Roman ports and Mediterranean connectivity

Roman ports and Mediterranean connectivity Andrew Wilson, Katia Schörle & Candace Rice INTRODUCTION T his paper examines several interrelated aspects of maritime connectivity within the Roman Mediterranean. First, it considers Fulford’s suggestion of a divide between eastern and western trading zones along the north African coast in the light of new evidence discovered in the twenty years since he wrote; second, it considers the glass industry as a case-study illustrating the geographical integration of the Roman economy through maritime networks between ports; and third, it examines how port structures, capacities and facilities might relate to the patterns of maritime trade in antiquity. AN EAST–WEST DIVIDE IN NORTH AFRICAN MARITIME TRADE? In two seminal papers in the late 1980s, Fulford drew attention to the high degree of economic interdependence between cities of the Roman Mediterranean that was suggested by quantified and provenanced analysis of the ceramic assemblages at several port sites, chiefly Carthage, Berenice (Benghazi) and Ostia (Fulford 1987; 1989). At these sites, between 20% and 40% of the pottery was imported from outside the local region or province during the first to fourth centuries AD; and at Ostia the proportion of imports rose sharply during the second century from 20% to over 85%. Fulford argued that the pottery assemblage might be representative of general trading patterns in other, perishable, goods and therefore could be used to provide a rough picture of trading connections. In his 1989 paper he pursued this approach further, looking at the different patterns of connections exhibited by ports along the north African coast to the west and east of the Gulf of Sirte. Carthage, to the west, looked mainly to Italy for its imports; so (with a smaller sample size) did Sabratha; Cyrenaica to the east (represented by Berenice) looked to Crete and the Aegean, with some imports from Italy. But there seems to have been relatively little traffic east–west along the north African shoreline, as — he argued — north African imported amphorae at Berenice are of minor significance in the assemblage there. Fulford explained the marked differences between the import patterns of Africa Proconsularis and Tripolitania, on the one hand, and Cyrenaica, on the other, by a combination of currents, prevailing winds and the treacherous waters of the Gulf of Sirte. Despite the enormous potential of the comparison of port ceramic assemblages to which Fulford drew attention, twenty years further on relatively little has changed. Although new fieldwork has been done at several ports between Carthage and Berenice that might help fill out the picture — for example Leptiminus, Meninx, Lepcis Magna —, only the amphora assemblage from Meninx is published so far in a quantified form that enables such comparative analysis (Fontana, Ben Tahar and Capelli 2009). New data are becoming available for some ports in other regions, but for north Africa we are still reliant on the reports for Carthage, Sabratha, Berenice and now Meninx. Moreover, Fulford’s 1987 comparisons aggregated the sherd counts for fine-wares, coarse-/cooking-wares and amphorae, although we might suspect that the distribution of these categories need not respond to the same economic logic; disaggregation of the figures by ceramic category (below) enables more detailed patterns to be detected. CYRENAICA AND THE ADRIATIC In some respects, the picture sketched by Fulford still seems to hold good some two decades later, and has even been reinforced by new evidence. His picture of Cyrenaica’s northward links was based on 368 WILSON, SCHÖRLE & RICE FIG. 20.1. Distribution of Cyrenaican Mid Roman 8 amphorae. (A. Wilson/K. Schörle.) imports alone, but since he wrote we can now distinguish one Roman-period type of Cyrenaican amphora — and one only — that seems to have been exported outside Cyrenaica, albeit in limited quantities, and its distribution supports his view that Cyrenaica traded primarily with regions to the north rather than to the west. Riley’s Mid Roman 8 amphora was produced during the first half of the third century AD in at least three of the Cyrenaican port cities — Berenice (Benghazi), Taucheira (Tocra) and Apollonia (Marsa Souza).1 On grounds of morphology, with its wide mouth, the Mid Roman 8 is probably to be seen as an olive-oil amphora, although it could have carried salted fish — no residue tests have been performed yet. The limited distribution of this type includes the expected sites of Ostia and Rome, of course, but is otherwise confined to the Adriatic and northern Italy — it is found at Aquileia, at Milan, and at Altino, Oderzo and Concordia Sagittaria in the Veneto region, and also at Zaton near Zadar in Dalmatia (Fig. 20.1).2 The export of Cyrenaican olive oil to the Adriatic is confirmed in fact by a passage of the late second-/early third-century jurist Scaevola in the Digest (19.2.61.1), broadly contemporary with the period of Mid Roman 8 amphora production, which refers to a contract for shipping 3,000 metretai of olive oil and 8,000 modii of wheat from Cyrenaica to Aquileia. This suggests some north–south axis of trade between Cyrenaica and the Adriatic, although Cyrenaica had other links as well, to which we shall return. Although the Cyrenaican exports are not well recognized yet and their distribution elsewhere may be under-reported, the scale of exports from Cyrenaica was limited by comparison with, for example, olive-oil exports from Tripolitania, which may suggest differences also in the intensity between eastern and western connections from these two regions. RETURN CARGOES: DISTRIBUTION OF BRICKS AND BUILDING MATERIALS The distribution of Italian stamped bricks also supports the idea of a closer relationship between Italy and Africa west of the Gulf of Sirte than between Italy ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY 369 FIG. 20.2. Distribution of Italian stamped bricks in north Africa. (A. Wilson/K. Schörle.) and Cyrenaica (Fig. 20.2).3 Bricks whose stamps identify them as products of Rome or the Tiber valley are found at most major north African ports between Cherchel and Lepcis Magna, but not further east. These are interpreted as return cargoes carried on ships that had sailed from Africa to Portus with exports of grain, olive oil and fish products, for example. Most of the Italian bricks at these sites are found in bathhouses or related large reservoir cisterns; we do not know whether they arrived as products sold on the open market or as specially commissioned loads because a magistrate involved in the construction of a bath-house at a north African town had lands in Italy, or had a close connection with a landowner with a brickworks there. Either way, it does not affect our model; the important point is that the export trade from Africa Proconsularis to Portus/Rome subsidized the return carriage of bulk cheap goods in a way that the trade between, for example, Cyrenaica and Rome could not. The same point can be made for Campania; bricks of T. Claudius Felix, a Campanian producer with figlinae at Salerno, are found at Hippo Regius, Leptiminus and Thapsus (Wilson 2001a). The baths at Leptiminus also used imported pumice from Pantelleria (Lancaster et al. 2010), presumably picked up on a return voyage from Italy or Sicily. But while Italian bricks are found also in Sardinia and southern Gaul (Parker 2008), at a few ports around the mouth of the Rhône, they are not found in Cyrenaica, or anywhere else in the eastern Mediterranean. Fulford’s model of the separation of trade to east and west applies principally to communications along the southern shore of the Mediterranean, because of winds and currents. Of course, there were trading connections between the eastern and western basins of the Mediterranean, but much of this took more northerly routes, between the Aegean and southern Italy rather than along the African coast. The annona traffic between Alexandria and Rome took routes up the Levantine coast via Cyprus and southern Turkey, then across to Italy, often returning by a more direct openwater route (cf. Arnaud 2007). Trace element analysis has shown that the concrete breakwaters of Herod’s harbour at Caesarea Maritima in Judaea were built 370 WILSON, SCHÖRLE & RICE with pozzolana imported from the bay of Naples, presumably as part of Roman technical assistance to its client king (Oleson and Brandon 1992: 56–60). The export of the large quantities of pozzolana necessary would have been greatly facilitated by the fact that the Alexandrian grain ships of several hundred tons’ burden returning from Puteoli (since they could not dock at Portus, which had not yet been constructed at this date) would have had vast amounts of spare cargo capacity and, indeed, the need for some kind of return cargo if they were not to make the return voyage in ballast. Pozzolana from the Puteoli region has been identified in harbour works at Chersonesos on Crete also, presumably another of these targeted shipments for a particular project facilitated by the grain fleet’s need for return cargoes (Hohlfelder 1999: 158–9; Votruba 2007: 326–7). But it is likely that there was also a trade in pozzolana outside these routes — in addition to its main cargo of Italian wine amphorae, the Madrague de Giens wreck (c. 60– 50 BC) was carrying a complementary cargo of volcanic sand (Liou and Pomey 1985: 562) that may have been intended for sale, perhaps for use in a harbour construction project, rather than simply as ballast. wares, c. 35% are imported, and 15% of the total are Punic imports from what is now Tunisia or Tripolitania (Wilson 2005; Swift 2006; Wilson forthcoming). Punic amphorae constitute about 5% of the total amphora assemblage (Göransson 2007). Moreover, Cyrenaican amphorae can now be recognized in the published material from Punic levels at Sabratha. This might be taken to indicate that the physical discouragements to east–west trade along the African coast — winds, currents, the lethal shallows of the Gulf of Sirte — have been overstated, and that if any east–west divide existed in the Roman period it was not so much determined by sailing conditions as by different cultural or economic factors. Indeed, Kenrick’s analysis of the fine-wares at Berenice casts some doubt on the notion of an east– west divide for the Roman period as well (Kenrick 1985; 1987). To deal with problems of residuality, his figures relate not to particular assemblages, some of which were highly residual, but to datable forms, which were then aggregated within broad periods (Table 20.1). At Berenice, in the early Imperial period, from c. 25 BC to c. AD 100, some 60% of imported fine-ware (by sherd count) was Italian terra sigillata (ITS) (including the Campanian production once referred to as Tripolitanian sigillata); apart from a further 1% Campanian black gloss and less than 1% South Gaulish wares, the remainder is from the eastern Mediterranean. By the early second century AD, the Italian contribution has shrunk to 13%, and 55% is now accounted for by African red slip ware (ARS); by the third century ARS dominates the assemblage to the virtual exclusion of everything else. Fulford suggested that perhaps the ARS came to Berenice ‘through established networks in the Aegean area’ (Fulford 1989: 180); however, if this were the case, we might expect to find considerably more Aegean fine-wares. Moreover, if one wanted to argue for an indirect source, one might prefer the idea of redistribution back from major Italian emporia such as Portus or Puteoli.5 However, we certainly cannot rule out some EAST–WEST CONNECTIONS ALONG THE NORTH AFRICAN COAST POTTERY ASSEMBLAGES AT MAJOR PORTS But if these data largely seem to support Fulford’s view of a divide in exports between east and west, there are some new complicating factors that point to a level of trade along the north African coastline.4 The first is provided by the recent excavations at Euesperides, the predecessor to the site of Berenice (Sidi Khrebish, Benghazi), which provided one of Fulford’s key datasets. Here, the quantification of the early Hellenistic pottery (325–250 BC) indicates considerable east–west trade across the Gulf of Sirte, between Hellenistic Cyrenaica and Punic north Africa. Of the cooking TABLE 20.1. Quantification by period of the fine-wares from Berenice (Benghazi). Total sherds per period: 25 BC–AD 100: 3,824; second century AD: 1,574; early third century AD: 3,811. Based on the data provided by Kenrick (1985; 1987). 