1A little over forty years ago, the eminent historian H. I. Marrou, in the company of the distinguished geographer M. Le Lannou, standing on an elevated spot in Sardinia, and gazing at the broad plain below, remarked « vue de loin, vue des livres, cette acquisition (i. e., of Sardinia) peut paraître sans grande importance: mais quand nous avons vu, dans les immenses campagnes du Campidano, le blé vert onduler à l’infini, floridum mare, sous le vent léger, nous avons réalisé brusquement tout ce que la conquête de la Sardaigne avait pu représenter pour la République romaine » (Marrou 1951, 143; cf. Le Lannou 1979, 353). Of course, it is highly unlikely that anyone standing there in or about 238 BC would have been presented with the same panorama Marrou and Le Lannou saw or one can see today. It is probable, but not at the moment provable, that something like that floridum mare would have been visible to a visitor in AD 100 or 200, a subject to I which I shall return below. Nor, it must be emphasized, does Sardinian grain seem to have been a major cause of the First Punic War or even of Rome’s seizure of the island in 238/7 BC: security and control of mineral resources were probably more important considerations (Rowland 1985a and 1992b, 167)*.

2This is not the place to trace the evolution of Sardinian society or agricultural history, but a few points need to be introduced by way of background to the Roman period. Although the Upper Palaeolithic hominids attested at Corbeddu Cave-Oliena may have survived and although there may have been human communities on the island in the mesolithic, neither has left any substantial traces in the archaeological record as it is currently known (Hofmeijer 1987; Sondaar 1988; Webster 1990). The first neolithic settlers of Sardinia, characterized by their Cardial Impressed pottery, arrived probably from Τuscany by way of Corsica early in the sixth millenium BC, and there is early neolithic evidence of Triticum monococcum and Triticum dicoccum along with grinding implements (Lilliu 1988, 42). Barley, Hordeum Hexasticum, enters the archaeological record in the middle neolithic (Bonu Ighinu phase) (Lilliu 1988, 60); although as yet unattested in earlier levels, it was probably cultivated then as well. In the sequel, barley and grain, and of course other products (legumes, for example), continued to be cultivated, but in what proportions cannot be known. There were surpluses, at least from time to time and in some places, undoubtedly accruing to the benefit of an incipient class of chiefs or headmen, perhaps those who controlled access to plow animals and whose final resting places were the more lavishly decorated rock-cut tombs (domus de janas).

3By early in the third millenium a fundamental change occured. Previously, settlements such as the late neolithic village at S. Gemiliano-Sestu (Atzeni 1961) had been without defensive installations. During the Chalcolithic, a new type of dwelling and a more dispersed pattern of settlement came about: the dwellings, variously called pseudo-nuraghi, gallery nuraghi, proto-nuraghi or corridor nuraghi, were essentially platforms on which huts were constructed (Lewthwaite 1986; for the cronology, Castaldi 1984, 142-148; Thomas in press). By the late third millenium these began to be replaced by the classic corbelled or tholos nuraghi: at first, these latter were simple towers, then more complex structures of two and three storeys and, finally, true nuragic complexes, what Lilliu (1959) once called proto-castles, with bastion walls and subsidiary towers. Typically, proto-nuraghi and many classic nuraghi are located on ridges above river and stream valleys or at least near a perennial spring. The distribution of classic nuraghi is much more widespread: they are rather thick in some landscapes and scarce or non-existent in others (Contu 1974, 145-163, especially 153).

4What is particularly striking about any distribution map of nuragic densities (Lilliu 1961, 62; Atzeni 1980) is that they are scarce or non-existent in precisely the region that was to become the principal grain growing region of Sardinia, the Campidano between Cagliari and Oristano: conventional wisdom holds that not many were constructed there because of the lack of suitable stone and that many of those which had been built were dismantled by subsequent generations of farmers, especially in the nineteenth century at the time of the enclosures (Contu 1974, 198). On the other hand, nuraghi may in fact have been scarce or non-existent in this zone in antiquity (cf. Lewthwaite 1986, 22). For town after town in the Campidano, Vittorio Angius, writing from the 1830’s to the 1850’s, speaks of the bad air, the bad water, the terrible heat, the swamps, the marshes: except for his figures on crop yields, his (and others’) descriptions of that zone have led me to the conclusion that, in the absence of a cash-crop or tax-crop economy, it is anachronistic to think of any widespread exploitation of these now fertile fields (Rowland 1992 and 1991).

5In contrast, in those zones where large numbers of nuraghi are extant, we find good water, often in abundance, numerous huntable animals, wood and forest products, and average crop yields that are often higher than those of towns in the Campidano, particularly for barley. Sedilo, for example, was an important nuragic center with 34 or 35 good springs, some streams and access to the Tirso River. In 1846, Sedilo had a population of 2 326 persons: 450 farmers produced 5 313 600 liters of grain and barley. Subtracting seed for the next year and 250 liters per inhabitant as subsistence, 4 289 300 liters remain, surplus for 17 157 persons-738 percent of Sedilo’s population. Thus, I have argued (Rowland 1992a and 1991) that nuragic agriculture was not focused on the crop and in the areas that Roman agriculture was to be, but was based on the cultivation of barley in the optimal zone 200 to 500 meters above sea level, from which it spread into the marginal uplands and lowlands; thus, the nuragic folk were not the “pastori-guerrieri” of conventional wisdom, but “contadini-guerrieri”. As in Iron Age Britain (Cunliffe 1974, 173), there seems to have been a relative increase in the numbers of sheep in the Campidano in the first millenum, probably linked to the spread of downland arable – which itself may be linked to the arrival of the Phoenicians in the ninth century, if not earlier (Rowland 1992).

