingredients.

Activities 31-34: No significant pottery elements were found, except some fragments of Forum ware (SU 1220, activity 31), dating to the eighth to ninth century. Activity 35:

Partial collapse of the frigidarium structures

SU 433: A certain amount of pottery material was found, consisting of residues dating to the early imperial period. Among these are a Curia 61-type bottle rim and the VH-CC/3-type olla (fig. 2), comparable to examples from Settefinestre and from Cosa.Confirming the dating are one fragment of a small jug base of Forum ware of unidentifiable type and several fragments of bowls of VH-CC/34 (fig. 8) and VH-CM/27 type (fig. 17). These are both characterized by deep hemispherical bellies and rounded rims marked all the way around with a groove, a characteristic shape of the common pottery of late antiquity and adopted in the heavy-glazed production.

Activity 36: No significant pottery elements were found. The only identifiable find (a fragment of white, monochrome majolica of the seventeenth-eighteenth centuries) must be considered to have been interpolated as result of agricultural activities in the last century. Activity 37: Episodes of ransacking and destruction

SU 265: The level contains only pot sides, of which the identifiable finds are one fragment belonging to a casserole of African kitchenware, possibly of the Hayes 197 type (datable to between the end of the second and the end of the fourth or beginning of the fifth century) and a fragment of amphora side, whose fabric shows it to be of the Late Roman 3 production, datable to between the end of the fourth and the end of the sixth century.

SU 277: Among the recognizable ceramic finds there are only residual materials, including a fragment of an Italic terra sigillata cup, Dragendorff 24/25 type (dating to between the Tiberian period and the first quarter of the second century), and an example of a VH-CM/23-type basin, with thickened rim, rounded on the upper side, vertical lip and truncated ovoid belly (fig. 17).

SU 278: The ceramic material recovered is mostly residual. This includes one fragment of an olla (VH-CC/5: fig. 3), generically attributed to the first century, and one fragment of thin-walled cup rim of the Atlante 1/33 type, bearing decorations with beads and petals of the Atlante 117 type, dated between the time of Tiberius and that of Claudius. Among the important fragments, there is also, as part of the common pottery, a group of fragments of ollae, VH-CM/14-15 type (fig. 14) and basins, VH-CM/21-22 (fig. 15). Noteworthy kitchenware includes the pan with rim a tesa, VH-CC/32 type (fig. 7), which has distinct analogies at Settefinestre, from strata of the first half of the second century, together with several fragments of ollae, which are datable to the Flavian age on the basis of comparisons and which are characterized by a fabric very rich in augite and mica (VH-CC/22-23, VH-CC/30-31: figs. 5 and 7). This seems characteristic of the local ceramic production in the late Republican and early imperial periods found in this context. Of uncertain date are one example of an olla with turned-out rim, slightly rounded lip and short neck (VH-CM/2: fig. 12); an olla with thickened rim, flattened on top and bent outwards, separated from the body by a “throat” (VH-CM/11, fig. 14); and a basin (VH-CM/21-22, fig. 15). One example of a censer (VH-CM/33: fig. 18) can be dated on the basis of comparisons from between the end of the first to the third century.

SU 287: This stratum, although datable to medieval times, contains almost only residual material. Finds include a VH-CM/8-type olla, with thickened rim of quadrangular cross-section, rounded on the upper surface and slightly hollowed on the inside (fig. 13), which is comparable to examples from Amelia, and one with short neck and cylindrical, or cylindrical-ovoid, body (cf. SU 278, VH-CM/14-15, fig. 14). Of the same period are some fragments of a pan, with rim a tesa, Albimintilium 109 type (from strata of the early Augustan Age), and some fragments of Dressel 2-4-type amphorae. Dating to a later period, presumably the early Middle Ages, is a fragment of a bowl with the rim simply rounded and grooving on the lip (VH-CM/27: fig. 17).

SU 414: Among the common kitchenware are three bowl fragments, VH-CC/34 type (cf. SU 433, activity 26), and one olla rim (VH-CC/2 and CC/24: fig. 2), which is datable to the late Republican period. Among the coarse ware is a type of pitcher with rim bent slightly outwards, flattened and oblique lip, and short neck, on which a wide handle a nastro is fixed (VH-CM/3: fig. 12). Among the residual material of note are two examples of ollae, datable between the late Republican and early imperial period, and a plate-lid, VH-CM/34 type (fig. 18), with rim thickened and turned outward, rounded lip, and a shallow belly shaped like a truncated cone. Among the later material, there is one bowl fragment of African red slip “D” ware, datable to between the middle of the fourth and the beginning of the fifth century. The dating is provided by two joining bowl fragments and one handle, of closed shape, in Forum ware (VH-FW/1: fig. 22).

SU 434: The residual material yields an example of a rim of Pélichet 47-type amphora, dating to between the middle of the first and the third century. Providing the date are a fragment of common pottery olla, with incised “wave” decoration, similar to examples found in Rome at the Crypta Balbi, dating to the seventh century, and a type of small amphora, probably produced locally, with a slightly thickened rim, grooving on the lip and separated from the neck by a little “throat,” and with a wide handle a nastro (VH-CM/6a: fig. 12); this is found also in other versions (cf. SU 447, below).

SU 447: Here we have the rim of a jug, VH-CM/1 (fig. 12), comparable to some examples from Ostia dating to the first century A.D. Among the cooking ware are one Hayes 196-type plate rim and one Hayes 197-type casserole bottom. Dating is provided by a type of small amphora, probably produced locally, with slightly enlarged rim, grooving on the lip and separated from the neck by a small “throat” (VH-CM/6b: fig. 12).

SU 448: Here there is only a fragment of the spout of an unidentifiable common pottery jug.

SU 450: The dating is provided by heavy-glazed pottery of the Forum ware type (VH-FW/2: fig. 22), found in large quantities (91 fragments in all). Among these are one fragment of base joining other fragments from SU 433 and SU 451 (VH-FW/3: fig. 22), and one cup, almost completely reassembled (there were 28 fragments: VH 047, fig. 25). Among the amphora material, there are one fragment of Keay LII-type amphora (VH-A/9: fig. 1) and one fragment of “carrot-shaped” amphora, Crypta Balbi 1 type (VH-A/10: fig. 1). Among the African red slip ware there are three fragments of Hayes 103-type bowl (beginning to the third quarter of the fifth century) and two joining fragments of Hayes 50-type bowl (third to fourth century).

The fine ware includes a type of small amphora, with a thickened and rounded rim (VH-CM/5: fig. 12), while among the coarse ware were found: one fragment of Lugnano 126-type olla; one fragment of Curia 387-type bowl (from early medieval strata); two fragments of carinate bowls (VH-CC/47-48: fig. 10); and two other fragments of the same type, but of an unusual shape (deriving from Hayes 61B?), possibly datable to the middle fifth to the sixth century (VH-CC/43-44: fig. 8).

SU 451: This level does not contain typologically important material, but the classes present (including red-painted pottery and Forum ware: VH-FW/3, fig. 22) are consistent with those found in the other SU of the same activity, particularly SU 450.

SU 822: This stratum includes one fragment of a rim of a casserole, with straight rim and deep carinate body (VH-CC/1: fig. 2), datable by comparison to between the end of the fourth and the sixth century, and one fragment of a common pottery jug (VH-CC/20: fig. 4), which has precise analogies with examples from the seventh century context of the Crypta Balbi.

Activities 38-40: No significant pottery elements were found.

Activity 41: No significant pottery elements were found. Activity 42: Strata of the collapse of the Roman and early medieval structures

SU 428: In addition to a fragment of Hayes 91A-type vase a listello (mid-fourth to fifth century), there is a type of small amphora, with thickened and rounded rim and handle a nastro attached to the neck (VH-CM/4: fig. 12); this too is datable to late antiquity.

Activity 43: No significant pottery elements were found. Activity 44: Strata of collapse of medieval structures of room 50

SU 269: The only recognizable fragments are one fragment of olla (VH-CC/27: fig. 6) and a pan rim (VH-CC/41: fig. 8), dating to the early imperial age.

Activity 46: Strata of collapse and abandonment in the level plane of room 50

SU 202: This stratum yields a few fragments of African red slip “D” ware (largely pot sides; just one fragment of plate rim, of Hayes 33 form, can be identified (dating to 200-250 A.D.). Included in the common kitchenware is one fragment of olla rim a tesa, of the Lugnano 83 type, found also in other strata (cf. SU 275).

Activity 48: Modern archaeological interventions

Material uncovered in the course of the excavation of these strata is obviously all to be considered residual. Note, however, some finds which are particularly significant in that they are evidence of little-documented phases of the life of the site (especially the late Republican and early medieval periods).

Among the oldest ceramic material is one fragment of black-glazed cup base, Morel 320 type, with a central rosette stamp (SU 231), attributable to the atelier des petites estampilles.

The VH-CM/34-type plate-lid (fig. 18), with rim thickened and turned out, vertical lip, shoulder slightly carinate and shallow body, is datable to the early imperial period.

Notable among the material from late antiquity are an example of olla, with turned-out rim, short neck, and carinate shoulder, with “wave” decoration incised before firing (VH-CM/16: fig. 14), as well as a small jug (VH-CC/19: fig. 4), which has exact analogies with examples from Carminiello ai Mannesi.

The only evidence of the medieval period is the fragment of a handle a nastro belonging to a glazed pottery jug (SU 205, VH-FW/4: fig. 22). It is difficult to attribute the fragment to the sparse-glazed “A” production or to the Forum ware production, as it is precisely at the handle, as at the bottom or on the rim, that there is most commonly attenuation of the coating, not always intentional. However, the characteristics of the clay seem to assign the fragment examined to a fairly late production phase (end of the tenth to the first half of the eleventh century), during which the Forum ware and the sparse-glazed productions still coexisted.

Activity 49: Formation of strata of humus

SU 1000: This layer does not produce much pottery material (amphorae, common kitchen- and tableware, African cooking ware), and contains also one fragment of earthenware, building material and modern nails. Note the presence of an African red slip ware plate, Hayes 105 type (one fragment of rim and another fragment, which join), datable to the end of the sixth to the middle of the seventh century. This piece, in fact, although coming from the most superficial stratum, is one of the very few pieces of pottery found in the area of the villa which can be definitely attributed to this period.

SU 1200: Among the important pieces are a fragment of a thin-walled cup, Atlante 2/408 type, and a fragment of olla, VH-CM/9 type (fig. 13), with cylindrical-ovoid body, thickened and rounded rim, and shoulder marked with grooving. The latter has analogies with examples from Luni and Albintimilium that are datable to the early imperial age (between the first and the third century) and were fairly widespread in Flavian times. Among the early medieval elements, note the olla with triangular cross-section rim and vertical lip with “wave” decoration incised before firing (VH-CM/17: fig. 15).