25 BC–AD 100 Second century AD Early third century AD ITS ESA ESB ARS Çandarli ware Misc. Total 61% 27% 3% 0% 1% 8% 100% 13% 2% 14% 55% 11% 5% 100% 0% 0% 0% 98% 0% 2% 100% ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY 371 TABLE 20.2. Quantification by period of the amphorae from Berenice (Benghazi). Total sherds per period: Augustan: 101; early to mid-first century AD: 154; mid- to late first century AD: 481; early to mid-second century AD: 556; late second century AD: 148; early third century AD: 615; late third century AD: 613. Based on the data provided by Riley (1979: 402–42). Local Italian Spanish North African Aegean Misc. imports Total 31% 6% 0% 5% 4% 54% 100% Augustan Early to mid-first century AD 7% 9% 0% 1% 12% 71% 100% Mid- to late first century AD 11% 5% 2% 2% 10% 70% 100% 13% 2% 2% 3% 24% 56% 100% 8% 0% 1% 6% 21% 64% 100% 5% 2% 0% 8% 41% 44% 100% 7% 0% 0% 12% 26% 55% 100% Early to mid-second century AD Late second century Early third century Late third century AD AD AD more direct contact between Tripolitania and Cyrenaica of the sort that appears to be attested for the third century BC at Euesperides. The amphorae, which formed a major part of Fulford’s argument, tell a less clear-cut story, as between 44% and 70% in any one period are largely unidentified imports, thus potentially masking other trends (Table 20.2). Moreover, Riley’s quantified tabulation of them by period does not attempt to deal with the problems of residuality highlighted by Kenrick for the fine-wares, although they come from the same contexts (Riley 1979: 402–42). Nevertheless, they suggest an increasing reliance on imports over local production, with Aegean imports growing over time to reach 41% by the early third century; the early Italian contribution is perhaps replaced by examples from north Africa, but these only achieve 6–8% in the late second/early third century. The proportions of Aegean to African amphorae are now the opposite of those for the fine-wares. The pottery from the Circular Harbour at Carthage also shows clear differences in the composition of the amphora and fine-ware assemblages (Tables 20.3 and 20.4) (Fulford 1994). These were analysed by phased contexts, not by datable sherds as Kenrick did for Berenice, and residuality is clearly a problem, as shown by the high percentages of Republican blackgloss wares in contexts of the first and second centuries AD. In an attempt to lessen this effect, several contexts that clearly contained a superabundance of residual material have been removed from consideration.6 In the late first century BC, Italian fine-ware is dominant, followed by Eastern sigillata A (ESA) (with 30%); thereafter ESA dropped out and ARS became progressively more important (Table 20.3). Its apparent failure to reach the dominant levels seen at Berenice is probably a result of the highly residual nature of the contexts examined (note residual Punic, black-gloss, thin-walled and probably residual ITS accounting in total for almost half of the second-century figures). By contrast the amphorae are overwhelmingly African at all periods (Table 20.4) — though this is likely to include imports from all along the Tunisian coast —, but with 7% identifiable Italian imports in the late first century BC and the first/early second century AD, and 10% Spanish in the period AD 1–125. The sample size for the amphorae is very large, but TABLE 20.3. Quantification by period of the fine-wares from the Circular Harbour at Carthage. Total sherds per period: first century BC: 906; AD 1–125: 428; AD 125–200: 1,378. Based on the data provided by Fulford (1994). First century BC Black gloss ITS ESA ARS Black gloss unknown Thin walled ware Punic Misc. Total 25% 2% 30% 0% 40% 1% 1% 1% 100% AD 1–125 4% 49% 0% 10% 14% 14% 0% 9% 100% AD 125–200 0% 20% 0% 43% 17% 5% 4% 11% 100% 372 WILSON, SCHÖRLE & RICE TABLE 20.4. Quantification by period of the amphorae from the Circular Harbour at Carthage. Total sherds per period: first century BC: 73,195; AD 1–125: 13,937; AD 125–200: 11,122. Based on the data provided by Fulford (1994). Italian Spanish North African Misc. imports Total 7% 1% 81% 11% 100% First century BC AD 1–125 7% 10% 74% 9% 100% AD 125–200 0% 0% 81% 19% 100% likely to suffer from the same problems of residuality identified for the fine-wares, making it difficult to draw detailed conclusions. What is clear, though, is that fine-ware distribution was driven by different economic considerations from that of amphora-borne commodities. This casts some doubt on the degree to which either type of pottery distribution on its own — or even both together — can be used as a simple proxy of trading connections. The recently published data from excavations at Meninx again show that the amphora assemblage there is dominated by African amphorae, mostly locally produced (Table 20.5).7 The imports are predominantly western rather than eastern. In the second and first centuries BC, 30% of the amphorae are imports, mainly from Italy; imports drop to under 25% in the first century AD, although are still dominated by Italian material, and then to a maximum of 9% in the early second century, and a maximum of c. 10% in the later second to mid-fourth centuries. There is uncertainty over the provenance of the Benghazi Mid Roman 1 amphorae in the sample; production sites are attested on the Tunisian coast, for example at Thaenae, but fabric analysis of samples in the assemblage suggests that the examples found at Meninx were produced in Sicily. Only three sherds from the Aegean region are present in the second- to fourth-century assemblages. The overall picture for Meninx must be seen against a general pattern of declining imports of amphoraborne goods to the island of Jerba as the region developed its own export-oriented production of wine and olive oil; but its overseas links clearly were mainly with Italy and the central Mediterranean, and not with the east. The amphorae from the Terme del Nuotatore at Ostia show a wide variety of sources (Table 20.6).8 Initially substantial contributions from Italy and Spain shrank after the middle of the second century, to be replaced by greater imports from first Gaul (24% in the early second century) and then north Africa (38% in the mid- to late second century), and then the Aegean (41% in the early third century), with Africa in second rank at 30%. The prevalence of north African imports (55%) is shown also in the late third- to late fourth-century destruction level. We do not have the fine-ware quantification for these deposits, and are not told anything about residuality; but it is interesting to note that there were more Aegean than African amphorae in the century when ARS exports peaked. More recent amphora evidence from the Deutsches Archäologisches Institut–American Academy in Rome (DAI–AAR) excavations at Ostia complements and enhances the picture.9 As in the Terme del Nuotatore, Spanish and Italian products dominate the assemblage between AD 50 and 100. The second century varies between the two excavation areas; whereas Gallic and Spanish amphorae are present in relatively similar quantities from the Terme del Nuotatore, Spanish amphorae solidly dominate the assemblage from the DAI–AAR excavations (44% of the amphorae). Mid-second to mid-third-century levels TABLE 20.5. Quantification by period of the amphorae from the excavations at Meninx, Jerba. Total sherds per period: 150–1 BC: 40; AD 1–100: 111; AD 100–50: 146; AD 150–350: 108. Based on the data provided by Fontana and his colleagues (2009: tables 16.15, 16.18 and 16.19). Italian Spanish Sicilian? North African Aegean Total 150–1 BC 27% 0% 0% 70% 3% 100% AD 1–100 18% 3% 3% 76% 0% 100% AD 100–150 3% 1% 3% 92% 1% 100% AD 150–350 0% 0% 6% 91% 3% 100% ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY 373 TABLE 20.6. Quantification by period of the amphorae from Ostia, Terme del Nuotatore. Total sherds per period: AD 80–90: 430; AD 90–155/60: 419; AD 155/60–190: 379; AD 190–235/40: 2,228; late third century to late fourth century AD: 1,129. Based on the data provided in: Carandini and Panella 1973: 463–619; Carandini and Panella 1977: 116–262, 359–83. Italian Gallic Spanish North African Aegean Misc. imports Total AD 80–90 23% 16% 25% 7% 2% 27% 100% AD 90–155/60 12% 24% 21% 9% 6% 28% 100% AD 155/60–90 5% 12% 12% 38% 15% 18% 100% AD 190–235/40 3% 10% 4% 30% 41% 12% 100% 6% 7% 6% 55% 20% 6% 100% Late third century AD to late fourth century AD were absent from the DAI–AAR excavations, missing the peak in Aegean amphorae visible in the baths. The DAI–AAR excavations have produced more substantial evidence on the later levels at Ostia. From AD 280 onwards, north African imports dominate (50% of imports AD 280–300; 61% AD 350–475). Eastern Mediterranean amphorae do, however, account for 64% and 55% of all wine amphorae respectively in these two phases. CONNECTIONS BETWEEN COMMUNITIES If elements of the more detailed ceramic analysis, notably the fine-ware analysis from Berenice, undercut the notion of an east–west divide along the north African coast, some other evidence does the same. At Carthage, a group of Greek-speaking worshippers of Sarapis was formed probably of resident Alexandrian merchants. The cult of Sarapis was, of course, of Alexandrian origin, and the worshippers at Carthage appear to have comprised a distinct group of people; the known dedicators all had Greek cognomina and the dedicatory inscriptions are either in Greek or bilingual in Greek and Latin (Rives 1995: 212–14). Several dedications show clear Alexandrian connections, most importantly a bust of the Egyptian priest Manetho, who Plutarch says assisted in the foundation of the cult in Alexandria (Plutarch, De Iside et Osiride 28). The presence of a group of Alexandrians at Carthage is certainly plausible and even to be expected. It is possible that a similar situation existed at Lepcis Magna. A group of 30 dedications to Sarapis in Greek (still unpublished) were found inside the temple to Sarapis. These are apparently similar to five inscriptions published in IRT (310, 310a, 311, 312, 313), three of which bear the names of men with the Latin nomen Aurelios with Greek cognomina. Out of the group of 30 unpublished inscriptions, twelve apparently have the nomen Aurelios, suggesting a date after Caracalla’s grant of universal citizenship in AD 212. Thirteen different dedicators are mentioned in total, but all are reported as having Greek cognomina (Brouquier-Reddé 1992: 