8%&([FDYDWLRQVRIWKH5RPDQ9LOODDW*HUDFH6LFLO\ 5HVXOWVRIWKH6HDVRQ

5-$:LOVRQ

Mouseion: Journal of the Classical Association of Canada, Volume 12, Number 2, 2012, LVI -- Series III, pp. 175-230 (Article)

3XEOLVKHGE\8QLYHUVLW\RI7RURQWR3UHVV

For additional information about this article http://muse.jhu.edu/journals/mou/summary/v1012/1012.2.wilson.html

Access provided by University of British Columbia Library (17 Sep 2015 21:18 GMT) Mouseion, Series III, Vol. 12 (2012) 175–230 © 2015 Mouseion (published in 2015)

UBC Excavations of the at Gerace, : Results of the 2013 Season

R.J.A. Wilson

Abstract/Résumé

The paper describes results from UBC excavations at the Roman villa of Gerace, Sicily, in 2013. Geophysical survey in 2012 demonstrated that some half a dozen further structures once existed here, in addition to the small villa-like building partially inves- tigated by others in 1994 and 2007. In 2013 excavation concentrated on rooms 1 and 2 of the latter residence. The former, provided with a bench, a work-top and an earth floor, may have been a kitchen. Room 2 had a white mortar floor and plastered walls. A small portion of -paved corridor outside these rooms was also investigated. The building was erected not before ca. ad 370 and perished in a fire in the second half of the fifth century. Immediately to the east a building suggested by geophysics to have been some 50 m long was trial-trenched. The part excavated was paved with an intact floor of stone flags. The paving extended up to 2.20 m beyond the structure’s east and west exterior walls, possibly to ensure it was kept dry. It may have been the estate gra- nary or storebuilding. Probably built in the first half of the fourth century, replacing an earlier structure below, it had a short life: it collapsed dramatically, perhaps in the earthquake of ad 361/363, and was never rebuilt. A heated room, possibly belonging to a bath-suite in the villa-like building, was found in the west end of the trench; part of the granary’s roof collapse and west wall were removed to provide space for the later structure’s stoke-hole. Among the finds were 99 stamps on roof tiles from the villa-like building. Ten dies were recorded, eight of them varieties of the name Philippianus, who may have been the owner of the Gerace estate in the later fourth century.

Cet article présente les résultats des fouilles menées en 2013 par l’UBC à la villa romaine de Gerace en Sicile. Une prospection géophysique avait révélé en 2012 l’existence de près d’une demi-douzaine de structures, en plus du petit bâtiment de style villa, par ailleurs déjà partiellement exploré en 1994 et 2007. En 2013, les fouilles de ce bâtiment se sont concentrées sur les chambres 1 et 2. La première, munie d’un banc, d’un plan de travail et d’un plancher en terre battue, pourrait avoir été une cuisine. La deuxième possédait un plancher en mortier blanc et des murs recouverts de plâtre. Une por- tion du corridor pavé de mosaïques à l’extérieur des chambres a également été exa­ minée. L’édifice fut construit au plus tôt en 370 après J. C. et détruit par les flammes durant la seconde moitié du Ve siècle. Immédiatement à l’est, une structure, évaluée à 50 m de long selon l’examen géophysique, a fait l’objet d’un sondage par tranchée, qui a révélé un pavement intact de dalles en pierre. Ce pavement se prolongeait jusqu’à 2,2 m au-delà des murs extérieurs à l’ouest et à l’est du bâtiment, dans un but probable

175 R.J.A. Wilson de régulation de l’humidité. Il pourrait s’agir du grenier à grain ou de l’entrepôt du domaine. Construit vraisemblablement durant la première moitié du IVe siècle, en remplacement d’une structure antérieure sous-jacente, il fut en usage peu longtemps ; il s’effondra brutalement, peut-être lors du tremblement de terre de 361-363 après J. C., et ne fut jamais reconstruit. Une salle chauffante, qui devait jouxter la salle de bain de la villa, a été découverte à l’extrémité occidentale de la tranchée ; une partie du toit du grenier effondré et du mur occidental avaient été retirées pour faire place au foyer de la chaufferie (praefurnium) de cette pièce. Parmi les découvertes figuraient 99 tuiles de toit provenant de l’habitation principale et marquées d’étampes, pour un total de dix coins, dont huit montraient des variantes du nom de Philippianus, probable pro- priétaire du domaine de Gerace à la fin du IVe siècle.

Introduction Contrada Gerace lies in the heart of Sicily in well-watered rolling country- side ten km due south of Enna. It is situated just south and east of Strada Provinciale 78, which leads towards Barrafranca, a further 17 km by road to the south-west. The site lies 15 km (as the crow flies) north-west of Sicily’s most famous late Roman villa, that in contrada Casale near (fig. 1). Contrada Gerace takes its name from the highest peak in the district,

Figure 1. Location map of Gerace, also showing the position of other major fourth-century villas in Sicily Map drawn by Eric Leinberger (Vancouver)

176 UBC Excavations at Gerace, 2013

Figure 2. Gerace, topographical survey, showing the location of Areas A and B. G1 and G2 represent the area of trenches dug in 2000 by Dott. Lorenzo Guzzardi. The aqueduct marked on the plan is modern, part of the water supply built to serve the city of Caltanissetta Map prepared by Alan Weston (BC Institute of Technology, Vancouver)

Monte Gerace, which rises to a height of 775 m about 500 m north-east of the area of archaeological interest. The latter, which covers about 3 ha, sits in a sheltered location at around 650 m above sea level (fig. 2), backed to the north and east by an amphitheatre of low hills but opening up to the west and south, where the site enjoys an extensive panorama over the neighbour- ing countryside (fig. 3). Today this comprises rich farming land, engaged in polyculture—in addition to grain, there are vines, olives, vegetables, and numerous fruit trees, especially citrus and almonds. That a Roman villa existed in contrada Gerace was first brought to the attention of the local Soprintendenza alle antichità in 1994, when erosion of the sides of a water channel swollen with winter rains exposed a portion of a geometric polychrome mosaic pavement.1 Subsequent excavation by

1 The gully originally ran on the west side of the fence line running from near the north-east corner of fig. 2 in a south-westerly direction. Later, to protect the building

177 R.J.A. Wilson

Figure 3. View looking south-west from the west side of Area A Photo by R.J.A. Wilson

Dott.ssa Enza Cilia Platamone revealed a small building, designated Area A on fig. 2, consisting of an apsed room measuring 9 m x 6 m (6), four adjacent rooms to the west (1–4) and corridors of unequal widths on the south and the west sides (fig. 4). The plan was revealed by surface clearance of the top of the walls, but limited trenching also demonstrated that the west end of the south corridor and the apsed room were paved with geometric and that a small marble fountain was a feature in the apse of the latter.2 Dott.ssa Cilia Platamone suggested that the Gerace villa was not built earlier than the beginning of the third century, and suggested that it had been abandoned, perhaps because of flooding, at some later date, which she tentatively sug- gested might have occurred at the end of that century. Dott.ssa Cilia Plata- mone’s excavations also revealed that, overlying the ruins of the Roman villa, was a later building, consisting principally of a single rectangular room (5 on

identified in 1994, it was diverted westwards near Point G2 on fig. 2 (labelled “ditch”). The following abbreviations of Sicilian provincia names are used below: AG: Agri- gento; CL: Caltanissetta; CT: ; EN: Enna; ME: Messina; PA: Palermo; RG: Ragusa. 2 Cilia Platamone 1996; 1997. Its existence was kindly indicated to me by Dott.ssa Enza Cilia Platamone (personal communication). For the wider context of the Enna region in the Roman period, see most recently Valbruzzi 2012.

178 UBC Excavations at Gerace, 2013

Figure 4. Plan of Area A (2007). The single-roomed structure at an oblique angle (“Room 5”) post-dates the destruction by fire of the main building in the second half of the fifth century— how much later is uncertain Plan courtesy of the Regione Siciliana, Assessorato Beni Culturali e Identità Siciliana, Soprintendenza di Enna fig. 2) with a portion of an adjacent paved yard to the south. Dott.ssa Cilia Platamone proposed that this structure may have belonged to somewhere between the tenth and the thirteenth century ad.3 A further excavation at Gerace was conducted by Dott. Lorenzo Guz- zardi, then of the Enna Soprintendenza, in 2000: this did not concern the part identified by Dott.ssa Cilia Platamone, but opened two new small areas some 60 m to the north (Area G1 and G2 on fig. 2). Further struc- tures interpreted as early medieval were revealed, with Roman layers beneath, but the excavation area was not large enough to yield detailed results.4 No further excavation was conducted at Gerace until 2007, when

3 On the proposed dating, Cilia Platamone 1996: 1687; 1997: 273–275. 4 Guzzardi 2001–2002: 582–584.

179 R.J.A. Wilson

Dott.ssa Carmela Bonanno, then director of the Sezione Archeologica of the Soprintendenza per i Beni Culturali e Ambientali of Enna Province, returned to the structures identified by Dott.ssa Cilia Platamone and con- ducted limited excavations at the west and east ends of the south corridor; she also uncovered most of the mosaic in the apsed room, except for the part overlaid by the medieval building at the higher level (fig. 4). Her work also showed the existence of a further room at the north-west corner of the excavated area (11); the possible existence of a north corridor between it and Rooms 3 and 4; and parts of further rooms to the west (7–10). In all cases, the walls of these rooms were exposed only superficially, with floor levels not being reached. Dr. Bonanno dated the villa and its mosaics to the late second century or early third century on the basis of the earli- est pottery found on the site, and postulated occupation until the end of the fourth century or the beginning of the fifth century, with fire destruc- tion after a period of abandonment. She also identified a further phase of occupation in the early Byzantine period (to which she ascribed the burial 20 m north of the main block). Dr. Bonanno also detected some medieval activity, on the basis of pottery (but not structures) which she dated to the eighth or ninth century.5

Research in 2012: The Geophysical Survey While this earlier work helped to underline the potential of the site at Gerace, it was clear that much still remained to be learned about its archae- ology. In particular, surface evidence, while not plentiful, suggested that the Roman site was much larger than the areas investigated up until 2007. With that in mind, the present writer commissioned a program of geophysical research, conducted on his behalf in May 2012 by a team of geophysicists led by Dr. Sophie Hay of the British School of Rome.6 This revealed that the Roman site is likely to have covered about 3 ha, although one field on the west and south-west sides of the building investigated by Dott.sse Cilia Platamone and Bonanno was unavailable for research because of a stand- ing crop.7 Elsewhere, discoveries were made in four principal areas (fig. 5): (a) a series of scattered buildings to the north-west (7–10); (b) an amorphous area of building to the north-east of the excavated villa (3–4); (c) a large and well-defined building (2), some 50 m long and perhaps 15 m wide, imme- diately east of the known “villa,” with at least some of its walls predicted as

5 Bonanno et al. 2010, especially 261 (the pottery reported there comprises sur- face finds from the site, especially from the adjacent almond orchard); Bonanno 2012; 2013a: 183 and 185 (for the dating); 2013b; 2014; forthcoming. 6 Wilson 2014a. 7 Subsequent geophysical research was conducted here by a British School at Rome team led by Sophie Hay in October 2013, and it revealed that ancient structures did not continue into this field.

180 UBC Excavations at Gerace, 2013

Figure 5. Interpretative results of the geophysical survey (2012): 1 the 1994 villa-like building (Area A); 2 the elongated structure; 3 farm buildings (or earlier villa?); 4 ancillary buildings (or a continuation of 3?); 5 and 6 location of kilns; 7–10 outbuildings Plot prepared by Dr. Sophie Hay (British School of Rome/University of Southampton)

181 R.J.A. Wilson being approximately one metre wide; and (d) a group of kilns lying on the southern escarpment of the site where the ground falls sharply away: five in total were identified by the geophysical work (5 and 6 on fig. 5). The discov- ery of kilns in such a location is not a surprise: making use of the natural slope ensures that the structures are naturally insulated by the rising ground into which they are built, minimizing heat loss while the kilns are in use. Not all of the structures and buildings, of course, need have been in use simulta- neously: only excavation will be able to establish their precise chronological sequence. All of the structures revealed by the geophysics, however, are on the same alignment as the Roman structures exposed in Area A (fig. 2), and there is therefore at least a prima facie case for treating them all as belonging to the Roman period.

Goals of the First Excavation Season, 2013 The first season of excavation at Gerace under the direction of the writer was carried out over a period of four weeks between 20 May 2013 and 15 June 2013. This set as its principal goals: (1) to determine whether other rooms in Area A, apart from the south corridor and the apsed room, had mosaic floors or whether they served merely as service quarters for the higher-status spaces that lie adjacent; (2) to confirm the date of the building in Area A; and (3) to establish if possible the function (residential or utilitarian?) and the date of the large rectangular structure identified by the geophysical survey (2 on fig. 5) and to determine its relationship, if any, to the villa-like building that lies immediately to the west.

The Villa-Like Building The excavations of Dott.ssa Cilia Platamone demonstrated the ground plan of Rooms 1–4 (fig. 4) to the west of the apsed room (6) but had not investi- gated them in depth, although her plan suggested that small trial trenches had been made in two of them (the north side of 1 and the doorway area of 2). It was not therefore certain at the outset of the 2013 season whether the mosaic pavements known to lie in Room 6 and in the south corridor also extended to other rooms in the building. The new work therefore con- centrated on excavating the southern part of Room 1 and the eastern half of Room 2 in the hope of discovering more both about them and the building’s history. A small section of the south corridor was also sampled (fig. 6). The irregularity of the south wall of rooms 1 and 2 is striking (see also fig. 4): a change of direction occurs halfway along the south wall of Room 2 and appears to be due solely to a surveying error, as there is no reason to think that Rooms 1 and 2 are other than contemporary.8

8 The doorway in the middle of the south wall of Room 2 was blocked in a sec- ondary alteration (see pp. 185–186), but otherwise this part of the villa seems to have been laid out in a single phase of construction.

182 UBC Excavations at Gerace, 2013

Figure 6. Area A, Rooms 1 and 2, aerial photograph at the close of excavation in 2013. Room 1 is on the left, where the southern half was excavated; of Room 2 (on the right), the eastern half was excavated. The photograph also indicates the part of the southern corridor (bottom) uncovered in 2013 Photo by F. Ianni, Regione Siciliana, Assessorato Beni Culturali e Identità Siciliana, Soprinten- denza di Enna

Area A, Room 1 Room 1 proved to have been a service room, possibly a kitchen. Its walls were not plastered, and two utilitarian structures had been built against the south wall. One of these structures, to the east of the entrance into the corridor, was a platform 82 cm high, 174 cm long and 46 cm deep (fig. 7). It made a straight joint with the walls it abutted but was probably an inte- gral part of the room from the start. It may have served as a kind of “work station” designed for the preparation of food, if the interpretation of the room as a kitchen is correct. To the west of the doorway was a low bench 43 cm high, 130 cm long and 51 cm deep. The floor of the room was of beaten earth. There was a good deal of animal bone (mostly of cattle) laying flat on it, but very little pottery. A trial slot dug in front of the bench confirmed that the floor did not lie at any lower level (the bench structure does not continue downwards), although the detection of the precise levels of earth floors in the dry conditions of Sicily is never easy. The destruction deposits inside the room were clearly definable. There was a good deal of burning near the floor and, above that, a thick layer of tile fall. Above that again,

183 R.J.A. Wilson

Figure 7. Area A, Room 1, south-east corner from the north, during excavation, showing the structure built against the south wall (scale 2 m). Photo by R.J.A. Wilson

many blocks from the upper parts of the walls had fallen in over time on top of the tile fall: a layer of wind-blown soil separated the top deposit of tiles from the stone blocks above. In the debris were found two lamps in African red slip ware (ARS), both of a smaller than usual size (see figs. 17–18 below). They belong to a sub-group of African lamps found elsewhere in deposits predominantly of the second half of the fifth century, as will be discussed more fully below (pp. 203–206). The destruction of the room cannot have occurred before ca. ad 450 when this category of lamps first came to be made, and if they were no longer in circulation in the sixth cen- tury, as present evidence suggests, the end of occupation here is unlikely to have occurred much later than ca. ad 500.