Activity 1: The strata relating to this phase did not produce any ceramic material of importance. Activity 2: SU 7042: Relating to the activity, there are some fragments of common pottery, among which is an olla of Dyson type PD 126, and a rim belonging to an olla of the Luni type XXX, both datable to the late Republican period.

Activity 3: SU 7021: The only recognizable fragments are those of a cup in common tableware (VH-CM/28: fig. 18), characterized by a thickened and rounded rim and shallow body, comparable to an example from Lugnano, and an olla with rim thickened and triangular in cross-section (VH-CC/16: fig. 4), which has analogies in examples from Cosa.

SU 7022: This layer contains material on the whole unimportant, except for the rim of a van der Werff-type amphora, of “Punic tradition” (VH-A/1: fig. 1), datable to between the second century and the beginning of the first century B.C.

SU 7027: This SU produces some fragments of a small thin-walled pitcher, Atlante 1/102 type, three fragments of an olletta with rounded rim, a fragment of Olcese 37-type olla with out-turned and hollowed rim, and some walls from amphorae of African production (probably of “Punic tradition”).

Activity 4: The strata relating to this phase yield no significant pottery material. Activity 5: SU 7019: The only identifiable fragments belong to a cup with thin “egg-shell” walls, decorated with a wheeled motif (VH-PS/2: fig. 19), analogous to the Atlante II type XXX, datable to the period between the Flavians and Hadrian.

SU 7026: The ceramic finds are mostly residual. Among them are one fragment of a small jug, with thickened and slightly out-turned rim, and a fragment of an olletta with rounded rim; both are datable to the period between the second and first century B.C.

SU 7040: This stratum contains a large quantity of fragments belonging to a single example of an amphora of small dimensions, of an unidentified type, perhaps attributable to Italic production on the basis of the impasto. The rest of the ceramic material (particularly the thin-walled ware) also points to a similar dating. An olletta (VH 148, fig. 24; see also Macaulay, ), cylindrical-ovoidal in shape, reused as a flower-pot, finds fairly close analogies in late Republican and early imperial contexts. In addition there is a certain quantity of residual material (common and black-glazed pottery) datable to the late Republican period.

SU 7041: This SU contains a small quantity of ceramic material, including some walls of Iberian-production amphorae (probably of the Beltran II type, already attested in other locations at the site) and a fragment of a lamp with volutes, generically datable to the Julio-Claudian period.

Activities 6-9: These strata do not contain significant ceramic material.

Activity 10: SU 7037: The only identifiable piece is a rim fragment of African kitchenware, of the Hayes 182 type, datable to between the middle of the second and the middle of the third century.

SU 7038: Among the significant pieces are some rims of pans of African kitchenware of the Hayes 181 type (second half of the second to the middle of the third century) and of the 197 type (late second to the middle of the third century). The presence in the stratum of the coin VH 144, however, would lead to the lowering of the dating to the late fourth to fifth century (see Buttrey, , no. 12).

SU 7039: The pottery present in this stratum is all residual. Among the identifiable sherds there is one fragment of a lamp with volutes and one of an olla with thickened and slightly turned-out rim, Sutri F20, A12 type, both dating to the early imperial age. There is, however, a follis of Constantius Chlorus dating to 297-298 A.D. (VH 123; see Buttrey, , no. 4), an element which would lower by about a century the dating provided by the pottery.

Activity 11: SU 7004: the dating material is a handle of glazed Forum ware; among the residues is to be noted one fragment of pan rim, of Hayes 196-type African kitchenware.

SU 7010: The stratum yields exclusively residual ceramic material, including some examples of African red slip ware (of the Hayes 8 and 27 types, datable to between the last decades of the first and the second century), and a pot rim in African kitchenware (or an imitation of it), Hayes 197 type, dating to between the second and the fifth century. Among the common kitchenware is attested a type of pan with rim a tesa (VH-CC/27: fig. 6), which finds an exact analogy in examples from Lugnano and Albintimilium, both datable to the Augustan Age; an olla with thickened rim of triangular cross-section (VH-CC/17: fig. 4); and an example of an olletta with out-turned rim (VH-CC/6: fig. 3), analogous to the type Luni II CM 6943; and a type of pan with banded rim and parallel grooving, slightly hollowed on the inside and with hemispherical body (VH-CC/46: fig. 9). Notable among the common tableware are two examples of cups (VH-CM/25-26: fig. 17), characterized by a thickened, banded and molded rim and hemispherical body, comparable with the Luni II type CM 2302.

SU 7035: There is a certain amount of residual pottery of the late Republican and early imperial age. The dating material consists of some fragments of common pottery, among which are a fragment of an olla with flattened rim and globular Curia 317-type body; two fragments characterized by a rim hollowed on the inside; a fragment of pan rim, thickened and slightly out-turned; and a fragment of pan, with rim thickened and flattened on the upper side, all datable to between the end of the fourth and the sixth century. Especially noteworthy are two fragments of pan with fillet under the rim, an imitation of the Hayes 91D form. There is also one fragment of a small cup in glazed pottery, whose state of preservation does not permit certain identification. The late antique chronology is confirmed, however, by two small coins datable to the fourth century (VH 092-093; see Buttrey, , nos. 6 and 11).

SU 7036: The stratum contains only residual material, including one fragment of a bowl of Hayes 185-type XXX (datable between the last two decades of the first and second century); one fragment of Dressel-Lamboglia 16 lamp; and one handle of Dressel 2-4 amphora. These last are attributable to the Julio-Claudian period.

Activity 12: SU 7017: The stratum contains exclusively residual and mostly unidentifiable ceramic material, except for the bottom of a late Italic terra sigillata cup, with relief decoration, portraying a dog racing rightward.

SU 7033: This SU yields material that for the most part is not identifiable, except for the wall of a Forum ware vessel, datable to the early medieval period (eighth to ninth century).

SU 7034: This stratum contains material that is mostly unimportant and residual, among which are some fragments of ollae of common kitchenware, of the types Settefinestre 28, 14 and 29, 20-22, generally attributable to between the first and the second century.

Activity 13: SU 7001: The SU contains some fragments of earthenware and glazed kitchenware. Among the residual material, to be noted are a rim of a black-glazed, Morel 2286-type cup, and a fragment of a bowl of Italic terra sigillata (VH-TS/2: fig. 20), attributable to the Atlante II, form XIX, 13 type.

During the excavations of “Horace’s Villa” at Licenza undertaken from 1999 to 2000, three ollae perforatae, or planting pots, were found in the garden area in front of the central staircase of the residence (Area 24). There is another pot known from the site, which is housed in the museum in Licenza. It apparently comes from one of the earlier excavations in the twentieth century. All four ollae perforatae will be discussed in this section.

Sector VII.1, Area 24, SU 7040

This pot is almost entirely preserved, missing only a small section of its rim (figs. 1-3). The pot measures 14 cm in height. Its diameter measurements are: rim, 11 cm; mouth, 9.2 cm; base, 5.2 cm; and base hole, 2.5 cm. There are no side holes. It is red in color (Munsell 5 YR 6/6) and is made of porous, moderately coarse clay. In the fabric, there are black and white inclusions, as well as small bits of grog (less than 1 mm). The pot appears to have been fully fired, as the red color remains constant and uniform throughout the wall of the vessel.

The pot is slender in form with a small flared rim. This slender form is similar to pots that have been found at Pompeii. It is clear that the pot was wheel-made, as smooth wheel-run ridges are present on both the interior and exterior. The pot has a single hole in its base. Extra material located just inside the hole in the base of the pot indicates that the hole was placed before firing. The potter took little care when making this pot; although the pot can stand up and does not wobble, it is lopsided. Additionally, extra material on the side of the base was not smoothed out before firing. The pot was not broken before it was placed in the ground; it was found intact. The flower pot dates to the last third of the first century A.D., which accords well with the Flavian garden context in which it was found. For the context, see Gleason, .

Sector VII.1, Area 24, SU 7051

This fragment of a flower pot base (figs. 4-5) was found in the rubble level above the ancient garden soil in SU 7051. It was identified as a flower pot base because it is clear that there was a purpose-made hole in the base. The fragment of the base measures 1.2 cm across. Based on a comparison with VH 148, the diameter of the base can be estimated to have been about 4.2 to 4.7 cm. The base hole was placed before firing, as extra material is present on the interior of the hole. The fabric of this pot is similar to the fabric of VH 148, though it is slightly lighter in color (Munsell 5 YR 6/8). The fabric, like that of VH 148, contains grog, small black inclusions and small light inclusions, and it is slightly porous. It cannot be determined whether or not there were other holes in the side of the pot. Because the two pots are of such similar fabric, it may be that VH 187 also resembled VH 148 in design and had only one hole in the base. Likewise, VH 187 may perhaps also date to the Flavian period. But a date cannot be assigned with absolute certainty, as this fragment was discovered in the rubble level.

Sector VII.1, Area 24, SU 7067

The diameter of the base is 3.5 cm (figs. 6-7); the base hole, 1.9 cm; and each of the two side holes, ca. 1.5 cm. The base is roughly made. The fabric is neither as coarse nor as porous as the fabric of VH 187 and VH 148. Its color is yellow-buff (Munsell 2.5 Y 8/2), and it is fully fired. There are at least three holes in this pot: one in the base and at least two on the side. One of the side holes is only partly preserved. There were probably at least three holes on the side of the pot, for in general flower pots that have multiple holes on the side have three holes. The extra material inside the base hole suggests that the hole was placed before firing. The holes on the sides also appear to have been placed pre-firing. Moreover, there are no known cases of side holes being placed on the side of a pot after firing. Generally, side holes are placed before firing, and this seems likely with VH 203. The form of the pot and the base is generally that of an inverted bottleneck. This pot was found turned upside down in the top level of the Flavian garden context to the right of the entrance from the atrium into the garden, on axis with the complete flower pot (VH 148). The pot dates to the last third of the first century A.D.

This flower pot (figs. 8-11) was probably found during earlier excavations, although there is no mention of it in the surviving documentation of the 1911-14 and 1930-31 excavations. It is impossible for us to know when and where the pot was found. It is 15 cm in height; the rim diameter at its widest point is 10.9 cm, and the mouth is 9.6 cm. The base is misshapen; as a result, the base diameter ranges from 3.5 cm to 3.7 cm. The base hole is also misshapen, and its diameter ranges from 1.7 cm to 2.1 cm. The size of the side holes range from 1.2 to 1.5 cm.

The design of this flower pot does not exactly match that of any of the other pots found at Horace’s Villa. This pot has four holes, three on its side and one in its base. The fabric is neither porous nor micaceous, but it has a few black inclusions and bits of grog (no bigger than 1 mm). It appears to be completely fired. The color is an orange-red (Munsell 5 YR 5/8). The body of the pot is slender, like many flower pots from Italy. Owing to the potter’s carelessness, it is slightly fuller on one side, and it wobbles when standing up. A similar slender form has been observed at Pompeii.The pot has a noticeably long neck (2.5 cm) and an upright-collar rim. While the rim thickens outward to 3 cm, it remains straight on the inside. All of the holes were placed before firing. The side holes are placed about half-way up the vessel. It has light wheel ridges on both the inside and outside. The pot appears to have broken on the top quarter and was restored from fragments.