103). While monuments to Sarapis are fairly common in north Africa, there are only four cities that have verifiable temples or sanctuaries: Carthage, Sabratha, Lepcis and Lambaesis, with a possible fifth at the port city of Gigthis in southern Tunisia (Kater-Sibbes 1973: 136).10 Whereas the sanctuaries at Carthage and Lepcis contained Greek dedications, the dedications at Lambaesis, in inland Numidia, are primarily in Latin. Indeed, the presence of Greek inscriptions in north Africa west of Cyrenaica is extremely slight before the Byzantine reconquest of north Africa. The two cities with the greatest concentration of Greek inscriptions are Lepcis Magna and Carthage; other cities that contain more than a few are Oea, Sousse and Cherchel, all major port cities.11 These communities of Greek speakers, probably merchants, further illustrate east–west trading connections. Lepcis Magna also may have had resident communities of other groups of merchants; as a former Phoenician colony, the city retained links with its mother-city of Tyre, and dedicated at Tyre a statue personifying that city with a bilingual Greek and Latin inscription at some time between the reign of Trajan and AD 198, possibly close to that later date (Rey-Coquais 1987). In the reign of Septimius Severus, after AD 198, a statue-base was dedicated to Geta in the Forum Vetus at Lepcis Magna by Septimia Tyros colonia metropolis Phoenices et aliarum civitatium (IRT 437). Whether or not these examples can be 374 WILSON, SCHÖRLE & RICE taken also to suggest communities of traders at Lepcis and Tyre, they may provide some evidence for continued east–west connective links between two former Phoenician emporia. Trading communities of residents of other cities are of course known at other large port cities; the Piazzale delle Corporazioni at Ostia provides a very familiar example, but there was also a community of Tyrians at Puteoli, who were having difficulty paying rent on their statio in AD 174, perhaps as a result of a decline in their numbers due to the Antonine plague (Mommsen 1843: 57–62; CIG III 5853; IG XIV 830). Puteoli had also acquired a sanctuary of Sarapis as early as 105 BC, and a temple of Dusares, a Nabataean god, by the mid-first century BC; both cults doubtless were imported by traders from Alexandria and Arabia respectively (Tchernia 1997: 128–9). MARITIME NETWORKS AND THE INTEGRATION OF THE ROMAN ECONOMY Such trading communities at major ports suggest regular mercantile connections on established routes, with knowledge about general market demand (if not actual price levels) at distant ports, and sometimes contacts between known trading partners. Direct maritime trading connections between emporia with major port facilities helped enable a considerable degree of geographical integration of the Roman economy, at least as far as the availability of goods in the Mediterranean provinces was concerned. It is less clear that prices for the same commodities in different regions were closely integrated,12 but it can be shown that the intensity and regularity of maritime export flows affected and even determined the articulation of production in certain goods, which is particularly clear in the glass industry. Primary production of raw glass took place often at a very considerable distance from the centres that transformed that raw glass into vessels; the different stages of production were linked by long-distance maritime trade-routes. THE ROLE OF MARITIME NETWORKS IN THE ROMAN GLASS INDUSTRY It is now recognized that, before the widespread (re)introduction of soda-rich plant ash as an alkali flux in the ninth century AD, the Roman and Byzantine production of blown glass was a two-stage business (Foy and Nenna 2001).13 The first stage was the primary production of raw glass in large kilns by heating sand with natron as a flux. The principal sources of natron in the whole empire were the Wadi Natrun, south of Alexandria, and the Beheria region of Egypt; Lake Pikrolimni in northern Greece seems to have been the source of what Pliny calls ‘Chalestricum’ natron, and there may have been some lesser sources elsewhere but they do not seem to have been significant in antiquity (Dotsika et al. 2009; Pliny, Naturalis Historia 31.46.106–9).14 Raw glass was produced in slabs weighing perhaps 8–10 tons each, and sometimes up to 25 tons, as in the case of the first- to second-century AD furnace excavated at Beni Salama in the Wadi Natrun, close to the source of the natron flux (Nenna 2007: 127–8). These massive slabs were then smashed up into ingots or chunks of raw glass, which were used by secondary production centres where the raw glass was heated and blown into vessels. Although secondary production centres (glassblowing workshops) are found all over the Roman world, there were only a few primary production centres (Table 20.7). Chemical analyses have shown that Roman glass — oddly, and in contrast to glass of earlier and later periods — is remarkably homogeneous in composition, implying a very limited range of geographical sources for the sand used for primary production, located chiefly in the southeastern Mediterranean (Foy and Nenna 2001). Primary production centres have been identified archaeologically in Egypt in the Wadi Natrun for the early Imperial period and to the south of Lake Mareotis for the late Roman period, and in the Levant — for example Bet Eli’ezer in Israel, where a battery of seventeen Byzantine furnaces of the sixth to seventh centuries AD was found (Foy and Nenna 2001: 34–9). Although no early Roman primary furnaces have been found yet in the Levant, most Roman glass has a chemical signature matching the sand from the Syro-Palestinian coast, which was especially well suited to glass-making with a natron flux; this region must have been the overwhelmingly dominant supplier of raw glass for the secondary production centres all over the Roman world (Foy 2003b: 26). By contrast, chemical analysis suggests that material from the primary production centres in Egypt was not on the whole exported across the Mediterranean, but used at secondary production centres within Egypt (Foy and Nenna 2001: 37). Primary production also seems to be indicated at Carthage by the find of large quantities of raw glass in a cistern reused as a primary furnace (as in the Levant), but the discovery was made by a bulldozer in the 1970s and details are scant; it is ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY 375 TABLE 20.7. Primary glass production centres. Site Country Date Reference Bir Hooker (Wadi Natrun) Egypt Early Imperial Foy and Nenna 2001: 36; Nenna 2003 Zakik (Wadi Natrun) Egypt Early Imperial Foy and Nenna 2001: 36; Nenna 2003 Beni Salama (Wadi Natrun) Egypt First–second centuries Taposiris Magna Egypt Late Roman Philoxénité Egypt Fifth–eighth centuries Hermoupolis Egypt Eighth–ninth centuries Apollonia/Arsuf Israel Late Roman? Beth Eli’ezer (Hadera) Israel Sixth–seventh centuries Beth She’arim Israel Ninth century Tyre Israel Early medieval Kingsley 2004: 21 Carthage Tunisia Roman or late Roman Foy 2003c Hambach (near Cologne) Germany Roman Foy 2003c: 35 assumed to date before c. AD 800 (Foy 2003c: 35). If this is the case, it also implies the import of natron to Carthage from Egypt, which would be consistent with the suggestion of Alexandrian merchants at Carthage, and further undercuts the idea of an east–west divide in north African trade. Primary production is also suspected in Italy, Spain and Gaul on the basis of remarks in Pliny (Naturalis Historia 36.66), and receives some support from recent chemical analysis of the strontium and neodymium signatures of a number of Roman vessel-glass samples from Maastricht and Bocholtz (Netherlands), Tienen (Belgium) and Kelemantia (Slovakia), which do not match eastern Mediterranean signatures and therefore indicate a source in the western empire (Degryse and Schneider 2008). Chemical analysis has proved production of raw glass also from sands at Cologne in the first and second centuries AD and in the nearby Hambach forest to the west in the third to fifth centuries AD (Fremersdorf 1965–6; Rottländer 1990; Gaitzsch et al. 2000; Wedepohl and Baumann 2000).15 The magnesium and potassium concentrations in glass produced here are too low for potash fluxes and seem to imply the importation of natron to the German provinces (Wedepohl and Baumann 2000: 130). However, no primary furnaces have been located yet elsewhere in the western Mediterranean, and the dominance of chemical signatures typical of Syro-Palestinian glass in most of the Roman glass samples that have been AD Foy and Nenna 2001: 35–6; Nenna 2003 Foy and Nenna 2001: 34 AD AD Foy and Nenna 2001: 39 Foy and Nenna 2001: 39 Foy and Nenna 2001: 38 AD AD Foy and Nenna 2001: 37–8; Foy 2003a Foy and Nenna 2001: 38–9 analysed indicates that most of it was produced in the Levant. Such primary production centres as did exist in the west were less important, and shorter-lived, than the Levantine centres that produced the great majority of the empire’s raw glass (Foy and Nenna 2001: 37). Although glass could also be, and was, recycled, the implication of the limited number of primary production centres is that the Roman glass-making industry was massively dependent on the efficient long-distance movement of raw materials: firstly, of the natron flux from Wadi Natrun in Egypt (the main source in antiquity) to the primary production centres in Wadi Natrun, the Levant, and, to a much lesser extent, in the western Mediterranean; and secondly, of the raw glass chunks or ingots from the Levant to nearly all the secondary production centres around the empire (Fig. 20.3). This large, bulk transport of raw glass was essentially a maritime phenomenon. The maritime transport of raw glass ingots between ports is attested amply in several wrecks dating between the third century BC and the fifth century AD (Table 20.8; Fig. 20.3), such as Les Sanguinaires A and the Jeaune-Garde, Mljet, Bourse, Mellieha, Ognina and Port Vendres 1 shipwrecks; the most substantial glass cargo comes from the Embiez Ouest wreck, which contained at least 18 tons of raw glass ingots (Foy and Nenna 2001: 101–12; Nenna 2007: 131). These are also present in several port 376 WILSON, SCHÖRLE & RICE FIG. 20.3. Roman primary and coastal secondary glass production sites, with the distribution of raw glass in wrecks and at ports. (A. Wilson/K. Schörle.) ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY 377 TABLE 20.8. Raw glass in wrecks. Site Country Date Reference Les Sanguinaires A France Second half of the third century Lequin 2 France Late third/early second century La Jeaune-Garde A France c. 100–25 Mljet Croatia Late first century Embiez-Ouest France Second half of the second century Marseilles (Bourse) France AD Mellieha Malta First half of the third century AD Foy and Nenna 2001: 111 Ognina Italy (Sicily) First half of the third century AD Foy and Nenna 2001: 112 Port Vendres 1 France Early fifth century BC Foy and Nenna 2001: 102 Foy and Nenna 2001: 102 BC Foy and Nenna 2001: 103 BC Foy and Nenna 2001: 109 AD 190–220 AD Foy and Nenna 2001: 110 Foy and Nenna 2001: 110 assemblages, where they may suggest the existence of secondary workshops (Table 20.9; Fig. 20.3). Raw glass has been found at the exporting port of Apollonia/Arsuf in Israel (close to the primary production centres of the Near East), and at Apollonia in Cyrenaica, Sabratha in Libya, Carthage in Tunisia, Pompeii in Italy,16 and in France at Marseilles, Marseillan, Narbonne and the Golfe du Fos. Outside the Mediterranean, glass ingots have been discovered at Clysma at the head of the Red Sea (Bruyère 1966: 49, 115),17 where they presumably had been shipped down Trajan’s canal and were either awaiting working in the town or further export down the Red Sea coast or even to India, as mentioned in the Periplus Maris Erythraei (chapters 7, 10, 49). Secondary production Foy and Nenna 2001: 112 AD centres (glass-blowing workshops) might be located either at port cities or inland; Foy and Nenna’s survey of French glass production centres collects abundant evidence for glass-blowing at inland sites (Foy and Nenna 2001: 40–66), and indeed it is likely that by the second century AD most Roman cities, coastal or inland, may have had glass-blowing workshops. But since the inland workshops must have imported much of their material — the raw glass from eastern Mediterranean sources, though not recycled material — via port cities, the ports must have functioned as key nodes in the organization of supply networks for the glass industry. Unsurprisingly, therefore, glass workshops also developed at a number of the major Mediterranean port cities (Table 20.10; Fig. 20.3). TABLE 20.9. Raw glass at ports. Site Country Date Reference Apollonia/Arsuf Israel Foy and Nenna 2001: 106 Apollonia Cyrenaica Foy and Nenna 2001: 106 Sabratha Libya Roman Carthage Tunisia Fourth century Pompeii Italy AD 79 Verità 1999 Narbonne France AD 30–50/60 Foy and Nenna 2001: 107 Golf de Fos France Foy and Nenna 2001: 108 Marseillan France Foy and Nenna 2001: 106 Marseilles France Clysma Egypt Wilson 1999: 50 Tatton-Brown 1994: 288 AD First/second century AD Foy and Nenna 2001: 25, 106 Bruyère 1966 378 WILSON, SCHÖRLE & RICE TABLE 20.10. Coastal secondary production centres (glass-blowing workshops). Site Country Date Reference Beth She’arim Israel Jalame Israel Apollonia/Arsuf Israel Beth She’an Israel Late sixth century Caesarea Maritima Israel Late antique Philoxénité Egypt Foy and Nenna 2001: 34 Alexandria Egypt Foy and Nenna 2001: 34 Zakik (Wadi Natrun) Egypt Foy and Nenna 2001: 34 Delos Greece Late second/early first century BC Foy and Nenna 2001: 35 Lepcis Magna Libya Early second century Ioppolo 1969–70: 232, 234 Iol Caesarea Algeria Roman CIL VIII 9430. Marseilles France Fourth–seventh centuries AD Foy and Nenna 2001: 43 Aquileia Italy Puteoli Italy Salona Croatia Rome Italy Imperial Wilson 2001b Portus Italy Late Roman Simon Keay, pers. comm. Foy and Nenna 2001: 34 Late fourth century AD Foy 2003a: 29 Foy and Nenna 2001: 34 Foy 2003a: 29 AD Kingsley 2004: 136. AD CIL III 9542; Clairmont and von Gonzenbach 1975: 58–63 Alexandria, Aquileia and Puteoli all had particularly famous glass-making industries. Besides the ports with glass ingots already mentioned, on the north African coast, there is direct evidence for secondary glass-working at Lepcis Magna, and at Iol Caesarea where an inscription mentions a vitriarius;18 elsewhere, Salona19 and Marseilles20 have produced archaeological evidence of glass production, and recent excavations at Portus have discovered evidence for late Roman glass-working there.21 Overall, the Roman glass industry exhibits a considerable degree of panMediterranean integration, and glass ingot distribution required entrepôt as a means of articulation between the supply of raw material and secondary production. Indeed, the Embiez Ouest wreck, a small ship of 20– 5 tons capacity carrying nearly 18 tons of raw glass as its primary cargo, has been interpreted as engaged in this kind of redistribution, probably from a major Italian entrepôt such as Portus or Puteoli towards a port in southern Gaul, such as Arles (Jézégou 2008). This picture of the organization of the glass industry being dependent on maritime trade links is supported by what happens from the fifth century onwards. With the disintegration of the empire in the west, the production and usage of glass in northwest Europe, remote from the primary source regions in the Levant, drops significantly. Cologne and the surrounding region continued into late antiquity as the major source of glass in the fifth and sixth centuries, and Frankish glass was produced and traded as far as Anglo-Saxon England, but not in the mass-produced quantities that had characterized the Roman period. The reduced access to eastern Mediterranean glass sources as a result first of the disintegration of the Roman Empire and then, from the seventh century, the formation of new Islamic states, meant that glass production and consumption in the western Mediterranean and northwest Europe diminished substantially from the fifth century onwards. From the fifth century, more varied chemical compositions of glass also suggest a growing diversity of silica sources, again probably as a result of reduced access to Levantine raw glass production following the collapse of the western Roman Empire, although natron still seems to ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY have been used as a flux (Aerts et al. 2003). In the ninth century, however, a technological shift towards potash fluxes, rather than natron, removed the industry’s dependence on a primary source region; the potash fluxes could be produced anywhere, and worked well with a wider range of sand deposits than did natron. The result of this new flux technology was the emergence of a much more decentralized pattern of glassmaking, with a wider range of sources of raw glass and with secondary glass-blowing centres closer to the primary production centres, removing the previous dependence on Egypt for natron and the Levant for the raw glass, and reducing the importance of long-distance maritime connections between primary and secondary production centres. TRADING CONNECTIONS AND HIERARCHIES OF PORTS The evidence of pottery assemblages and of merchant communities in port cities, and the importance of maritime trade in the organization of the glass industry, accords with recent research on patterns of trading connections argued on the basis of cargo composition from shipwrecks, which emphasize for the Roman period the predominance of direct, regular connections between ports over a pattern of casual tramping.22 Nieto (1997) argued for a model in which direct longdistance connections between principal ports or emporia were supplemented by coastal connections, with secondary ports supplying or receiving goods from the main emporia. Boetto (Chapter 8) develops this picture by identifying five major patterns of trading voyages: direct voyages with single cargoes between principal ports (emporia or entrepôt); voyages with mixed cargoes loaded at an entrepôt and conveyed to another principal port; mixed cargoes loaded at a major entrepôt and redistributed towards a secondary port; homogeneous cargoes transported between ports as the result of a specific order; and casual tramping from port to port. While Nieto’s regional collection model implies a simple two-level hierarchy of ports and harbours, with emporia in one category and secondary ports in the other, the more complex set of possibilities argued for by Boetto might lead us rather to suspect the existence of a multi-level hierarchy, with small ports serving as the central place for several anchorages and coastal villas, and feeding in turn into a regional emporium. These questions have yet to be addressed by research on ancient ports, but we offer some thoughts as to how one might proceed in 379 constructing regional hierarchies, based on an analysis of harbours, mainly along Italy’s western façade maritime.23 RECONSTRUCTING PORT HIERARCHIES At a basic level, one could compare ports on the basis of their enclosed harbour area, length of wharfage and depth. Figures for wharfage length and depth are available for only a few Roman ports, and since most Roman ships could be accommodated in a depth of 3 m or so, the latter variable may not be very indicative. Rough harbour areas more often can be obtained or estimated from published plans; these should be considered a rough indication of scale rather than scientifically accurate data.24 Figure 20.4 plots the available data for harbour areas along the Italian coast from Cosa to the bay of Naples (cf. Table 20.11). Several points are immediately apparent: the dominance of Portus and Puteoli, which comes as no surprise; the existence of a second rank that includes Nero’s harbour at Antium (Anzio) and the Trajanic harbour at Centumcellae; and a host of much smaller harbours for lesser towns and villas. In particular, we should note that the port of Cosa, at 2.5 ha, is smaller than the harbours of the villas at Torre Astura (7.8 ha) and Torre Valdaliga, although the recent identification of what appears to be a second port for Cosa on the western side of the headland, associated with kilns for Dressel 1 amphorae (Fentress 2009), may more or less double Cosa’s total port capacity. When we consider that by the Imperial period the port of Cosa appears to have become the port for a coastal villa, and that the town itself had declined, this is less surprising. It is possible that the eastern port is now to be seen as associated with the export of fish products from the fishery, and the western port with the export of wine, bottled in the amphorae produced at the adjacent kilns. Nevertheless, the example of Cosa highlights the fact that a maritime villa of the first or second century AD might possess better and larger harbour facilities than some towns of the second century BC. While both ostentation and utility were probably factors in the development of such villa harbours, their size, together with other aspects of export-oriented villa production, is easier to reconcile with the model of export-directed trade through the medium of emporia than with the world of casual cabotage. Such exports from villa harbours might go to secondary ports for regional collection, or even directly to distant emporia, or to other ports as 380 WILSON, SCHÖRLE & RICE FIG. 20.4. Roman harbours along the Italian coast from Cosa to the bay of Naples, with relative sizes where known. (K. Schörle.) directly ordered cargoes. This perhaps can be exemplified by the port on Giglio associated with the granite quarries from which columns were exported; both may have been linked originally with the nearby villa, which belonged to the Domitii Ahenobarbi before becoming imperial property (Bruno 1998: esp. p. 128; Ciampoltrini and Rendini 2004: 137–42). To put these Roman villa harbours into perspective by comparison with earlier periods, the Punic cothon harbour of Motya (Sicily) encloses just 0.18 ha (with less than 170 m of wharf space), and the cothon of Mahdia (Tunisia) encloses 0.78 ha (with less than 370 m wharf space). One could also try to move beyond the simple quantification of harbour basin sizes to a more nuanced analysis of the relative importance of harbours in a region by considering other factors — the size of the associated port city, its legal status and range of public buildings — to come up with a kind of Central Place Theory ranking of functions and services, with larger centres providing a greater variety of goods and services over a larger geographical range.25 Figure 20.5 represents an impressionistic attempt to do this for the Minturnae region, where harbour size data are not available, but there is other information on settlements, public buildings and inscriptions documenting trading connections, which enables a crude ranking of the importance of the ports (Ruegg 1988; Schörle 