The villa at Matrice lies a few kilometres to the north east of Campobasso and south of the municipium of Fagifulae, to whose territory it presumably belonged (Lloyd, Reference Lloyd1991a: 261). The site dominated a fertile valley just below the southern Biferno watershed, an area the subject of intensive fieldwalking by the Biferno valley survey between 1974–78 (Barker, Reference Barker1995). The site was first discovered in the late 1970s during the construction of a road to the north of the church of Santa Maria della Strada. Excavations, directed by John Lloyd, were undertaken by a joint team from the Universities of Aberdeen and Sheffield between 1980 and 1984 (Lloyd and Rathbone, Reference Lloyd and Rathbone1984).

The site was first inhabited during the late Bronze Age/early Iron Age, but it was in the later Samnite period (c. 200 BC) that a much larger structure built of roughly trimmed limestone blocks was constructed which has been interpreted as a substantial farmstead (Lloyd, Reference Lloyd, Barker and Lloyd1991b: 184). The advent of Roman control in the area seems to have affected the villa considerably. The density of habitation in the area decreased considerably and Matrice emerged as the main or only site at the head of the valley.

In the early first century AD a substantial restructuring and enlargement of the villa took place, with many new rooms built in opus reticulatum. This larger villa contained a suite of pressrooms and sunken dolia, devoted most likely to the production of wine. However, the paucity of amphorae from the site suggests that this was not a large-scale operation, which agrees with the modesty of the press installations (Lloyd, Reference Lloyd1991a: 262). Imported amphorae, obviously reflecting consumption rather than production, had been mainly from Puglia during the Republic, but in the Empire came from Campania and later from North Africa.

The villa continued to flourish into the late second century AD: the excavations discovered a refuse deposit containing pottery and glass from Etruria, Campania and North Africa – over a third of the vessels in the deposit were non-local fine tablewares. Even in the later empire, the 4^th and early 5^th centuries AD, quantities of fine pottery from North Africa and fine quality Red Painted tablewares were common. Coin and pottery evidence suggests that occupation of the villa continued, though possibly on a reduced level, into the early sixth century. The end of the villa's life may be indicated by the burial of a small child c. AD 550 in one of the drains running through the building. No further activity is detectable on the site, and it appears to have been abandoned, until c. AD 1600, when a substantial building, probably of agrarian character, was erected in its ruins.

In 2017 the Ashmolean Museum, King's College London and British School at Rome applied for initial funding to bring the excavations to publication. The fieldwork began with a geophysical survey in October 2017, funded by the Roman Society, of the areas not previously investigated, in particular a field bordering the excavations to the west as well the modern road that bisects the site. The aim of the prospection was to define the limits of the site, checking the hypothesis that the northern and southern walls of the Imperial villa had been correctly identified by the earlier excavations. The gradiometer survey to the west of the road recorded several structures, in particular a well-defined rectangular feature on a similar alignment to the previously excavated walls. Ground-Penetrating Radar (GPR) was used to investigate the modern road as well as inside some of the unexcavated rooms to the east and several small open areas in the woodland to the east of the villa.

The discovery of structures to the west of the road allowed the opportunity to explore further parts of the site through excavation, in order to refine the chronology suggested by the earlier excavations as well assess the limits of the site and identify further rooms within the villa.

The 2018 seasons focused upon five areas (Fig. 1):

Fig. 1. Plan of the Roman villa at Matrice together with the location of the 2018 trenches.

The excavation in Area 1 revealed three distinct chronological phases, from the Republican period through to occupation in the Imperial phase and finally the construction of a drain in late antiquity. The lower walls of the Republican phase appear to have been demolished in the Imperial period when the villa was expanded, as the walls were cut to allow the construction of a wider wall slightly further to the north.

The previous excavations of the villa had identified a phase of Samnite occupation at the northern limit of the site (Lloyd, 1991: 261, fig. 29) characterised by a series of regularly aligned limestone blocks. The 2016 GPR survey indicated a further wall leading westward which appeared to intersect with the previously excavated structure. The 2018 excavation (Fig. 1: Area 2) revealed a further seven limestone blocks which were bonded to the north-south wall with a coarse yellow mortar. The associated material, including black glossware and dolia, suggest an occupation associated with these walls to the 2^nd century BC.

The strong magnetic anomaly recorded in Area 3 immediately to the west of the modern road was shown to be a large cistern (measuring 8.42m in length, 4.22m in width and 1.42m in depth), constructed with tile and stone, that was lined with a red cocciopesto. The eastern end had a lower internal wall with a rounded top also covered in cocciopesto (Fig. 2). The fill of the structure indicates that the structure was roofed, with a layer of collapsed roof tiles forming the first layer of the fill above the floor. The subsequent higher contexts show that the structure was out of use by the 3^rd century AD. Whilst the previous excavation had recorded a network of drains and rooms indicating agricultural practice, it had been unclear where the water source was to allow these processes. A brick and tile channel leading from the cistern towards the east indicates that the water was distributed from this cistern that lay at the highest north-western corner of the site.

Fig. 2. The northern internal wall of the cistern (Area 3).

Areas 4 and 5 (Fig. 1) were excavated with the aim of examining both the northern limit of the site as well as establishing the chronology within a further room along the northern range of the villa. Area 4 to the west of the modern road indicated traces of the wall, but deep ploughing and the construction of the road had disturbed much of the area. To the east of the road (Area 5) the excavation recorded a significant occupation level dating to the 5^th century AD, similar to that which had been recorded by the excavations in the 1980s in the adjacent room.

The geophysical survey and new excavations of the villa at Matrice have revealed the full extent of the site and confirmed several hypotheses put forward by John Lloyd. However, a number of questions still remain regarding the Samnite phase, in particular the precise nature of the settlement where the walls are substantially constructed from large limestone blocks. The excavation also revealed that there was some remodelling of the site between the Republican and Imperial phases, as well as an important phase of settlement in late antiquity.