In Search of the Roman Frontier in Sardinia

I. In the present state of our knowledge, which is admittedly not very secure, there are in Sardinia as elsewhere (Greece and Iberia, for example) no sites with anything resembling massive fortifications before the Copper Age.¹ Besides being unfashionable to speak of invaders (Chapman et al. 1984: 2.274), there is very little evidence to support an invasion hypothesis: the Bell Beaker Culture does make its first appearance in Sardinia at this time (Ferrarese Ceruti 1981: lv-lxv),² but its traces are too few and too scattered to account for the widespread appearance of defensive architecture; rather, we must think of internal pressures generated by a growing population and by an increase in accumulated wealth which required protection (Rowland 1991: 87-117). This is not the place to trace the evolution from the late fourth millennium of the native cultures which, by the late third millennium, have become recognizable as the nuragic culture, so called from the large stone towers (nuraghi), which are the single most characteristic feature of Sardinian prehistory (Lilliu 1962) and of the actual Sardinian landscape.

Before the end of the second millennium (ca. 15th-13th century), a number of these structures had evolved into true castle fortresses, undoubtedly the centers of chiefdoms, and this evolutionary process continued through the Iron Age: not enough firm chronological indicators exist as yet to allow us in each case to disentangle the architectural history of each structure; but, our present purposes render that defect irrelevant. By the time of the Carthaginians and then of the Romans, these multi-towered (or “polylobate”) nuraghi had arrived at their fully developed complexity, indicators of surplus labor at the disposal of the chiefs and accumulators of surplus capital. Contacts with the outside world may have served as a stimulator to the acceleration of the evolutionary process; this is not yet clear. We now know of contacts with the Mycenaean world (Ferrarese Ceruti 1981a: 605-612; Harding 1984: 252-255 and passim; Jones and Day 1987: 257-269; Ugas 1996) which, so far as we can tell at present, remained relatively slight; but in at least two sites, the evidence for these contacts seems to be associated with a very early use of iron (Rowland 1985: 17-20; Lo Schiavo 1989: 83, n. *, 89). Hard on the Mycenaean contacts followed contacts with early Phoenicians, perhaps with both the Levant and with Cyprus, perhaps at first with Cyprus alone (Ridgway 1979-80: 54-58; Barreca 1982: 57-61; Bernardini 1993: 29-67). There is some clear evidence of late second millennium trade relationships between Phoenicians and Sardinians, including Cypriote or Cypriote-derived bronze objects: a joint U.S.-Israeli-Italian underwater team spent part of 1984 looking for direct traces of Phoenician shipping in Sardinian waters (Barreca et al. 1986). By the ninth century, the Phoenicians had established commercial entrepots in several coastal sites, but there is no evidence for their having engaged in widespread territorial expansion into the interior before about the seventh century, expansion which was followed around the turn of the sixth and fifth centuries by a very determined and energetic Carthaginian expansion on a number of fronts (Moscati et al. 1997).

At several sites not far from the coast, such as Monte Sirai in the southwest only 5 km. inland, we have Phoenician colonial outposts; it would be excessive to speak of a Phoenician frontier system. Monte Sirai itself was burned, presumably by natives on the counterattack around 550 BC, but was soon rebuilt and strengthened; it would be out of place here to trace in detail what little we know of the subsequent Carthaginian advance; suffice it to say that by the middle of the fifth century they developed a coherent frontier system and an interior line of defense. As I have previously noted (Rowland 1982: 20-39), this system segregated the mountainous interior from the coastal exterior, and controlled the principal lines of communication between the two worlds, the largely but not exclusively pastoral world of the interior, roughly east of the line, from the agricultural and maritime world of the exterior to the west and south. The forts of this system served as centers of control and as points of contact between the two civilizations, the urban civilization of the Carthaginians and the village and tribal-center world of the nuragic folk. From this system we can immediately recognize one principal difference between the Carthaginian control of Sardinia and the Romans’ modality of control, namely, the organization of space: without denying that Carthaginian imperialism was brutal and sanguinary, it at least had a limit; the Romans’ compulsion to subdue and organize everything, to control all by domination, stands in sharp contrast with the Carthaginians’ recognition of limits and their ability and willingness to conduct warfare by other means.

The history of the Romans’ first involvement with Sardinia, that is, the cooperative resistance to Romans by Carthaginians and natives joined together, along with the Carthaginians’ success in provoking resistance during the first generation of the latter’s conquest, speaks eloquently of the mutual advantages deriving from conquest Punic style.