2011). The colony of Minturnae had a large sheltered river port, where fifteen Roman ship-sheds have been discovered, and it appears also to have been a centre for shipbuilding, especially for dolia ships used in the wine trade (Schörle 2011).26 Minturnae may have played a role as a subsidiary ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY 381 TABLE 20.11. Sizes of selected harbour basins. Site Portus (total) Claudian basin Trajanic hexagon Darsena Alexandria, Portus Magnus Puteoli (total) Harbour area (ha) Wharfage length (m) 234 c. 13,890 c. 200 33.3 1.08 c. 2,860 2,100 >226 12,380 67.9 Portus Iulius 53.9 Portus Baianus 14 Antium 25–30 Ephesus c. 18–24 Reference Keay (Chapter 2: n. 64); Morelli, Marinucci and Arnoldus-Huyzendveld 2011 Wharfage figure includes various canals Keay (Chapter 2, this volume) Calculated from plan in Goddio and Fabre 2008: 38 Calculated from plan in Brandon, Hohlfelder and Oleson 2008: 376 fig. 1 Calculated from plan in Brandon, Hohlfelder and Oleson 2008: 376 fig. 1 Calculated from plan in Brandon, Hohlfelder and Oleson 2008: 376 fig. 1 Felici 1995: 61 Calculated from Google Earth Caesarea Maritima (outer basin) 20 Oleson 1988: 152 Hadrumetum 20 Bartoccini 1958: 12 Centumcellae 14 Carthage (circular and rectangular harbours) 14 Romanelli 1925: 92 Terracina 11 Calculated from plan in De Rossi 1980: 100, fig. 25 Lepcis Magna 10.2 Torre Astura 7.8 Calculated from Marzano 2007: 49, fig. 5 Kenchreae (Corinth) 3 Kingsley 2004: 140 Cosa 2.5 Gazda 1987: 75 Giglio Porto No more than 2,000 1,200 c. 2 Calculated from plan in Caruso, Gallavotti and Aiello 1991 Bartoccini 1958: 12–13 Calculated from plan in Ciampoltrini and Rendini 2004: 138 fig. 6* La Mattonara 1.24 Calculated from plan in Higginbotham 1997: 94 fig. 18 Villa port at San Simone 0.84 Degrassi 1955: 136 Ventotene (Pandateria) 0.7 Franco 1996: 297 *The units of the scale bar of this plan are not specified and the plan has clearly been greatly reduced from the stated 1 :20,000 scale; checking against Google Earth indicates that the scale bar must represent 30 m in 2 m and 10 m units. hub via which cargoes from larger emporia (Naples, Puteoli) might be redistributed to local smaller ports (Formia and Gaeta) and maritime villas (Gianola). Nevertheless, Minturnae’s replacement of Sinuessa as the main port for the export of wine from the Ager Falernus, and its role as the home port for dolia ships trading wine between Italy, Gaul and Spain in the first century AD, also suggest that it had direct longdistance links that bypassed the emporia of the bay of Naples (Schörle 2011). In some instances we may be able to deduce something about a port’s role in wider trading networks 382 WILSON, SCHÖRLE & RICE FIG. 20.5. Roman harbours in the Minturnae region, with a rough estimate of relative importance expressed by the size of the anchor symbol. (K. Schörle.) even when there is no archaeological evidence of port facilities. The case of Baelo Claudia (Bolonia, Cádiz) is instructive here — a coastal town, with several fish-salting factories implying export production, although no port facilities have been found yet. One possibility would be the existence of wooden jetties, or perhaps ships were loaded and unloaded by stevedores wading through the shallows (Rickman 1985: 111), as shown on a third-century mosaic from Sousse apparently depicting the unloading of firewood on a beach.27 In either case, though, Baelo can only have been frequented by small ships, which presumably collected the export produce into, for example, Gades (Cádiz) or Iulia Traducta in Algeciras Bay for transshipment and re-export. The fourth-century ostraca recording shipments of olive oil into Carthage show this process of regional collection at work. An official describing himself as the mensor olearius at Carthage recorded a series of consignments of 200–20 amphorae each, arriving on ships, which he then inspected for quality, and weighed (Peña 1998). This was part of a state-directed operation and seems to reflect the collection of oil at Carthage before onward shipment to Rome as part of the oil annona. The ships arriving into Carthage were coming from ports along the northern Tunisian coast, and the ships themselves must have been pretty small — the second-century Grado wreck, a 20- to 25tonner, was carrying three times as many amphorae. CAPACITIES OF HARBOURS AT MAJOR EMPORIA Ideally, we want to estimate harbour capacities not simply in terms of area, but of ships, for which we need to know figures on wharf length and constraints on ship sizes.28 Table 20.11 lists some Mediterranean harbour sizes and, where they can be measured, wharf lengths. As expected, Alexandria and Portus top the list, far larger than other major harbours; but some of the other well-known harbours, such as Carthage and Lepcis Magna, look relatively small. The harbour at Lepcis Magna, so well preserved because it was silted up and abandoned, is about half the area of the harbours of Caesarea Maritima and Hadrumetum. The area figure for Carthage, 14 ha as reported by Romanelli (1925: 92), looks small by comparison with several other large harbours in Tunisia and Italy, but it refers only to the Punic circular and rectangular harbour basins (which were reused in the Roman period), and cannot represent the full extent of Carthage’s harbour facilities. Indeed, recent work by Hurst (2010) suggests that the stretch of coastline between Falbe’s Quadrilateral and the Bordj Djedid hill, some 2,000 m, was developed in the Roman period with warehouses that must have been fronted by continuous wharfage, and that much of the merchant shipping for Carthage may have docked here.29 Indeed, if Hurst’s hypothesis is correct, it would be logical to see the Roman development of the natural spring known as the Fontaine aux mille amphores at the foot of the Bordj Djedid hill as connected with the need to supply water to ships along this stretch of docks. In addition, Saint Augustine (Confessions 5.8) departed from Carthage by ship from a point near the memoria of Saint Cyprian, usually identified with a basilica overlooking the bay of Dar Saniat (the site of the modern Hotel Amilcar) to the north of the city; Augustine’s ship presumably left from this bay. Nevertheless, comparison of the columns for area and wharf length shows that area only gives a very rough guide to harbour capacity; a more directly relevant (but often less easily available) figure is wharf length, and this bears no constant ratio to area. The Claudian harbour at Portus is now estimated to have enclosed an area of c. 200 ha with perhaps 2,860 m of wharf length (Morelli, Marinucci and ArnoldusHuyzendveld 2011; Keay, Chapter 2: n. 64); the Trajanic hexagonal basin, 33.3 ha in size, added another 2,100 m of wharfage (Keay, Chapter 2: n. 64) — seventenths as much, but for only a sixth of the surface area, reflecting the more efficient design of the Trajanic harbour. The wharfage length is our best indicator of how many ships could dock; in the Mediterranean, ships traditionally moor perpendicular to the quayside, in the so-called ‘Mediterranean moor’.30 That similar ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY arrangements were also standard in antiquity is indicated both by the alignment of seven sunken fifthor fourth-century BC wrecks at Heracleion-Thonis, found parallel to each other and apparently moored end-on to a wooden jetty (Goddio 2007: 111, 113 fig. 3.83, 114), and of fifteen fifth-century AD wrecks burnt while moored in the harbour at Olbia, bow to the shore between wooden jetties (D’Oriano and Riccardi 2004). We therefore need to divide the length of available wharfage by the amount of space required for a ship to moor — its beam, plus extra room for clearance between ships (although see below for the suggestion of broadside-on mooring in the canals at Portus). Fortunately, evidence from Portus helps here: the sides of the Trajanic hexagon each measure 358 m and had mooring blocks every 14–15 m (on side V at least), which would accommodate ships of 10 m beam with 4–5 m clearance between pairs of ships (Testaguzza 1970: 162–3). This would allow up to 24 or 25 ships on a side, but in order to avoid interference between ships at the angles where two sides met this maximum number may not have been reached, and indeed a column at the angle of sides III and IV, inscribed with the numeral XXIII (Testaguzza 1970: 163), presumably indicating the numbering of individual berths, suggests that we should imagine 23 ships for each of the five sides I– IV and VI. With a further sixteen ships for side V (which includes the entrance), the Trajanic hexagon therefore might have provided docking space for 130 large merchant ships each of several hundred tons.31 A ship of 10 m beam might be some 40 m long (at a breadth:length ratio of 1:4); for comparison, the Madrague de Giens wreck of 300–400 tons was 9 m in the beam and c. 40 m long (Pomey 1982: 145).32 To the figure for the Trajanic hexagon, we need to add the Claudian harbour. Testaguzza assumed that the 2,500 m of quayside there allowed for the docking of 250 ships, evidently calculating only 10 m per ship (including clearance). New coring work in the Claudian basin shows that it was larger than Testaguzza thought, and Keay estimates the wharfage at 2,860 m (Morelli, Marinucci and Arnoldus-Huyzendveld 2011; Keay, Chapter 2: n. 64); but if we assume the same arrangements as for the Trajanic hexagon (we have no direct evidence for the spacing of mooring stones in the Claudian basin), we would need to revise this figure downwards to berths for c. 190–205 ships of the same size. Both basins, however, provided (relatively) sheltered anchorage for more ships that could wait at anchor in the harbour awaiting their turn to dock at 383 the quays. The protection thus offered was not total, however — Tacitus records the loss of nearly 200 ships at anchor in the Claudian harbour of Portus in AD 62 (Annals 15.18), much the same figure as our estimate for the docking capacity. It is unlikely that every ship in the harbour was wrecked in this storm, and there were probably therefore over 200 ships in port at the time, which is entirely conceivable if there were c. 200 docked at the quays and others at anchor in the basin awaiting their turn to dock. Besides the two main harbour basins at Portus, there were other quays too — Testaguzza estimated the available wharfage of the Darsena basin and the connecting canal between the Claudian and Trajanic basins at a total of 1,950 m (Testaguzza 1970: 161). The connecting canal had perhaps 1,425 m of quay space,33 but mooring here might have impeded through traffic. The Darsena must have accommodated smaller ships than in the main basins; it measured 240