Area A, Room 2 Room 2 produced a very similar sequence. Here the burnt layer on top of the floor was even more dramatically well preserved, up to 4 cm thick: it was very obvious that life in the villa-like building had been brought to an abrupt end by fire. Above this layer was a thick layer of roof tiles. Both here and in Room 1, some of the tiles bore stamps, with the name Philippianus recurring in different forms; these too will be further discussed below (pp. 207–216). The roof tiles were all of the curved variety, so the roofing system used here,

184 UBC Excavations at Gerace, 2013 as was frequently the case during the Roman Imperial period in Sicily, was the so-called “Laconian” style, with each alternate roof tile in the side-by- side sequence being inverted.9 Above the extensive tile fall, there were again numerous stone blocks fallen from the upper part of the walls. In this room, the floor was not of beaten earth but consisted of white mortar (fig. 8), into which small pieces of red tile and some pottery sherds had been pressed. The floor was not very well laid, another feature that does not suggest a high-status room; its uneven thickness (up to 4 cm but often considerably less) was demonstrated by a small trial excavation at a point in front of the doorway where the surface had already been damaged. There was no sign in this test pit of an earlier floor beneath, only scattered small stones and other make-up for the single, preserved floor. The walls of this room, in contrast to those in Room 1, had been given a coating of white plaster. On the east wall, on either side of the doorway into Room 6, the plaster reaches floor level; but in the excavated parts of the north and south walls it does not, as large blocks near the base of the wall, never dressed off flush with the rest of the wall, project slightly into the room (see top of fig. 8). This was clearly deliberate (the plaster finishes above these projections and so respects them); the blocks may have served as narrow ledges or miniature shelves but, if so, only for very small objects.10 The plaster on the excavated part of the north wall (the eastern portion), and on the east wall north of the doorway, is smooth if somewhat damaged, caused in part by contact with collapsing stone when the walls of the room crumbled. In addition, both near the top (on the east wall) and near the base (on the north wall), blackened scorch marks from the intense fire that destroyed the build- ing are still clearly visible. South of the doorway, however, the east wall and the excavated eastern part of the south wall have a different quality of plaster, much more broken up and adhering only poorly to the wall. It is clear that this plaster was secondary, not only because portions of the earlier plaster are visible beneath in places, but also because it covers a blocked doorway in the south wall. When this part of the villa was first built, there was access into Room 2 from the south corridor through a doorway, which was later blocked. This necessitated the replastering of this southern part of the room. Perhaps

9 Wilson 1979: 20–23, Type B, with further thoughts, especially on chronology, in Wilson 2000a: 538. On Greek roof tiles, including the so-called “Laconian” type (the earliest curved tiles and, ultimately, the prototype for later ones of Roman date), cf. Hellmann 2002: 298, fig. 402 and especially 306–308. 10 A block protrudes into the room near the north-east corner from the north wall by 10 cm; two blocks do so in the south wall, by 10 cm and 23 cm respectively. Similar blocks (but more regularly fashioned and projecting into the room by up to 15 cm) were noted in the north and east walls of Room 2 in Building 6 at (, RG) of ca. ad 580/600–625 (Wilson 2010: 128).

185 R.J.A. Wilson

Figure 8. Area A, Room 2, from the north, at the end of excavation (scale 1 m) Photo by R.J.A. Wilson

having two doorways created an unwelcome through-draft to anyone using this room, and so a change was thought necessary, possibly soon after the original construction. Despite the amount of destruction debris in both Rooms 1 and 2, there were virtually no ceramic finds: both seemed to have been kept remarkably clean right up to the moment of the fire. This is less surprising for Room 2 than it is for Room 1, since buildings that have solid floors are relatively easy to keep clean, and a well-maintained house might be expected to have uncluttered floors. The paucity of ceramic finds is also likely to indicate that the building in Area A perished suddenly in a major conflagration while still in use. In particular, the absence of garbage on top of the floor, or of a layer

186 UBC Excavations at Gerace, 2013 of wind-blown soil between the floor and the charred evidence of the fire (which rests directly on the floor), makes it clear that there was no period of abandonment before the conflagration that destroyed the building. One critical piece of ceramic evidence was forthcoming for the date of the laying of the floor in Room 2. A rim sherd of ARS, Hayes Form 67, produced in northern Tunisia at El Mahrine and no doubt elsewhere, was discovered with the white mortar still adhering to it on all sides. Although damaged, its form is unmistakable. It is an early version, thin-walled and with a vertical rim (Type A in Bonifay’s classification, his Form 41), of a widely circulated and long-lived form of which this variant (Type A) belongs to the second half of the fourth century. Hayes dates its start to not earlier than ca. ad 360.11 Since the piece was structurally part of the white mortar floor of Room 2 and because the floor is primary to the building, the implication is that the structure as a whole in Area A is also unlikely to be earlier than that date. We will return below (pp. 197 and 219–220) to the chronology of the building in the context of its relationship with the structures uncovered in Area B.

Area A, the South Corridor The corridor on the south side of Rooms 1 and 2 had been excavated in part in 1994 and 2007, but the central portion had not. It had been the inten- tion in 2013 to study at what point the floor design changed between that recorded at the west end and that found at the east, and also to see how the mosaic overcame the problem of the uneven width of the corridor (see fig. 4). A trench uncovering a strip 5 m long and 1.30 m wide along the north side of this corridor found an extensive tile fall in position on top of the mosaic. A rim fragment of an ARS bowl of Hayes Form 81, a form current in the second half of the fifth century, was discovered in the debris. This pro- vides further evidence for the date of the destructive fire, for both this pot- tery form and the African lamps found in Room 2 are unlikely to have been made before ca. ad 450.12 There was some evidence of the fire in the corridor, especially near the entrance into Room 1. The mosaic itself, however, was found to be covered by a thick calcified deposit, which colleagues from the Centro di Restauro in Palermo instructed should be left in place until a full professional restoration of the mosaics could be funded and carried out. Two small areas of laurel scroll were uncovered, but little can be said about them

11 See further Appendix, no. 6, with references. One of the earliest examples of the form is in a destruction deposit in the Roman villa at Favara (AG), 55 km south- west of Gerace, which was thought to have been caused by the earthquake of “ad 365” (more likely that of ad 361/363: see note 17 below): Castellana and McConnell 1990: 32, fig. 7, no. 84/341 and 41. 12 Hayes 1972: 128.

187 R.J.A. Wilson

Figure 9. Area A, south corridor, detail of the mosaic pavement on the north side near the doorway into Room 1, from the south (scale 20 cm) Photo by R.J.A. Wilson

in detail (fig. 9). Detailed examination of the corridor mosaic will therefore have to await future work. Two aspects of it were, however, clear from even the small areas not covered by incrustation: first, the very irregular nature of the tesserae, both in shape and size; and second, the cavalier attitude toward colour, where colours of different hues are sometimes indiscriminately used within areas of what is meant to be uniform colour. No independent evidence for the dating of the mosaic in the corridor was possible, because the pavement appears to be intact and no area of it was therefore available for a sondage beneath. If, however, the mosaics were laid at the same time as the construction of the building in Area A, as seems very likely, they presumably belong to the second half of the fourth century, when the primary floor in Room 2 was laid. The building in Area A would therefore seem to share a building history similar to that at Caddeddi on the River Tellaro in south-east Sicily (fig. 1), which was also constructed in the second half of the fourth century and also destroyed by fire before the end of the fifth century.13

13 Voza 2003: 40 and 44; Wilson 2014b and forthcoming. The mosaics at Cad- deddi also share the features of irregular tessera size and cavalier use of different 188 UBC Excavations at Gerace, 2013

Area B1: The Westernmost Structure The villa-like building (Area A) explored by Dott.sse Cilia Platamone and Bonanno possessed, as we have seen, an apsed chamber (6 on fig. 4), located on the eastern edge of the excavated area. In the architecture of Roman vil- las, apsed rooms, where they occur, normally (although not invariably) lie at the centre of a suite of rooms, not at one side of them, and this suggests that a further set of contiguous rooms in all probability awaits discovery imme- diately east of the apsed hall. Indeed a wall, visible in section on the eastern side of the former drainage gully, suggests as much. It was with that in mind that a new exploratory trench (20 m by 3 m) was laid out in the almond orchard on higher ground to the east (Area B1), specifically in the hope of detecting part of the continuation of the villa-like building and of exploring its relationship with the 50 metre-long structure identified by the geophys- ical survey. At the western end of Area B1, a wall was uncovered (1 in fig. 10), approx- imately 58 cm wide, which in all likelihood marks the eastern extremity of the villa-like building: the distance between it and the east wall of the apsed room, and between the west wall of the latter and the west wall of Room 1, is almost identical. It was pierced by an arched opening, of which the key- stone and perhaps two further voussoirs are missing. Excavation behind the opening in the small portion that lay inside the south-west corner of Trench B1 revealed not only extensive deposits of burning, but also a portion of smooth white mortar floor at a higher level (fig. 11). The latter was resting on a flat limestone slab supported on a stack of smaller square slabs: set at a slight angle to the external wall, they were designed to support the white mortar floor and create a void beneath it for the circulation of hot air. It became clear, therefore, that the arched opening was a stoke-hole, and that the square slabs represented one of the pilae of a hypocaust. In order to see whether the exterior wall continued further south, a separate test trench (B2), 3 m square, was dug (fig. 10). It soon became clear that the wall did not continue there. Therefore, if the stoke-hole was set at the centre of its external wall, and if the wall turns a corner at the north baulk of Trench B1, both of which seem likely, the wall was originally about 2.25 m long and probably formed the eastern side of a rectangular recess. Since a heated living room is improbable in a Sicilian context, a working hypothesis is that the recess with the white mortar floor belonged to a small hot-­water bath. It is possible, therefore, that a small bath-suite awaits discovery between Area B1 and Area A. The opening of the stoke-hole had been closed off completely, after use, by a large vertical slab, perhaps associated with the subsequent early Byzantine activity in the immediate vicinity (pp. 202–203).

shades of colour within what are meant to be uniform stretches of mosaic using a single colour. 189 R.J.A. Wilson

(rubble collapse). (rubble Photo by F. Ianni, Regione Siciliana, Assessorato Beni Culturali e Identità Siciliana, Soprintendenza di Enna Siciliana, Soprintendenza Identità e Beni Culturali Assessorato Siciliana, Regione Ianni, F. by Photo wall. Remains of an earlier structure are visible west of 4; the area between it and 3 in the northern half of the trench represents unexcavated fill unexcavated represents trench of halfthe northern the and 3 in it between area ofthe 4; west visible are structure earlier an of Remains wall. A; 2 and 3 define a paved building, which has subsidiary paving on the outside to both west (especially in clear in B2) and east; 4 is interpreted as a terrace terrace a as interpreted east; 4 is and B2) in clear in (especially west both to outside the on paving subsidiary has which building, paved a define and 3 A; 2 Figure 10. Areas B1 and B2, aerial photograph at the close of excavation in 2013. Wall 1 probably belongs to the late fourth-century building excavated in Area Area in excavated building fourth-century late the to belongs probably 1 Wall 2013. in excavation of close the at photograph aerial and B2, B1 Areas 10. Figure

190 UBC Excavations at Gerace, 2013

Figure 11. Area B1, showing the east wall of the proposed bath-suite, with the voussoirs of a hypocaust stoke-hole (foreground) and part of a white mortar floor (far left), from the south. The wall visible on the right (2 on fig. 10) belongs to the earlier paved building. The scale is 0.50 m long. Photo by R.J.A. Wilson

The Large Rectangular Building

Description In the rest of Area B1, three further walls were uncovered that were associ- ated with the substantial structure whose presence had already been indi- cated by the geophysical research (2, 3 and 4 on fig. 10). On the second day of the excavation, a massive wall, 1.37 m wide, was discovered close to the modern ground-level at the east end of the trench (4 on fig. 10); indeed its upper surface had been partly damaged by the plough. It was very solidly built of mortared rubble, with conspicuous amounts of hard white mortar binding it together. Its west face, composed of approximately seven coursed rows of limestone blocks (of varying sizes), was excavated down to its foun- dations: it was preserved to a total height of 1.25 m (fig. 12). At first, this object was interpreted as the east wall of the rectangular building identified in the geophysical survey, but, as we shall see, there are problems with this interpretation, and it may rather have been an independent terracing wall built at the same time as the rectangular building adjacent: it is referred to as a terrace wall in the discussion that follows.