II. Early on in the first Punic war, the Romans came to realize the strategic importance of Sardinia: Zonaras (8.10) records that in the winter of 263/2, the Carthaginians conveyed most of their army to Sardinia with the idea of attacking (presumably Italy) from that island; and a couple of years later L. Scipio (cos. 259) invaded the northeastern corner of the island, around Olbia (MRR 1.206), which is only a bit more than 100 nautical miles from Ostia. Our sources also speak of naval battles, and it is easy to imagine that the Carthaginians’ strategy for conducting the war included harassing Roman shipping if not actually invading Latium itself. In 258, Sulpicius was active in the southwestern part of Sardinia and is said to have “overrun” the country before being defeated by Hanno (Zonar. 8.12). Two years later, Sardinia was one of the prizes demanded by Atilius Regulus (Dio, frg. 43.22-23, not accepted by all scholars) when he arrogantly and prematurely presented the Carthaginians with a series of conditions by which they could end the war, acknowledging Rome’s supremacy; Carthage disagreed with Regulus’ view of affairs and, with the assistance of Xanthippus, regained a level of equality. At this point, Sardinia disappears from accounts of the war, nor was it included in the truce terms of 241. In addition to its strategic value, Sardinia offered mineral wealth (especially in the forms of argentiferous lead, copper and iron), salt beds, and considerable agricultural potential, some of which at least had already been actualized by the Carthaginians. Whatever the reasons for the Romans’ not laying claim to Sardinia in 241, they were able quickly to add it to their still nascent overseas empire by seizing it while the attention of Carthage was distracted by the mercenary revolt, simultaneously, as Polybius says (3.9), greatly aggrieving the Carthaginians and laying the foundation for the long-protracted conquest of the entire island. The meager evidence we possess for the history of the wars of conquest in the remainder of the third and through the second century confirms that the Carthaginians had long ago been accurate in their judgement that conquest of the entire island was not worth the price.

There is nothing to be gained in this paper from cataloguing a list of names of proconsuls, propraetors and triumphatores. But some points of particular concern need recalling. The consul of 238, Ti. Sempronius Gracchus, with a fleet and troops took possession, probably in 237, of Sardinia “without combat,” that is, he most likely established control over the Punic cities of the littoral. As would often happen on the frontiers of empire, a change of overlords provoked rebellion, not without the encouragement of Carthage (Zonar. 8.18). The consul of 231, M. Pomponius (MRR 1.225f.), we are told, was unable to find many of the inhabitants who, as he learned, had slipped into “caves of the forest, difficult to locate, so he summoned keen-scented dogs from Italy, and with their aid discovered the trail of both men and cattle and cut off many such parties.” This is the first clear indication we have that the wars had by now moved into the mountains -- assuming that the forests were in the mountains (which is the same thing as saying that this was the first penetration by the Romans into the interior: we simply cannot tell). We likewise cannot know which mountains or forests, nor can we be certain that the caves were actual caves or if they were in fact nuraghi: at least from the time of the Prometheus Bound, savage barbarians can be found dwelling in caves whatever their actual habitations. During this same early phase, in the north, C. Papirius Maso (Zonar. 8.18) drove the Corsi from the plains, perhaps around Tempio or at least in the hinterland of Olbia, if these really are Sardinian Corsi, not Corsicans -- scholarly opinion is divided. Wherever, and Sardinia does make much more sense for these operations than Corsica does (Dyson 1985: 249),³ the Romans were repeatedly ambushed when they attempted to pursue the Corsi into the mountains, and a limited success was achieved only with great difficulty: Papirius was in the event denied a triumph by the senate. We know that the revolt of Hampsicoras in 215 early in the second Punic war focused around Cornus on the central west coast and the Campidano (Mastino 1979: 33-36). It is also clear that the natives of the interior (whom Livy [23.40.3] calls Sardi pelliti, “skin-wearing Sardinians”), with whom the Carthaginians had had, as we have seen, a centuries-long symbiotic relationship, joined the old invaders in resisting the new ones.

Alas, we have no early Roman field camps, forts or other material evidence to allow precise interpretation of the earliest phases of the conquest -- how one envies British archaeologists who can argue endlessly about Agricola’s route to Scotland and such! One type of evidence might cast some light or at least some insight into these earliest operations: i.e., the find-spots of the earliest, pre-denarii coins found in a certain profusion around the island (Rowland 1985a: 4.99-117). For the next phase, Michael Crawford’s precise redating of denarii allows us to hypothesize one direction of Roman penetration into the interior (Crawford 1969). The discoveries of Crawford number 38 (217-215 ) at Bolotana, in the mountains northeast of Cornus, a hoard of denarii closing with nearly mint-condition coins of 195 at Burgos, a few km. from Bolotana, and an example of Crawford number 133 (194-190) at Paulilatino, almost due east of Cornus, suggest that this region was one of the foci of Roman penetration into the interior after the revolt of Hampsicoras had been quashed. Excavations by teams from Pennsylvania Statue University under the direction first of Joseph Michels then of Gary S. Webster at several nuraghi in the territory of Borore might provide some confirmation, for these nuraghi were unused from the end of the Punic period to about 80 BC (Michels and Webster 1987).