The flower pots from the new excavations are on axis with a planting pit and with an amphora reused as a planting pot. It is common to find planting pots and reused amphora together in a Roman garden context; reused amphorae and planting pots, for example, have been found together along the Canopus at Hadrian’s Villa. The variation in vessel types probably indicates that they contained different types of plants, or plants brought from different nurseries. The pots run parallel to the stairs and path in the Flavian garden. Though the base fragment came up in rubble levels, it was also found on axis with the ollae perforatae, a planting pit, and half of a reused amphora. As the olla perforata VH 203 was also found turned upside down, it is possible that the tilling of soil during the medieval period upturned pot VH 203 and destroyed pot VH 187.

The three ollae perforatae found at Horace’s Villa in 1999-2000 are typical of the range of forms of planting pots found in Italy. The flower pot in the museum of Licenza (SAL 39344) is very similar in form to those found at Pompeii; its body is slender, and its side holes are smaller (1.4 cm) and cleanly made. However, its fabric does not match the typical micaceous Pompeian fabrics. The slender body and flared rim of pot VH 148 is also similar in form to the flower pots that have been found at the House of the Ship Europa and other locations at Pompeii, and at Boscoreale. But again the fabric of VH 148 does not match the typical Pompeian fabric.

It seems probable that olla perforata VH 148 was not a pot used for aerial layering, because it lacks side holes to provide drainage and air for the new roots. Rather, it probably housed a sapling or plant that was brought from a nursery. However, the planting pot in the museum and the buff base fragment may have been used for layering and then planting, because they have side holes. The side holes enabled air to reach the plants’ roots when being layered and planted.

Nothing can be said about the place of manufacture of these pots. The range of fabrics found at Horace’s villa is unusual, as compared to the ensembles of flower pots found at sites such as Pompeii, Herculaneum, and Prima Porta. The rule of homogeneity is not surprising: Italian ollae perforatae were generally made locally and used locally or at most regionally. There is no ancient kiln yet known from the Licenza area. The range of fabrics from Horace’s Villa may mean that these pots—and the plant materials contained inside them—came from nurseries in different locations throughout the region, or it could indicate that they came from the same nursery at different times. At the moment, the question must be left a non liquet. One way of addressing the matter in the future could be through new excavations of the garden and other nearby villas and their gardens; perhaps the heterogeneity found thus far is deceptive and with more data, the Licenza site, too, will revert to the norm of homogeneity.

Figure 2 shows the system for classification of bricks and roof-tiles. After the typological naming of the object and the codification of the fragment, the words “+ cut” mean that the piece (brick or tile) has a side that is sawed and may therefore have been used in a brick facing. The dimensions are given in the following order: height x width x thickness. In the case of fragmentary pieces, a portion of an original side that is preserved is considered as the width of the object for the purposes of measuring. The shape of the stamp follows: rectangular, horseshoe (semicircular with disc partially set in), and orbicular (circular with large, medium or small-sized disc, an orbiculus, set in). For the rectangular and horseshoe stamps the first measure is the height, the second the length; for the orbicular stamps, the diameter of the stamp is given first, and then that of the orbiculus. The height of the letters follows, and then the number of auxiliary lines which delimit the lines of the text, beginning from the outside. The number which precedes the transcription of the text refers to CIL XV; in the case of the unpublished pieces, “N.” indicates the place where the unpublished piece would have appeared in the CIL. The term “compl.” explains that the fragmentary or badly preserved text of the stamp as given in CIL XV or N has been completed.

1. (LP2) fig. 3. Fragment (E) of brick, cm. 15.6 x 10.5 x 4-3.8. Orbicular stamp, cm. 7.5?: disc 3.6; letters 1.1, 0.9; aux. lines 2, 2.1. CIL XV.129 Tert(ius) D(omitiae) L(ucillae) [ex f(iglinis) Can(inianis) op(us) dol(iare) fe(cit)] Tit(iano) [et Gall(icano)co(n)s(ulibus)] [((ramus palmae))]

2. (LP1) fig. 4. Fragment of brick (?+ cut), cm. 21 x 23.8 x 4.6. Orbicular stamp, cm. 9.3; disc 3.6; letters 1.1, 0.9; aux. lines 2, 2.1. CIL XV.129 Te[rt(ius) D(omitiae L(ucillae) ex] f(iglinis) Can(inianis) op(us) dol(iare) fe(cit) Ti[t(iano) et] Gall(icano) co(n)s(ulibus) ((ramus palmae))

3. (LP6) fig. 5. Fragment (H) of tile?, cm. 19.6 x 15.8 x 3. Horseshoe stamp, cm. 7.4, 3.5; disc 3.2; letters 0.8 – 1.3. N. 890/1 L(uci) Braeti

4. (LP3) fig. 6. Fragment (E) of tile?, cm. 10 x 6 x 2.7. Horseshoe stamp, cm. 6.7?, 2.8?; disc 2.5?; letters 1.1 – 1.3? N. 890/1 L(uci) Braeti

5. (LP5) fig. 7. Fragment (E) of tile?, cm. 25.5 x 19.5 x 2.5. Rectangular stamp, cm. 7.3?, 3.3; letters 1, 0.9. N.933/4 Claud[iae] Epich[aris]

6. (L10R) fig. 8. Fragment (E) of tile?, cm 15.8 x 12.4 x 2.2. Rectangular stamp, cm. 3.6?, 3.4; letters 1, 0.9. N.933/4 [Claud]iae [Epich]aris

7. (VH 204) fig. 9. Fragment (G) of tile, formed of two parts joined together, cm. 37.9 x 33.8 x 2.5-2.8. Rectangular stamp, cm. 8.9?, 3.3; letters 1, 0.9. N. 933/4 Claudiae Epicharis

8. (L4R) fig. 10. Fragment (E) of brick, cm. 19.5 x 16 x 4.9-4.2. Orbicular stamp, cm. 10; disc 5.1; letters 1.1, 1.1; aux. lines 1, 2, 2. CIL XV.1033 [Op(us) d(oliare) Doryph]or(i) Domit(iae) P(ubli) f(iliae) Lucill(ae) [Paet(ino) et A]pro(niano) c[o(n)s(ulibus)

9. (L3R) fig. 11. Fragment (E) of brick, cm. 18.2 x 13.6 x 4. Orbicular stamp, cm. 9.8; orb. 4; letters 1.1-1.3, 1; aux. lines 1, 2, 2. CIL XV.1210 C(ai) Comin(i) Proc(uli) [dol(iare) ex p]r(aedis) Iuli Step(hani) Titian(o) et Squil(la) co(n)s(ulibus)

10. (LP4) fig. 12. Fragment (E) of brick, cm. 16.4 x 8.5 x 4.1. Rectangular stamp, cm. 6.3+, 3.7-4; letters 1.3, 1.1. CIL XV.1211 C(ai) Comin(i) [Proc(uli)] fec(it ) Feli(x Caric(us)]

11. (L1R) fig. 13. Fragment of brick (bipedalis or sesquipedalis), cm. 29.3 x 34.2 x 4.5. Rectangular stamp, cm. 11+, 3.9; letters 1.3, 1.1. CIL XV.1211 [C(ai) Com]in(i) Proc(uli) fec(it) Felix Caric(us)

12. (L2R) fig. 14. Fragment of brick, cm. 29.3 x 14.6 x 5-4.8. Rectangular stamp, cm. 7.1+, 2.4; letters 1.8. [---]ILLI.

Two identifications can be proposed. In the first case, we may have to do with the stamp CIL XV.1340a [L. Ope]illi on the basis of the letters and the rectangular form with one line of text; the use of the dipthong ei for i (=Opilli) is frequent in the Republican period. The second integration derives from CIL XV.2381b=XIV.4092.8 [C. Caecili Bathy]ylli, and is based solely on the interpretation Y of the first preserved letter (of which only the lower part of the straight stroke remains), since the form of the only reference copy is not transmitted.

13. (LP7) fig. 15. Fragment of brick, cm. 17.2 x 23 x 3.2. Rectangular stamp, cm. 8.4, 3.6; letters 3.2-2.7. N. 1370/1-1371 L(uci) Pomp(---)

14. fig. 16. Rectangular stamp CIL XV, 1972 a compl. [M(ani)] Naevi

15. (LP8) fig. 17. Fragment (C) of tile, cm. 21 x 34 x 3.1-2.9. Rectangular stamp, cm. 15.7, 2.5: letters 2-1. CIL XV, 1972 a compl. M(ani) Naevi

16. (L7R) fig. 18. Fragment (H) of tile, cm. 17.3 x 14.5 x 3-2.5. Rectangular stamp, cm. 15+, 2.5; letters 2-1. CIL XV, 1972 a compl. M(ani) Naevi

17. (L6R) fig. 19. Fragment (H) of tile cm. 14.1 x 14.5 x 3.2. Rectangular stamp, cm. 12.2+, 2.6; letters 2-1.5. CIL XV, 1972 a compl. M(ani) Naevi

18. (L8R) fig. 20. Fragment (H) of tile cm. 14.3 x 11.7 x 2.9-2.7. Rectangular stamp, cm. 13+, 2.6; letters 2-1.7. CIL XV, 1972 a compl. M(ani) Naevi

19. (L5R) fig. 21. Fragment (H) of tile cm. 17.5 x 24.4 x 3.1. Rectangular stamp, cm. 13+, 2.5; letters 2.2-1.7. CIL XV, 1972 a compl. M(ani) Naevi

20. (L9R) fig. 22. Fragment (H) of tile cm. 15.8 x 8.6 x 3-2.8. Rectangular stamp, cm. 7.5+, 2.2+; letters 2.2. CIL XV, 1972 a compl. M(ani) N[aevi]

21. (VH 125) fig. 23. Fragment (H) of tile cm. 22 x 17.2 x 3-2.6. Rectangular stamp, cm. 15.5, 2.6-2.4; letters 1.7-1.5. CIL XV, 1972 a compl. M(ani) Naevi

22. (VH 185) fig. 24. Fragment (H) of tile cm. 9 x 13.5 x 3.4-3. Rectangular stamp, cm. 8.7+, 2.5; letters 2-1.7. CIL XV, 1972 a compl. [M(ani)] Naevi

23. (VH 201) fig. 25. Fragment (H) of tile cm. 10 x 12 x 3. Rectangular stamp, cm. 5.3+, 2+; letters 1.7. CIL XV, 1972 a compl. [M(ani) Na]evi

24. (VH 136) fig. 26. Fragment (H) of tile cm. 22.6 x 13.5 x 3-3.2. Rectangular stamp, cm. 6.8+, 2.6; letters ? CIL XV, 1972 a compl. [M(ani) Na]evi