191 R.J.A. Wilson

Figure 12. Area B1, “terrace wall,” west face from west, detail of walling; the scale is 0.50 m high. Photo by R.J.A. Wilson

Two other walls, both parallel to the “terrace wall,” were also found in Area B1. Both of these (2 and 3 on fig. 10) were also solidly built with exten- sive use of white mortar. Each is 0.69 m wide (exactly half the width of the terrace wall further east (4 on fig. 10) and survived to a maximum visible height of 0.60 m. They define a long narrow building, internally 6.55 m wide and, on the basis of the geophysical results, as much as 50 m long. The floor throughout was of square or rectangular stone flags, all intact except for one small one (40 cm by 30 cm) that is missing (fig. 13). The slabs closest to the east wall are all of the same width, and there is a similar “spine” of equal- width slabs in the middle of the floor, but, otherwise, paving stones of differ- ent dimensions were used to provide some variety in the floor arrangement. Its surface is now partly undulating, but the paving was surely laid flat when the building was originally constructed. A curious feature of the slab paving is that it continues outside the building, if the walls 0.69 m wide (2 and 3 on fig. 10) are correctly identified as the limits of the roofed structure. On the east side, the paving extends for 2.21 m before ending in a neat straight edge. In section, it can be seen that the slabs were all carefully cemented into position, lying on a substantial bed of white lime mortar. Trench B2, dug to test the southward continuation of the putative bath structure described above, revealed that the same feature

192 UBC Excavations at Gerace, 2013

Figure 13. Area B1, detail of floor paving, from west; scale 2 m. The sole missing slab in the floor, exposing a layer of white mortar, is visible in the foreground. Photo by R.J.A. Wilson

occurred also on the west side of the long building: the external paving con- tinued here further west for approximately 2.30 m before stopping, with a more or less straight edge (fig. 14). In both cases, but more noticeably on the west side, the stone flags are deliberately set at an angle that slopes down markedly from the exterior wall of the building. On the eastern side of the building, there is a gap of 1.65 m between the point where the exterior paving stops and the thick “terrace wall” at the east end of the trench. Here excavation revealed walling belonging to parts of two rooms of a demolished structure below, presumably part of an earlier phase of building on the Roman villa-estate at Gerace (fig. 15). One of the rooms, the more westerly, had white plaster on the walls, but time prevented exca- vation to its floor level. Underneath the east–west wall of this earlier struc- ture (which continued below the footings of the thick terrace wall), there are three courses of a further wall on the same alignment, projecting south of the structure above it by 12 cm. It is too substantial and too well built to have been originally intended as a foundation (although after partial demo- lition of the rest of the wall to which it belonged, it served as such); rather it provides evidence for a still earlier phase of building beneath (Phase 1). The uppermost course of the Phase 2 structure stands slightly higher than the

193 R.J.A. Wilson

Figure 14. Area B2, detail of exterior paving and the west wall of the long building, from the north-west (scale 2 m) Photo by R.J.A. Wilson

Figure 15. Area B1, detail (from south) of an earlier building (Phase 2) (centre), demolished when both the terrace wall (right) and the paving outside the storebuilding (on the extreme left) were constructed; scale 2 m. Photo by R.J.A. Wilson

194 UBC Excavations at Gerace, 2013

Figure 16. Area B1, detail of roof tile collapse within the storebuilding, from south; scale 2 m. Photo by R.J.A. Wilson flagged surface adjacent (Phase 3), but was presumably concealed beneath earth when the later building was in use. There was an almost complete absence of finds within the (Phase 3) paved building: very little pottery of any sort was found on top of the flagged floor. It had certainly been kept clean, and there was clearly no abandon- ment phase when refuse might have been expected to have gathered inside a disused building. It seems likely, therefore, that the building collapsed suddenly and unexpectedly: the roof tiles lay as they had fallen, directly on top of the flagged floor right across the building (fig. 16), and also, but to a lesser extent, on the paving outside to the east. In addition, substan- tial quantities of stonework, collapsed from the upper part of the walls of the building, were found on top of the tiles. None of these roof tiles bore name-stamps. The only distinctive feature about them was the frequent recurrence of either one, two or occasionally three finger-marks, inserted prominently in the rim of numerous examples; presumably they served as some sort of tally-mark.14

14 I know of no precise parallel for this type of mark, although finger-marks scor- ing a horizontal line rather than a vertical depression, and on the body of the tile rather than on the edge, are of course commonplace. On tally-marks generally, see Brodribb 1987: 131–135.

195 R.J.A. Wilson

Dating evidence for the Phase 3 building is exiguous. In the absence of lifting the paving of the floor in order to search for sealed deposits beneath (which would compromise a monument perhaps one day to be put on public display), the best evidence for chronology came from the make-up immedi- ately adjacent to and to the west of the eastern terrace wall. This deposit is interpreted as part of the levelling up and backfill of this part of the site after demolition of the Phase 2 building and on completion of the construction work for the Phase 3 one. The pottery here includes five pieces of ARS, the forms of which are those generally circulating in the first half of the fourth century: one is an early version of Hayes Form 61A and is unlikely to be ear- lier than ca. ad 325.15 A date therefore in the second quarter of the fourth century for the new building seems likely. If the destruction of the paved building was sudden and violent (and it was clearly so total that subsequent rebuilding was not considered feasible), one possibility is that it was destroyed in an earthquake. In that context, a devastating earthquake attested by Libanius as having occurred in Sicily sometime during the reign of Julian (ad 361–363) might have been responsi- ble for the sudden collapse and abandonment of the building.16 Repair work attributed to the aftermath of an earthquake around that time has been surmised at the nearby Roman villa of Piazza Armerina,17 and probably also at the road statio of Philosophiana, 6 km to its south;18 while the rural site near Favara north-east of is also believed to have been destroyed by an earthquake at this time.19 If the Gerace building had pottery in the

15 For Hayes Form 61A, see Hayes 1972: 100–107; for the early version, cf. Mack- ensen 1993: 401, Form 4.1–2 with Taf. 53 (approximately second quarter of the fourth century). For this and the other four pieces, see the Appendix below (p. 223). 16 Libanius 18.291–293 (specifically including “the greatest cities of Sicily” as among the areas ruined). The earthquake occurred before the death of Julian and so is not to be confused with that of ad 365 (see next note). The earthquake that destroyed and the nearby villa at Patti Marina (Voza 1989, although no date is there given) is prob- ably a quite separate one that should be ascribed to the early fifth century ad. 17 Cf. most recently Pensabene 2010–2011: 193, who (following Carandini, Ricci and De Vos 1982: 376) attributes it, I think wrongly, to the events of ad 365, as does an influential paper by Antonino Di Vita (1972–1973, especially 256–257; cf. also Di Vita 1982). However, the epicentre of the earthquake in ad 365 is believed to have been located in the eastern Mediterranean, probably near Crete, and it was a devastating tsunami that affected Sicily in that year, on 21 July, as indeed the wording of Jerome implies: mare litus egreditur et Siciliae multarumque insularum innumerabiles pop- ulos opprimit (Hieron. Chron. p. 244; cf. Guidoboni 1989: 607 and 678–680 and cf. also 569, fig. 320). This would have affected the east coast of Sicily and might have caused widespread destruction, but only in coastal areas. On this event, see now the discussion of Wilson 2014c: 695–696. 18 La Torre 1994: 38. 19 See note 11 above; Castellana and McConnell 1990: 41.

196 UBC Excavations at Gerace, 2013 adjacent make-up with a terminus ante quem non of ca. ad 325, the building could easily have been constructed a decade or two later than this, as late, for example, as ad 350. If it perished in the earthquake of ad 361/363, it may have therefore had a very short life indeed. One other chronological implication of the excavation in Area B also emerged clearly in 2013. After the collapse of the building, most of it was left as it had fallen, but on its west side the tile debris was cleared away and the west wall (2 on fig. 10; that on the right in fig. 11) was demolished for a short stretch down to the level of the paved floor. The purpose of this was clearly to make room for the operation of a hypocaust when the stoke-hole in the later wall (1) was erected immediately to the west. That the latter structure was indeed later than the paved building was also indicated by its construction in part on top of the pre-existing stone “apron” paving to the west of wall 2 (fig. 11, between the two walls, on the right of the pho- tograph), without removing the slabs to make proper foundations for the new building.

Interpretation What was the paved structure used for? It was clearly utilitarian and not residential in character, as both the stone flagged floor and the absence of any plaster on the walls clearly indicate. There are two main possibili- ties: that it was a granary or, alternatively, a stable. To discuss the second suggestion first: as will be seen from the preliminary examination of the tile-stamps below (pp. 214–216), there is at least a possibility that the rais- ing of horses for the circus games, whether for use in Sicily or elsewhere, may have been part of the economic basis of this rural estate at Gerace in the later fourth century. The building under discussion, however, if it was a stable, dates from earlier than the tile-stamps, and when it collapsed around the middle of the fourth century it was not replaced, at least not in the same position. The use of the space, in the small portion of it uncov- ered in 2013, does not seem appropriate to horses, although of course a single section only 3 m wide may not be representative of the building as a whole, and in any case, partitions to create horse stalls could have been constructed of timber and so have left no trace. A building that can be identified, with good justification, as a stable on the rural estate of Oued Athménia in Algeria has square compartments (some with mortar floors) arranged on at least three sides of a central court, and a fourth side with a central row of over 40 troughs, probably mangers for horses.20 A soil sample

20 Berthier 1962–1965: 7–12; cf. also Rossiter 1992; Matter 2012. Gentili (1999: vol. 1, 50) postulated that the large aisled hall beside the entrance to the villa at Piazza Armerina (see further below, pp. 199–200) was the dominus’ stables, but produced no evidence in support of such a contention.

197 R.J.A. Wilson from the floor of the Gerace building, sent to Palermo for phosphate analy- sis, provided no trace of urine, such as one might expect to find in a stable;21 nor was there provision, at least in the section excavated, of water basins for the sustenance of animals, or for water runoff, drains and the like that might be expected to assist in the mucking out of such buildings.22 The hypothesis that the building served as a stable, is, therefore, unconvincing. A much more plausible interpretation is that the building was a granary or storehouse. The archaeological identification of rural granaries in the Roman world is notoriously difficult.23 Soil samples were regularly taken from the building in the hope of identifying carbonized seeds through flotation; the floated material awaits detailed analysis. Certainly, if there were grain bins lining the walls, these would have been of timber and would have left no trace. In any case, we do have only a small cross-section of the building; we do not know at this early stage of the excavation, for example, if the paved floor con- tinues throughout. If the building’s outer walls prove to have had buttresses, whether internal or external, that would greatly strengthen its identification as a granary. Either internal or external buttresses (or both) are present, for example, in Sicily’s earliest known granaries, the Hieronian examples (of ca. 250 bc) at ,24 and later in Roman rural storehouses at Casalotto near Acireale (CT) and at San Luca near Castronovo di (PA).25 On the other hand, buttresses are not an essential feature of storebuildings: indeed elsewhere in the Roman world, they are more common in military granaries

21 I am grateful to Dr. Cosimo Di Stefano of the Centro Regionale per la Proget- tazione e il Restauro e per le Scienze naturali ed applicate ai Beni Culturali in Pal- ermo, who visited Gerace when the excavations were in progress in 2013 and who took a soil sample from on top of the paved floor for analysis. 22 A good example is outbuilding D in the Roman villa at Bondorf in Baden-Württemberg, identified as a stable by just such a provision (a criss-cross of drains): Gaubatz-Sattler 1994: 84–87, with discussion on 126–127 (citing with refer- ences similar structures with drains, also interpreted as stables, in Roman villas at Wittlich in Rheinland-Pflaz and Lauffen am Neckar in Baden-Württemberg). 23 Cf. Spurr 1986: 81: “Granaries could be of different shapes, sizes and degrees of sophistication . . . Such diversification renders the identification of granaries in the excavation of rural buildings notoriously difficult . . . sizable, undecorated rooms are often designated [as such], but certainty is rare.” 24 Morgantina: cf. Deussen 1994; for their originality, Wilson 2013: 163–166, no. 129. Hieron’s granary in Syracuse was clearly impressive (Livy 24.21.11–12). For a summary of Sicilian granaries in the Hellenistic period, cf. Costanzo 1996 (to which should be added the storebuilding at : Parra 1995); cf. now also Soraci 2011: 23–24. 25 Casalotto: Libertini 1922 (internal buttresses); Wilson 1990: 191, with fig. 159a. S. Luca: Vassallo 2009 (external).

198 UBC Excavations at Gerace, 2013 than in civilian ones.26 It is tempting to think that the large (external) base in the centre of the building’s north wall, interpreted from the geophysics plot (2 on fig. 5), might represent a loading platform, an especially common feature in military granaries; but the inferred detail needs verification by the spade.27 The long, thin nature of the Gerace building (6.55 m by ca. 50 m (?), a ratio of approximately 1:7) is in keeping with the elongation of the Morgantina granaries, and might represent local conservatism in the Sicilian design of such structures—a feature shared also with what is plausibly another Roman storebuilding in Sicily, at the Piazza Armerina villa (EN), and, with less cer- tainty, the one at Casalotto (CT), where the original length is unknown.28 It might be thought that the absence of a raised floor argues against the iden- tification of the Gerace structure as a storebuilding; but not all granaries had raised floors, even after the second century ad when the practice first started in stone.29 The example at San Luca mentioned earlier, for example, which was 21 m long and 6.5 m wide and identifiable as a storehouse/granary by the

26 Buttresses: Rickman 1971: 79, 131, 231, 247–248; Gentry 1976: 4, 7–8. For exam- ple, none of the (mostly urban) granaries identified in Roman North Africa have buttresses: Virlouvet 2007; Drine 2007; Papi and Martorella 2007a; 2007b. On Spain (and also more generally), cf. Arce and Goffaux 2011, and on military granaries in the western provinces as a whole, cf. now also Salido Domínguez 2011. 27 On loading platforms, see Rickman 1971: 233; Gentry 1976: 12. 28 Morgantina: see note 24 (one is 92.85 m x 7.60 m (about 1:12); the other is at least 32.90 m long (original length uncertain) and 7.50 m wide). Piazza Armerina: Pensabene 2010–2011: 185, fig. 22 (87.50 m x 15.50 m, a ratio of about 1:5.5). Casalotto: see note 25 (12 m x 36 m, but not fully excavated; more therefore than 1:3). 29 First raised floors in stone: Rickman 1971: 293–297 (but the notion of sus- pended granary floors goes back much further—not only to earlier Roman military granaries built of timber, but also to civilian contexts as far back as the late fifth century bc, for example those with stone support walls at Moleta del Remei near Tar- ragona in Spain: Alonso 1995: 95 with Abb. 2–3). On granaries without raised floors: Rickman 1971: 226, which lists military granaries lacking raised floors at Caerhun, Caersws II and Brecon Gaer in Wales; Gentry (1976: 9) adds a further Welsh example at Caernarfon and one at Templeborough in England. The west granary inside the third-century stone fort at Vindolanda is another example, recently excavated and now fully published (Birley 2013, especially 58–61), and the west end of the north granary at Birdoswald was also relaid with a solid flagged floor in the later third century (Wilmott 1997: 122–125, fig. 91; cf. also 136 for other examples at Bar Hill and Benwell). For civilian examples from Britain where the floor was not raised, see Black 1981. Birley (2013: 60), commenting on the presence of two contemporary “granaries” at Vindolanda side by side, one with raised floor and one without, speculates (to explain the distinction) that the latter was “a stores building, perhaps for barrels, smoked goods, amphorae, hay for the horses or even other military supplies” rather than grain; but in Sicily, where rainfall levels were so exiguous by comparison with those of northern England, there is no reason why grain cannot have been stored along with other agricultural produce in the Gerace building.

