Italy
A clear pattern exists in the data of very high pig percentages in western central Italy. The sites plotted in Fig. 1 are of late Republican and Imperial date and show that the reg ion comprising Etruria, Latium and Campania differs both from sites to the north (i.e. the Arno and Po valleys), which tend to have a lower percentage of pig bones and a higher representation of cattle, and also from sites to the south, which are largely in the old Magna Graecia territory and display a high sheep/goat pattern. This, as will be shown below, conforms with other areas that came under the influence of the western Greeks. The differences between these three regions of Italy can also be seen cle arly in the means and standard deviations for the data, given in Table 1.
It was not always the case that western central Italy had a dietary pattern of this sort. Greek, Etruscan and other pre-Roman sites in Italy, including many from Etruria and Campania, tend to have significantly fewer pig bones, having more sheep/goat and cattle instead (Fig. 2). Many display roughly equal percentages of the main food-providing domesticates, which in practice meant a meat diet that was more reliant on beef than other meats. The high pork pattern, therefore, was a feature of the diet that emerged in the late Republic. The chronological parameters are clearest in the Bay of Naples area (Fig. 3), where prior to the 1st century BC, a cattle-dominant pattern was in place, to be replaced in the Imperial period by the high pig pattern. It is likely, from available data, that Latium saw the emergence of a pig-dominant pattern earlier than Campania; in the 3rd/2nd centuries BC.
To help explain this, an interesting differentiation can be seen at Settefinestre villa, where intra-site analysis indicates that the higher status parts of the site had a higher pig percentage than the zones interpreted as being occupied by lower status inhabitants, such as the slaves (Fig. 4; King 1988a , 52-3). This appears to show that pork, particularly young pork and suckling pig, was considered to be a desirable and high-status dietary element. Villas such as Settefinestre were production sites, supplying the urban centres with pork and other meats, so that it is not surprising that a correspondence can be seen between the producer and consumer sites in terms of assemblage statistics. Rome, Ostia and the other towns tended to have pork-rich meat diets, presumably as a result of cultural preference, and were able to obtain supplies from the villas, which in turn were able to raise the animals in a fairly intensive manner (King 1985, 299). This is an idealized model, that presupposes a relatively perfect market mechanism for satisfying demand, but nevert heless, appears to be appropriate for interpreting this type of slowly-changing dietary pattern.
Sheep and goat bones are consistently poorly represented in assemblages from western central Italy at this period, and it is generally assumed that these anima ls were not so prominent in the agricultural economy as pigs. Interpretation of the classical agronomists (who, of course, were largely writing about conditions in this particular region) suggests that arable agriculture, including vine and olive cultivation, dominated farming activity. Pig herding could have been fitted into this, especially if the animals were kept in sties, as at Settefinestre, but little opportunity for extensive sheep farming would have been left, unless they were grazed away from the arable in summer while the crops were growing. This could have taken place within a system of localized movement of flocks, as well as the better known transhumance systems that are documented for the Apennines (Frayn 1984). Despite the likelihood that sheep and goats were more common in the hills of central Italy, it remains the case that the dietary pattern indicated in the bone assemblages favours pork over mutton.
Cattle bones are always third in rank order of percentages in all the early Imperial asse mblages in the region, reflecting an apparently low importance for beef in the meat diet. This may be due in part to religious factors (Prina Ricotti 1988; Dosi & Schnell 1984), but it is also the case that the usual agricultural regimes of the period wer e not such as could be easily compatible with large-scale cattle rearing. Establishment of low-level pasture would have competed directly with the arable land, effectively meaning that more cattle would have resulted in less cereals for humans. This may ha ve been a critical factor if there was population pressure on Rome and its hinterland from late Republican times. In addition, dry Mediterranean conditions during the summer produce few favourable areas for good pasture land.
The pork-rich diet seems from the data presented above to be the remarkably consistent dietary pattern that was normal and desirable in the region around Rome itself, in part due to the agricultural conditions of the period, but mainly due to cultural preference. As such, it is the ' Roman' dietary pattern in the late Republic and early/middle Empire. The obvious question therefore arises; was this pattern exported as a 'Romanization' of the diet in other regions of the empire?
Figs 1 and 2 provide data to start answering this question . Southern Italy, as mentioned above, may display a pattern that owes more to pre-Roman Greek and indigenous influences than to 'Romanization'. In overall terms, northern Italy also retained a distinctive dietary identity, possibly a more conservative pat t ern that conformed more to pre-Roman percentages. Commentary by Riedel (1988, 323) appears to confirm that north-western Italy had relatively stable percentages from prehistory through to the Middle Ages. The mountain areas tended to have more sheep and g oats, because these species were better adapted, whilst the plains tended to have more pigs, since they were more suited to the warmer winter climate near the Adriatic coast.
One site, Aquileia, which by most political and cultural yardsticks would be rega rded as a highly 'Romanized' town, in fact has a markedly different dietary pattern from that of western central Italy. It has approximately 80% cattle bones, and few of other species (Riedel 1994). Interestingly, this is like sites in the northern provinces (see below), and it may be the case that the town's pivotal position on communication routes to the north exposed it to dietary patterns brought in by traders and the army.
The chronological pattern in Fig. 3 shows that in the late Roman period, the hi gh pig percentage drops, in favour of sheep and, increasingly, goats. Several urban sites, such as Naples, Ostia and Rome, in fact continue the early Imperial pattern, at least for the duration of the 4th century and into the 5th century. However, the later groups from Rome and Naples show a shift away from high pig percentages, so that by the 6th century in Naples, sheep and goat bones were more common than those of pig. This trend is also seen in the rural sites, but there are hints of possibly crucial di fferences in the timing of the changes. Settefinestre, for instance, sees a gradual decline in pig numbers by the 4th century, that is earlier than comparable changes in urban assemblages. At Monte Gelato (King 1994), late Roman percentages are significant ly shifted towards sheep and goats at a time when Roman and Ostian preferences were still orientated towards pig meat. For villas in southern Italy, S. Giacomo (Albarella 1993) and S. Giovanni di Ruoti (Steele 1983; MacKinnon forthcoming), the 4th and 5th centuries see sheep and goat dominant assemblages.
What this signified in general economic and dietary terms is not easy to determine. However, it is known from literary and archaeological sources that Rome was supplied with basic foodstuffs from other parts of the Mediterranean, notably North Africa. Local supply may have declined after the 2nd century, possibly because production costs were cheaper in the provinces. This may have resulted in more autonomy on the rural sites after this, albeit at a lower level of productivity. The result appears to have been a drift to less pork and more sheep and goat meat on the rural sites, possibly coincident with a trend to raising sheep for wool, as suggested for Monte Gelato. For the 4th century and part of the 5th, therefore, urban centres were able to sustain the traditional dietary pattern, whilst rural sites had changed. Dislocation of the Mediterranean supply routes in the mid-5th century, however, may have upset this pattern, and it is probably after this time that urban and rural assemblages move into line again, on a pattern that had less pork and more mutton and goat meat, probably locally supplied.
This picture holds good with the exception of S. Giovanni di Ruoti in 5th/6th century. Here, a midden with very high pig percentages appears to contradict the model presented above. It is a special case, and has been interpreted by Whitehouse (1983) and Barnish (1987) as a site supplying the Roman market with pigs and pig meat. This is probably an example of specialist cash-cropping in order to satisfy the urban market demand, and also the annona distribution of pork that may have taken place in Rome at this time. However, this does seem exceptional: the general trend in Italy, and indeed in the western Mediterranean as a whole, was towards a late Antique diet with less pork and more mutton and goat meat (see below).