25. (VH 035) fig. 27. Fragment (H) of tile cm. 12.5 x 14.2 x 3.5. Rectangular stamp, cm. 6.5+, 2.6; letters 2. CIL XV, 1972 a compl. M(ani) N[aevi]

26. (VH 189) fig. 28. Fragment (H) of tile cm. 14 x 9.2 x 3. Rectangular stamp, cm. 6.5+, 2.6; letters 1.8. CIL XV, 1972 a compl. [M(ani) Na]ev[i]

27. fig. 29. Fragment of dolium cm. 22 x 36 x 11.5; height of lip 8, width 19.5. Two separate rectangular stamps: a) cm. 3.2 x 12.5; letters 2.1-1.6 b) cm. 3.5 x 12.6+; letters 2-1.7 N. 2437/8 Favoniae C(aii) f(iliae) ((palma ds.)) Q(uinti) Fabrici Fel[icis]

28. (AAR 5142) fig. 30. Fragment (H) of tile cm. 14 x 9.2 x 3. Rectangular stamp, cm. 9.5+, 2.5; letters 1.9-1.8. CIL XV, 1972 a compl. [M(ani) Na]ev[i]

29. fig. 31. Fragment (H) of tile cm. 14 x 9.2 x 3. Rectangular stamp, cm. 15.6, 2.5; letters 2-1. CIL XV, 1972 a compl. M(ani) Naevi

On the basis of the lesser or greater area of diffusion of the products, it is possible to identify a local market, the so-called municipal market. This includes productions of privati domini and smaller figlinae, which were distributed over a limited radius, and which rarely reached Rome. This market is to be distinguished from the wider “urban” market, of Rome and Ostia, the largest centers for the arrival and distribution of goods in the middle Tyrrhenian area in Roman times (fig. 35).

The brickstamps, for the most part, distinguish in a fairly detailed way the commercial activity of individuals linked by economic interests at a specific time and place of production; thus, our primary objective is the investigation of the workshops and the contextualization of the stamps with reference to the territory in which these workshops occur.

Previous studies on Rome and Lazio have been complicated by the ambiguity caused by the contrived grouping of the texts gathered in CIL XV, published in 1891. The large number of finds in Rome, resulting from excavations and systematic searching in the second half of the nineteenth century, in comparison with Latium vetus and the surrounding regions, has resulted in the use of the term “urban” opus doliare to indicate not only that the place of finding was in the city, but also that they may have been produced there.

Recently, however, the investigation of the stamped bricks and their topographical distribution in the valley of the Tiber has shown that the greater part of “urban” production took place along the valley at a distance sometimes more than one hundred kilometers north of Rome. The finding of one or more brickstamps in an area further away than any previous finds provided two important results: one, it demonstrated that the market extended at least for a radius as wide as that evidenced by these localities; and two, it enabled scholars to localize some of the factories.

It must be noted that if the presence of a stamped brick at a particular site is not sufficient on its own to localize the factory, being a product of the “industrial” type with the stamp of an activity aimed at a fairly extensive market, the absence of such a brick is not proof of the contrary.

The fact that the brickstamps, which allow the identification of the producer/owner, the place, and/or the period of production, come from known factories gives the inscriptions their importance, both in support of and against accepted theories. In some cases the name of the producer/owner corresponds with the dominus of the villa where the brick was found.

The topographical location of the workshops and kilns, therefore, is fundamental for a clearer understanding of the production and the commerce of the bricks, for our knowledge of the gentes active in making them, and for better use of the epigraphical data.

As is well known, the brick industry is closely linked to the natural components of the territory (raw materials, watercourses, wood), and to the human resources available (including the economy of the fundi, manpower, industrial and marketing activities, roads, and transport). Any villa was, by and large, in a position to produce tiles, for which highly specialized manpower was not necessary; bipedales and dolia, on the other hand, were usually produced in the larger workshops. From an economic point of view, if a piece of land (praedium) was available, even a small one, as long as it had a bed of clay under it and a plentiful supply of water, the plant would have no expenses other than wood and manpower.

Since brick products are heavy, transporting them over land could markedly increase their price in comparison with the lower cost and greater speed of water transport. Thus, it has been observed that in the Tiber valley, kilns were generally positioned upstream and near the principal watercourse or one of its subsidiaries, in order to transport the bricks downstream. Downstream river transport was easier than upstream animal haulage, which was very expensive for heavy loads.

Natural resources and environment

Clay material adequate for satisfying the demands of a principally local market is available more or less everywhere in the middle Anio valley, from Tivoli to Subiaco, even if in limited quantities. These small deposits were occasionally used by local kilns even up to the beginning of the last century. The abundant forests of the area provided the necessary quantities of fuel for baking the bricks. However, the clay deposits were not sufficiently large to be exploited economically for industrial purposes.

Unlike those in the valley of the Tiber, the clay deposits of the Anio valley are often in logistically difficult areas and are not suitable for high-quantity production and commerce. In fact, the morphology of the territory in the area is characterized by rough and tortuous paths, leading down to the via Valeria, which in Roman times was well maintained, but whose use for the heavy transportation of bricks was expensive. Water transport was not a possibility, because above Tibur, the watercourses, consisting of the Anio and its tributaries, were not navigable because of the small volume of water, which was mostly in the form of torrents.

The production of bricks took place, therefore, where there were deposits at hand sufficient for meeting a purely local market. This is shown by the positioning of the kilns, which was linked to the vicinity of the settlement (villa, vicus or castrum).

Territory and documentary sources

The geographical area in which “Horace’s Villa” lies is the valley of Licenza (vallis Digentiae, a tributary of the middle Anio valley). Licenza belonged administratively to the territory of Varia, situated between Tibur, Carsioli and Trebula Suffenas, through which passed the via Valeria. Considering the vastness of the area under consideration – the Anio valley from Castel Madama to Subiaco – the number of samples of stamps in relation to the number of unmarked bricks present in the territory seems very small. In fact, there have been very few finds other than those of “Horace’s Villa.” Some of these finds follow.

From S. Maria dei Morti near Marano Equo, on the left bank of the Anio, halfway between Vicovaro and Subiaco, comes a fragment of tile with an unprecedented brickstamp, C(ai) Aufili (fig. 32). The gentilicium Aufilius, which has no precedents locally and originates in the central Italian area, appears in an inscription from Hadria (Atri, Teramo) and in the variants Aufillius from Corfinium and Afilius from Praeneste, all from the Republican age. The palaeography and the onomastic formula indicate dating to the middle of the first century B.C. The formula is typical of the “one-name” stamps of private producers.

The territory of Anticoli Corrado, from which come some important Latin inscriptions, has yielded the dolium stamp DEFRV.

Nero’s villa in Subiaco has not produced any stamped bricks so far. We also have little knowledge about the brick production in areas adjacent to the one under consideration, with the exception of the ager Tiburtinus.

From Trajan’s villa in the upper Anio valley at Arcinazzo, which is outside the territory under consideration but part of the district of Subiaco, products have been found from the age of Trajan also known in Rome, Grottaferrata and the ager Praenestinus. From a geographical point of view, this area was better linked to Praeneste and Labicum in Roman times.

In the ager of Trebula Suffenas, bordering on our area and very similar from a geomorphological point of view, several brick facilities and two dolium stamps are known. Relevant in this regard are the figlinae owned by C. Biculeius Priscus, a member of the ruling class in Tibur. Two kilns have also been identified in the territory. The first, identified by C. F. Giuliani about 500 meters to the east of Ponte di S. Cecilia, was attributed by him to the gens Caecilia (as is known, the present-day town of Ciciliano derives from the fundus Caecilianus), whose products reached Castel Madama. No stamped products have been found at the second, discovered by F. Sciarretta near the Fontana dell’Acquaone.

In the territory of Carsioli, CIL IX (1883) records 13 “urban” brickstamps of uncertain provenance, preserved at Arsoli, as well as two local productions.

The ager Tiburtinus, including Hadrian’s Villa, can be considered a peripheral area for the “urban” brick market. The products found here were used locally but also formed part of the great Tiberine transportation and supply system that provisioned the city of Rome. The area was easily accessible because of particular topographical conditions such as the navigability of the Anio and its direct link with the Tiber, as well as good road connections.

Our stamps, even if quantitatively few, are representative for the research and topographical reconstruction of the kilns of the territory. Since we lack a systematic study of the brickstamps, and because little adequate archaeological research has been conducted, the absence from the territory of a certain product cannot be considered decisive for excluding the possibility that it originated here.

The first Roman kiln in the area was identified by Z. Mari, on the basis of much brick waste, blackish and over-baked, near the confluence of the Roscio stream and the Licenza river, in the locality known as Piani a Otto, a site previously noted by Lugli.

The stamps of Manius Naevius are of strictly local production, distinct from the contemporaneous products of the Naevii of the areas of Tusculum and Praeneste. Many unstamped lateres, found in the masonry of “Horace’s Villa,” are of an impasto similar to that of the production of Naevius. On the basis of the very limited radius of distribution for the products, the kiln could have been in the Licenza valley. In this connection, it is worth noting that, at the beginning of the twentieth century, near the spring in the locality known as Le Moglie, very near “Horace’s Villa,” there was a functioning kiln for the production of bricks, in which vitrified waste materials of baking have been found. This kiln (fig. 33) could have been active in antiquity and may have belonged to the same fundus of the nearby Villa of Horace, making it likely that the kiln was prepared and used for the specific needs of one or more owners of the villa.

Another possible production site has been identified on the other side of the valley, near the top of Colle Franco, in the locality known as I Limiti. There a bed of clay is visible on the surface and a large quantity of tiles and coppi (semicircular cover-tiles), concentrated in a very small area, has come to light in the course of agricultural work.

The evidence of three stamps examined in our study helps us to narrow down the chronology of the semicircular form. The production of C. Aufilius, on the basis of onomastic and of palaeographic arguments, takes the use of the horseshoe design back to the Republican age and locates its occurrence outside the area of Rome. Mari’s convincing identification of Erasinus Aug(usti) l(ibertus), mentioned in the stamp of the Arcinazzo villa, with Trajan’s imperial freedman known from a funerary inscription on the via Appia, demonstrates the continued use of the semicircular form of stamp in the age of Trajan (98-117). On the basis of these considerations, the stamp of L. Braetius should be placed between the first and the early second century A.D.

From the first quarter of the second century A.D., there is evidence at Licenza of a change in the provisioning of bricks. The owner of the property seems no longer to use local products, or the kiln on the grounds of the fundus, but turns to the market supplied by bricks produced at industrial levels like those of Iulius Stephanus and of the Domitii, located in the central Tiber valley. Since these goods came from a distance and would have incurred transport costs, the choice of such a product must have depended on their higher quality and lower price when compared with the local products.

The paucity of the samples available so far does not allow us to establish whether such a purchase of bricks involved only one class of opus doliare, as, for instance, the bipedales, which must have been well manufactured to be used in the brick work, nor whether it was related to a crisis in the local “artisan” workshops as a result of competition from the “industrial” workshops in quality and price.