In the early 170′s Livy (41.6.5-7) tell us, “the Ilienses with the support of the Balari had invaded the pacified part of the province, nor could resistance be mounted with the army weakened and to a great extent carried off by disease.” From this derives a long series of wars between the Romans and those tribes; indeed, both Diodorus (5.15.6) and Livy (40.34.13) affirm that they were still unpacified in the later first century, Livy by using the present tense -- which, from the context, is clearly not the historical present. We have no real information about the locations of those tribes: it is not very helpful to be told that they lived in the mountains, as Pausanias (10.17.9) and Strabo (5.2.7) tell us -- but a recently discovered boundary marker from the territory of Berchidda records Balari scarcely a dozen km. into the mountains from the later Roman fort at Castro-Oschiri (which we will consider shortly). The inscription simultaneously locates the tribe, suggests an approximate date for the establishment of the fort or a predecessor, and provides a reason for its location, namely, the protection of the road to Ozieri, Chilivani, Mores, Ploaghe, and beyond -- all prime agricultural territory. A coin hoard published by Taramelli in 1918 also came from the territory of Berchidda, from the region Sa Contrizzola, 2 km. west of the modern village; when re-examined in the light of Crawford’s dates for the denarii, this hoard is extremely informative (Rowland 1990-91: 301-310; cf. Gasperini 1992: 579-589): the coins were found in a nuragic pot and consist of an almost unbroken series of denarii from 178 to 82 BC. The most economical hypothesis is that this zone was the home of the Balari, rather than that it was a reservation into which they were herded by the Romans; the Ilienses, then, can be located to the south, in the Marghine and Goceano mountains, i.e., to the north and east of the focus of penetration spreading inland from around Cornus.

Ti. Sempronius Gracchus, father of the Gracchi, fought the Ilienses in 177-6 (triumphing on 23 February 175, though this did not prevent his return to Sardinia in 163-2) as the first consular commander since the early days of the second Punic war (MRR 1.398). In addition to bringing back his army “safe, sound, and full of booty” and dedicating a famous depiction of the island in the temple of Mater Matuta, he reported that more than 80,000 had been killed or captured. In separate passages, Livy (41.12.5; 17.2) records 12,000 and 15,000 killed, which could mean, if there is any validity at all to the figures, that perhaps more than 53,000 were sold into captivity. It is usually believed, or at least appears to be believed by historians, that these captives were shipped to Rome or Italy to be sold; it seems to be much more likely that some, if not the vast majority of them, were sold in Sardinia itself;⁴ other victorious commanders, although perhaps less spectacularly, will have also acquired captives who were transformed into slaves. That is, Rome’s victories substantially, but not entirely, depopulated the island’s mountain fastnesses and provided the workforce which radically altered the Sardinian landscape, transforming the island into one of the three subsidia frumentaria of the Roman state (Rowland 1988: 243-247). The figure of 80,000 is astonishing; besides representing what Dyson (1985: 256-257) calls “an impressive military feat,” it represents about 1/4 of the 1728 population of the entire island and about 1/8th of the 1881 population.⁵ There continued to be wars in the second century -- C. Gracchus returned to Rome from campaigning in Sardinia to stand for the tribunate of 123, and M. Caecilius Metellus triumphed in 111, the last triumph celebrated over Sardinia. As Dyson rather wryly remarks (1985: 259), “the Romans seem to have accepted that there would always be some turbulence in Sardinia and that minor actions of frontier pacification hardly deserved triumphal honors.”