199 R.J.A. Wilson presence of external buttresses, had only a beaten earth floor,30 and the same may also be true of the long Piazza Armerina storebuilding, where the wide spacing of the square piers shows that they are roof, not floor, supports.31 Both here and at San Luca, however, there may have been independent floors of timber within the building on wooden supports of which no trace is left. The size of some of these buildings is also notable. The geophysics, as we have seen, suggests that Gerace’s length is about 50 m; the Piazza Armerina store- building is over 87 m long; and a Hellenistic one at Morgantina measures 93 m in length.32 Even utilitarian buildings were constructed on a scale designed to impress and, in a private context, to send the message that the visitor had entered an estate with very significant agricultural productivity.33 The really crucial feature of a good granary is an ability to keep the con- tents dry and free from vermin, and for this purpose the building needed not only a sound roof but also well-built walls, which the Gerace building with its liberal use of mortar certainly has. Palladius, the Roman author whose Opus agriculturae was published only a few years later than the Gerace building’s construction (ca. ad 400), recommends that “all the ground should be paved with two-foot [square] or smaller bricks, which we should set on an underly- ing floor of crushed-tile mortar”;34 there is no mention of raising the floor. At Gerace stone slabs instead of bricks and white rather than pink mortar were used, but otherwise the builders followed these guidelines quite closely. In particular, the care with which the stone flagging was laid on a thick mortar bedding suggests a deliberate attempt to minimize the potential problem of rising damp; and taking rainwater away from the roof, rather than let- ting it stand in pools along the foot of the granary’s outer walls, is the most logical explanation for what otherwise seems an unnecessary and expensive feature—­the “apron” of stone paving that continues for over 2 m on either side of the building.35

30 Vassallo 2009: 673; it was divided by a central longitudinal wall into two parts. Further structures with cross walls immediately to the north were also interpreted as being for the storage of agricultural produce. 31 References in notes 20 and 28, but few details of this structure have been published. 32 See note 28. 33 Cf. Brown 2013: 14: “. . . great granaries were a sight calculated to trigger disquiet in late Roman minds. Large Roman villas flaunted the wealth of their owner by being built on solid granaries, endowed with heavy locked gates.” 34 Palladius 1.19.1. The translation is that of the excellent new edition, Fitch 2013: 50. Columella (De re rustica 1.6.12–13) recommended a similar procedure, with the additional instruction to apply the lees of olive oil (amurca) to the floor during con- struction to deter vermin. Palladius (1.19.2) also mentions amurca, but only as a coat- ing for the granary walls after construction had been completed. 35 I know of no precise parallel for this, but the baths at the site of Cassinomagus (Charente) in central France are provided with an “apron” of stone paving round the

200 UBC Excavations at Gerace, 2013

Apart from the examples mentioned earlier at Casalotto, Piazza Armer- ina and San Luca, very few storebuildings of the Roman period have been identified in Sicily in recent years.36 Internal pillars at Casalotto probably indicate that it had a raised floor, as did another fourth-century granary, a comparatively small structure (5 m by 7 m) that has recently been identified at Pietrarossa (CT).37 Carbonized seeds of durum wheat (triticum durum), emmer (triticum dicoccum) and barley (hordeum vulgare) have been iden- tified from it. In the road settlement built along the bay at (ME), one wall of a storebuilding 50 m long was uncovered that was used in part as a repository for dolia, with evidence of some internal partitioning; it was believed to date from the first century bc or first century ad. Two smaller rectangular buildings further west, both apparently tripartite, were also ten- tatively identified as storehouses.38 Some structures excavated long ago at San Leone, Agrigento’s harbour settlement, may have been storehouses, but secure details are lacking.39

The “Terrace Wall” The “terrace wall” (4 on fig. 10; fig. 12; fig. 15, right) was at first taken to be the east wall of the building: the interpretative plan of the geophysics sug- gested that what is attested here was a basilical building with a “nave” and two “aisles” (fig. 5, no. 2). This proved not to be the case, at least in the sector excavated: for example, the expected west wall separating the “nave” from the west “aisle” never materialized, and the thick west exterior wall proved to be two adjacent walls of different periods. So the triple-aisled building expected from the geophysics turned out instead to be a single, long, undivided rect- angular space. Parallel to it, 3.86 m to the east, was the very substantial wall best interpreted as terracing for the slightly higher ground that lies further east. It was presumably built to shield the granary from soil slippage and fur- ther damp from this direction. This interpretation removes several difficul- ties: (1) the building no longer has exterior walls of two different thicknesses, a most irregular arrangement; (2) it no longer has an off-centre longitudinal dividing wall (3 on fig. 10), which presupposed an eccentric (a-centric) roof; exterior walls (well preserved on the north: Brethenoux 2012: 61), but here it incorpo- rates a built drain running along the middle axis (personal observation, May 2014). 36 Soraci 2011: 197–198. 37 Arcifa 2008a: 51 (excavation by Dott.ssa L. Maniscalco). 38 Lentini 2001: 26–29, fig. 30. 39 Gabrici 1925: 423–425. This is not the place for a consideration of granary com- paranda from elsewhere in , but cf. in brief Rossiter 1978: 57–59; Arce and Goffaux 2011: 8–11, 16–17 and 67–95; and for granaries in the Veneto, cf. Busana 2002: 193–202, most with raised floors since the region is one where “le esigenze geomorfologiche e climatiche rendevano necessaria l’adozione di un pavimento rialzato” (194; unlike in Sicily). For an alleged granary at Settefinestre, cf. Carandini 1985: vol. 2, 189–203. For grain storage in general in the western provinces, see Heimberg 2011: 113–117.

201 R.J.A. Wilson

(3) the building does not have to be regarded as unfinished, with paving unlaid in the part of the “floor” immediately adjacent to the thick east “exte- rior” wall, even though the roof was complete; and (4) the demolition gangs who prepared the site for the new building do not have to be accused of fail- ing to do their work properly, by leaving walls of an earlier structure standing proud of the floor level of the new building. If it lay outside the new granary building, the exact height of this remaining stump of earlier walling would not have mattered. Against this interpretation are the geophysics results, which appear to show that the “terrace wall” is integral to the building as a whole, and the exterior wall on the short north-west side seems to join it at the north corner. This latter area will be a focus of future excavation: the conclusion presented here must therefore be regarded as preliminary.

Early Byzantine (?) Activity As noted earlier, the building interpreted as a granary or storehouse was not rebuilt after collapse. At one point within its remains, an oval area of pre- pared clay embracing a central hollow was laid on top of the flattened roof- tile collapse. Measuring 2.30 m across (east-west) and extending 1.45 m in from the north baulk of Area B1,40 it shows manifest signs of intense burning in its bright orange-red fired clay. The latter feature suggests that it was used as a hearth; for what precise purpose, whether domestic or industrial, has not been established. Nor was its date, except that it must post-date the collapse of the building, suggested earlier as occurring in the ad 360s. It could there- fore be contemporary with the occupation of the adjacent villa-like building (between the late fourth and the late fifth century), or it could be later. In one context, immediately above the hearth and associated with it, and in two other contexts nearby (the rubble collapse from the walls of the storebuilding, and the large area of ashy deposit east of the hypocaust flue in the separate structure to the west, post-dating the latter’s use),41 several pieces of cooking pottery were found, of a distinctive, coarse black fabric with large white calcite grits. The surface decoration consists of either small punched circles or long straight lines, the latter looking like straw impres- sions made while the clay was still damp; as a result, this category of pottery, which has been detected elsewhere in central and eastern Sicily, is known as the a stuoia (“wickerwork”) type.42 This is currently thought of as belonging to ca. ad 800, because a deposit of it near Mineo (CT) was found associated

40 Its easternmost point is 1.50 m west of the east wall of the building (3 on fig. 10). Not all of the feature was excavated; more of it lay beneath the north baulk of Area B1. 41 Contexts 9, 13 and 17 in the site records. 42 Drawn finds are P. 22a, 35, 46 and 50 in the site records and will be published in the final report. For earlier finds (unstratified) of this pottery at Gerace, see Bonanno et al. 2010: 264–265, figs. 12–13; also Bonanno 2013a: 193–194.

202 UBC Excavations at Gerace, 2013 with two coins of the emperor Michael I Rangabe (ad 811–813).43 Even if that is correct, it is not impossible that production started earlier in the Byzan- tine period. At Gerace, one of the contexts containing it clearly post-dates the abandonment of the hypocaust stoke-hole, which probably occurred when the building to which the latter belonged was burned down toward the end of the fifth century. The pottery a stuoia could, therefore, belong to as early as the sixth or seventh century rather than later,44 and the hearth is probably contemporary with it. Possibly supportive of this earlier chronol- ogy is a fragment of ARS lamp found immediately on top of the hearth (and so not securely stratified), which belongs to the very last series of African lamps produced in the seventh century.45 Certainly, the stratigraphy of the deposits in Area B does not suggest that, after the fire that destroyed Area A, there was a gap in occupation at Gerace of a full three centuries before this “early medieval” activity occurred—the inevitable conclusion if the a stuoia pottery here is ascribed to a period as late as the ninth century. Whether the early medieval building identified by Dott.ssa Cilia Platamone is contempo- rary with this Byzantine activity or later can only be resolved by further work.

Selected Finds

1. ARS Lamps Two African-made red-slip lamps, one complete (Lamp 2) and the other nearly so, were found in Room 1 amid the debris associated with the fire. It seems likely that they were on a shelf in the room and were knocked off when the roof fell in. Both are of Atlante Type XA, Group C3, or Lamp Type 55, “de petite taille,” in Bonifay’s classification.46 They are smaller than the usual African lamp, with a width of less than 7 cm (African lamps are usually 8 cm wide or more). Both our examples have two filling holes and the usual long nozzle and blackened wick hole, although that of Lamp 1 is damaged. The base of both (not illustrated here) has an identical mark within the base ring, consisting of two close-set concentric circular grooves. This type of lamp occurs typically in contexts of the second half of the fifth century ad.47

43 Arcifa 2008b: 295, 303–305. For this pottery, see also Arcifa 2010a: 120–121 and especially 2010b: 67–77 (with distribution map at 77, fig. 3). Arcifa (2010a: 121) notes its absence in eighth-century levels at Catania (S. Agata), but, in any case, this was at the eastern limits of its distribution, and earlier production elsewhere is not pre- cluded, especially in view of the production of very similar pottery from the sixth and seventh century onwards in northern Italy (Arcifa 2010b: 74–76, fig. 4). 44 An amphora rim of “African” (or Sicilian imitation) type (P. 62) from context 13 seems unlikely to be later than the seventh century. 45 The fragment has very worn decoration on the rim of the discus of squares and hearts set at irregular angles to one another, as Bonifay 2004: 410–413, Lamp Type 69, no. 8. 46 Bonifay 2004: 382–386; Atlante 1981: 198–203. 47 Bonifay 2004: 386.

203 R.J.A. Wilson

Figure 17. ARS lamp, from Area A, Room 1, bird on top of the Greek letter iota, ca. ad 450/500; breadth 68 mm (scale 5 cm) Photo by R.J.A. Wilson

Lamp 1 (width 6.8 cm) has a depiction on the discus of an elaborate single Greek letter iota, standing for the first letter of Jesus Christ’s name in Greek, Iesos (fig. 17). It has scroll decoration in the centre and projecting “volutes” at top and bottom. At its base is a pelta; the top of the letter is also curved, on which stands a bird facing right, flapping its wings. This is usually described as a dove or a swan.48 The dove is of course a symbol of the Holy Spirit and, therefore, an appropriate Christian symbol; but the bird on our lamp has a prominent neck that a dove lacks. There are examples of lamps where a dove appears on top of the iota, and in such cases the bird is very clearly different from the Gerace example.49 It is possible that, rather than a swan, a goose (the neck is appropriate) was intended by the lamp maker, but

48 E.g. Bussière 2007: 143, no. 1010. 49 E.g. ibid., with plate 64, C1009, also in the same category of small lamps.

204 UBC Excavations at Gerace, 2013 a certain identification of the species depicted is elusive.50 On the shoulder of the lamp are squares, triangles, concentric circles and clover leaves.51 A lamp from the identical mould to that used for the Gerace example is in the museum at Hippona (Annaba) in Algeria.52 The discus decoration showing the iota accompanied by the same bird is attested on several other lamps, but all have different shoulder decoration.53 A version with two iotas side by side is also known, with the bird set at the top between the two Greek letters.54 Lamp 2 (width 6.8 cm) is in the same category of small African lamps of the second half of the fifth century (fig. 18). The discus is decorated with a chi-rho symbol in the form of a Greek cross. Bonifay notes55 that the mono- grammatic cross (as opposed to the standard chi-rho symbol with saltire chi) is extremely rare in this category of lamps (his Type 55), citing only a single example in Rome, the chi-rho symbol being there decorated by a series of raised dots or “pearls.”56 Another example, with the same discus design and same shoulder decoration, is known from Mimiani (CL) in Sicily and is from the identical mould to that used for the Rome example, as is another from the region of Aquileia, now in Trieste.57 The Gerace lamp was intended to have had the same type of pearled monogram with pendent alpha and omega: some of the raised dots can indeed just be made out at the top of the rho, and the broken (v-shaped) cross-bar of the upside-down alpha is just visible to the

50 Cf. also Garbsch and Overbeck 1989: 154, no. 153 = Badisches Landesmuseum Karlsruhe 2009: 301, no. 238, for four well-preserved examples of the same bird on a single circular lamp-handle attachment. One might have expected the phoenix, a well-known Christian symbol, to have been a candidate for the identity of the bird, especially because it is usually shown flapping its wings (van den Broek 1972); but the bird on the lamps lacks the solar crown that is its identifying attribute, and past identification of the phoenix on lamps is probably erroneous (e.g. Ennabli 1976, nos. 508–510, which look identical to no. 507, “aigle”). On the difficulty of distinguishing between eagle and phoenix in early Christian art, cf. Bisconti 1979; 1981. 51 Ennabli 1976: A10, D7, E2; the clover leaf, not included in Ennabli’s classification, is Bussière’s P1 (2007, pl. 138); the others are his A9a, D2 and E1c (2007, pls. 133–134). 52 Bussière 2007: C 1010, with plate 64. 53 E.g. in Trieste, from Aquileia (?): Graziani Abbiani 1969: 150, no. 482 with fig. 92; in Hannover: Mlasowsky, Uhsemann and Wolff 1993: 390, no. 375 (where the bird is described as an “Ente (?)” [duck]); in Rome, Barbera and Petriaggi 1993: 247, no. 208; and also in Sicily, from Mimiani (CL): Bonacasa Carra and Panvini 2002: 262, fig. 6 (“volatile”). 54 In the Musée du Bardo, Tunis: Badisches Landesmuseum Karlsruhe 2009: 299, no. 232, where the iotas are improbably described as columns and the bird as a chicken. 55 Bonifay 2004: 382. 56 Barbera and Petriaggi 1993: 166, no. 123. 57 Mimiani: Bonacasa Carra and Panvini 2002: 261, fig. 5 (but found in a tomb with seventh-century lamps); Aquileia region: Graziani Abbiani 1969: 137, no. 434 with fig. 83.