Certainly, the small local formations of clay were enough to satisfy the local market even in modern times, as is shown, for example, by the brick kiln C(asa) M(assimo) Arsoli, halfway between Vicovaro and Subiaco, set up by Clement X in 1670 and active until the end of the nineteenth century (fig. 34).

The revetment of walls using slabs of marble or other colored stone material (crustae: CIL VI.10237; Sen. Benef. 4.6.2, Dial. 1.6.4, Epist. 86.6; Plin. NH 35.3, 36.48; Ps. Quint. Decl. 9.17), often of varied sorts and arranged so as to form geometric shapes or complex figured motifs, was known as incrustatio in antiquity; this we know from numerous literary sources (Varro Men. 533; Proc. Dig. 8.2.13.1; Paul. Dig. 50.16.79.2) and inscriptions (CIL III.6671, IX.451, XII.935). Today, however, the expression commonly used is “parietal opus sectile.”

The custom of lining walls with slabs of marble, already widespread in the Hellenistic world from the second century B.C. (Plin. NH 36.47), is documented in Rome from at least the first century B.C., as again attested by Pliny (NH 36.48: Primum Romae parietes crusta marmoris operuisse totas domus suae in Caelio monte Cornelius Nepos tradit Mamurram). Since what Pliny writes of here probably took place at the end of the Gallic Wars (51 B.C.), when Mamurra, praefectus fabrum and friend of Caesar, was very greatly enriched, we may date the introduction to Rome of the technique of parietal revetment in marble to around the middle of the first century B.C.This dating seems to be confirmed by a passage in the Menippean satire of Varro, written between 45-43 B.C. (Varro Men. 533). It is in this period, therefore, that panels began to take the place of the painted socles typical of pictorial styles I and II, often executed in imitation of the more costly polychrome marble revetments.

A profound change came about, probably in the Augustan age, with the massive exploitation of marble quarries (above all lunense, but also giallo antico, pavonazzetto and africano, which appear abundantly in the great monuments of Rome, such as the Forum of Augustus). It is unfortunately rare to find the most ancient remains of wall revetment in simple slabs, because marble was a readily recyclable material, rarely surviving the sacking that inevitably followed abandonment. However, several extant Augustan monuments (in addition to the Forum of Augustus with the temple of Mars Ultor, mention must also be made of the Temple of Concord) attest the widespread use of marble wall revetment in plain panels during in that period, although it was primarily employed for the decoration of monumental public buildings.

The increasing diffusion of marbles between the Augustan and Flavian ages led to the wider use of marble revetments, not just for most walls but also for floors, which had previously been created with other stones or in mixed materials. Probably in this period the use of wall decorations in simple marble slabs became widespread in private homes as well, even in “middle class” houses, as seen in numerous examples in the Vesuvius area (among these note the House of the Relief of Telephus at Herculaneum).

Revetments consisting of simple slabs become ever more elaborate from the second century onwards: evidence of this are great monuments such as the Pantheon and, in private architecture, the Villa Adriana at Tivoli. In these examples it is already evident that the panels are better delimited and the internal arrangement is enlivened with geometrical intarsia, or by moldings or simple raised edging.

Simple slabs continued to be used in parietal revetments until late antiquity, but ever more frequently alternating with a type of intarsia with complex motifs. These could be geometrical or with animal and human figurations, in accordance with a taste whose expression reaches its zenith in the last decades of the fourth and the fifth century A.D., with the extraordinary examples of the Basilica of Junius Bassus on the Esquiline and that of the so-called “Building outside Porta Marina” at Ostia. In these, the figured compositions, in arrangements that are also symbolically complex, predominate over the smooth parts or the simple geometric sections.

In the excavation of an ancient building, it is very rare to find in situ marble wall coverings surviving whole. This is chiefly due to the depredations in the periods following the abandonment of the monument, but is sometimes the result of collapse due to factors such as lack of maintenance (bear in mind the weight of the marble slabs themselves, positioned vertically) or to natural disasters (fires, earthquakes, etc.).

It is much more usual, therefore, that the presence of marble wall revetments in an archaeological excavation is documented indirectly or must be reconstructed on the basis of clues. The traces left by the slabs in the mortar in which they were set and the holes of the metal clips used to anchor the slabs to the walls, as well as the numerous fragments and splinters of the slabs themselves, frequently found in the excavated soil, mark the presence of revetment panels. These fragments, however, although sometimes present in the excavated stratigraphies in quantities much greater than other more “canonical” classes of material (such as pottery, metal items, glass, etc.), are on the whole ignored. Even if they are recorded in the publication of an excavation, they are normally included in very broad categories, such as that of “non-sculptural marble elements,” or even more generically as “construction materials.” It is obvious that in this way further study of the overall corpus is hampered; not only is direct information on the decoration lost, but also indirect information on the function and the qualitative level of the rooms.

Such an example is the Villa of Horace. The fragments of small stone crustae–marble and non-marble–presented here are only a minuscule quantity compared with the exceedingly abundant material recovered in the course of Pasqui’s excavations of 1911-14. Those finds, at the moment of excavation (or, rather, of unearthing), were included among the construction materials or mixed up indiscriminately (and then stored) with the simpler architectonic or sculptural elements, together with the shapeless splinters, large slabs, steps, brackets etc.

The only information on these materials is supplied by Lugli, who, in the rapid listing of the finds from the excavations of the villa, also mentions “many crustae of colored marbles for the covering of socles and many fragments of moldings in rosso antico, giallo antico, bardiglio and in various marbles.” Some of this material was put together by Pasqui on four panels and displayed in the old Antiquarium in Licenza, without even a generic mention of their provenance.

Principally for this reason and because of the impossibility of reconstructing the stratigraphies to which they belonged, we have preferred to forego detailed analysis of this material. It has only recently been divided up, classified and stored in some thirty boxes, today held inside the storehouse of the Soprintendenza Archeologica per il Lazio at the Santuario di Ercole Vincitore in Tivoli (fig. 1).

In the present study, therefore, only the material coming from the 1997-99 excavation seasons is discussed, with the exception of some fragments of capitals from pilasters and slabs with relief decoration from the excavations of Pasqui, now displayed in the Museum at Licenza (figs. 2-4). These fragments, taken together, lead us to conclude that there was a decorative program of considerable value at the villa.

In the catalogue that follows, all those elements presumed to belong to parietal coverings are analyzed, namely molded slabs or those with relief decoration, pilasters, and moldings, but also simple slabs. This term is taken to cover all the whole or fragmentary marble elements worked simply, in such a way as to have one or more flat faces.

In addition to specific information about the find-spots, the pieces under discussion are supplied with details of measurements (thickness, maximum and minimum dimensions, weight); type of marble (with complete identification in the case of the colored ones, and more generically in the case of the grey and white ones); and relative disposition and treatment of the flat surfaces (principal faces, edges). The dimensions are expressed in centimeters in the case of linear dimensions and are to be taken as indicative, except in the case of the thickness, which is always taken to the nearest millimeter, even for the thinnest crustae. The weight, where measurable, is expressed in grams, and taken to the nearest 10 gr, except for the smallest crustae.

Capitals (Ca)

Greco scritto Ca-1. Inv. 75230 Pilaster capital of Corinthian inspiration Chipped along the edges. Broken on the two upper corners, just above the volutes. Back of the panel polished, edges finished with fine-ended chisel. Dimensions: 29 x 25.5 x 2.3 Ca-2. Inv. 75231 Pilaster capital, of Corinthian inspiration Whole example, slightly chipped along the edges. Back of the slab polished, borders finished with fine-ended chisel. Dimensions: 29 x 23-24 x 2.3 Ca-3. Inv. 62976 Pilaster capital, of Corinthian inspiration Example in three fragments, reassembled. Back of the slab polished, borders finished with fine-ended chisel. Dimensions: 30 x 20 x 1.4

Grey, medium-grained marble Ca-4. Inv. 62973 Pilaster capital, of Corinthian inspiration Whole example, slightly chipped along the edges Back of the panel polished, borders finished with fine-ended chisel. Dimensions: 29.5 x 20-28 x 2

The examples Ca-1/4, probably to be attributed to the decoration of a single room, unfortunately not identified, were summarily published by Lugli, who included a photograph of them. The same pieces have recently been reexamined, also rather rapidly, by Reggiani.

Analogous capitals, with central calyx and lateral leaves serving as volutes, are attested at Ostia and datable to the last years of the first and the first decades of the second century. There are also rather precise analogies with an example from the Museo Nazionale Romano, datable to the age of Hadrian; in this example, the morphology of the central calyx, which has expanded extremities, is similar to that of the present ones. The type of relief and the structural characteristics are different, however; they are rather more simplified in the Licenza example.

On the basis of analogies, dating to between the last decades of the first and the first years of the second century seems plausible.

Pilasters (Le)

White, fine-grained marble, with greenish veining (Pentelic?) Le-1. Sector VII.1, Area 24, SU 7001 Broken on all sides. One face seems to preserve traces of molding. The tool marks are no longer recognizable because of the corrosion of the surface. Dimensions: 4.4 x 2.2 x 0.7; weight: 20

White, medium-grained marble Le-2. Sector I.3, Area 37, SU 428 (fig. 11) Broken on all sides. The surface is molded, with the form of a segment of circle (diameter 4.8) and fillets (width 1), and is polished; the back face (in contact with the wall) is smoothed. Dimensions: 7.3 x 7.3 x 1.5; weight: 380 Le-3. Sector I.3, Area 37, SU 414 Broken on all sides. Two matching fragments, similar to the preceding ones. Molded polished surface, reverse face smoothed. Dimensions: 16 x 6.2 x 1.2; weight: 575 Le-4. Sector I.3, Area 37, SU 414 Broken on all sides. Similar to the preceding ones. Surface molded and polished; reverse face smoothed. Dimensions: 7.4 x 7.2 x 13; weight: 190

Slabs molded and/or with relief decoration (Lm)

White, fine-grained marble Lm-1. Sector I.2, Area 50, SU 220, VH 019 (figs. 5 and 11) Broken on two sides. Two flat and smoothed faces, of which the one at the back still shows the raised bit left by the saw cut and traces of the mortar used to fasten the slab to the wall. Upper edge polished, in which there are two holes for metal clips (diameter 0.5 and depth 1.1; diameter 0.7 and depth 2.2).The different dimensions and the closeness of the two holes probably indicate maintenance work on the wall covering, or else reuse of this slab. Dimensions: 10.8 x 7.6 x 3.6 (max)–1.9 (min); weight: 640 Lm-2. Sector I.2, Area 50, SU 220 (fig. 11) Broken on four sides. Upper edge polished, one face molded with two cymae reversae and smoothed. Dimensions: 8 x 3.7 x 2.2; weight: 170 Lm-3. Sector I.2, Area 50, SU 220, VH 019 (figs. 5 and 11) Broken on four sides. One face flat and smoothed; upper edge with traces of fine-ended chisel (perhaps indicating a reworking of the artifact). The face, molded with fillet and cyma reversa, is polished. Dimensions: 11.6 x 9 x 2.5; weight: 450 Lm-4. Sector I.7, Area 35, SU 1200 (fig. 11) Broken on two sides. Two of the original edges remain, both vertical and finished, one with a gradine and the other smoothed. Two flat faces, of which one is smoothed and decorated with rectangular panelling, bordered with a double molding (fillet, cyma reversa). The back face, smoothed, preserves traces of the mortar (grey with volcanic elements in it) used to fasten the slab to the wall. Dimensions: 11.5 x 10.5 x 2.5; weight: 1830