45 m (1.08 ha) (Testaguzza 1970: 173), and its 9 m wide entrance will have limited the ships that could enter it. Since it is only 45 m wide it is difficult to imagine vessels more than 15 m long mooring opposite each other; this area was most probably used to load up barges taking goods up the Tiber to Rome. If we allowed 9 m mooring per vessel (for a beam of 5 m plus 4 m clearance), 26 ships could be accommodated along each of the long sides (240 m long), making 52 in total (assuming that none was moored against the short sides, as most of the space here would be taken up by the endmost ships moored on the long sides). But vessels may have moored broadside-on in this more restricted basin, which would have reduced the number substantially.34 There was at least a further 3,710 m of available wharfage along the Fossa Traiana (50 m wide),35 but here we probably need to imagine vessels mooring broadside-on to the quays. The marble yards of Portus were located along the southern side of this canal, and column shafts, blocks and other items of cargo weighing several tons must have been regularly offloaded from sea-going ships and loaded onto vessels heading upstream to Rome, operations that must have involved the use of cranes. This could hardly have been done across the forequarters of ships docked prow-on to the quay, and must have necessitated broadside-on mooring, with ships tethered at bow and stern. Indeed, it was probably the opportunities offered by the Fossa Traiana for broadside-on mooring that determined the location of the marble yards here. Large ships with cargoes of marble could enter the Fossa 384 WILSON, SCHÖRLE & RICE Traiana without having to pass through the Claudian basin or the connecting canal by the Darsena, and dock side-on in the canal so that their cargoes could be unloaded by cranes. Allowing 50 m length per ship (including space for manoeuvering), in theory up to 74 large ships could have been accommodated on the southern side of the Fossa Traiana by the marble yards. Was the other side of the canal used for barges loading up for Rome with goods from the warehouses by the Trajanic hexagon immediately to the north? Including other canals and basins, the total wharfage of Portus after the Trajanic improvements may have been as much as 13,900 m (Keay, Chapter 2: n. 64), sufficient for some 330 large ships in the two main basins and several hundred small to medium vessels berthed at the quays, as well as dozens of others waiting at anchor in the main harbour basins. For comparison, the Severan harbour at Lepcis Magna had c. 1,200 m of wharfage, which Bartoccini classified into 590 m of major wharfage (on the north and east sides of the harbour basin) and 610 m of minor wharfage (on the south and west sides) (Bartoccini 1958: 12–13). On the west side, there are flights of steps 3 m wide at intervals with 9 m between each set; if each set of steps corresponded to one berth, this would allow 12 m per ship in the minor wharfage area: c. 50 ships on the south and west sides, and perhaps another 50, or rather fewer if the ships in the major wharfage area were bigger, on the north and east, for a maximum total of around 100 ships. At Alexandria, the eastern port, or ‘Portus Magnus’ enclosed over 226 ha (precise measurement is impossible because the modern city covers the southwest part of the basin by the heptastadion). Ongoing underwater topographical research by Goddio’s team has started to reveal the extent of moles and jetties that projected into this basin to provide sheltered subharbours and extend the available wharfage length, which may be estimated at at least 12,380 m.36 The Eunostos and Kibotos harbours would have provided further docking space, increasing this figure considerably for the amount of shipping that could use Alexandria’s Mediterranean-facing harbours. The very large emporia of the Roman world thus provided purpose-built facilities for the simultaneous loading and unloading of scores or even hundreds of large ships at a time, with dockside warehouses providing entrepôt storage. These emporia were linked directly to each other by regular shipping routes, and often by established trading arrangements between groups of merchants. But they played a no less important role in articulating between the longdistance shipment of goods and local collection from and redistribution to lesser ports in their foreland regions. CONCLUSION The amphora imports to ports along the north African coast do indeed support Fulford’s idea of two largely separate trading zones with little direct interaction along the north African coast, but this is undercut to some extent both by the fine-ware evidence (ARS at Berenice in the third century; Aegean wares at Carthage) and by the evidence for eastern trading communities at ports in Tripolitania and Africa Proconsularis. The analysis of the ceramic assemblages of major north African ports suggests both that amphorae and fine-wares were distributed through different trading mechanisms, and that there was rather more connection between Africa Proconsularis and Cyrenaica than Fulford’s argument allowed. Although some of the ARS in Cyrenaica may be explained as return cargoes from major entrepôt in Italy, the evidence for Alexandrian and other eastern trading communities at some of the north African ports supports the idea of some trading connections along the north African coastline, even though the major sources of amphora-borne goods in Tripolitania and Cyrenaica were in the western and eastern Mediterranean respectively. The evidence for primary production of glass at Carthage also implies the import of natron from Egypt. The widespread distribution of relatively low-value ceramics such as ARS or African cook-wares is explained by the patterns of maritime trade based on connections between major emporia and local redistribution. We see the virtual capitulation of some regional fine-ware networks in the face of overwhelming competition from productions that are both organized on a large scale and have exceptionally good access to distribution networks through major ports. The wide pan-Mediterranean distribution and even market dominance of a few ceramic types is explained by large-scale flows of trade into key emporia, and the redistribution from these both as return cargoes to other emporia and locally along coastal hinterlands. In the case of ARS, the pan-Mediterranean distribution may be explained first by its travelling as part-cargoes into key Italian ports such as Portus, and then back to all parts of the Roman world as part of a return cargo to almost any region with which Portus was in regular ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY contact (Fentress and Perkins 1988: 213; Bonifay 2003; Fentress et al. 2004: 157–8). Direct trading links between emporia, with local coastal redistribution, also explains the distribution of Cyrenaican Mid Roman 8 amphorae up the Adriatic, and it, or specially commissioned orders, also accounts for the distribution of Italian bricks in north Africa. Analysis of harbour capacities and facilities sheds some interesting light on trading connections, and helps flesh out the picture of voyaging patterns suggested by Boetto’s analysis of wreck cargoes in Chapter 8, to give a fuller image of the nature of Mediterranean connectivity. The bigger Roman harbours and large port cities, in their infrastructure — which in north Africa as at Thapsus may include 500 m long or more jetties to enable large ships to dock on the shallow shelving coast — and their market and warehousing facilities, are entrepôt or emporia rather than just another stop in a succession of ports of call in a world of cabotage tramping. We may begin to construct hierarchies of harbour sizes and facilities, although data on both basin size and wharfage length are often incomplete. They do suggest, for example, that some of the traffic from villa harbours like Gianola or small ports such as Formia fed into regional ports like Minturnae, which might in turn form part of the network feeding major hubs like the emporia of Naples or Puteoli. But it would be false to assume that all trading voyages need necessarily work through all the steps of this hierarchy; the apparent production of dolia ships at Minturnae and its connection with the wine trade to Gaul and Spain suggest that a middling regional port like Minturnae might see voyages departing for and returning from distant emporia or indeed distant smaller ports (Schörle 2011), and the fact that villa harbours of the Roman Imperial period might compete with or even exceed in size the harbours of some cities suggests that they were capable of accommodating not just coasters but also large sea-going shipping that could trade directly with distant emporia via voyages across open water. Morley concluded his recent survey of trade in classical antiquity with the sentence: ‘The ‘decline’ of late antique trade can also be seen as a return to the normality of small-scale, short-haul cabotage after the exceptional level of activity, and exceptional degree of dependence on traded goods, in classical antiquity’ (Morley 2007: 102). The Classical, Hellenistic and, particularly, the Roman periods were marked especially by long-distance maritime trade between entrepôt, and it was this coordinated access to remote and larger 385 markets that helped stimulate high levels of urban production at port sites, not only in marine products (salt, fish products, Murex purple dye), but also amphorae, cook-ware production and metalworking, either geared to an export market or reliant on the import of raw materials (Wilson 2002). The geographical separation of primary and secondary production in the glass industry, and the consequent need for integration by trading networks, seems to confirm this relationship. 386 WILSON, SCHÖRLE & RICE NOTES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. For production at Tocra, see: Riley 1974–5; 1976–7; 1979–80. Berenice: Riley 1979: 193. Apollonia: personal observation (A. Wilson) of a sherd dump by the shore of the harbour. Distribution compiled from data in: Ferrarini 1993: 157–61; Cipriano, Mazzocchin and Paspore 1997. For Zaton: Smiljan Gluščević pers. comm. (email of 13.02.2007). Compiled from data in CIL XV and in: Bloch 1947; plus: Foucher 1958; Foucher 1965; Di Vita 1966; Hirschland and Hammond 1968; AE 2000, 1711 and 1712; Wilson 2001a. Cf. Hartley 1973; Tomber 1987. The comparison of pottery quantifications from Berenice, Carthage and Ostia presented here is explored in more detail elsewhere (Rice 2011). Indeed, for an earlier period, Kenrick (1985: 251) noted that 20–30% of the ITS at Berenice came from Puteoli. The second period, that of AD 1–125, as analysed here, consists of deposits 4.4, 4.7a, 4.7b, 4.7c, 4.7e, 4.8, 4.10, 4.12a and 4.12b. Deposits 4.6, 4.14a and 4.15c, dated to this period stratigraphically, have been removed from consideration owing to the high degree of residual pottery. The contexts from AD 125–200 analysed here are deposits 4.16a, 4.16b and 4.18. Deposits 4.12a, 4.14b and 4.17 were removed because of residual contents. The figures are based on Fontana’s quantification of amphorae from excavated contexts at Meninx: Fontana, Ben Tahar and Capelli 2009: tables 16.15, 16.18, 16.19. The amphorae figures are taken from: Carandini and Panella 1973. See also the analysis of Antonine deposits at Ostia by Rizzo in Chapter 4. The results are published in detail by Martin (2008). A head of Sarapis found in the forum at Gigthis beside the main forum temple, and a relief of scenes possibly connected with the cult of Isis and Sarapis on the podium of the temple, led to the suggestion that the main temple there also may have been dedicated to Sarapis (Constans 1916: 29–32). For some other evidence for the use of Greek by communities in port cities of north Africa, see: Desanges 1999. Kessler and Temin (2008) have argued that grain prices at Rome reflected those in the provinces plus the costs of transport, but cf.: Bowman and Wilson 2009: 24–7. Analysis of prehistoric glass may suggest the use of sodium salts and potash as a flux in iron age Europe (Henderson 1985). Other sources of natron are known today in England, Hungary, Switzerland, Italy (Campania) and