6As early as the mid-seventh century, the Phoenician coastal settlements began to expand the territory under their control. The fort at Monte Sirai-Carbonia, soon followed by Pani Loriga-Santadi, was established as a military colony by Sulcis to secure control over the coastal plain. It would seem that the Phoenician levels at S. Sperate attest to a colony sent out from Cagliari: Phoenician material at Cuccuru Nuraxi-Settimo S. Pietro and M. Olladiri-Monastir (where Marrou and Le Lannou stood) might be evidence for immigrant settlement or for commerical and other contacts between native and indigenous cultures. The settlement at S. Sperate was a “proto-urban” agglomeration of houses with rectilinear mud-brick walls with stone foundations and numerous rooms. At Cuccuru Nuraxi, a sacred well was constructed over the ruins of a nuragic tower; associated material includes a large amount of Phoenician, protocorinthian, Corinthian and Etruscan pottery along with locally made wares. At Monastir, the assemblage revealed by the plow included bricks, imported pottery and amphoras, and carbon: the area could have been a sacred zone, an incineration cemetery (therefore not indigenous) or one or more structures which were burned. Other archaic imports in the region are most likely objects of trade. S. Sperate is adjacent to the Mannu River, and the settlement there can be seen as representing the first stage of an expansionary process along this key route to the interior; it was followed by the establishment of forts at S. Brai-Furtei in the sixth century and at Su Nuraxi-Barumini in the fifth, both of which formed an integral part of the Carthaginian frontier system (Rowland 1982).

7In the middle of the sixth century BC, Carthage directly intervened in Sardinia, sending an expeditionary force there; over the course of the next century, the Carthaginians vigorously pursued a policy of active imperialism which resulted, by around 450, in the establishment of an interior frontier system. By the late sixth century, the Carthaginians had been sufficiently successful that, in their first treaty with Rome (Polyb., III, 22, 8-9), they could assert control over all trade by the Romans in Libya and in Sardinia. After the establishment of the Carthaginian frontier, Punic influences on the native populations, both within their sphere of control and beyond, intensified. If the Romans and others could trade in Sardinia only in the presence of a Carthaginian official, then it seems likely that internal trade might also have been similarly restricted and that the Punic forts, in addition to the obvious military function, served as the nodes of exchange between Sardo-Carthaginians and natives and between diverse groups of indigenes. There may also have been non-military sites, such as S. Maria di Monserrato-Serramanna (Rowland 1988b, 854), which functioned as emporta. From the fifth century on, imported goods at native sites increased dramatically, but there is to my mind insufficient evidence to follow Barreca in postulating widespread colonization of the countryside by the Carthaginians (Rowland 1992). Although there surely was some colonization, most of the material evidence is best explained by the effects of punicization on those nuragic folk who lived within the widening zone of Carthaginian influence. For our purposes, the most important effect of this influence is the vast increase in the cultivation of grain both for trans-shipment to the coastal cities and, surely, for export.

8So, one result of Carthage’s domination over the Sardinian lowlands was the transformation of much previously uncultivated, even uncultivatable, soil and a vast increase in the amount of grain produced there, not nearly as much as was to be produced by the time of the Roman Empire, certainly — not that we have any precise, quantifiable data for either period – but inferences can be drawn. A lengthy survey of citations from ancient texts documenting Rome’s exploitation of Sardinian grain would be tedious and out of place. A few examples will suffice (cf. Meloni 1974, chapters 4, 6, 8). The revolt of Hampsicoras in 215 BC was caused by Rome’s ruling the Sardinians “harshly and greedily”, crushing them with an oppressive tribute und an unjust requisition of grain; after the revolt had been put down, the communities which had supported Hampsicoras and Carthage paid cash and grain in proportion to their resources orto their guilt. The cash was given to the quaestors in Rome, the grain to the aediles; in 204 an immense quantity of grain was sent from Sardinia to the army in Africa; and, in the two following years, 100 transports with clothing and grain were sent, while in the latter year (202 BC), grain from Sardinia and Sicily was so abundant in Rome that the price collapsed.