One form of such pacification is described by Strabo (5.2.7), who could have been writing about any and all periods between the arrival of the Romans and his own day: “the mountain tribes live in caves and if they possess any sowable land, they do not sow it conscientiously, but rather plunder the lands of those who do cultivate, whether in their own neighborhood, or by sailing against those in the harbor…. The generals who are sent are sufficient for some of them, but inadequate against others, and since it does not pay to maintain an army continually in plague-ridden regions, the only recourse is to plan certain stratagems. And so, keeping close watch over some one of the barbarian tribes, who hold festivals for several days after a raid, they attack them at that time and capture many.”⁶ Diodorus Siculus (5.15.6) and Varro (RR 1.16.2) both mention Sardinian banditry (latrocinium), which could just as well mean guerilla warfare or resistance to romanization, and Varro emphasizes the deleterious effects on agriculture: “it is not worthwhile to cultivate many fertile fields on account of the banditry caused by people dwelling nearby as is the case in Sardinia with certain ones near Oelies” (perhaps modern Dolia or, following Cichorius’ emendation, Usellus).⁷ By AD 6, the situation had become so bad that Augustus was forced to take over the governance of Sardinia (Meloni 1990: 139-141): not only were the natives harassing the plains, but they were also practicing piracy in the Tyrrhenian Sea, even posing a threat to Pisa.⁸ In AD 19 Tiberius sent 4000 persons suspected of practicing Jewish or Egyptian rites to put down brigandage -- a famous (or infamous) passage, coercendis illic latrociniis et, si ob gravitatem caeli interissent, vile damnum (Tac. Ann. 2.85.5). Dyson is, I believe, absolutely correct in seeing these exiles not as conscripted soldiers, but as forced settlers in exposed farms and villages of the frontier zone (Dyson 1985: 261). A very, very slight archaeological confirmation of this hypothesis might be found first in the discovery more than a century ago of a gold ring with a Hebrew inscription, in the territory of Macomer, and more recently, finds at Sa Idda Eccia-Isili: a lamp with a menorah, and a tombstone reading [---]o Iudaeus / vix(it) an(n)is VIII (RRS 59; Corda 1992-93: 479-486).⁹

III. Now it is precisely in the age of Augustus that our search for the Roman frontier in Sardinia can move to a new level. To the extent that denarii and aureii arrived in Sardinia as pay for troops, the large numbers of those coins minted during the reign of Augustus and found in Sardinia indicate an increase in military pay, hence of soldiers, and therefore of military activity during his reign (Rowland 1978: 87-112). There are more gold and silver coins from the reign of Augustus than from any other reign through that of Heraclius: the Augustan peace and concomitant prosperity no doubt account for some of these coins, as does the length of his reign. On the other hand, there are more Augustan gold and silver coins than there are from the period from Trajan to Severus Alexander -- a century and a third! There is a sharp increase in coins minted under Vespasian, possibly indicating a rise in military activity during his reign, perhaps to be connected with the Civil Wars,¹⁰ and there is a quantum leap in the later third century, through the reign of Diocletian. And it is precisely in 227 that we can clearly see a change in the career patterns of the governors of the island as viri militares replaced the essentially civilian governors of prior years. Apparently, in the latter part of the reign of Severus Alexander, Sardinia was as disturbed by unrest and uprisings as was much of the rest of the Roman world, unrest which, on the island, was obviously not directly caused by barbarian pressures or incursions. The number and pattern of the coin hoards deposited between 235 and 284, and not recovered, clearly demonstrate that this unrest continued throughout the period of anarchy, and we can postulate serious efforts to restore tranquility under Gallienus, Claudius II, and Probus. There are four fourth century coin hoards, all in the interior: at Perdasdefogu, Orgosolo, Siddi, and Nurallao, which suggest continued or renewed unrest in these areas; the one at Orgosolo, which closed with coins of Julian, is particularly interesting, for the site appears to have been a religious center which could have served as a focus for rebellion: it was destroyed at about the time the hoard closed.

For the imperial period, there is a useful body of epigraphical material, which can carry us a long way forward in our search for the Roman frontier (Rowland 1978a: 166-172). The earliest such inscription is the epitaph found at Praeneste of a prefect of the first cohort of Corsi and of the civitates of Barbaria in Sardinia (CIL 14.2954 = ILS 2684; Zucca 1987: 349-373; Mastino 1992: 23-44; Mastino 1993: 457-536); the date is early imperial, not far removed in time from that of the dedication made at Fordongianus by civitates Barbariae (Sotgiu 1961a: 188), most likely during the reign of Tiberius, perhaps precisely in AD 19 (Corda 1999: 149). Fordongianus is strategically located on the Tirso river: it was the seat of the military commander of Sardinia during the Byzantine period, when it was renamed Chrysopolis (Boscolo 1978: 34). Numerous remains were visible in the sixteenth century, and bits and pieces of the Roman city, including part of the aqueduct and of an amphitheater have been reported in modern times. The major monument remaining is the baths, but there is nothing to indicate that Fordongianus played any military role in the high empire, i.e., after about the Julio-Claudian period: its name from the second century, Forum Traiani, suggests that it did not (RRS 46-47; Dyson and Rowland 1992: 207-209; Dyson and Rowland 1998: 313-328). A secure date for the recently discovered altar and temple (26 x 20 m.) dedicated to Juppiter on Monti Onnariu-Bidoni, some 20 km. upstream from Fordongianus, would be a useful indicator of the chronology of the process of pacification (Zucca 1998: 1205-1211).¹¹ Inscriptions recording Sardinian cohorts in and around Cagliari tell us nothing about military affairs, and the ones from Grugua¹² suggest that soldiers were used as guards in the mines here as elsewhere in the Roman world.