Sicily (http://www.mindat. org/min–2858.html [last accessed 26.04.2011]), but there is no evidence for their exploitation in antiquity. We thank Tyler Franconi for these references. Cf. raw ingot SAP inv. 13111 (cat. no. 259): Ciarallo and De Carolis 1999: 202, no. 259; Verità 1999: 109; De Carolis 2004: 72. The identification of a supposed glass-blowing works with nine furnaces (p. 62) is, however, dubious. Lepcis Magna: Ioppolo 1969–70: 232, 234; Pisani Sartorio 1969–70: 250–5. Carthage: Tatton-Brown 1994: 288; Freestone 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 1994. Sabratha: Wilson 1999: 50. Caesarea: CIL VIII 9430 ¼ ILS 7649. CIL III 9542 for a vitriarius working at Salona; Clairmont and von Gonzenbach (1975: 58–63) for the excavation of the glass workshop and furnace at Salona. See: Foy and Nenna 2001: 107 — four workshops dated between the fourth and seventh centuries AD. Simon Keay, pers. comm.; excavations by the Portus Project of the University of Southampton and the British School at Rome (www.portusproject.org). See: Parker 1990a: 342–3; Parker 1990b; Parker 1992: 20–2; Nieto 1997; Tchernia 1997: 124–7; Arnaud 2005; Arnaud 2007; Boetto (Chapter 8); Bonifay and Tchernia (Chapter 16). These evidence-based studies contradict the assumption of Horden and Purcell (2000: 143–52, 365–70) that coastal tramping (‘cabotage’ in their terms) was as or more important than direct connections (‘le grand commerce maritime’). Bang’s view (2008: 141) that cabotage tramping was the norm is uninformed by the archaeological reality. These ideas are explored in greater depth by Schörle (2011). We used the program TakeOff Live, which enables the measurement of areas of irregular polygons from any digitized plan with a scale; harbour size estimates are therefore reliant on the accuracy of previously published maps. For Central Place Theory, see: Christaller 1933; cf. Evans and Gould 1982. CIL X 5371, from Interamna Lirenas 10 km upstream from Minturnae, is the tombstone of an architectus navalis who may have practised his trade at Minturnae or Interamna (or both). Sousse mosaic: Du Coudray la Blanchère and Gauckler 1897, 10, pl. I.6. Bardo Museum, A.6; Foucher 1960: 77–8 no. 57.169 and pl. XLIa. Foucher interpreted the mosaic as representing the unloading of lead ingots, but the objects being carried off the boat and weighed on shore in fact look nothing like Roman lead ingots in either size or shape. On the original mosaic they are yellowish brown in colour, and even in Foucher’s black-and-white photograph it is clear that they have knobbly projections. They are better understood as logs, perhaps intended for firewood, sold by weight, as Gauckler (Du Coudray la Blanchère and Gauckler 1897: 10) and Meiggs (1982: 529–30) recognized. For a colour picture, see: Fantar et al. 1994: foldout picture at 115–18. See Boetto (2010) for a first attempt to do this for Portus, but she assumes side-on mooring rather than end-on mooring (see below). Cf. also Heinzelmann (2010: 5–6) for estimates of the capacity of the river port at Ostia. For some of the concrete structures (without interpretation of their function), see: Yorke and Davidson 1985. Boetto (2010: 122 fig. 11) assumes broadside-on mooring only; the berth capacity in the Trajanic hexagon implied by this figure should be increased considerably. This is fewer than the 200þ ships estimated by Testaguzza (1970: 161), who seems simply to have divided his figure ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY 32. 33. 34. 35. 36. (2,028 m) for the total wharfage length for the hexagon by 10 m per ship. The original total length of c. 40 m is estimated from the preserved length of 35.1 m. 1,950 m less the 525 m calculated for the Darsena. Cf. Boetto 2010: 122, fig. 11. Keay, Chapter 2: n. 64, revising Testaguzza’s lower estimate (1970: 179) of 2,000 m. Calculated from the plan published by Goddio and Fabre (2008: 38). REFERENCES Abbreviations AE ¼ L’Année Épigraphique. CIL ¼ Corpus Inscriptionum Latinarum (1863–). Berlin, Georg Reimer/Walter de Gruyter. CIG ¼ Corpus Inscriptionum Graecarum. IG ¼ Inscriptiones Graecae. IRT ¼ Reynolds and Ward Perkins 1952. Aerts, A., Velde, B., Janssens, K. and Dijkman, W. (2003) Change in silica sources in Roman and post-Roman glass. Spectrochimica Acta Part B 58: 659–67. Arnaud, P. (2005) Les routes de la navigation antique. Itinéraires en Méditerranée. Paris, Éditions Errance. Arnaud, P. (2007) Diocletian’s Prices Edict: the prices of seaborne transport and the average duration of maritime travel. Journal of Roman Archaeology 20 (1): 321–36. Bang, P. (2008) The Roman Bazaar. A Comparative Study of Trade and Markets in a Tributary Empire. Cambridge, Cambridge University Press. Bartoccini, R. (1958) Il porto romano di Leptis Magna (Bollettino del Centro Studi per la Storia dell’Architettura. Supplemento 13). Rome, Azienda Beneventana Tipografica. Bloch, H. (1947) The Roman brick-stamps not published in volume XVI of the ‘Corpus Inscriptionum Latinarum’. Harvard Studies in Classical Philology 56–7: 1–128. Boetto, G. (2010) Le port vu de la mer: l’apport de l’archéologie navale à l’étude des ports antiques. In S. Keay and G. Boetto (eds), Ostia and the Ports of the Roman Mediterranean. Contributions from Archaeology and History. In H. Di Giuseppe and M. Della Riva (eds), 17th AIAC International Congress of Classical Archaeology: Meetings between Cultures in the Ancient Mediterranean. Rome, 22–26 September 2008: 112–28. Rome, Ministero per i Beni e le Attività Culturali. http://151.12.58.75/ archeologia/bao_document/articoli/9_Boetto_paper.pdf [last consulted 03.08.2012]. Bonifay, M. (2003) La céramique africaine, un indice du développement économique? Antiquité Tardive 11: 113–28. Bowman, A.K. and Wilson, A.I. (2009) Quantifying the Roman economy: integration, growth, decline? In A.K. Bowman and A.I. Wilson (eds), Quantifying the Roman Economy: Methods and Problems (Oxford Studies in the Roman Economy 1): 3–84. Oxford, Oxford University Press. 387 Brandon, C., Hohlfelder, R.L. and Oleson, J.P. (2008) The concrete construction of the Roman harbours of Baiae and Portus Iulius, Italy: the ROMACONS 2006 field season. International Journal of Nautical Archaeology and Underwater Exploration 37: 374–92. Brouquier-Reddé, V. (1992) Temples et cultes de Tripolitaine (Études d’antiquités africaines). Paris, Éditions Centre National de la Recherche Scientifique. Bruno, M. (1998) Isola del Giglio, la cava di granito del Foriano presso Giglio Porto. In P. Pensabene (ed.), Marmi antichi II: cave e tecnica di lavorazione, provenienze e distribuzione (Studi miscellanei 31): 119–40. Rome, ‘L’Erma’ di Bretschneider. Bruyère, B. (1966) Fouilles de Clysma-Qolzoum (Suez) 1930–1932 (Fouilles de l’Institut Français d’Archéologie Orientale du Caire 27). Cairo, Imprimerie de l’Institut Français d’Archéologie Orientale. Carandini, A. and Panella, C. (1973) (eds) Ostia III. Le Terme del Nuotatore. Scavo degli ambienti III, VI, VII. Scavo dell’ambiente V e un saggio nell’area SO (Studi miscellanei 21). Rome, De Luca. Carandini, A. and Panella, C. (1977) (eds) Ostia IV. Le Terme del Nuotatore. Scavo degli ambienti III, VI, VII. Scavo dell’ambiente XVI e dell’area XXV (Studi miscellanei 23). Roma, De Luca. Caruso, I., Gallavotti, D. and Aiello, M. (1991) Civitavecchia e il suo territorio. Rome, Quasar. Christaller, W. (1933) Die Zentralen Orte in Süddeutschland. Jena, Gustav Fischer. Ciampoltrini, G. and Rendini, P. (2004) Il sistema portuale dell’ager Cosanus e delle isole del Giglio e di Giannutri. In A. Gallina Zevi and R. Turchetti (eds), Le strutture dei porti e degli approdi antichi: II seminario ANSER, Roma–Ostia antica, 16–17 aprile di 2004: 127–50. Rome, Rubbettino Editore. Ciarallo, A. and De Carolis, E. (1999) (eds) Homo Faber. Natura, scienza e tecnica nell’antica Pompei (Studi della Soprintendenza Archeologica di Pompei 6). Milan, Electa. Cipriano, S., Mazzocchin, S. and Paspore, P. (1997) Nuove considerazioni sui commerci del territorio patavino in età imperiale. Analisi di alcune tipologie d’anfore da scavi recenti. Quaderni di Archeologia del Veneto 13: 99–109. Clairmont, C.W. and von Gonzenbach, V. (1975) The excavations. In C.W. Clairmont (ed.), Excavations at Salona, Yugoslavia (1969–1972): 38–82. Park Ridge (NJ), Noyes Press. Constans, L.A. (1916) Gigthis. Étude d’histoire et d’archéologie sur un emporium de la Petite Syrte. Paris, Imprimerie Nationale. De Carolis, E. (2004) Il vetro nella vita quotidiana. In M. Beretta and G. Di Pasquale (eds), Vitrum. Il vetro fra arte e scienza nel mondo romano: 71–80. Florence, Giunti. Degrassi, A. (1955) I porti romani dell’Istria. In Anthemon. Scritti di archeologia e di antichità classiche in onore di Carlo Anti: 119–69. Florence, Sansoni. Degryse, P. and Schneider, J. (2008) Pliny the Elder and Sr-Nd isotopes: tracing the provenance of raw materials for Roman glass production. Journal of Archaeological Science 35 (7): 1,993–2,000. 388 WILSON, SCHÖRLE & RICE De Rossi, G.M. (1980) Lazio meridionale. Rome, Newton Compton. Desanges, J. (1999) Quelques considérations sur l’usage du grec dans les ports de l’Afrique romaine. In M. Reddé (ed.), Toujours Afrique apporte fait nouveau. Scripta minora: 173– 9. Paris, De Boccard. (¼ Graeco-Arabica 6 (1995): 27–36.) Di Vita, A. (1966) La Villa della ‘Gara delle Nereidi’ presso Tagiura. Un contributo alla storia del mosaico romano ed altri scavi e scoperte in Tripolitania (Supplement to Libya Antiqua 2). Tripoli, Directorate-General of Antiquities, Museums and Archives. D’Oriano, R. and Riccardi, E. (2004) A lost fleet of ships in the port of Olbia. In S. Kingsley (ed.), Barbarian Seas: Late Rome to Islam. Encyclopaedia of Underwater Archaeology: 89–95. London, Periplus. Dotsika, E., Poutoukis, D., Tzavidopoulos, I., Maniatis, Y., Ignatiadou, D. and Raco, B. (2009) A natron source at Pikrolimni Lake in Greece? Geochemical evidence. Journal of Geochemical Exploration 103: 133–43. Du Coudray la Blanchère, R.M. and Gauckler, P. (1897) Catalogue du Musée Alaoui (Musées et collections archéologiques de l’Algérie et de la Tunisie). Paris, Ernest Leroux Éditeur. Evans, S. and Gould, P. (1982). Settlement models in archaeology. Journal of Anthropological Archaeology 1: 275–304. Fantar, M.H., Picard, G.-C., Ben Mansour, S., Jeddi, N., Slim, H., Khanoussi, M., Yacoub, M., Foucher, L., Béjaoui, F. and Ghedini, F. (1994) La mosaı̈que en Tunisie. Paris, Éditions Centre National de Recherche Scientifique. Felici, E. (1995) Anzio: un porto per Nerone. Archaeologia Viva 52: 56–63. Fentress, E. (2009) Review of R. Hohlfelder (ed.), The Maritime World of Ancient Rome. Proceedings of the ‘Maritime World of Ancient Rome’ Conference Held at the American Academy of Rome 27–29 March 2003 (The University of Michigan Press, Ann Arbor 2008). American Journal of Archaeology 113 (1) Online Book Reviews http://www.ajaonline.org/ online-review-book/599 [last consulted 03.08.2012]. Fentress, E. and Perkins, P. (1988) Counting African Red Slip ware. In L’Africa romana. Atti del V convegno di studio: 204–14. Sassari, Università di Sassari. Fentress, E., Fontana, S., Hitchner, R.B. and Perkins, P. (2004) Accounting for ARS: fineware and sites in Sicily and north Africa. In S.E. Alcock and J.F. Cherry (eds), Side-by-side Survey: Comparative Regional Studies in the Mediterranean World: 147–62. Oxford, Oxbow Books. Ferrarini, F. (1993) Osservazioni su due tipologie di anfore della media età imperiale da Altino. Quaderni di Archeologia del Veneto 9: 157–64. Fontana, S., Ben Tahar, S. and Capelli, C. (2009) La ceramica tra l’età punica e la tarda antichità. In E. Fentress, A. Drine and R. Holod (eds), An Island through Time: Jerba Studies 1. The Punic and Roman Periods (Journal of Roman Archaeology Supplementary Series 71): 241–327. Portsmouth (RI), Journal of Roman Archaeology. Foucher, L. (1958) Thermes romains des environs d’Hadrumète (Notes et documents, nouvelle série 1). Tunis, Institut National d’Archéologie et d’Arts. Foucher, L. (1960) Inventaire des mosaı̈ques. Feuille no. 57 de l’atlas archéologique: Sousse. Tunis, Imprimerie Officielle. Foucher, L. (1965) La Maison des Masques à Sousse. Fouilles 1962–1963 (Notes et documents, nouvelle série 6). Tunis, Institut d’Archéologie. Foy, D. (2003a) Les ateliers primaires syro-palestiniens. In D. Foy (ed.), Cæur de verre: production et diffusion du verre antique: 28–31. Gollion, Infolio. Foy, D. (2003b) Une chaı̂ne de fabrication segmentée. In D. Foy (ed.), Cæur de verre: production et diffusion du verre antique: 26–7. Gollion, Infolio. Foy, D. (2003c) Quid de l’occident? In D. Foy (ed.), Cæur de verre: production et diffusion du verre antique: 34–5. Gollion, Infolio. Foy, D. and Nenna, M.