White, medium-grained marble Lm-5. Sector IV.2, Area 23, SU 4201 (fig. 12) Broken on two sides. Two flat faces, of which one is polished and molded (with fillet and cyma reversa) and the other smoothed. Dimensions: 17.5 x 5.9 x 2.2; weight: 1120

Greco scritto Lm-6. Sector I.5, Area 38, SU 822, VH 072 (figs. 6 and 12) Broken on three sides. A curvilinear border worked with marteline. Two flat faces, of which one is polished and the other smoothed. On the right margin is still visible part of a circular motif, cut with the chisel. Dimensions: 9.6 x 7.4 x 1.5; weight: 380 Lm-7. Sector I.3, Area 37, SU 424, VH 021 (figs. 7 and 12) Broken on all sides. Two flat faces, of which one is decorated with three curvilinear concentric grooves, and the opposite one is polished. Dimensions: 10.5 x 6.9 x 1.2; weight: 190 Lm-8. “Surface find” (fig. 12) Broken on all sides, surface very corroded. Two flat faces, of which one is decorated with two straight grooves, of different widths (0.1 and 0.7), and a curvilinear groove (width 0.5); the opposite face is polished. Dimensions: 9.8 x 5.5 x 2.1; weight: 240 Lm-9. Sector I.2, Area 50, SU 285, VH 073 (figs. 8 and 12) Broken on all sides. Two flat faces, of which one is polished and decorated with a rectangular panel, of which one corner survives, bordered by a molding (fillet and cyma reversa). The back face is just rough-hewn. Dimensions: 10.5 x 9 x 1.8; weight: 285 Lm-10. Sector I.3, Area 37, SU 448, VH 069 (figs. 9 and 13) Broken on all sides. Two flat faces, of which one is polished and decorated with lozenge-shaped concentric panels, bordered by simple grooves of varying widths (0.3–0.7). The back face is just rough-hewn. Dimensions: 8 x 9.2 x 1.5; weight: 280 Lm-11. Sector I.3, Area 37, SU 400 (fig. 13) Broken on three sides; only a border is left, bevelled and finished with chiselling. Two flat faces, of which one is polished and decorated with straight and parallel grooving; the opposite face is smoothed. Traces of mortar on the decorated surface and the type of workmanship observed on the only surviving border indicate reuse of the slab of which it was a part. Dimensions: 8.5 x 6.6 x 1.1; weight: 86 Lm-12. Inv. 75227 Slab in two fragments, recomposed, broken on all sides. Two flat faces, of which the back one is smoothed and the front one has relief decoration. The decorated area seems to have been divided into metopes (of dimensions which cannot be reconstructed), delimited by an astragal (height 1.1); the only surviving metope is decorated with two crossed shields and bordered at the bottom with a pelta. Dimensions: 24 x 29 x 1.5 Lm-13. Inv. 75228 Slab in two fragments, recomposed, broken on three sides. Only one of the original borders is left, worked with the chisel. Two flat faces, of which the back one is smoothed and the front one has relief decoration. The decorative area, bordered at the bottom with an astragal (height 1.1), is divided into rectangular metopes of various sizes (28.5 x 22.5, 13.5 x 22.5), decorated respectively with a shield and hexagon with inflexed sides. Dimensions: 52 x 35 x 1.8 Lm-14. Inv. 75229 Broken on all sides. One face flat and one moulded, both polished. On the vertical side of the slab, from bottom to top: cyma reversa, astragal, strip, astragal with cyma recta. Part of a slab molding, decorated with panels. Dimensions: 20 x 22.5 x 2.1

The examples Lm-5/14 belong to a typology of parietal revetment panels that was fairly widespread in imperial Roman times, but has not so far been the subject of an overall study. The scarcity of attestations of elements in situ or of finds associated with datable stratigraphies has meant that such items have been attributed to a very wide chronological span, ranging from the early imperial age all the way to the paleo-Christian period, as we shall see.

If we analyze the data in our possession in chronological order, we observe that slabs similar to those presented here are still visible in situ at Pompeii, in the base of the lararium of the House of Caecilius Iucundus. At Ostia, numerous fragments of slabs of greco scritto with decoration in rectangular panels and inscribed lozenges with inflexed sides come from the late first century strata of the of the Thermae of the Swimmer. Other examples are known in Rome, in the domus over the Sette Sale and at the Villa dei Quintili (unpublished example displayed in the antiquarium), at Luni, and at the Villa di S. Vincenzino near Cecina. From these data it seems clear that the artifacts under examination were already widespread at the beginning of the imperial period, perhaps as early as the first half of the first century. Similar dating is also suggested by the marble revetment, so far unpublished, of some public buildings of the forum of Carsulae, in Umbria (consisting of two different series of bardiglio slabs with decoration very analogous to those above), whose dating seems no later than the Flavian period.

Another indication of the antiquity of these slabs is their very frequent presence in parietal revetments of late antiquity, where they are clearly recycled. Note, for instance, the domus of the Nymphaeum of Ostia, where the fountain is lined with slabs decorated with simple rectangular panels in greco scritto, visibly cut and adapted to new use. Also at Ostia, in the Nymphaeum of the Erotes, there is a slab of greco scritto decorated with a lozenge, inserted almost as an emblem in the very homogeneous parietal revetment in white and grey marbles, underneath the apsed niche of the wall at the back.

Numerous whole or fragmentary examples of such artifacts, perfectly analogous to those just described, are found reused in the catacombs and other paleo-Christian buildings. These have been published in various volumes of the Corpus della Scultura Altomedievale. They are generally interpreted as plutei and dated to the first half of the sixth century on stylistic grounds, on the basis of presumed analogies with the decorative motifs found on the slabs of the presbytery enclosures from Constantinople surviving in many Roman churches, in particular at S. Clemente.

The proposed chronology can be contested for a number of reasons. First, the moldings that frame the slabs are of classic type, completely different from those found on artifacts of Byzantine provenance or tradition, which have a characteristic band sloping obliquely outwards, unknown in Roman architectonic sculpture. The dimensions of the slabs, too, in particular the height (between 25 and 95 cm) and the thickness (mostly of 2-3 cm, but also of 0.5 cm), seem too reduced to belong to plutei (see, for a comparison, the dimensions of the series of S. Clemente, of a homogeneous thickness between 5.5 and 7 cm and heights varying between 113.5 and 117 cm). Furthermore, some details of the decoration, for instance the urceus which appears on a panel from S. Lorenzo fuori le mura, or the peltae on a fragment from SS. Giovanni e Paolo, do not belong to the Byzantine iconographic repertoire, as the authors of the volumes of the Corpus themselves observe. On the other hand, the Byzantine slabs have some decorative elements that are wholly absent in those examined here, in particular the fleur-de-lys endings to the corners of the lozenges and the constant presence of an internal motif.

It seems evident at this point that the chronology of the fifth to the sixth century proposed in the Corpus must perhaps be taken as that of the reuse of the pieces. We cannot exclude the possibility, in fact, that the widespread reuse of these slabs in paleo-Christian architectural complexes was in some way encouraged by the contemporary diffusion of the Byzantine slabs, which have in common with ours the taste for the geometric division of the surface and the simplicity of the ornamentation.

Another issue to be resolved is the use of the slabs with relief decoration in parietal revetment. A primary classification seems possible on the basis of the dimensions. Slabs up to 40 cm high, varying in length (cf. for instance our Lm-12/14) and between ca. 1.5 and 2 cm thick (such as Lm-5/14), were probably used as a fascia for dividing horizontally, above the socle, or else as a fascia running along the top. The same spatial arrangement survives, much later, in the highly refined decoration in opus sectile of the “Building outside Porta Marina” at Ostia. Here the slabs with incised decoration are substituted by complex marble intarsia, in which the motif of the horizontal band decorated with peltae, lozenges or discs is repeated twice in its entirety. Another possible use for slabs of these dimensions is for revetting architraves or jambs (in particular for the fragments with a decoration of rectangles linked together in a series, although this is not attested at Licenza).

Larger slabs (one meter or more in height and 2.2-2.5 cm thick, such as Lm-1/5) may have been used in the lower or central part of the wall. These would be set next to each other (particularly when more complex sculpted architectural elements, such as capitals, are present in addition to the panelling), or they would alternate with pilasters.

Moldings (C)

Bardiglio C-1. Sector I.2, Area 50, SU 288 Broken on two sides. One face is polished and molded with cyma reversa, the upper and lower bases and the back face are smoothed. Dimensions: 11 x 3.3 x 2.8-2.4; weight: 220

White fine-grained marble C-2. Sector VI.1, Area 24, SU 6007 Broken on two sides. One face is molded, with two cymae reversae. The considerable corrosion of the surfaces prevents analysis of the tool marks. Dimensions: 8.1 x 2.2 x 1.3-2.2; weight: 170 C-3. Sector VI.1, Area 24, SU 6007 Broken on two sides. One face is polished and molded with cyma reversa, the upper and lower bases are smoothed. The back face does not survive. Dimensions: 5 x 3.2 x 2.3; weight: 87

Giallo antico C-4. Sector I.7, Area 35, SU 1242 (fig. 13) Broken on two sides. The molded face (with two cymae reversae) and the opposite one are polished, the upper and lower bases are smoothed. Dimensions: 10.7 x 3.2 x 2.8; weight: 380 C-5. Sector I.7, Area 35, SU 1242 (fig. 13) Broken on two sides. The molded face (with two cymae reversae) and the opposite one are polished, the upper and lower bases are smoothed. The back face still has the raised bit left by the saw cut and traces of mortar used to fix the molding to the wall. It is not perpendicular to the lower edge of the molding, but slightly inclined, forming an acute angle with it. Dimensions: 4.7 x 3.2 x 1.6-2.1; weight: 87 C-6. Sector I.7, Area 35, SU 1242 (fig. 13) Broken on two sides. The molded face (with two cymae reversae) and the upper base are polished, the lower base and the back face are smoothed. This last, which still shows the raised bit left by the saw cut, is not perpendicular to the lower edge of the molding, but slightly inclined, forming an acute angle with it. Dimensions: 5.2 x 2.6 x 1.9-2.4; weight: 65 C-7. Sector VII.1, Area 25, SU 7001 (fig. 13) Broken on two sides. All the faces, including the molded one (with two cymae reversae), are smoothed; the back face, furthermore, is not perpendicular to the lower edge of the molding, but slightly inclined, forming an acute angle with it. On the upper edge there is a circular hole (diameter 0.45, depth 1.3) for the insertion of a metal clip. Dimensions: 17.5 x 4.2 x 2.5; weight: 480