9The wars fought by Ti. Sempronius Gracchus in the mid-nineteenth170’s may have been important for Sardinian agriculture as well as for the military conquest of the interior: on the tablet which he caused to be erected in the temple of Mater Matuta, he claimed that more than 80 000 Sardinians had been killed or captured; subtracting the 27 000 earlier reported by Livy as having been killed, more than 53 000 were captured and sold into slavery, and I have suggested that most of these were destined to be agricultural slaves in Sardinia itself, simultaneously depopulating the defensable mountain fastnesses and providing the workforce which accelerated the transformation of Sardinia’s agriculture, particularly in zones where Punic influence had been slight or non-existent or where soggy clay and marl soils required considerable effort, for example Marmilla, Trexenta and Northern Logudoro (Rowland 1988a; Dyson 1985, 257). In the area where Steven Dyson and I conducted surface surveys in the summer of 1990, a particularly fertile zone whose towns regularly produced surpluses in the mid-nineteenth century, around the site later known as Colonia Iulia Augusta Uselis (which was evidently first settled around the middle of the second century BC), we found eleven nuragic sites with Dressel 1 or Dressel 1 type amphoras; eight of those sites also yielded black glazed pottery, as did others, attesting to their having been occupied in the late republican period (Rowland and Dyson 1991a).

10In that period, Sardinia is often linked whith Sicily, with Africa, and with both (Rowland 1984, 45). In a well-known passage, Cicero calls all three frumentaria subsidia rei publicae (Leg. Man., 34), and Varro (RR, II, 1, 3) lamented that «we hire people to bring us from Africa and Sardinia the grain to fill our bellies». For the empire, Strabo (V, 2, 7), Appian (BC. II. 40) and Pausanias (X, 17, 1) attest to Sardinian fertility, and the author of the mid-fourth century Expositio Toti us Mundi (210 Rougé) speaks of the island as frumentifera et ditissimafructibus et iumentis. In an earlier paper (Rowland 1984), in which I demonstrated that Sardinian grain continued to be important to the Roman grain supply throughout the high empire, I said that «a considerable amount of fieldwalking and excavation is still required» before we will have an adequate understanding of how extensive rural settlement in Sardinia was, both spatially and chronologically. In the interval, some of that work has been carried out (Tore 1987 and 1988; Zucca 1987, 115-147; Zucca 1991. 155-166; Sanluri 1982; Villaspeciosa 1984; Gesturi 1985). The results of these projects and the one Steven Dyson and I have been conducting (Rowland and Dyson 1988;1989;1990;1991a;1991b;1991c; 1992a; 1992b; Dyson 1990) are converging to provide a much clearer, but still inadequate, picture of rural Sardinia in the Roman period.

11The first phase of our project centered on the area around Roman Forum Traiani, modern Fordongianus, including the communes of Bauladu and Paulilatino. Particularly impressive was the frequency with which we found republican black glazed ware at sites in the territory of Bauladu with good access to the valley of the Mannu River and its tributaries. Most of these sites (9 of 11) also had early imperial arretine and/or thin ware; conversely, all of the sites which had early imperial pottery also had republican, suggesting a strong continuity here. Ten of the eleven republican sites plus an additional nine sites had African Red Slip ware of the high and later imperial periods, suggesting both continuity at the republican period sites and expansion to more marginal ones. More in the interior, only nine of ninety-one nuragic sites had republican pottery, and six of those nine had early imperial wares. Eight of those nine yielded African Red Slip, and so did twenty-eight additional sites, again suggesting strong continuity at the republican period sites and a virtual explosion of sites in the high and late imperial periods. In the Usellus region, where we had by February 1991 spent only one season, twenty-six sites yielded republican pottery; twelve of those also had early imperial, and twenty-two had mid to late imperial pottery, while another twenty-four had mid to late imperial pottery, without republican antecedants. Again, the evidence suggests continuity from the Republic through the Empire, although it is around Usellus less marked than around Fordongianus, and a similar great increase in the number of sites in the imperial period. In both zones, Roman period settlement was even more extensive than this brief summary suggests, for many sites had roof tiles, common, utilitarian and storage wares which cannot be dated with the same precision; for example, six sites in the Fordongianus area and sixteen sites in the Usellus area provided roof tiles but no dateable pottery.

12It is becoming increasingly apparent, therefore, to use Rickman’s phrase (1980, 106-107), that Sardinia was heavily studded with settlements during the Republic and extremely heavily studded with settlements during the high empire. Recently, I suggested that Sardinia could have sent a minimum of seven and one-half million modii of grain to Rome per year (Rowland 1990), that is, nearly half of what Garnsey (1983, 120) reckoned that Africa produced. I should here like to stress the word “minimum”: I think that he figure was actually higher, but did not want to overstate the case. In the later Empire, when Egyptian grain was diverted to Constantinople, Sardinian grain became even more important for Rome, and we have been accumulating an impressive amount of data for late Roman rural settlements: a preliminary analysis of only some of our African Red Slip pottery (249 sherds) by Philip Perkins shows that some 12% date to later tha ca. AD 400 and 34% to later than ca. AD 300 (Perkins 1991). If natural and man-induced destructive processes do not destroy more of the evidence beforetime, one hopes that further surveys will fill in some of the remaining gaps in our knowledge.

13We are very ill informed about how Sardinian grain was collected and transported to storage and trans-shipment points at the coast; that it arrived in Rome as part of the flotilla which originated in Africa seems to be a likely hypothesis (Rowland I 984; cf. De Salvo 1989).