205 R.J.A. Wilson

Figure 18. ARS lamp, from Area A, Room 1, crudely scratched monogrammatic cross, ca. ad 450/500; breadth 68 mm (scale 5 cm) Photo by R.J.A. Wilson left of the right-hand filling hole. In our case, however, the monogram clearly emerged from the mould in a very indistinct state, and rather than throw the lamp away, the lamp maker clarified the shape of the monogram by cutting into the leather-hard clay a series of short, stabbed, oblique lines. For good measure, he also added a triangle of oblique strokes below the chi-rho sym- bol. Had the moulding process been more successful, the filling holes would have been moved higher up the discus in order not to interfere with the pen- dent alpha and omega, an integral part of the design. This is an unusual case of a craftsman making alterations to an individual mould-made lamp once it had left the mould and before it was fired, giving our example a unique, if rather crude, appearance. The shoulder decoration has two patterns alternat- ing: a double square enclosing two concentric circles and a circular pattern in which a “wreath” with oblique strokes encloses an inner circle.58

58 Bussière 2007: A6 and F10b; the first of these also occurs on the Rome and Mimiani parallels (cited in notes 56–57).

206 UBC Excavations at Gerace, 2013

2. Tile-stamps Of great interest are the 75 fragmentary roof tiles found in the 2013 exca- vations that bore stamps—in several cases, more than one: 99 individual stamps were catalogued in total.59 Sixty of the stamped tiles (80%) come from Area A, of which seven came from Room 1 (12%) and 53 (88%) from Room 2. Of the remaining 15 examples, three are surface finds and 12 (16%) come from Area B. Of this latter dozen, two were from surface layers, and all the remaining ones were found in abandonment contexts of the large rect- angular building, principally in the rubble levels of the tumbled walls. Sig- nificantly, not a single stamped tile came from the spectacular tile fall of this building’s own roof, which lay sealed below the rubble fill (fig. 16). Rather, the stamped tiles are all later than the demise of the rectangular building, and were therefore clearly made for the construction of the villa-like build- ing, which has been assigned to the second half of the fourth century ad. This is a rare example of Sicilian stamps on the so-called Type B Roman roof tiles (the “Laconian” type) being closely datable.60 Ten different varieties of stamps were identified. The most striking (Type 4) is a circular one naming PHILIPPIANI in a circular inscription round the edge; the last and first letters are separated by a crown with attached ribbons (fig. 19). In the centre of this stamp is depicted a racehorse facing left, with a plume attached to its head and a palm branch in front of it. This stamp, which survives in five examples (of which only two are entire), differs from all of the other stamps found at Gerace in having its letters and deco- ration executed in relief, implying that the die was incuse. The name always occurs, on this stamp and on the other Gerace stamps, in the possessive gen- itive and so ends in –I. A second stamp (Type 2, of which there are four exam- ples) is rectangular with the profile of a horse’s head and neck facing right, flanked to the left and right by ligatured letters also standing for the name Philippianus (fig. 20). A third stamp (Type 5), showing a standing racehorse facing left, also with plume and with a victory crown above, is designed like a

59 This does not count the two fragmentary dolphin stamps where no letters sur- vive: see caption to table 1. For a fuller discussion of these stamps than there is space for here, see Wilson 2014d. 60 For earlier discussions, Wilson 1979: 23–26; 1990: 216–217; 2000a: 537–542. The GALB(a) tiles from Piano della Clesia (CL) appear to be associated with second- and third-century ad material, while the stamp-type ECNAT[i] at the same site comes from fourth- and fifth-century levels (Bonacasa Carra and Panvini 2002: 81–82). It seems likely that this (the second half of the fourth century ad) represents the final period of stamping roof tiles in Sicily. An extensive tile fall in a building at Campa- naio (AG), constructed ca. ad 375/400, yielded only a single stamped tile (Wilson 2000b: 347), which was taken to be contemporary but it might have been re-used from an earlier structure; and no stamps were found at all among the extensive fallen roof tiles in Building 6 at Kaukana, of ca. ad 580/600 or a little later (Wilson 2010).

207 R.J.A. Wilson

Figure 19. Tile-stamp of Philippianus (Type 4), with horse at centre, palm branch in front of it and a crown with ribbons in the legend, marking the beginning and end of the name; diameter 66 cm (scale 5 cm) Photo by R.J.A. Wilson

Figure 20. Tile-stamp of Philippianus (Type 2), with horse to right in profile at the centre and the name Philippiani ligatured in two parts, to the left and right; length 80 mm (scale 5 cm) Photo by R.J.A. Wilson 208 UBC Excavations at Gerace, 2013 coin, in that the legend, which appears to read PHILI[P]/PIANI, is inscribed in small letters around the standing horse at the centre; it is arranged in two parts, interrupted by the head of the horse and the victory crown (fig. 22). This stamp occurs in just three examples (two of them occurring on a single tile). The horse is also featured on two much smaller circular stamps, both without accompanying letters: one depicts a horse facing left, the other to the right, and in both cases the horse is wearing a plume on its head (Types 6 and 7). The former occurs only twice, both on the same tile; the second appears on a single tile (see below). The inscription PHILIPPIANI also occurs on three other versions in this series. One, attested only once, is also small and has the letters of the name all ligatured into a single design (Type 8). The other two, one rectan- gular and the other circular, are the most commonly occurring of all the tile stamps found at Gerace in 2013: there are 22 examples of the rectangular stamp (Type 1) and 31 of the circular variety (Type 3). The first (fig. 21) reads PHILIP/PIANI in two lines, with the space at the end filled by what looks like a pomegranate (with two stalks), a symbol of life and fertility and, by exten- sion, of good luck, but it is probably a poorly drawn ivy leaf, the standard epi- graphic space filler. The other is a circular stamp, FILIPPIANI (which has an initial letter “F” in place of “Ph”), the letters framed between two concentric circles; an ivy leaf separates the last letter from the first (fig. 22). In the centre of this stamp, the tile-maker had depicted what looks to have been intended to represent hoof marks, presumably a further equine allusion. Although preserved in varying degrees of clarity, these marks appears to be exactly the same position on all examples and therefore formed part of the design in the original die.61 In two cases only, the centre is not marked in this way but contains instead, as an addition, one of the stamp types already referred to—in one case, a plumed horse in a small circle facing right (Type 7, its only occurrence) and, in another, the “coin-type” of stamp with horse facing left, crown above, and the letters PHILIPPIANI (indistinctly) around (Type 5). In the latter case, the stab marks are also present (fig. 22). Only in a couple of examples where enough of the central circle is preserved is it certain that the central area was left completely plain. The final pair of stamps do not name or allude to Philippianus. One has the word SALVS, “[your] good health,” inscribed within the shape of a dolphin (16 examples, including two set of pairs, in which the same stamp is duplicated side by side on the tile edge). The second has TVTELA, “pro- tection,” i.e. “take care,” which is also within the form of a dolphin (14 exam- ples). A small four-petalled fleuron precedes both. A majority of the dolphin

61 I had originally taken these “stab” marks to have been added by hand to each tile; but Dr. Philip Kenrick, whom I thank most warmly, has pointed out to me, by overlapping and so comparing photographs of different occurrences of the same stamp, that the marks must be part of the original die.

209 R.J.A. Wilson

Figure 21. Tile-stamp of Philippianus (Type 1), PHILIP/PIANI with ivy leaf (?); 62 mm by 28 mm (scale 5 cm) Photo by R.J.A. Wilson

Figure 22. Two tile-stamps of Philippianus: one (Type 3) reads FILIPPIANI with ivy leaf; stabbed marks around the inner circle (diameter 75 mm); a second stamp (Type 5) within, with horse to left, wreath above and letters around (diameter 26 mm) (scale 5 cm) Photo by R.J.A. Wilson

210 UBC Excavations at Gerace, 2013 ) ( Continued + Type 3 (2) Type + 9 (1) Type + 10 (12) Type + 9 Type with tile 9 (1 + 1 Type + twice) stamped 4 (1) Type + Combination with other other with Combination in of occurrences stamps (no. parentheses) + Type 1 (2) Type + 5 (1) Type + 7 (1) Type + 9 (2) Type + + Type 8 (1) Type + + Type 3 (1); the other 2 examples examples 2 other the 3 (1); Type + tile same the on are 5 ofType 2 examples on same tile same on examples 2 + Type 3 (1) Type +

B B B Position on tile on Position B = body) rim; (R = B B B B B 22 4 1 No. of No. stamps 31 5 3 2 1 mm mm

mm

62 mm × 28 mm 62 × 29 80 mm × 12 16 mm Diameter or or Diameter length/breadth (approx.) 75 mm 75 66 mm 26 mm 17 mm 17 18 mm Shape Rectangular Circular Rectangular Circular Circular Circular Circular Oval Description PHILIPPIANI ligatured; ligatured; PHILIPPIANI r. to profile head in horse FILIPPIANI PHILIP/PIANI Racehorse to l. + to Racehorse PHILIPPIANI Racehorse to l. + PHILI[P]/ + l. to Racehorse type) (“coin” PIANI Racehorse to l. to Racehorse Racehorse to r. to Racehorse PHILIPPIANI ligatured PHILIPPIANI Type 2 Type ( fig. 20 ) 3 Type ( fig. 22 ) Type 1 Type ( fig. 21 ) Type 4 Type ) ( fig. 19 Type 5 Type ( fig. 22 ) Type 6 Type Type 7 Type Type 8 Type Table 1. Stamp types 1. Stamp Table

211 R.J.A. Wilson Combination with other other with Combination in of occurrences stamps (no. parentheses) + Type 1 (2) Type + above) note see 2 (2; Type + 3 (2) Type + + Type 1 (12) Type +

Position on tile on Position B = body) rim; (R = B (8);R (8) R No. of No. stamps 16 14 Diameter or or Diameter length/breadth (approx.) 52 mm 50 mm Shape Dolphin Dolphin Description SALVS TVTELA Type 9 Type Type 10 Type Table 1. Continued Table no dolphin, but of fragment 10 (a or 9 type either of examples two are included here Not Note: tile. same the on 8 Type a and 4 Type a with other the 3, Type a with surviving),one letters

212 UBC Excavations at Gerace, 2013 stamps are found on the raised edge of a tile, often in close association with one of the Philippianus stamps: for example, 12 of the 14 TVTELA stamps are found in close association with the rectangular PHILIP/PIANI, which is carefully placed at right angles to the rim, in line with where the dolphin stamp occurs. Half of the SALVS stamps are found on the body of the tile, but TVTELA by contrast is stamped invariably on the thickened edge of the tile. Despite the variety of dies, it is clear that all were of the same fabric and that they were produced as part of the same tile-making process. This pro- cess in all likelihood took place in one of the on-site kilns identified by the geophysical research. There are tile wasters at Gerace, so tile manufacture certainly took place there, but a tile waster with a stamp has not yet been found to confirm that the Philippianus tiles were part of that production. The sheer frequency of the stamps and the uniform fabric nevertheless make local production very probable. They were presumably manufactured for the building of the new villa (after the mid-fourth century destruction of the old one, the location of which is still to be ascertained; so far, as we have seen, only its granary is known); each design exists in only a single die, and there are enough cross-links between the dies (because of multiple stamping, with some individual tiles bearing up to three separate stamps) for their con- temporaneity to be clear. For example, it might be thought that the simple rectangular stamp (22 examples) would have preceded the simple circular variety (31 examples) or vice versa; but there are two instances where both the circular and rectangular stamps occur together on the same tile, demon- strating that both dies were in use simultaneously. Quite apart from the close association of TVTELA with the rectangular stamp, several other pairings are also attested (summarized in table 1). Two of the stamp types, the circular FILIPPIANI and the dolphin design containing the word SALVS, are attested at the site of Piano della Clesia near Caltanissetta, 13 km to the west of Gerace (as the crow flies), where another Roman villa is located.62 It is not clear at present whether this is a case of a villa owner elsewhere acquiring a consignment of tiles surplus to the requirement from the tile producer (probably, as we have seen, at Gerace

62 Adamesteanu 1963: 272, fig. 28; Wilson 1990: 216, fig. 176, nos. 21–22; Bonacasa Carra and Panvini 2002: 256, no. 57, fig. 39 (for the FILIPPIANI stamp). I incorrectly read the indistinct fleuron on the Piano della Clesia SALVS stamp as a K (Wilson 1990: fig. 176, no. 22); this should now be discarded in light of the new Gerace discov- eries. Bonanno (2013a: 195) states that the villa here is of the second century ad, but its surface finds date from between the first and the fifth century ad (Wilson 1990: 214), so the site clearly had a long life. The FILIPPIANI stamp here is from the same die as those at Gerace, and so also should belong to the second half of the fourth century. Pensabene (2010–2011: 181) claims that the Philippiani stamp is also attested “nel territorio di ,” citing Fiorentini 1984: 49 (in fact 1985: 51), but the latter is a reference to the Piano della Clesia example.

213 R.J.A. Wilson itself), or whether the presence of identical tiles at both sites indicates that both villas were under the same ownership. Certainly, the recurrence of the Gerace stamps at two separate places a short distance apart (with no other examples so far known elsewhere) is intriguing. If future fieldwork reveals a more extensive distribution of Philippianus’ tiles, it will become likely that the Gerace property, having invested in tile-making facilities, was producing and selling tiles to those estates that did not have such facilities. This hypo- thetical surplus production would have represented an additional economic activity and source of income for the owner of the Gerace estate.63 The name Philippianus surely refers to the dominus of the estate at Ger- ace in the second half of the fourth century: it seems unlikely, in the context of a rural estate producing its own tiles, that the tile-maker actually making the tiles, or a middleman such as a contractor, would have put his own name on them rather than that of the estate owner.64 There are two possible expla- nations to account for the imagery of the victorious racehorse. The first is that its presence on several of the dies is nothing more than a play on the man’s name, a “speaking” signum. Philippianus is of course cognate with the name “Philippos,” “lover of horses.” As a signum of this type, it would be unique in Sicily, but such visual puns are attested, albeit extremely rarely, in the instru- mentum domesticum of the Rome brickworks: M. Rutilius Lupus used the design of a wolf to accompany his stamps, C. Iulius Stephanus used a crown and M. Flavius Aper a running boar.65 These were all issued in the first half of the second century, more than 200 years before the Philippianus tiles were made. In this interpretation, the palm branch and the crown as victory sym- bols at Gerace need to be interpreted metaphorically, as good luck emblems for a happy, successful and prosperous life—much as the image of the victo- rious charioteer can sometimes have a similar apotropaic meaning (the “race of life”) rather than literally signifying personal involvement in circus ludi.66 The alternative explanation is that the owner of the Gerace estate was indeed proud to be in the business of raising racehorses, and that the plume, the crown and the palm branch were intended to indicate either that

63 In view of the fact that no stamps bearing other names are attested, and that tile-making (albeit of unknown date) is attested at Gerace, it seems improbable that the tiles were all imported from another source, although that is, of course, theoreti- cally possible if no tile making was being done at Gerace at the time of the construc- tion of the late Roman villa-like building. 64 I am here in fundamental disagreement with Manganaro 1988: 32–34, who states that all Sicilian tile stamps record the names of officinatores or conductores and never domini; cf. Wilson 1990: 390, note 109 (one of the names records a woman, “Popilia Paetina, daughter of Marcus,” an improbable conductor or tile-maker). 65 Bodel 2005. I am grateful to Prof. Christer Bruun (Toronto) for drawing this paper to my attention. 66 Dunbabin 1982, especially 82–84.

214 UBC Excavations at Gerace, 2013

Philippianus’ horses had already achieved victory in circus games, whether in Sicily or further afield, or at the very least were being raised with this hope and intention. If so, it would seem to be too great a coincidence if the domi- nus, born with Philippianus as one of his names, developed an interest in raising and training horses when he was an adult. More plausibly, Philippi- anus may have been an additional cognomen or supernomen added to his other names, in recognition by friends and family of this particular individ- ual’s passion for horses or, alternatively, that he had been given this name by his father to reflect the latter’s own interest in horse breeding, which his son inherited.67 The name Philippianus is in fact uncommon,68 and only occurs twice in Sicily.69

67 For the practice of giving additional cognomina, agnomina and signa, cf. Kajanto 1963: 24–49; 1966, passim; Salway 1994: 127–128 and especially 136–137 (for the Late Empire). Livy (30.45.6) recalls that Sulla’s agnomen Felix first started “with the flattery of his friends” (ab adsentatione familiari). The term supernomen occurs rarely and not before the third century ad (Kajanto 1966: 5 with note 1). An agnomen should, sensu stricto, be preceded by qui et, “who [is] also [known as],” but a second cognomen, even without this specification, is sometimes referred to as an agnomen without this indication. If an additional name granted in his lifetime, Philippianus is probably an example of this phenomenon. It is unlikely to be a signum, another (espe- cially late Roman) form of additional name (sometimes specifically preceded by the word signo), since the vast majority of these end in –ius: of Kajanto’s (1966: 76–90) list of 383 signa, only three examples end in –us (Marianus, Rusticulus and Secundinus). 68 Cf. M. Aurelius Philippianos Iason at Prusias ad Hypium in Asia Minor in the first half of the third century ad (SEG 30.1442); Tiberius Claudius Secundus Filippi- anus, imperial freedman and coactor (a financial operator), on the via Appia 5 km south of Rome (CIL 6.1860); two senators, T. Flavius Secundus Philippianus, a gov- ernor of Gallia Lugdunensis ca. ad 209 (PIR2 F 362) and his son T. Flavius Victorinus Philippianus (PIR2 F 400; cf. ILS 1152, Lyon); and Flavius Philippianus, a centurion in the Fifth Cohort of Legio II Traiana (ILS 2304, near Alexandria). There is also one example (third/fourth century ad) of Philippianos, at Perinthus-Heraclea (Fraser, Matthews and Catling 2005: 344) and seven in Asia Minor, six of them in Bithynia, all closely datable to the period ad 198/221, and one in Lydia, at Sardis (Fraser, Mat- thews, Catling and Ricl 2010: 448). There are also two men called Philippianos (or -us) of uncertain rank in Rome, one of the second century ad, the other of the first half of the third century (Solin 2003: 226). Mention of a Philippianus also occurs in a letter of Libanius (7.90). 69 The only example in Sicily recorded by Fraser and Matthews (1997: 451) is this one (based on the Piano della Clesia stamp); but cf. PHILIPPIANORVM on a brick- stamp at Tindari on the north coast (CIL 10.8045.16), also with a horse and surely related to the same Philippianus production; it came however from Baron Astuto’s collection at Noto (in Palermo since 1830), and the Tindari provenance may not be reliable. This stamp is unfortunately now lost and personal autopsy is not there- fore possible: I am grateful to Dr. Francesca Spatafora, Director of Palermo’s Museo Archeologico Regionale, for trying to locate it for me.