Rosso antico C-8. Sector I.7, Area 35, SU 1200 (fig. 13) Broken on two sides. The molded face (with two cymae reversae) is polished, the upper and lower bases are smoothed. The back face does not survive. Dimensions: 4.7 x 3.4 x 1.05-2.1; weight: 52 C-9. Sector I.3, Area 37, SU 400 (fig. 13) Broken on two sides. The molded face (with two cymae reversae) is polished, the upper and lower bases are smoothed. The back face does not survive. On the upper base there is a circular hole (diameter 0.5, depth 1.2) for the insertion of a metal clip. Dimensions: 5.4 x 2.8 x 1.2-2.1; weight: 72 C-10. Sector I.2, Area 50, SU 200 Broken on two sides. The molded face (with two cymae reversae) and the lower base are smoothed, while the upper one shows traces of being worked with a gradine. Dimensions: 11.5 x 3 x 4; weight: 420 C-11. Sector I.7, Area 35, SU 1220 Broken on two sides. The molded face (with two cymae reversae) and the upper base are polished, the back face and the lower base are smoothed. Dimensions: 7.5 x 3.5 x 2.9; weight: 170

The moldings, which served to divide the marble revetment into horizontal panels, are always characterized here by the most thorough workmanship; the two bases are smoothed or polished, and the external face molded with fillet and cyma reversa. The back face is generally smoothed but may be just rough-hewn, vertical or slightly oblique (in some cases, even sharply inclined), probably with the intention of helping the layer of mortar behind to adhere better. In some cases (cf. C-5) there is a raised bit, due to the saw cut; this shows clearly that these elements were cut from slabs of great thickness. Among the examples presented here, the moldings in giallo antico (C-4/7) and rosso antico marble (C-8/11) predominate; rarer are those in white marble or bardiglio (C-1/3). It must be emphasized that giallo antico and rosso antico, which are generally the materials most frequently used for making such artifacts, are found in almost canonical association with slabs of white or veined marble, and, respectively, with cipollino or giallo antico, which are also those most often found at “Horace’s villa” (see below).

Fillets (Li)

Palombino Li-1. Sector VII.1, Area 24, SU 7001 Broken on two sides: one side is flat and polished. Two edges are ground and slightly bevelled. Dimensions: 4.5 x 0.4 x 0.6; weight: 12

Giallo antico Li-2. Sector VII.1, Area 24, SU 7001 Broken on two sides. Two flat faces, of which one is polished. Two smoothed edges, worked with fine-ended chisel but not ground, slightly bevelled. Dimensions: 2.3 x 1.5 x 0.9; weight: 26 Li-3, 4. Sector VII.1, Area 24, SU 7001 Two similar fragments. Broken on two sides. Two flat faces, of which one is polished. Two smoothed edges, worked with fine-ended chisel, but not ground, slightly bevelled. Made from a pre-existing small panel (or fillet), because the visible surface of the panel—indicated by the inclination of the edges—is smoothed, while the reverse is polished. Dimensions: a) 2.2 x 1.4 x 0.6; weight: 18; b) 4.3 x 6.1 x 0.7; weight: 86 Li–5. Sector I.7, Area 35, SU 1218 Broken on two sides, two original edges survive, ground and slightly bevelled. Dimensions: 5.2 x 1.7 x 0.7; weight: 25

Pavonazzetto Li–6. Sector VII.1, Area 24, SU 7001 (fig. 14) Broken on two sides. Two flat faces, of which one is polished. One edge is ground and slightly bevelled and one is worked with marteline and finished with fine-ended chisel where it joins with the adjacent element. The different type of working of the two opposite edges indicates that the fillet was created by working a larger slab, of which one of the original borders was retained. Dimensions: 2 x 5.4 x 0.8 ; weight: 43 Li-7. Sector VII.1, Area 24, SU 7001 Broken on two sides. Two flat faces, of which one is polished. One edge is ground and slightly bevelled and one worked with marteline and finished with fine-ended chisel where it joins the adjacent element. Like the previous fragment (Li-6), it was probably reworked from a pre-existing slab. Dimensions: 3.6 x 2.8 x 0.6 ; weight: 32

Rosso antico Li-8. Sector VII.1, Area 24, SU 7001 Broken at the two ends. Two flat faces, of which one is polished. The edges are ground and slightly bevelled. Dimensions: 1.5 x 0.3 x 0.4; weight: 6

Elements for parietal intarsia (T)

Pavonazzetto T-1. Sector VII.1, Area 24, SU 7001 (fig. 14) Broken on two sides. Two flat faces, of which one is smoothed and one polished. Edges slightly bevelled and finished with fine-ended chisel. Belongs to a crusta originally triangular or rhomboid in form. Dimensions: 1.7 1.3 x 0.6; weight: 10

Giallo antico T-2. Sector VII.1, Area 24, SU 7001(fig. 14) Broken on two sides. Two flat faces, of which one is smoothed and one polished. Edges slightly bevelled and finished with fine-ended chisel. The crusta, irregularly triangular in shape, has one (or possibly two) sides curvilinear in line and a third side approximately straight. Dimensions: 2.3 x 2.5 x 0.4; weight: 8 T-3. Sector VII.1, Area 24, SU 7001 (fig. 14) Irregular edges, which are not enough to determine the overall shape of the crusta. Two flat faces, of which one is smoothed and one polished. Edges slightly bevelled and finished with fine-ended chisel. Fairly regularly triangular in shape, with two sides slightly inflexed. Dimensions: 2.9 x 2.6 x 0.35; weight: 7 T-4. Sector VII.1, Area 24, SU 7001 (fig. 14) Broken at one end. Two faces flat, of which one is smoothed and one polished. Edges slightly bevelled and finished with fine-ended chisel. The small crusta, irregular in shape, has one straight side, one slightly curvilinear and one sinusoidal. Dimensions: 3 x 1.2-1.9 x 0.41; weight: 9

It is difficult to determine whether the crustae T-1/4, characterized by their particular shape and extreme thinness (between 0.35 and 0.6), can be attributed to decoration in complex-motif opus sectile or to interraso marmore. However this may be, the thickness, while not a definitive proof, is an important indication for assigning such fragments to parietal decoration.

Slabs (La)

Bardiglio La-1. Sector I.2, Area 50, SU 231 Broken on all sides. Two flat faces, of which one is polished and one smoothed. Dimensions: 5.9 x 3.5 x 1.2; weight: 40

Breccia corallina La-2. Sector I.2, Area 50, SU 231 Broken on three sides. Two flat faces, of which one is smoothed and one is rough-hewn; the latter still has the raised bit left by the saw cut. Dimensions: 9 x 6 x 2.1; weight: 170 La-3. Sector VII.1, Area 24, SU 7001 Broken on three sides. Two flat faces, with no identifiable tool marks on account of the wear of the surfaces. One edge survives, ground and slightly bevelled. Dimensions: 3.2 x 2.5 x 0.4; weight: 20

Cipollino La-4. Sector I.7, Area 35, SU 1218 Broken on three sides. Two flat faces, of which one is polished and one smoothed; one edge, worked with marteline and bevelled, survives. Dimensions: 4.5 x 8 x 1.5; weight: 170 La-5. Sector I.7, Area 35, SU 1218 Broken on two sides. Two flat faces, of which one is polished and one smoothed; two borders, worked with marteline, finished with fine-ended chisel and slightly bevelled, survive. Dimensions: 6.1 x 7.6 x 1.1; weight: 150 La-6. Sector I.7, Area 35, SU 1218 Three fragments, broken on all sides, belonging to the same slab. Two flat faces, of which one is polished and one smoothed; no original edge survives. Dimensions: a) 2.1 x 4 x 1; weight: 70; b) : 5.5 x 4.1 x 1; weight: 80; c) 5.2 x 1.5 x 0.9; weight: 70

Giallo antico La-7. Sector I.2, Area 50, SU 231 Broken on all sides. Two flat faces, of which one is polished and one smoothed. Dimensions: 5.5 x 4 x 1.3; weight: 43 La-8. Sector I.2, Area 50, SU 288 Broken on three sides. Two flat faces, of which one is polished and one smoothed. Only one original edge survives, ground and slightly bevelled. Dimensions: 3.5 x 1.7 x 0.65; weight: 25 La-9. Sector VII.1, Area 24, SU 7001 Nine fragments with two flat faces, of which one is smoothed and one polished; no original edges survive. Dimensions: 0.45-0.9; weight: 340. La-10. Sector I.7, Area 35, SU 1218 Two flat faces, of which one is smoothed and one polished; no edge surviving. Dimensions: 3.4 x 2.5 x 0.45; weight: 35 La-11. Sector I.7, Area 35, SU 1218 Two flat faces, of which one is smoothed and one polished; two original edges survive, worked with chisel and slightly bevelled. Dimensions: 1.8 x 3.3 x 0.7; weight: 40

Greco scritto La-12. Sector I.7, Area 35, SU 1218 Broken on all sides. Two flat faces, of which one is smoothed and one polished; none of the original edges survive. Dimensions: 13.5 x 11.5 x 1.6; weight: 430 La-13. Sector I.7, Area 35, SU 1218 Broken on all sides. Two flat faces, of which one is smoothed and one polished; none of the original edges survive. Dimensions: 4.3 x 2.2 x 1.25; weight: 80 La-14. Sector I.7, Area 35, SU 1218 Broken on two sides. Two flat faces, of which one is smoothed an the other rough-hewn; two original edges survive, right-angled one to the other, ground and slightly bevelled. Corner of a revetment slab. Dimensions: 3.9 x 4.8 x 1.1; weight: 90 La-15. Sector I.7, Area 35, SU 1218 Broken on all sides. Two flat faces, one of which is smoothed and one rough-hewn; none of the original edges survive. Dimensions: 3.7 x 4.5 x 1.1; weight: 50