But now our quest for the Roman frontier can begin in earnest. In the territory of Oschiri, overlooking the modern Lago di Coghinas and the valley of the Riu Mannu di Ozieri and the broad plains to the west, stand the remains of the Roman fort still known as Castro (RRS 36, 92-93; Le Bohec 1990: 66-67; AE 1994, 795 [from Ruggeri 1994: 193-196], Manconi 1995: 32-33) -- the nearby medieval church is dedicated to N. Signora di Castro. The magnificent panoramic view of the underlying terrain and the protection to the plain afforded by this strategically located fort present extremely well the Romans’ conception of the functionality of a military site. And what a magnificent site it must have been in the last century and in the early decades of the present one! Remains of buildings, aqueducts, lead pipes, numerous incised stones, tombs (many containing glass objects), carnelians and jaspers, and mosaics are all found in the literature, but the actual site has only a few miserable traces of the fort and its defenders. Three inscriptions record that Castro was the base of cohorts of Sardinians and of Aquitanians early in the first century AD; a fourth found at Tula some 7 km. away records a signifer of a cohort of Ligurians which was probably also based at Castro. Some of the pottery found at the site, particularly African Red Slip ware, demonstrates that it was occupied through the imperial period; but whether or not it was always a military site or only sporadically such cannot at present be divined.

We know from Itineraries that Tempio, ancient Gemellae, was the site of a mansio and undoubtedly had a garrison; it provides an inscription of a soldier in an unnamed unit, and may have had a long history as a military site. At any rate, we know of pre-denarii coins found in the territory, which may derive from campaigns against the Corsi (RRS 133-134; Le Bohec 1990: 68-69). Through the empire, however, I would think of the garrison here more as a police caserna than a full-fledged fort, at least in the absence of any hard archaeological evidence to the contrary. Such establishments would serve not only to maintain peace and order but also to socialize the indigenes of the vicinity. Cassius Dio (56.18.2) beautifully summarizes the process: “The barbarians were adapting themselves to the Roman world. They were setting up markets and peaceful meetings, although they had not forgotten their ancestral habits, their tribal customs, their independent life, and the freedom that came from weapons. However, as long as they learned these different habits gradually and under some sort of supervision, they did not find it difficult to change their life, and they were becoming different without realizing it.” Just west of Gallura, local tradition and speculation would locate a Roman camp or camps on the plateaus of M. Franco and M. Seine a short distance from the town of Martis, either one of which would be an ideal location, mirroring Castro-Oschiri 26 km. away.¹³

Further in the interior, an inscription of another Aquitanian auxiliary soldier has been found in the territory between Bitti, Buddusò, and Osidda, in a place called Campu Sa Pattada (RRS 21; Rowland 1978a: 169; Le Bohec 1990: 73, noting the absence of a wall). It is located along a stretch of the Roman road and has an excellent view of the terrain to the north. None of the fairly extensive remains extant in the last century were visible in the summer of 1985, and there is no telling how long after the first century this remained as a garrison. It, too, should be counted more as a police than an army base -- a rather fine distinction when dealing with a Roman imperial province, but I use the distinction to imply normal, everyday control of a largely acquiescent populace, as opposed to an impressive display of military might to overwhelm still or potentially rebellious indigenes.

Well into the interior, almost precisely in the center of the island, from the terrain of the modern village of Austis -- Augustis in medieval documents --, came another military inscription, of a horn player serving in a cohort of Lusitanians; three other tombstones from the same site (named Perda Litterada) record children, one of whom was named Castricius, surely derived from castrum (CIL 10. 7885). A nineteenth century scholar tells us that the area was covered with the remains of ancient structures and materials; very little of what is available in the literature can be dated (RRS 15-16; Rowland 1978a: 169; Rowland 1994-95: 355-357; Le Bohec 1990: 72 [skeptical]). The inscriptions are all first century, and the only coins attested are a Republican denarius, two Augustan bronzes, and a bronze of Drusus. Austis may have had a long period of occupancy, and it may have been a military base for centuries; however, all that we can affirm with confidence is that it served as a military station during the reign of Augustus and for a period afterwards -- and it may have served a policing function along the new road across the interior of the island.¹⁴

Between Austis and Fordongianus was found the tombstone of a certain M. Valerius Germanus, a soldier for either twenty or twenty-six stipendia; it was one of a group of five stones all of persons named Valerius or Valeria (RRS 24; Rowland 1978a: 170). The findspot of these inscriptions, now under water, was near the bridge over the Tirso by which the interior road running past Austis was connected with the main north-south highway between Cagliari and Porto Torres. This may have been the location of a settlement of veterans who were given land in exchange for sharing in the protection of the lower Tirso valley; there are no other remains recorded from the site, so we have no possibility of estimating its extent or date. The inscriptions are probably from the first century AD and are very unlikely to be later than the early part of the third century.