215 R.J.A. Wilson

A dominus’ passion for horses can also be expressed in other ways. Pompeianus featured named horses on a mosaic in his villa’s baths at Oued Athménia in Algeria, as did Sorothus in his town-house at Sousse in Tunisia— two well-known examples of a widespread practice featuring named steeds on mosaics, especially in North Africa, but also elsewhere.70 It is logical to think that at least some of these depict horses owned by the dominus who commissioned each mosaic floor, and that the plumes that they wear and the trappings of victory often depicted suggest that they had competed in circus games.71 Gerace lies in an area of Sicily where horses are kept to this day, includ- ing on the modern Gerace estate. The well-watered terrain and the generally cooler climate make conditions for keeping horses more favourable than in many other parts of Sicily. The island had a very high reputation in the late Roman world for supplying both circus ponies and mounts. Ancient writers roughly contemporary with the Gerace tile-stamps, including Vegetius in his Ars Mulomedicina, the anonymous author of the Expositio Totius Mundi et Gentium and Symmachus in a letter of ad 401, all allude to Sicily as a valu- able source of supply.72 In view of this fact, it would not be surprising to find Philippianus operating in central Sicily as a breeder and supplier of horses at this time. The tile-stamp evidence is suggestive of such a possibility but cannot, of course, be absolutely conclusive.

3. Belt-buckle A well-preserved copper alloy belt-buckle plate with an intact fastening pin was discovered in a superficial level in Area B1 (figs. 23 and 24). The plate is not hinged in the middle but is rigid throughout: it consists of a rectangular

70 E.g. Torre de Palma (Portugal): Lancha and Beloto 1994; Darder Lissón 1996: no. 30; Santa Lucia di Aguilafuente (Spain): Darder Lissón 1996: no. 33; Oued Athménia, Sousse: Dunbabin 1978: 93–95; Rossiter 1992: 46–47; and for Sousse, see now in full Lavagne 2006, especially 1354–1386. For African mosaics with racehorse themes, Ennaïfer 1983; for a colour illustration of one (fourth century ad), from Djerba now in Blois, featuring a horse called Ispicatus with palm-branch plume, another in its tail and a third branch spread before it, see Lavagne 2001: 80–81. 71 See most recently a full discussion in Matter 2012: 64–66. One of the mosaic inscriptions at Oued Athménia, vincas non vincas te amemus Polidoxe (“win or lose, let’s love you, Polydoxus”), implies that Pompeianus raced horses competitively, and Sorothus’ horses have his name branded on their flanks, so his ownership of them is clear. 72 Vegetius, Ars Mulomedicina 3.6.2–4, (writing after ad 383 because he describes the emperor Gratian as divus and before ad 450 when a revision was made by another hand); Expositio 65; Symmachus, Ep. 6.33 and 42. Cf. Matter 2012: 63–64, who also believes that the mention of probatio equorum on a fragment of the Taormina con- sular Fasti of 28/19 bc (AE 1988: 625–626; 1996: 788) indicates that horse-trainers of repute had been around in Sicily long before the fourth century ad.

216 UBC Excavations at Gerace, 2013

Figure 23. Copper-alloy belt-buckle from Area B1, fifth century ad (length 49 mm); photograph taken before conservation (scale 5 cm) Photo by R.J.A. Wilson

Figure 24. Copper-alloy belt-buckle, from Area B1, fifth century ad Drawing by Sally Cann (Matrice) right-hand part and a D-shaped portion to the left. The pin is attached not at the centre, as is customary, but to the right-hand end of the buckle-plate. The central division is narrowed at the middle where it takes the form of a small tubular join of metal (now partly broken); this was done to accommo- date the chunky pin, allowing it to lie flat. As usual, the apex of the “D” is of reduced thickness, again to accommodate a pin-rest. The central part of the surface has two pi-shaped grooves framing decorative fields, each partly filled with close-set engraved circles; each of the latter has a dot at the cen- tre. A single row of the same decoration covers most of the rest of the flat

217 R.J.A. Wilson upper part of the buckle-plate, but not the pin; the spacing of the dots is somewhat irregular throughout. The object’s maximum length is 49 mm (of the pin, 51 mm); its maximum width (top to bottom) is 41 mm. This buckle-plate is unique in Sicily. Most types found in the island are hinged in the middle and are very different in appearance and shape. Of Maurici’s classification of the Sicilian finds into 13 types, only one (his Type 13) is of the fixed-plate variety, represented by a single example from .73 That, however, consists of two roughly equal halves, both rect- angular, very different from the D-shaped Gerace example. In addition, the pin is shorter on the Centuripe buckle-plate and affixed to a central hole. Circles with a punched central dot (rather larger and more prominent than on the Gerace object) do occur on Sicilian buckles,74 but the decoration is so commonplace over a wide range of personal objects across the entire late Roman and Byzantine world that this causes no surprise. The origin of this type of belt-buckle is military, but it also occurs on civilian sites as well. By the fourth century, such items were worn by high-ranking civilians as a status symbol, in affectation of military dress.75 The closest comparisons with the Gerace find come from North Africa. Examples in Tunisia, probably from Carthage, and in Algeria, at Tiddis (or Constantine) and Djemila, several of them also with circle and dot decora- tion, are very similar.76 Christoph Eger informs me that in his view the Gerace buckle is unlikely to be later than the fifth century; so it might have been worn and lost by someone living in, or visiting, the late Roman villa there (ca. ad 375 to ca. ad 500). It is not impossible that the item was brought from North Africa, but so little is known in Sicily at present about these much neglected objects that other parallels might lie undetected in Sicilian museum collections. Conclusions about where the Gerace buckle was man- ufactured are probably, therefore, premature.

Conclusions The research carried out by the University of British Columbia in both 2012 (the geophysics) and 2013 has greatly increased our knowledge of the rural settlement at Gerace. The villa-like building now looks likely to have been built in the second half, and probably the last quarter, of the fourth century and not earlier as originally thought, and it is also now known that it perished

73 Maurici 2002: 520 (Centuripe 2), with fig. 1.13 on 543 and fig. 7.4 on 547; Dann- heimer 1989: 36, no. 15, with Taf. 2.15 (some oblique incised lines as decoration). 74 For example, Dannheimer 1989: Taf. 2.13 and 17; Maurici 2002: 547–551, fig. 7.6–7; fig. 8.14 and 8.19–21; fig. 11.38 and 40. 75 Swift 2009: 169–179; Reece 2007: 155–157; Booth et al. 2010: 516–517. 76 Eger 2012: 132–136 (“U-förmigen Schnallplatten mit festen Rahmenbeschlag und Riemensteg”), and Taf. 3.1–2; 10.2–3; and 11.3–5 and 9, with catalogue entries on 356 and 365–368 (eight examples of which five show some circle-and-dot decoration).

218 UBC Excavations at Gerace, 2013 during a fierce fire, probably toward the end of the fifth century (or at any rate not earlier than ca. ad 450). Two of the rooms to the west of the apsed reception chamber (Room 6 on fig. 4) have been shown to have had either a plain earth floor (Room 1) or a simple white mortar surface (Room 2), rather than mosaic paving. In addition, close inspection of the remains of Rooms 3 and 4, cleared to the tops of the walls, shows that Room 3, like Room 1, had no wall plaster at all, while Room 4, like Room 2, probably also had simple white plaster. All four rooms here are best regarded as service rooms and not “show” rooms intended to be seen by visiting guests. For them, their route through the villa would have been along the south corridor and into the apsed reception/dining room, and so the owner’s resources were ded- icated to making those spaces as impressive as possible by flooring them with geometric mosaic pavements. The discovery of a recess that probably contained a hot bath at the eastern extremity of this part of the villa-like building (Trench B1) suggests that a modest bath-suite may lie adjacent to the apsed reception hall: exploring this will be one of the priorities of the next excavation season.77 If its presence is confirmed, the dominus would have been able to offer his guests the delights of first bathing and then din- ing in close proximity to one another. I have referred throughout this report to this part of the site (Area A) as the “villa-like” building, not least because the presence of mosaic pavements in the south corridor and in Room 6 suggests a well-appointed building of a type appropriate to a villa; but while Room 6 can be classified as being used for “reception,” it and the other adjacent rooms do not form part of the res- idential quarters sensu stricto of the villa owner, which must lie elsewhere. The plan of Area A shows further rooms (identified by surface clearance) opening off to the west of the western corridor, one of which (Room 7) has a width of 7.50 m, and it is possible that some of the rooms here were residen- tial in character; on the other hand, there is no sign (on admittedly limited evidence) that Room 7 was provided with wall plaster. Unfortunately, the continuation of the villa here to the west lies buried under the large spoil heap from the 1994 excavations and is presently inaccessible. Beyond them, further west again, lies a field where the view southwards is at its most splen- did (fig. 3), but it is now known that no further part of the late fourth-cen- tury villa or any earlier building lies buried in this field.78 To the east of Area A, the 3 m wide trench (Area B1) across the large build- ing revealed by geophysics has shown very clearly its non-residential char- acter. It was probably built not earlier than the second quarter of the fourth century ad, and it therefore preceded the villa-like building in Area A of ca. ad 375. Its interior was paved with stone flags, as was also a short stretch of

77 None was able to take place in 2014. 78 See note 7 above.

219 R.J.A. Wilson ground outside, and immediately adjacent to, both exterior walls. The build- ing was probably intended as a storehouse for grain and other foodstuffs. It may have perished in an earthquake, perhaps that of ad 361/363, as a result of which Libanius reported that “all the cities of Sicily lay in ruins.” Rather than rebuild it, Philippianus apparently chose to build a new villa immedi- ately to the west of the now-ruined long building, clearly not before ca. ad 365, perhaps ca. 375. Whether the storehouse was also replaced elsewhere by another structure awaits further research; so too does the location of the expected residential villa contemporary with the storebuilding: could it be the structure marked 3 on fig. 5, or do its demolished remains lie underneath the late villa-like building in Area A? The widths of the walls of the building in Area B are of some interest. At 69 cm wide, they are exactly half the width of the “terrace wall” (1.37 m), and it would appear that the architect was using the Ionian or “Samian” foot of 34.8 cm as the basis for measurement,79 in which case the granary walls are exactly two Ionian feet wide, and the thick wall on the eastern side of the trench is four Ionian feet wide. If correct, this would be an interesting illustration of the conservatism of building practice in Roman Sicily, where a long-established Greek unit of measurement was apparently still in use in the fourth century. By contrast, the walls of the later villa in Area A, approximately 59 cm wide, seem to have used the Attic or Cycladic foot (29.4/29.6 cm) for its unit of measure- ment, the walls here being two Attic feet wide. This was the same as the Roman foot (29.4/29.7 cm but generally around 29.6 cm). Whether the architect of this building at Gerace was thinking in Roman units of measurement or Greek (the Attic foot) is therefore open to speculation. Further research on the metrol- ogy of the Gerace buildings is planned in the hope of understanding better the working practices of the ancient architects who worked there. Evidence both earlier and later than the mid- and later fourth century ad was also discovered during the 2013 excavations. A small portion of an earlier structure (third or early fourth century?), which was destroyed when the large building in Area B was built, was found beneath and appears to rest on a still earlier building below. Although not found to be associated with it, a single sherd of Hayes Form 3a in the ARS series, and a single plate base of Italian sigillata ware, show at least some frequentation of the site in the first and second centuries ad, but material of this date was extremely rare. A single unstratified fragment of a black-glazed lamp (fourth century bc?) hints at still earlier occupation, or at least visitation, of the site. In addition, late occupation in the ruins of the large building in Area B included the con- struction of a hearth, associated with pottery that hitherto has been thought to belong to ca. ad 800, although an earlier date is not impossible.

79 Cf. Wilson Jones 2000, especially 75 (with bibliography in his note 16) for a recent discussion of the use of this and other units of measurement in the Greek period. For the Roman period, cf. Senseney 2014, especially 151–153.

220 UBC Excavations at Gerace, 2013

In summary, five principal phases of activity were identified during the 2013 excavations at Gerace:

Phase 1 the footings of a wall at the east end of Area B (undated; third century ad?); Phase 2 parts of two rooms, one with white plastered walls, of a build- ing on top of the Phase I structure (undated; third century or early fourth century ad?); Phase 3 the long building (Area B), probably a granary, with stone- paved interior floor (ca. ad 325/350?); Phase 4 the “villa-like” building (Area A) (ca. ad 375 or later); Phase 5 Byzantine activity in the ruins of the collapsed building in Area B1 (ninth century ad or earlier?).

The dates are provisional. Those for Phases 3 and 4, and for the destruc- tion of both, the first ca. ad 360 (?) and the second after ca. ad 450, are based on the forms of the very few rim sherds of ARS (and the two lamps) that the excavations produced. This chronology may stand in need of revision as fur- ther evidence emerges in future work.

Acknowledgments My first debt is to the landowner of Gerace, Dott.ssa Antonella Fontanazza Coppola, without whose permission to enter her property and conduct disrup- tive excavation, this research would never have been possible. For the many personal kindnesses shown by her and her late husband, Dott. Ettore Coppola, I shall always be extremely grateful. An equal debt of gratitude is due to the Regione Siciliana and, in particular, to its then Direttore Generale dei Beni Culturali e dell’Identità Siciliana, Dott. Sergio Gelardi, and to its then Asses- sore, Prof.ssa Mariarita Sgarlata, who kindly entrusted me with a concessione di scavo. This was through the agency of the Soprintendenza per i Beni Cultur- ali di Enna, where I would particularly like to acknowledge the practical help and encouragement of the then Soprintendente, Architetto Fulvia Caffo, and of her staff, especially Arch. Giuseppe Farina and Dott.ssa Francesca Valbruzzi. At Il Mandorleto, our delightful base throughout the project, Maurizio Stellino, his wife Saida Crea, their elder son Alessandro, and their now daughter-in-law Viorica Cantaro were a never-failing source of kindness, support, encourage- ment, patience and practical help. In addition, the meals that they cooked for us every day were without exception mouth-wateringly memorable (not least their younger son Marco’s ice creams). Their neighbour, Dott. Carmelo Fon- tanazza, also provided invaluable logistical support and showed a keen interest in the progress of the excavations throughout. A tower of strength throughout the project, as mentor and friend, has been Dott.ssa Enza Cilia Platamone, the first excavator of Gerace in 1994, a never-failing source of friendship, encour- agement and advice from the moment that I first expressed to her an interest in conducting research at Gerace. I owe more to her than words can express.