White medium-grained marble La-16. Sector I.2, Area 50, SU 231 Broken on all sides. Two flat faces, of which one is smoothed, with traces of whitewash, and one is polished, with remains of grey-beige mortar containing pozzolana elements. Dimensions: 9.2 x 6.2 x 1.4; weight: 120 La-17. Sector I.3, Area 37, SU 412 (fig. 14) Broken on two sides; two original edges survive. One edge is ground, slightly bevelled; the other edge has a regular undulate profile, worked first with marteline and subsequently ground on its upper half, nearer to the exposed face. Two flat faces, of which one is smoothed and one polished, both with traces of mortar, perhaps indicating reuse of the slab. Dimensions: 22.4 x 3.8 x 1.7; weight: 340 La-18. Sector I.2, Area 50, SU 231 Broken on three sides, one edge surviving, curvilinear and worked with marteline. Two flat faces, of which one is polished, with conspicuous traces of grey-beige mortar containing pozzolana elements and ferrous oxide sediment (probably due to the presence of fastening clips). The other face, smoothed, has traces of whitewash painted on with a brush (the brush-strokes are evident); in some areas, small patches of violet pigment are present. Dimensions: 18 x 15 x 1.8; weight: 520 La-19. Sector I.2, Area 50, SU 231 Broken on all sides. Two flat faces, of which one is smoothed and one polished. Dimensions: 19.7 x 7 x 1.8; weight: 480 La-20. Sector I.3, Area 37, SU 412 Broken on all sides. Two flat faces, of which one is smoothed and one simply rough-hewn, on which can be seen traces of grey-beige mortar containing pozzolana elements. Dimensions: 18.5 x 10 x 1.4; weight: 260 La-21. Sector I.3, Area 37, SU 412 Thirty-five fragments belonging to a single slab. Two flat faces, of which one is smoothed and one polished. Edges worked with marteline and finished with flat-ended chisel, only in the upper half nearest to the exposed face. Dimensions: max. 16.5 x 15 x 3.3; min. 1 x 1 x 2.5; weight: 3450 La-22. Sector I.2, Area 50, SU 320 Broken on two sides. Two flat faces, of which one is smoothed and one polished. One original edge survives, ground and slightly bevelled. Dimensions: 7.5 x 8 x 2.5; weight: 220 La-23. Sector I.2, Area 50, SU 261 Three slab fragments. Two flat faces, of which one is smoothed and one rough-hewn; no edge survives. Dimensions: a) 4.5 x 4 x 1.9; b) 2.8 x 2.9 x 1; c) 7.2 x 6 x 2.9; weight: 670 La-24. Sector I.2, Area 50, SU 288 Broken on one side. Two flat faces, of which one is smoothed and one polished. Two original edges survive, of which one is ground and slightly bevelled, and the other is smoothed with fine-ended chisel. Dimensions: 7.1 x 4.2 x 2; weight: 150 La-25. Sector I.2, Area 50, SU 261 Two flat faces, of which one is smoothed and one rough-hewn; no edge survives. Dimensions: 8 x 6.3 x 1.4; weight: 340 La-26. Sector I.2, Area 50, SU 261 Two flat faces, of which one is smoothed and one rough-hewn. Broken on two sides; only one original edge survives, worked with fine-ended chisel. Dimensions: 8 x 6.2 x 1.3; weight: 310 La-27. Sector I.3, Area 37, SU 412 Two flat faces, of which one is smoothed and one rough-hewn; no edge survives. Dimensions: 14 x 6 x 1.1; weight: 132 La-28. Sector I.2, Area 50, SU 320 Two flat faces, of which one is smoothed and one polished. Broken on two sides; two of the original edges survive, of which one is worked with marteline and one is worked with fine-ended chisel. Dimensions: 12.6 x 9.7 x 2; weight: 310 La-29. Sector I.7, Area 35, SU 1218 Two flat faces, of which one is smoothed and one polished. Broken on three sides; only one edge survives, worked with marteline and finished with flat-ended chisel. Dimensions: 5.4 x 4 x 1; weight: 240 La-30. Sector I.7, Area 35, SU 1218 Two flat faces, of which one is smoothed and one polished. Broken on two sides; two edges, worked with marteline, survive at right angles to each other, on which are visible traces of mortar. Dimensions: 11.6 x 10 x 1.4; weight: 410

Pavonazzetto La-31. Sector VII.1, Area 24, SU 7001 Two fragments belonging to the same slab but not joining. Broken on three sides. Two flat faces of which one is polished and one smoothed. One edge survives, first worked with a marteline and then half-ground, towards the visible face of the slab, to permit better adhesion to the next slab. Dimensions: a) 8.5 x 6.9 x 0.8; b) 7.3 x 5.8 x 0.8; weight: 360 La-32. Sector VII.1, Area 24, SU 7001 Broken on all sides; no original edge survives. Two flat faces, of which one is smoothed and one polished. Dimensions: 3.4 x 2.3 x 0.7; weight: 19 La-33. Sector I.2, Area 50, SU 320 Two flat faces, of which one is smoothed and the other polished. Broken on three sides; only one original edge survives, ground and slightly bevelled. Dimensions: 9.6 x 5.4 x 1.05; weight: 110 La-34. Sector I.7, Area 35, SU 1218 Broken on all sides, no original edge surviving. Two flat faces, of which one is smoothed and one polished. Dimensions: 5.4 x 3 x 1.6; weight: 120

Portasanta La-35. Sector VII.1, Area 24, SU 7001 Two flat faces, one of which is polished, with many traces of mortar (perhaps indicating reuse of the fragment) and one is smoothed; no edge survives. Dimensions: 5 x 5.7 x 1.2; weight: 67

Rosso antico La-36. Sector VII.1, Area 24, SU 7001 Broken on all sides, no edge surviving. Two flat faces, on which no traces of tool marks can be identified due to the worn condition of the surfaces. Dimensions: 1.6 x 2.5 x 0.4; weight: 12

On the basis of the marble types here present, the principal observation is that the variety is rather limited; there is in fact a prevalence of white or veined marbles (particularly Lunense), grey (bardiglio) and colored of the commonest types (giallo antico, pavonazzetto, cipollino, etc.).

Among the white marbles, fine-grained ones that may come from Carrara, and medium-grained ones, which are on the whole imported from the Aegeo-Anatolian area, are especially predominant. Among the latter, fragments characteristic of Proconnesian and Parian marble have been frequently identified as well.

Of the commonest colored types, absolutely the most abundant are the giallo antico, pavonazzetto, and cipollino, which are among those most widespread and most often used in revetments from the earliest years of the empire, and the “greco scritto” marble, with white or grey ground, of the type from Tunisia that was introduced slightly after the beginning of the imperial age. Another marble found at the villa, but not in abundance, is breccia corallina. Surprisingly rare, however, is portasanta, one of the most widespread marbles from the early imperial age onwards.

The complete absence of Egyptian red porphyry and Greek green porphyry (used with a certain lavishness from the last decades of the first century) must be noted, as well as the so-called breccia “verde antico di Tessaglia,” which was exported only from the time of Hadrian onwards, of the “broccatello di Tortosa” and of the “fior di pesco” marbles, which were also mostly used in late imperial times.

As for the working techniques, the slabs are mostly characterized by having one face polished, which was intended to be visible. The opposite face sometimes shows the same treatment; more frequently, it is simply smoothed or left rough, with clear signs of the wire-saw cutting. The marks left by this instrument are easily recognizable: there is a series, more or less serried, of parallel lines, slightly oblique or curvilinear. On some slabs a raised bit, more or less thin, can be seen; this indicates the furthest point reached by the saw. To break off a slab, it was enough to reach 0.5-1 cm from the lower end of the block (2-3 cm for the thicker slabs), in order that the cut part could be separated with relative ease; this would leave a sort of listel with an irregular surface at the base.

Most of the fragments have clean cuts, with edges polished and slightly bevelled; in at least one case an undulate edge was found. The use of undulate joints is certainly not new in Roman architecture; note those of the granite columns of the Pantheon, of the Temple of Venus and Rome or of the Basilica Ulpia, or the columns revetted with marble crustae of irregular undulate shapes, such as found in the Fossa Traiana of Ostia. Examples such as these attest the very high specialization of Roman craftsmen in the cutting and working of these materials. The aim of this technical artifice was to make the join of two differently patterned elements as pleasing to the eye as possible. Such manipulation was often necessary after some technical problem had occurred, such as breaking during transport or mounting, or in those cases in which only reused and/or waste (i.e., not homogeneous) marble was available. Undulate joints in wall or floor revetments, particularly for the joining of slabs of homogeneous colour and type, seems so far to be little documented. They have been found particularly at Brescia, in the floor of the left cell of the Capitolium and at Capua, in the floor of a domus of late antiquity that is largely made up of recycled material. Again at Brescia, they are found in the parietal revetments, still unpublished and under study by the present author, of the Roman theater, probably to be attributed to a decorative phase of the Flavian period, contemporary with that of the Capitolium already mentioned.

Overall, the chronology of the slab fragments we have presented would seem, in synthesis, not to go beyond the middle of the second century and may perhaps be earlier by several decades (i.e., belonging to the Flavian period), given the absence of the inserts of porphyry and the presence of other technical characteristics, such as the typology of the cuts, which are always very clean.

Whereas many of the rooms in the so-called Villa of Horace show obvious traces of mosaic pavement, the remains of sectilia pavimenta are quite scarce. As is well known, this kind of flooring came into use as early as the first century B.C., when marble was not yet employed and when the patterns were small. Its use continued until late antiquity with the development of more or less complex decorative patterns and with the employment of various kinds of marble.

In the case of the earlier excavations at the Villa of Horace, the only relevant published notice is given by Lugli, who mentions a fragmentary pavement “with slabs of palombino and white marble, combined in a cross pattern” originating in “atrium A” (i.e., Area 8) and “from room B1,” (i.e., room 7). These were subsequently detached and reintegrated by Pasqui on a panel that was long stored in the Licenza museum. Following the reorganization and remodeling of the old Antiquarium at Licenza, in the early 1990s, this panel—together with a great deal of other material found in the Villa of Horace—was transferred to the storehouse of the Archaeological Superintendency of Lazio at the Santuario di Ercole Vincitore in Tivoli, where it is found today.

The panel (fig. 10) measures 40.5 cm x 42.0 cm and contains a fragment of pavement in small format, consisting of squared tiles, 14.5 cm per side, set into the pavement with an axis of rotation of 45° in accordance with a design generally denoted Q. Due to the limited space afforded by the panel, only five complete tiles are preserved, of which the central one is palombino and four are made of a fine-grained white marble with blue veining. Each side of the central tile is contiguous to one of the sides of the tiles in white marble.

This ensemble was the original nucleus of the pavement, which repeated the same pattern over most of the surface of the floor. When the panel was created and the tiles were repositioned (set on a bed of plaster), the empty spaces and frame were filled in with eight foreign triangular pieces, of which four are made of a white-veined marble, two of palombino and two of greco scritto. These eight tiles were disposed without respect for what can be assumed to have been the floor’s original pattern.

The decorative motif of this pavement is the simplest of all known motifs. It is at the very boundary of the definition of the term opus sectile as it is commonly understood, since it is made of materials that are essentially the same color.

We do not know whether the pavement has been correctly reintegrated. In any event, such a sectile can be easily associated with the checkerboard pattern well known in the late Republic, especially in versions that are not entirely made of marble (e.g., slate and palombino in sharp chromatic contrast). Such versions are documented in the famous frescoed triclinium of the Villa of the Mysteries and in the house of Holconius Rufus in Pompeii. At Rome, it is found in the House of Augustus on the Palatine and in the Domus Pactumeiorum. At Tivoli, it is found in the mensa ponderaria.