Dyson (1985: 261-262; cf. Panciera 1992: 325-340) points to a military diploma found at Anela as “evidence of continued need for settlers to defend frontier regions even in the first century AD; the soldiers were legionaries retired from the First Adiutrix, and the inscription dates from the reign of Galba.” The site, called Charchinargiu, in 1873 showed extensive traces of Roman habitations (RRS 13), and may well have been such as Dyson interprets it. On the other hand, First Adiutrix was formed of sailors, and a large number of known sailors came from Sardinia (Sotgiu 1961: 78-97; Le Bohec 1990: 46, 86, 120; Mastino 1993: 489-491; Nonnis 2001). This diploma and others like it found in the interior of Sardinia may well be evidence of retired veterans returning home after their terms of service. Another diploma was found inside a nuraghe just to the east of Sorabile-Fonni (below), where an extensive Roman site was excavated in 1879 and 1881.¹⁵ There was a number of buildings, one of which was 51 meters long; a bath (about 42 x 60 m.); mosaic floors; gold, silver, and bronze objects; and, discovered fifty years later, an inscription, a dedication to Silvanus by C. Ulpius Severus, Procurator Augusti and Praefectus Provinciae Sardiniae (Sotgiu 1961a: 221). Coins of the first century undoubtedly provide the terminus post quem for the fort -- for this was clearly more than a police station -- and republican coins, one from 146 BC, elsewhere in the zone might indicate earlier military activity in this region well into the interior. Alas, there is nothing in the original excavation report, subsequent discovery, or autopsy in the summer of 1985 to allow even a wild guess about the fate of Sorabile in the late Empire. Both Meloni¹⁶ and Sotgiu¹⁷ date the governorship of this Ulpius to the early second century; however, a certain M. Ulpius Victor was Procurator Augusti and Praefectus Provinciae Sardiniae under Philip around 244,¹⁸ which strikes me as a much more satisfactory timeframe for our dedicator. He presumably visited the fort when he made the dedication so it was surely still functioning sometime in the mid-third century: the diploma found at the nuraghe Dronnoro about 1 km. east of Sorabile dates to 214-217 （Notizie degli Scavi 1882, 440-441; CIL 10.8325). Future epigraphical discoveries might allow us to fit into this picture the Cohors I Nurritanorum, which surely originated in Sardinia though it is recorded only on inscriptions from North Africa (one a diploma of AD 107) and from Umbria (Benseddik 1982: 59; Mastino 1984: 46, with n. 103): a boundary marker from Orotelli records a people abbreviated NURR, which has been suppleted as Nurr(enses?) but should now with greater likelihood be read as Nurr(itani) (Meloni 1990: 315).

Which brings us finally to Nuragus, where Germanus Nepotis filius, a soldier who had served for seven years, was buried by his mother Fausta (CIL 10.8323). The cemetery which included this burial yielded coins ranging from the reign of Claudius to that of Hadrian (RRS 74-76), and the presence of his mother scarcely suggests major military activity. We can envision the same sort of garrison protection here, with serving soldiers, that Dyson postulated for Anela and that we saw for Gemellae and Campu Sa Pattada. Pliny (NH 3.7.85) and Ptolemy (3.3.6) call the inhabitants of this place Valentini, leading editors of Ptolemy to correct Valeria to Valentia in his list of interior cities (Forbiger 1848: 3.826); the zone will have been around the church of S. Maria ’e Alenza (i.e., S. Maria di Valenzia), which still preserves the name. If the people are Valentini in Pliny, settlement must have been at least an Augustan foundation: Valence in Narbonenesis was established by Caesar or by Augustus. One would rather like to think of Valencia in Spain, founded by Junius Brutus in 138 BC.¹⁹ Excavations in 1872 and 1874 provided only imperial material, with bronze coins of Augustus, Vespasian, Gordian and Philip (RRS 74-76, 146-147); but, on a visit to the site in the summer of 1986, in an area which had been excavated to lay irrigation pipes, Stephen Dyson and I saw some pottery of the Sullan period which supports an earlier rather than a later foundation date. About 1 km. or so away, a nuragic structure included third century coins, beginning with one of Alexander Severus and including six of Tetricus. And just down the hill to the south, a small nuragic well was repaved in the Roman period, with the new pavement covering a coin of Claudius Gothicus (RRS 74-76; Le Bohec 1990: 69-70). In all, a very slender thread, but it does look very much like there was a certain amount of activity here during the third century. The territory around Valenzia -- including the nuragic sites of Forraxi Nioi and S. Millanu and the church of S. Elia (with an adjacent Roman cemetery of the imperial period with numerous burials in crouched position [Notizie degli Scavi 1882, 306-308]) will repay detailed examination, not only because of its potential for understanding the military history of the region.