221 R.J.A. Wilson

The excavation would not have been possible without funding from the Social Sciences and Humanities Research Council of Canada, whose sup- port in the form of an Insight Grant is here gratefully acknowledged. Gen- erous funding was also provided by the Society for the Promotion of Roman Studies, the Association for the Study and Preservation of Roman Mosaics, the Faculty of Arts of the University of British Columbia, and the Research Fund of my Department of Classical, Near Eastern and Religious Studies. I am indebted to various colleagues whose expertise contributed enormously to the project, especially Sophie Hay and her team from the British School at Rome, whose skills in geophysics have contributed so much to expanding our knowledge of Gerace; Alan Weston (British Columba Institute of Tech- nology), who conducted the topographic survey in often difficult terrain with great expertise and never a complaint; Filippo Ianni (Villarosa), who took the two aerial photographs (figs. 6 and 10); Sally Cann (Matrice), who drew the finds with her customary skill and unfailing good cheer (figs. 24 and 25 are her work); Amanda Foley (Brockport), who lovingly extracted carbonized mate- rial from selected deposits by flotation; Carmela Franco (London), who read and advised on my description of the ARS material in the Appendix below; Jean Bussière (Pézenas), who has generously once again put at my disposal his encyclopaedic knowledge of late Roman lamps; Christoph Eger (Göttin- gen), who went out of his way to help me contextualize the belt-buckle plate; Michael MacKinnon (Winnipeg), who will study the faunal assemblages at Gerace in a later season, for identifying cattle bones in Room 1 of Area A from my photographs; Christer Bruun (Toronto), who kindly responded at length to my epigraphic questions about the tile-stamps; and Jim Russell (Vancou- ver), who commented on a first draft of this paper with his customary insight and eye for detail. Andrew Birley (Vindolanda) kindly presented me hot off the press with a copy of his report on the Vindolanda granaries, while Eliza- beth Greene (University of Western Ontario, London, ON) alerted me to its imminence and arranged for its transportation through the kindness of Alex Mayer; I am grateful to all three of them. My biggest debt is to my excavation supervisors, Danny Waterfall (Area A) (Ossett) and Antonietta Lerz (Area B) (London), both former students of mine who are now full-time professional archaeologists, who recorded their respective trenches with meticulous care and exemplary skill, and who were also brilliant teachers of archaeological method to their teams. The latter were composed almost exclusively of stu- dents from the University of British Columbia,80 who were happily joined for four days by Laura Mazzaglia, a graduate student of the University of Cata- nia. Last but far from least, my wife Charlotte Wilson has provided invaluable logistical support for this project from its inception. Without the dedication,

80 Bethany Brothers, Patricia Taylor (graduate students), and Marco Aldro- vandi, Elysia Allen, Mikaela Buscher, Chloe Martin-Cabanne, Kirsten Clarke, Kava- nagh Golka, Holly Hadley, Irene Lo, Charlotte Mackie, Liria Nair and Emilly Porter (undergraduates).

222 UBC Excavations at Gerace, 2013 ceaseless hard work and sheer technical skill of all of the members of my team, Gerace would never have yielded the quantity of new information presented above. I am truly grateful to them all.

Appendix ARS pottery finds listed below (fig. 25) are from contexts that are key to the chrono- logical framework outlined above.81

Context 51 Area B1, make-up fill between the exterior paving of the granary on its east side and the broad terrace wall.

1. [P. 17] D. 34 cm. Plate with low wall and sharply in-turned rim. Fine ware with bright lustrous slip on interior (“C” ware). Ca. 230/40 to ad 300 or later (Hayes ARS Form 49): cf. Mackensen 1993: 320, Abb. 110.1–2. 2. [P. 18] D. 22.4 cm. Near vertical rim, slightly rounded on outside, slight groove on inside. Early version of Hayes ARS Form 61 A, cf. Bonifay Type 37 (2004: 167); Mackensen Form 4.1–2, with Taf. 53, dated by him approximately in the second quarter of fourth century. 3. [P. 20] D. 25.0 cm. Bonifay (2004: 162) Type 24 (Sidi Abdullah, Tunisia); cf. also Mahrine Form 1 (Mackensen 1993: 398); Carandini 1981: 92, Tav. XL.5 (Sperlonga); related to Hayes Form 58B, produced between ca. ad 300/310 and ca. ad 375 (Mackensen 1993: 398). 4. and 5. [P. 44 and 45] D. of both uncertain. Hayes ARS Form 50 (later variety, with coarsish fabric), ad 300/360 (Hayes).

Context 60 Area A, embedded within the white mortar floor of Room 2.

6. [P. 85] Rim fragment, D. 21.7 cm, of Hayes ARS Form 67 (1972: 112–116), not before ad 360; Bonifay (2004: 171–173) Form 41A; Mackensen Form 9 (1993: 595–596, with 403–404 on dating). Early examples of the form are attested in destruction deposits associated with the earthquake of ad 365 in the eastern Mediterranean (Reynolds et al. 2011: 18; see also note 11 above).

Context 64 Area A, from within the tile fall on top of the mosaic pavement in the south corridor

7. [P. 51] D. 19.2 cm. Hayes ARS Form 81, cf. variant with slightly thickened rim, no. 8 (Hayes 1972: 128); second half of fifth century.

Context 23 Area A, ashy layer of burning in Room 1

8. [P. 19] D. 27.0 cm; near vertical outer rim; groove on inside. Hayes ARS Form 61B, ad 400/450; cf. Bonifay (2004: 167–170) Form 38, late variant B3, dated by

81 Hayes 1972. The P. number in square brackets refers to the drawn find number in the site inventory.

223 R.J.A. Wilson

Figure 25. Selected ARS from the 2013 excavations Drawing by Sally Cann (Matrice)

him to ad 420/500 (it has been noted at Carthage in levels down to the end of the fifth century).

Department of Classical, Near Eastern and Religious Studies University of British Columbia [email protected]

References Adamesteanu, D. 1963. “Nuovi documenti paleocristiani nella Sicilia centro- meridionale,” BA 48: 259–274. Alonso, F.G. 1995. “Producción y comercio del cereal en el N.E. de la Península Ibérica entre los siglos VI–II a.C,” Pyrenae 26: 91–113.

224 UBC Excavations at Gerace, 2013

Arce, J. and B. Goffaux (eds.). 2011. Horrea d’Hispanie et de la Méditerranée romaine. Collection de la Casa de Velázquez 125. Madrid: Casa de Valázquez. Arcifa, L. 2008a. “Facere fossa et victualia reponere: La conservazione del grano nella Sicilia medievale,” MEFRM 120: 39–54. ———. 2008b. “L’area del santuario dall’età bizantina all’XI secolo,” in L. Manis- calco (ed.), Il santuario dei Palici: Un centro di culto nella Valle del Margi. Collana d’Area. Quaderni 11. Palermo: Regione Siciliana. 291–309. ———. 2010a. “Indicatori archeologici per l’alto Medioevo nella Sicilia orientale,” in P. Pensabene (ed.), Piazza Armerina: Villa del Casale e la Sicilia tra tardoantico e medioevo. Rome: L’Erma di Bretschneider. 105–128. ———. 2010b. “Indicatori archeologici e dinamiche insediative nella Sicilia tardo bizantina,” in M. Congiu, S. Modeo and M. Arnone (eds.), La Sicilia bizantina: storia, città e territorio. Atti del VI Convegno di studi. Caltanissetta and Rome: Salvatore Sciascia Editore. 67–89. Atlante. 1981. Atlante delle forme ceramiche I: Ceramica fine romana nel bacino Medi- terraneo (medio e tardo impero). Rome: Istituto della Enciclopedia Italiana. Badisches Landesmuseum Karlsruhe (ed.). 2009. Das Königreich der Vandalen: Erben des Imperiums in Nordfrika. Karlsruhe: Badisches Landesmuseum Karlsruhe. Barbera, M. and R. Petriaggi. 1993. Le lucerne tardo-antiche di produzione africana. Rome: Libreria dello Stato. Berthier, A. 1962–1965. “Établissements agricoles antiques à Oued Athménia,” Bulle- tin de l’Archéologie Algérienne 1: 7–20. Birley, A. 2013. The Vindolanda Granary Excavations. Greenhead: Roman Army Museum Publications. Bisconti, F. 1979. “Aspetti e significativi del simbolo della fenice nella letteratura e nell’arte del Cristianesimo primitivo,” VetChr 16: 21–40. ———. 1981. “Lastra incisa inedita dalla Catacomba di Priscilla (con note di revisione critica sul metodo di individuazione della fenice nell’arte paleocristiana),” RAC 57: 43–67. Black, E.W. 1981. “An Additional Classification of Granaries in Roman Britain,” Bri- tannia 12: 163–165. http://dx.doi.org/10.2307/526247. Bodel, J. 2005. “Speaking signa and the Brickstamps of M. Rutilius Lupus,” in C. Bruun (ed.), Interpretare i bolli laterizi di Roma e della Valle del Tevere: produzione, storia economica e topografia. Rome: Acta Institiuti Romani Finlandiae 32. 61–94. Bonacasa Carra, R.M. and R. Panvini (eds.). 2002. La Sicilia centro-meridionale tra il II ed il VI sec. d.C. Catalogo della mostra, Caltanissetta–Gela, aprile–dicembre 1997. Palermo: Regione Siciliana, and Caltanissetta: Salvatore Sciascia Editore. Bonanno, C. 2012. “La villa romana di Gerace,” in S. Lo Pinzino (ed.), Studi, ricerche, restauri per la tutela del patrimonio culturale ennese. I Quaderni del Patrimonio Culturale Ennese. Collana Interdisciplinare del Servizio, Sorintendenza per i Beni Culturali ed Ambientali di Enna diretta da Fulvia Caffo 1. Palermo: Regione Sicil- iana. 88–92. ———. 2013a. “La villa romana di Gerace (EN),” in F.P. Rizzo (ed.), La villa del Casale e oltre. Territorio, popolamento, economia nella Sicilia centrale tra Tarda Antich- ità e Alto Medioevo. Atti delle Giornate di Studio (Piazza Armerina 30 settembre–1 ottobre 2010). Macerata: Edizioni Università di Macerata. 181–208. ———. 2013b. “La villa romana di Gerace,” in C. Bonanno and F. Valbruzzi (eds.), Mito e Archeologia degli Erei. Museo Diffuso Ennese. Itinerari Archeologici. Palermo: Assessorato Regionale dei Beni Culturali e dell’Identità Siciliana. 113–117. ———. 2014. “La villa romana di Gerace e altri insediamenti residenziali nel territorio ennese,” in P. Pensabene and C. Sfameni (eds.), La Villa restaurata e i nuovi studi

225 R.J.A. Wilson

sull'edilizia residenziale tardoantica. Atti del Convegno internazionale del CISEM di Roma, Piazza Armerina 7–9 novembre 2012. Bari: Edipuglia. 79–91. ———. Forthcoming. “Nuove esplorazioni in località Gerace (Enna): la conservazi- one dei mosaici pavimentali,” in Acts of the 10th Conference of the International Committee for the Conservation of Mosaics, Palermo 2008, in press. Bonanno, C., R. Carbella, C. Capelli and M. Piazza. 2010. “Nuove esplorazioni in località Gerace (Enna–Sicilia),” in S. Menchelli et al. (eds.), Congresso internazio- nale sulle ceramiche comuni, le ceramiche da cucina e le anfore della tarda antich- ità nel Mediterraneo. Archeologia e archeometria. British Archaeological Reports International Series 2185. Oxford: British Archaeological Reports. 261–272. Bonifay, M. 2004. Études sur la céramique romaine tardive d’Afrique. British Archaeo- logical Reports International Series 2185. Oxford: British Archaeological Reports. Booth, P., A. Simmonds, A. Boyle, S. Clough, H.E.M. Cool and D. Poore. 2010. The Late Roman Cemetery at Lankhills, Winchester. Excavations 2000–2005. Oxford Archaeology Monograph 10. Oxford: Oxford Archaeology. Brethenoux, J.-P. 2012. L’agglomération antique de Chassenon (Charente). Poitiers: Région Poitou-Charentes. Brodribb, G. 1987. Roman Brick and Tile. Gloucester: Alan Sutton Publishing. Brown, P. 2013. Through the Eye of a Needle: Wealth, the Fall of Rome, and the Making of Christianity in the West, 350–550 AD. Princeton, NJ: Princeton University Press. Busana, M.S. 2002. Architetture rurali nella Venetia romana. Rome: L’Erma di Bretschneider. Bussière, J. 2007. Lampes antiques d’Algérie II. Lampes tardives et lampes chrétiennes. Montagnac: Éditions Monique Mergoil. Carandini, A. (ed.). 1981. Atlante delle forme ceramiche I. Ceramica fine romana nel bacino mediterraneo (medio e tardo impero). Rome: Istituto della Enciclopedia Italiana. ———. 1985. Settefinestre. Una villa schiavistica nell’Etruria romana. 2. La villa nelle sue parti. Modena: Edizioni Panini. Carandini, A., A. Ricci and M. De Vos. 1982. Filosofiana. La villa di Piazza Armerina. Palermo: S.F. Flaccovio Editore. Castellana, G. and B.E. McConnell. 1990. “A Rural Settlement of Imperial Roman and Byzantine Date in Contrada Saraceno near Agrigento, Sicily,” AJA 94: 25–44. http://dx.doi.org/10.2307/505524. Cilia Platamone, E. 1996. “Recente scoperta nel territorio di Enna: l’insediamento tardo-romano in contrada Geraci,” in M. Khanoussi, P. Ruggeri and C. Vismara (ed.), L’Africa Romana 11.3. Ozieri: Editrice Il Torchietto. 1683–1689. ———. 1997. “Rinvenimenti musivi nel territorio di Enna tra passato e presente,” in R.M. Carra Bonacasa and F. Guidobaldi (eds.), Atti del IV Colloquio dell’Associazi- one Italiana per lo Studio e la Conservazione del Mosaico (Palermo 9–13 dicembre 1996). Ravenna: Edizioni del Girasole. 273–290. Costanzo, M.A. 1996. “Granai di età ellenistica in Sicilia,” in Atti delle Giornate di Stu- dio sugli insediamenti rurali nella Sicilia antica (Caltagirone 29/30 giugno 1992). Catania: Edizioni Greco. 67–70. Dannheimer, H. 1989. Byzantinische Grabfunde aus Sizilien. Christliches Brauchtum im frühen Mittelalter. Einführung und Katalog. Munich: Prähistorische Staatssam- mlung München, Museum für Vor- und Frühgeschichte. Darder Lissón, M. 1996. De nominibus equorum circensium: Pars occidentis. Barce- lona: Reial Acadèmia de Bones Lletres. Deussen, P.W. 1994. “The Granaries of Morgantina and the lex Hieronica,” in Le rivitaillement en blé de Rome et des centres urbains des débuts de la République

226 UBC Excavations at Gerace, 2013

jusqu’au Haut Empire. Actes du colloque international de Naples, 1991. Collec- tions Jean Bérard 11/Collections École française de Rome 196. Naples: Centre Jean Bérard and Rome: École française de Rome. 231–235. Di Vita, A. 1972–1973. “La villa di Piazza Armerina e l’arte musiva in Sicilia,” Kokalos 18–19: 251–261. ———. 1982. “Evidenza di terremoti del 306–310 e del 365 in monumenti e scavi in Tunisia, Sicilia, Roma e Cirenaica,” Africa 7–8: 127–139. Drine, A. 2007. “Les entrepôts de Méninx,” AntAfr 43: 239–251. Dunbabin, K.M.D. 1978. The Mosaics of Roman North Africa: Studies in Iconography and Patronage. Oxford: Clarendon Press. ———. 1982. “The Victorious Charioteer on Mosaics and Related Monuments,” AJA 86: 65–89. http://dx.doi.org/10.2307/504294. Eger, C. 2012. Spätantikes Kleidungszubehör aus Nordafrika I. Trägerkreis, Mobilität und Ethnos im Spiegel der Funde der spätesten römischen Kaiserzeit und der van- dalischen Zeit. Münchener Beiträge zur Provinzialrömischen Archäologie 5. Wies- baden: Reichert. Ennabli, A. 1976. Lamps chrétiennes de Tunisie (Musée du Bardo et de Carthage). Paris: Éditions du Centre National de la Recherche Scientifique. Ennaïfer, M. 1983. “Le thème des chevaux vainqueurs à traverse la série des mosaïques africaines,” MEFRA 95: 817–858. http://dx.doi.org/10.3406/mefr.1983.1393. Fiorentini, G. 1985. Gela. La città antica e il suo territorio. Il museo. Palermo: Asses- sorato Regionale Beni Culturali Ambientali e P.I., and Agrigento: Soprintendenza archeologica Sicilia centro-meridionale. Fitch, J.G. 2013. Palladius: the Work on Farming (Opus Agriculturae) and Poem on Grafting: A New Translation from the . Totnes: Prospect Books. Fraser, P.M. and E. Matthews. 1997. A Lexicon of Greek Personal Names IIIA: The Peloponnese, Western Greece, Sicily and Magna Graecia. Oxford: Clarendon Press. Fraser, P.M., E. Matthews and R.W.V. Catling. 2005. A Lexicon of Greek Personal Names IV: Macedonia, Thrace, Northern Regions of the Black Sea. Oxford: Clarendon Press. Fraser, P.M., E. Matthews, R.W.V. Catling and M. Ricl. 2010. A Lexicon of Greek Per- sonal Names VA: Coastal Asia Minor: Pontos to Ionia. Oxford: Clarendon Press. Gabrici, E. 1925. “Girgenti. Scavi e scoperte archeologiche dal 1916 al 1924,” NSA 1925: 420–461. Garbsch, J. and B. Overbeck (eds.). 1989. Spätantike zwischen Heidentum und Chris- tentum. Munich: Prähistorische Staatssammlung München, Museum für Vor- und Frühgeschichte. Gaubatz-Sattler, A. 1994. Die Villa rustica von Bondorf. Forschungen und Berichte zur Vor- und Frühgeschichte in Baden-Württemberg 51. Stuttgart: Konrad Theiss Verlag. Gentili, G.V. 1999. La villa romana di Piazza Armerina. Palazzo Erculio. 3 vols. Osimo: Fondazione Don Carlo. Gentry, A.P. 1976. Roman Military Stone-built Granaries in Britain. British Archaeo- logical Reports 32. Oxford: British Archaeological Reports. Graziani Abbiani, M. 1969. Lucerne fittili paleocristiane nell’Italia settentrionale. Bologna: Casa Editrice Prof. Riccardo Pàtron. Guidoboni, E. (ed.). 1989. Le terremote prima del Mille in Italia e nell’area mediterra- nea. Bologna: SGA Storia – Geofisica – Ambiente. Guzzardi, L. 2001–2002. “Attività della sezione archeologica della Soprintendenza di Enna negli anni 1997–2000,” Kokalos 47–48: 561–598. Hayes, J.W. 1972. Late Roman Pottery. London: British School at Rome.

227 R.J.A. Wilson

Heimberg, U. 2011. Villa rustica: Leben und Arbeiten auf römischen Landgütern. Mainz am Rhein: Philipp von Zabern. Hellmann, M.-C. 2002. L’architecture grecque: 1. Les principes de la construction. Paris: Picard. Kajanto, I. 1963. Onomastic Studies in the Early Christian Inscriptions of Rome and Carthage. Helsinki: Helsingfors. ———. 1966. Supernomina: A Study in Latin Epigraphy. Commentationes Humanarum Litterarum 40.1. Helsinki: Helsingfors. Lancha, J. and C. Beloto. 1994. Chevaux vainqueurs. Une mosaïque romaine de Torre de Palma, Portugal. Paris: Fondation Calouste Gulbenkian and Centre Culturel Portugais. La Torre, G.F. 1994. “Gela sive Philosophianis (It. Antonini, 88.2): contributo per la storia di un centro interno della Sicilia romana,” Quaderni dell’Istituto di Archeo- logia della Facoltà di Lettere e Filosofia della Università di Messina 9: 99–139. Lavagne, H. (ed.). 2001. Mosaico romano del Meditérraneo. Paris: Unión Latina, and Madrid: Museo Arqueológico Nacional. ———. 2006. “Sousse: la domus de Sorothus et ses mosaïques,” CRAI 2006: 1327–1392. Lentini, M.C. (ed.). 2001. Naxos di Sicilia in età romana e bizantina ed evidenza dai Peloritani. Bari: Edipuglia. Libertini, G. 1922. “Acireale. Scoperte a Casalotto,” NSA 1922: 491–499. Mackensen, M. 1993. Die spätantiken Sigillata- und Lampentöpfereien von El Mahrine (Nordtunisien), 2 vols. Munich: C.H. Beck. Manganaro, G. 1988. “La Sicilia da Sesto Pompeo a Diocleziano,” in H. Temporini (ed.), Aufstieg und Niedergang der römischen Welt. II.11.1. Berlin and New York: De Gruyter. 3–89. Matter, M. 2012. “Des chevaux du cirque : économie et passions à Rome,” in S. Lazaris (ed.), Le cheval, animal de guerre et de loisir dans l’Antiquité et Moyen Âge. Biblio- thèque de l’Antiquité Tardive 22. Turnhout: Brepols. 61–72. Maurici, F. 2002. “Ancora sulle fibbie di cintura di età bizantina in Sicilia,” in R.M. Carra Bonacasa (ed.), Bizantino-Sicula IV. Atti del I Congresso Internazionale di Archeologia della Sicilia Bizantina. Quaderni 15. Palermo: Istitito Siciliano di Studi Bizantini e Neoellenici. 513–557. Mlasowsky, A., A. Uhsemann and W. Wolff. 1993. Die antiken Tonlampen im Kest- ner-Museum Hannover. Sammlungskatalog 8. Hannover: Kestner-Museum. Papi, E. and F. Martorella. 2007a. “I granai della Numidia,” AntAfr 43: 171–186. ———. 2007b. “Il grano della Tingitana,” in E. Papi (ed.), Supplying Rome and the Empire: The Proceedings of an International Seminar Held at Siena–Certosa di Pontagnano on May 2–4, 2004. JRA Supplementary Series 69. Portsmouth, RI: Journal of Roman Archaeology. 85–96. Parra, M.C. 1995. “L’edificio ellenistico nella conca orientale,” in G. Nenci (ed.), Entella I. Pisa: Scuola Normale Superiore di Pisa. 9–76. Pensabene, P. 2010–2011. “Studi recenti sulla Villa del Casale: gli interventi della Sapienza – Università di Roma. II. La villa del Casale tra tardo antico e medievo alla luce dei nuovi dati archeologici: funzioni, decorazioni e trasformazioni,” Atti della Pontificia Accademia romana di archeologia. Rendiconti 83: 141–226. Reece, R. 2007. The Later : An Archaeology, AD 150–600, 2nd ed. Stroud: History Press. Reynolds, P., M. Bonifay and M.A. Cau. 2011. “Key Contexts for the Dating of Late Roman Mediterranean Fine Wares: A Preliminary View and ‘Seriation’,” in M.A. Cau, P. Reynolds and M. Bonifay (eds.), Late Roman Fine Wares: Solving the Problems of Typology and Chronology. A Review of the Evidence, Debate and New

228 UBC Excavations at Gerace, 2013

Contexts. Roman and Late Antique Mediterranean Pottery 1. Oxford: Archaeo- press. 15–32. Rickman, G.E. 1971. Roman Granaries and Store Buildings. Cambridge: Cambridge University Press. Rossiter, J.J. 1978. Roman Farm Buildings in Italy. British Archaeological Reports International Series 52. Oxford: British Archaeological Reports. ———. 1992. “Stabula equorum: Evidence for Race-Horse Stables in Roman Africa,” in Afrique du Nord antique et médiévale: spectacles, vie portuaire, religions. Ve Collo- que International sur l’histoire et l’archéologie de l’Afrique du Nord, Avignon 1990. Paris: Éditions du Comité des Travaux Historiques et Scientifiques. 41–47. Salido Domínguez, J. 2011. Horrea militaria. Aprovisionamiento de grano al ejército en el occidente del Imperio Romano. Madrid: Consejo Superior de Investigaciones Científicas, and Polifemo. Salway, B. 1994. “What’s in a Name? A Survey of Roman Onomastic Practice from c. 700 B.C. to A.D. 700,” JRS 84: 124–145. http://dx.doi.org/10.2307/300873. Senseney, J.R. 2014. “Plans, Measurements, Systems and Surveying: The Roman Tech- nology of Pre-Building,” in R.B. Ulrich and C.K. Quenemoen (eds.), A Companion to Roman Architecture. Oxford: Wiley Blackwell. 140–156. Solin, H. 2003. Die griechische Personennamen in Rom, 2nd ed. 3 vols. Berlin: W. de Gruyter. Soraci, C. 2011. Sicilia frumentaria: Il grano siciliano e l’annona di Roma. Rome: L’Erma di Bretschneider. Spurr, M.S. 1986. Arable Cultivation in Roman Italy. Journal of Roman Studies Mono- graph 3. London: Journal of Roman Studies. Swift, E. 2009. Style and Function in Roman Decoration: Living with Objects and Inte- riors. Farnham, VT: Ashgate. Valbruzzi, F. 2012. “Archeologia dei paesaggi: gli insediamenti rurali di età romana e tardoantica nel territorio degli Erei,” in S. Lo Pinzino (ed.), Studi, ricerche, restauri per la tutela del patrimonio culturale ennese. I Quaderni del Patrimonio Culturale Ennese. Collana Interdisciplinare del Servizio, Sorintendenza per i Beni Cultur- ali ed Ambientali di Enna diretta da Fulvia Caffo 1. Palermo: Regione Siciliana. 205–240. van den Broek, R. 1972. The Myth of the Phoenix According to Classical and Early Christian Traditions. Leiden: Brill. Vassallo, S. 2009. “La villa rustica di Contrada San Luca (Castronovo di Sicilia),” in C. Ampolo (ed.), Immagine e immagini della Sicilia e di altre isole del Mediterraneo antico, vol. 2. Pisa: Edizioni della Normale. 671–677. Virlouvet, C. 2007. “Entrepôts de stockage, entrepôts et marches: Pour une typologie des horrea dans l’Afrique du Nord antique,” AntAfr 43: 165–169. Voza, G. 1989. “I crolli nella Marina,” in E. Guidoboni (ed.), Le terremote prima del Mille in Italia e nell’area mediterranea. Bologna: SGA Storia – Geofisica – Ambiente. 496–501. ———. 2003. I mosaici del Tellaro. Lusso e cultura nel sud-est della Sicilia. Syracuse: Erre Produzioni. Wilmott, T. 1997. Birdoswald: Excavations of a Roman Fort on Hadrian’s Wall and Its Successor Settlements: 1987–92. Archaeological Report 11. London: English Heritage. Wilson, R.J.A. 1979. “Brick and Tile in Roman Sicily,” in A. McWhirr (ed.), Roman Brick and Tile. British Archaeological Reports International Series 68. Oxford: British Archaeological Reports. 11–43.

229 R.J.A. Wilson

———. 1990. Sicily under the Roman Empire: The Archaeology of a Roman Province, 36 B.C. – A.D. 535. Warminster: Aris and Phillips. ———. 2000a. “Iscrizioni su manufatti siciliani in età ellenistico-romana,” in G. Nenci (ed.), Sicilia epigrafica. Atti del Convegno di 15–18 ottobre 1998. Annali della Scuola Normale Superiore di Pisa4, Quaderni 1999. 1–2. Pisa: Scuola Normale Superiore di Pisa. 531–556. ———. 2000b. “Rural Settlement in Hellenistic and Roman Sicily: Excavations at Campanaio (AG), 1994–1998,” PBSR 68: 337–369. http://dx.doi.org/10.1017/ S0068246200003974. ———. 2010. “Life, Death and Dining in Early Byzantine Sicily: UBC Excava- tions at Kaukana, 2008–2010,” Mouseion 10: 119–167. http://dx.doi.org/10.1353/ mou.2010.0061. ———. 2013. “Sicily, c. 300 – 133 BC,” in C. Smith (ed.), The Cambridge Ancient His- tory, new edition. Plates to Volumes VII Part 2 and VIII. The Rise of Rome to 133 BC. Cambridge: Cambridge University Press. 156–196. ———. 2014a. “La villa romana di Gerace: primi risultati della ricerca geofisica,” in P. Pensabene and C. Sfameni (eds.), La Villa restaurata e i nuovi studi sull'edili- zia residenziale tardoantica. Atti del Convegno internazionale del CISEM di Roma, Piazza Armerina 7–9 novembre 2012. Bari: Edipuglia. 109–116. ———. 2014b. “La villa tardoromana di Caddeddi sul fiume Tellaro (SR) e i suoi mosaici,” in P. Pensabene and C. Sfameni (eds.), La Villa restaurata e i nuovi studi sull'edilizia residenziale tardoantica. Atti del Convegno internazionale del CISEM di Roma, Piazza Armerina 7–9 novembre 2012. Bari: Edipuglia. 37–46. ———. 2014c. “Considerazioni conclusive,” in P. Pensabene and C. Sfameni (eds.), La Villa restaurata e i nuovi studi sull'edilizia residenziale tardoantica. Atti del Convegno internazionale del CISEM di Roma, Piazza Armerina 7–9 novembre 2012. Bari: Edipuglia. 691–702. ———. 2014d. “Tile-Stamps of Philippianus in Late Roman Sicily: A Talking sig- num or Evidence for Horse-Raising?,” JRA 27: 472–486. http://dx.doi.org/10.1017/ S1047759414001366. ———. Forthcoming. Caddeddi on the Tellaro: a Late Roman Villa in Sicily and its Mosaics. Bulletin Antieke Beschaving Supplement. Leuven: Peeters. Wilson Jones, M. 2000. “Doric Measure and Architectural Design I: The Evidence of the Relief from Salamis,” AJA 104: 73–93. http://dx.doi.org/10.2307/506793.

230