There are other sites, too, where an optimist might find traces of third century activity; but, like so much else we must remain for the present rather vague. Surgical implements found at Perda Floris-Lanusei and around Oliena (RRS 57, 88) are suggestive of a military presence, but lack both context and date.²⁰ Speaking of a grave at S. Andrea Frius, Dyson (1985: 262) says that the “tomb can be dated to the late second century… a man was buried with his weapons. It seems that even during the height of the Pax Romana frontiersmen in Sardinia lived, died, and were buried with their weapons at their sides.” The grave goods (actually, of two men) are dated by coins, one of Trajan, the other of Antoninus Pius -- both pierced with a large hole therefore, perhaps, jewelry of a later age, viz., the third century, or even later.²¹ The material consisted of a 26 cm. long iron knife, two iron lance heads, a small iron axe head, a bronze bell with an iron clapper, a large fibula with geometric designs in relief, what must be an armlet (“un grosso anellone”) of bronze, and some other objects (including a bit of silver). This suggests a late imperial or even Byzantine site such as we know from other finds S. Andrea Frius to have been (Salvi 1989: 465-474; Rowland 1985b; Rowland 1992: 154-158).

The nuragic sacred area of S. Vittoria di Serri appears to have been used as a garrison post in the late empire: Roman structures and a street are perhaps to be associated with inhumations dated by coins of Gordian III, Aurelian, and Constantius (RRS 124); this site too seems to have been the site of a Byzantine garrison.²² Not far away, at Is Murdegus-Nurri (RRS 77-78), a deposit of votive figurines similar to one from Linna Pertunta-S. Andrea Frius can also be dated to the reign of Gordian. Coins in the S. Andrea Frius deposit continue to the reign of Constantine, and a seventh century fibula and pitcher, both of bronze, also came from Linna Pertunta (RRS 106); possibly (as we have by now come to expect) the place was a Byzantine military outpost (Spanu 1998: 169). Both votive deposits could perhaps be due to military presences.

Not far from Nurri stands another site that had some Byzantine troops, Biora,²³ where an inscription was set up in honor of the Numen Deus Hercules by a group called the Martenses (CIL 10.7858; cf. Meloni 1990: 309, 332, 400), who must be connected with the military. It is a strategically located site, 26 hectares in extent, along an important roadway, and might have had a long military occupation, or it might have had sporadic military occupation over a long time -- we simply cannot tell in the present state of the evidence. A garrison at Biora would make sense either before the time of the one at Valenzia or coeval with it, either in the second century BC or around the time of Augustus. Both sites would provide ample protection to the Trexenta, which the Romans made rich in grain. About 20 km. to the interior from both Biora and Valenzia, at Seulo, was found a diploma of a Sardinian ex-sailor (CIL 16.127). Was he there on garrison duty or in retirement? About another 20 km. into the interior, in the territory of Tonara, there is a region described by Taramelli (1911: 384-386) as being “sparso di abitati romani.” That archaeologist called one of those structures a villa “di qualche signore di età romana” -- a military station, which would yield materials similar to those of a villa, seems much more likely when one considers the locale and terrain (935 m. ASL). Part of still another military diploma was found within about 10 km. from Tonara, at Sorgono (RRS 130). An amulet of Zeus Serapis found at Sorgono along with a statuette of Isis found about 25 km. away at Asuni (Rowland 1976: 170) are most likely evidence for a Roman military or veteran presence in these locales so far from the coast. Austis is about 10 km. from both Sorgono and Tonara. About 20 km. or a bit less from Valenzia, Usellus, and Fordongianus is the somewhat enigmatic fortress known as Castello di Medusa-Samugheo, clearly a military site in the first and sixth centuries AD, but we haven’t a clue about its history in the interval.²⁴ The distance from Tonara to Sorabile-Fonni is about another 20 km. along the road; from there it is a bit more than 20 km. to Nuoro, then another 20-odd km. to Bitti, thence nearly 40 km. to Castro, but only about 20 km. to the environs of Ozieri (rich in Roman fine wares [RRS 94-96]) and from Ozieri about another 20 km. to the fort. Here at last is an outline of a frontier system: a few garrisons as anchors with police stations at more or less regular intervals about every 20 km. -- some of them purely hypothetical deduced from perhaps inadequate evidence.

Alas, there is far too much that we cannot tell at present -- and won’t be able to without more extensive exploration and excavation combined with detailed, accurate and timely reports. It is a great pity that, more than a century after the military base at Fonni was excavated, at a time when we still await reports of excavations at Castro-Oschiri, it is still premature to suggest more than the outlines of a coherent defensive system in the Sardinian interior, and to do even that on what amounts to not much more than educated guesswork. We can feel fairly confident about our deductions for the Augustan and early imperial ages, and we can be reasonably certain that there was an increase in military activity in the third century and later, with the establishment of small military posts, some of which can in fact be identified. During much of the Pax Romana, Sardinia seems to have been relatively peaceful -- and neither archaeological nor epigraphical evidence exists as yet to suggest the contrary.

There is one well-known piece of evidence which might serve to confirm the picture just sketched (Rowland 1985a: 110). A bronze tablet (CIL 10.7852; ILS 5947), called the Esterzili tablet from its findspot, recording events of the last portion of Nero’s reign, presents to us a people called the Gallilenses, “frequently resisting and not obedient to the decree,” who are warned, not for the first time, to be quiet and leave the estates of the Patulcenses and to hand over free (vacua) possession to the later; they are further warned not to persevere any longer in their stubbornness lest the authors of the sedition be dealt with severely. Subsequently, the Gallilenses, scarcely acting like the crude, savage bandits portrayed in modern discussions of the text (Mastino 1993a),²⁵ appealed the government’s order, saying that they would deliver a tablet pertaining to the matter from the emperor’s record office (tabula ad eam rem pertinens ex tabulario principis), and they later asked for further delay of the execution of the order to depart, asking forgiveness that the table had not yet been delivered. No further delays were granted. If they had not departed by a fixed dated from the boundaries of the Patulcenses (now called Campani), which they had seized by violence, they would pay the penalty for their long stubbornness (longa contumacia). These events are generally interpreted as an incursion into the agricultural zone settled by Campanian colonists; this zone was possibly around Dolianova (perhaps the [unemended] region Oelies mentioned by Varro as being harassed by bandits) and the perpetrators would have been unruly barbaricini from Gerrei, slightly to the northeast. Clearly, however, they are squatters, not invaders. The area where the inscription was found is in the fertile high plain of the Flumendosa, and it was discovered by a farmer while plowing (Spano 1867: 1-52; Angius 6.407-408). Conflicts between shepherds and farmers are admittedly a well-attested recurring phenomenon in Sardinian history (Le Lannou 1979: 125-136; Day 1987: 269-290):²⁶ reducing or eliminating such conflicts was presumably one of the primary functions of the Roman military/police in the Sardinian interior. But such troublesome shepherds pass on fairly quickly and do not, like the Gallilenses, settle down for a long time in someone’s fields and, presumably, farm them. And what sort of vagabond brigands claim that they can produce a document from the imperial archives, even if in the end they cannot carry through? Or have Lares (CIL 10.8063.1)? And, finally, if there was a numerous Roman military presence in the 60′s AD in the region — hardly a day’s march from Biora or Valenzia — why didn’t the army simply march out, evict the squatters, and deal with them appropriately?

* An earlier version of this paper was read as the Henry Rowell Memorial Lecture at the Johns Hopkins University in November 1985; I am delighted to have the opportunity now to present a revised and updated version in homage to my old friend, former colleague and office-mate, Gene Lane.

Key to Frontier Map

• Anela
• Asuni
• Austis
• Berchidda
• Bidoni
• Bitti
• Bolotana
• Borore
• Buddusò
• Burgos
• Cagliari
• Chilivani
• Cornus
• Dolia
• Esterzili
• Fonni
• Fordongianus
• Grugua
• Isili
• Lanusei
• Lasplassas
• Macomer
• Martis
• Monte Sirai
• Mores
• Nuoro
• Nuragus
• Nurallao
• Nurri
• Olbia
• Oliena
• Orgosolo
• Orotelli
• Oschiri
• Osidda
• Ozieri
• Ploaghe
• Paulilatino
• Perdasdefogu
• Porto Torres
• S. Andrea Frius
• Samugheo
• Serri
• Seulo
• Siddi
• Sorgono
• Tempio
• Tonara
• Tula
• Uselis