-D. (2001) Tout feu tout sable: mille ans de verre antique dans le Midi de la France. Marseilles, Edisud. Franco, L. (1996) History of coastal engineering in Italy. In N.C. Kraus (ed.), History and Heritage of Coastal Engineering: a Collection of Papers on the History of Coastal Engineering in Countries Hosting the International Coastal Engineering Conference 1950–1996: 275–335. New York, ASCE Publications. Freestone, I.C. (1994) Appendix: chemical analysis of ‘raw’ glass fragments. In H.R. Hurst (ed.), Excavations at Carthage. The British Mission II.1. The Circular Harbour, North Side (British Academy Monograph in Archaeology 5): 290. Oxford, Oxford University Press for the British Academy. Fremersdorf, F. (1965–6) Die Anfänge der Römischen Glashütten Kölns. Kölner Jahrbuch für Vor- und Frühgeschichte 8: 24–43. Fulford, M.G. (1987) Economic interdependence among urban communities of the Roman Mediterranean. World Archaeology 19 (1): 58–75. Fulford, M.G. (1989) To east and west: the Mediterranean trade of Cyrenaica and Tripolitania in antiquity. In D.J. Mattingly and J.A. Lloyd (eds), Libya: Research in Archaeology, Environment, History and Society 1969–1989. Libyan Studies 20: 169–91. Fulford, M.G. (1994) Quantification of the pottery. In M.G. Fulford and D.P.S. Peacock (eds), Excavations at Carthage. The British Mission II, 2. The Circular Harbour, North Side, The Pottery (British Academy Monograph in Archaeology 5): 97–114. Oxford, Oxford University Press for the British Academy. Gaitzsch, W., Follmann-Schulz, A.-B., Wedepohl, K.-H., Hartmann, G. and Tegtmeier, U. (2000) Spätrömische Glashütten im Hambacher Forst — Produktionsort der ECVA-Fasskrüge. Archäologische und naturwissenschaftliche Untersuchungen. Bonner Jahrbücher 200: 83–241. Gazda, E.K. (1987) The port and fishery: description of the extant remains and sequence of construction. In A.M. McCann (ed.), The Roman Port and Fishery at Cosa: a Center of Ancient Trade: 74–97. Princeton, Princeton University Press. ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY Goddio, F. (2007) The Topography and Excavation of HeracleionThonis and East Canopus (1996–2006) (Underwater Archaeology in the Canopic Region in Egypt). Oxford, Oxford Centre for Maritime Archaeology. Goddio, F. and Fabre, D. (2008) (eds) Egypt’s Sunken Treasures. Munich/Berlin/London/New York, Prestel. Göransson, K. (2007) The Transport Amphorae from Euesperides. The Maritime Trade of a Cyrenaican City 400–250 BC (Acta Archaeologica Lundensia 25). Lund, Lunds Universitet. Hartley, K.F. (1973) La diffusion des mortiers, tuiles et autres produits en provenance des fabriques italiennes. Cahiers d’Archéologie Subaquatique 2: 49–60. Heinzelmann, M. (2010) Supplier of Rome or Mediterranean marketplace? The changing economic role of Ostia after the construction of Portus in the light of new archaeological evidence. In S. Keay and G. Boetto (eds), Ostia and the Ports of the Roman Mediterranean. Contributions from Archaeology and History. In H. Di Giuseppe and M. Della Riva (eds), 17th AIAC International Congress of Classical Archaeology: Meetings between Cultures in the Ancient Mediterranean. Rome, 22–26 September 2008: 5–10. Rome, Ministero per i Beni e le Attività Culturali. http://151.12.58.75/archeologia/bao_document/articoli /2_Heinzelmann_paper.pdf [last consulted 03.08.2012]. Henderson, J. (1985) The raw materials of early glass production. Oxford Journal of Archaeology 4 (3): 267–91. Higginbotham, J. (1997) Piscinae. Artificial Fishponds in Roman Italy. Chapel Hill, University of North Carolina Press. Hirschland, N.L. and Hammond, M. (1968) Stamped potters’ marks and other stamped pottery in the McDaniel collection. Harvard Studies in Classical Philology 72: 369–82. Hohlfelder, R.L. (1999) Building Sebastos: the Cyprus connection. International Journal of Nautical Archaeology and Underwater Exploration 28: 154–63. Horden, P. and Purcell, N. (2000) The Corrupting Sea. A Study of Mediterranean History. Oxford, Blackwell. Hurst, H. (2010) Understanding Carthage as a Roman port. In S. Keay and G. Boetto (eds), Ostia and the Ports of the Roman Mediterranean. Contributions from Archaeology and History. In H. Di Giuseppe and M. Della Riva (eds), 17th AIAC International Congress of Classical Archaeology: Meetings between Cultures in the Ancient Mediterranean. Rome, 22–26 September 2008: 49–68. Rome, Ministero per i Beni e le Attività Culturali. http://151.12.58.75/archeologia/bao_document/ articoli/6_Hurst_paper.pdf [last consulted 03.08.2012]. Ioppolo, G. (1969–70) Introduzione all’indagine stratigrafica presso l’arco di Marco Aurelio a Leptis Magna (saggi del 1959 e del 1964). Libya Antiqua 6–7: 231–6. Jézégou, M.-P. (2008) L’épave Ouest-Embiez 1: proposition d’un modèle de réexportation de produits verriers et du vin à la charnière des IIe/IIIe siècles après J.-C. In J. Pérez Ballester and G. Pascual (eds), Comercio, redistribución y fondeaderos. La navegación a vela en el Mediterráneo. Actas V jornadas internacionales de arqueologı́a subacuática (Gandı́a, 8 a 10 de noviembre de 2006): 451–60. Valencia, Universitat de València. 389 Kater-Sibbes, G. (1973) Preliminary Catalogue of Sarapis Monuments. Leiden, E.J. Brill. Kenrick, P.M. (1985) Patterns of trade in fine pottery at Berenice. In G.W.W. Barker, J.A. Lloyd and J.M. Reynolds (eds), Cyrenaica in Antiquity (British Archaeological Reports, International Series 236): 249–57. Oxford, British Archaeological Reports. Kenrick, P.M. (1987) Patterns of trade at Berenice: the evidence of the fine wares. Rei Cretariae Romanae Fautorum Acta 25–6: 137–54. Kessler, D. and Temin, P. (2008) Money and prices in the early Roman Empire. In W.V. Harris (ed.), The Monetary Systems of the Greeks and Romans: 137–59. Oxford, Oxford University Press. Kingsley, S. (2004) (ed.) Barbarian Seas: Late Rome to Islam. (Encyclopaedia of Underwater Archaeology). London, Periplus. Lancaster, L.C., Sottili, G., Marra, F. and Ventura, G. (2010) Provenancing of lightweight volcanic stones used in ancient Roman concrete vaulting: evidence from Turkey and Tunisia. Archaeometry (2010) http://dx.doi.org/10.1111/j.1475–4754. 2009.00509. [last consulted 03.08.2012] Liou, B. and Pomey, P. (1985) Direction des recherches archéologiques sous-marines. Gallia 43: 547–76. Martin, A. (2008) Imports at Ostia in the Imperial period and late antiquity: the amphora evidence from the DAI–AAR excavations. In R.L. Hohlfelder (ed.), The Maritime World of Ancient Rome: 105–18. Ann Arbor, University of Michigan Press. Marzano, A. (2007) Roman Villas in Central Italy: a Social and Economic History (Columbia Studies in the Classical Tradition 30). Leiden/Boston, Brill. Meiggs, R. (1982) Trees and Timber in the Mediterranean World. Oxford, Clarendon Press. Mommsen, T. (1843) De Collegiis et Sodaliciis Romanorum. Kiel, Libraria Schwersiana. Morelli, C., Marinucci, A. and Arnoldus-Huyzendveld, A. (2011) Il porto di Claudio: nuove scoperte. In S. Keay and L. Paroli (eds), Portus and its Hinterland. Recent Archaeological Research (Archaeological Monographs of the British School at Rome 18): 47–65. London, British School at Rome. Morley, N. (2007) Trade in Classical Antiquity. Cambridge, Cambridge University Press. Nenna, M.-D. (2003) Les atéliers égyptiens à l’époque grécoromaine. In D. Foy (ed.), Cæur de verre: production et diffusion du verre antique: 32–3. Gollion, Infolio. Nenna, M.-D. (2007) Production et commerce du verre à l’époque impériale. Facta 1: 125–48. Nieto, X. (1997) Le commerce de cabotage et de redistribution. In P. Pomey (ed.), La navigation dans l’Antiquité: 146–59. Aixen-Provence, Édisud. Oleson, J.P. (1988) The technology of Roman harbours. International Journal of Nautical Archaeology and Underwater Exploration 17: 147–57. Oleson, J.P. and Branton, G. (1992) The technology of King Herod’s harbour. In R.L. Vann (ed.), Caesarea Papers. Straton’s Tower, 390 WILSON, SCHÖRLE & RICE Herod’s Harbour, and Roman and Byzantine Caesarea (Journal of Roman Archaeology Supplementary Series 5): 49–67. Ann Arbor, Journal of Roman Archaeology. Parker, A.J. (1990a) Classical antiquity: the maritime dimension. Antiquity. A Quarterly Review of Archaeology 64 (243): 335–46. Parker, A.J. (1990b) The pattern of commerce as evidenced by shipwrecks. In T. Hackens and M. Miró (eds), Le commerce maritime romain en Méditerranée occidentale. Colloque international tenu à Barcelone, Centre Européen pour le Patrimoine Culturel du 16 au 18 mai 1988. PACT 27: 147– 68. Strasbourg, Conseil de l’Europe. Parker, A.J. (1992) Ancient Shipwrecks of the Mediterranean and the Roman Provinces (British Archaeological Reports, International Series 580). Oxford, Tempus Reparatum. Parker, A.J. (2008) Artefact distributions and wreck locations: the archaeology of Roman commerce. In R.L. Hohlfelder (ed.), The Maritime World of Ancient Rome: 177–96. Ann Arbor, Michigan University Press. Peña, J.T. (1998) The mobilization of state olive oil in Roman Africa: the evidence of late 4th-c. ostraca from Carthage. In Carthage Papers: the Early Colony’s Economy, Water Supply, a Private Bath, and the Mobilization of State Olive Oil (Journal of Roman Archaeology Supplementary Series 28): 117–238. Portsmouth (RI), Journal of Roman Archaeology. Pisani Sartorio, G. (1969–70) Arco quadrifronte di Marco Aurelio a Leptis Magna. Saggio stratigrafico presso il pilone sud. Libya Antiqua 6–7: 237–79. Pomey, P. (1982) Le navire romain de la Madrague de Giens. Comptes Rendus des Séances de l’Académie des Inscriptions et Belles-Lettres (janvier–mars 1982): 133–54. Rey-Coquais, J.-P. (1987) Une double dédicace de Lepcis Magna à Tyr. In A. Mastino (ed.), Atti del IV convegno di studio, Sassari, 12–14 dicembre 1986 (L’Africa Romana 4): II, 597–602. Sassari, Dipartimento di Storia, Università degli Studi di Sassari. Reynolds, J.M. and Ward-Perkins, J.B. (1952) The Inscriptions of Roman Tripolitania. London, British School at Rome. Rice, C. (2011) Connectivity and ports. In D. Robinson and A. Wilson (eds), Maritime Archaeology and Ancient Trade in the Mediterranean: 81–92. Oxford, Oxford Centre for Maritime Archaeology. Rickman, G.E. (1985) Towards a study of Roman ports. In A. Raban (ed.), Harbour Archaeology: Proceedings of the First International Workshop on Ancient Mediterranean Harbours, Caesarea Maritima, 24–28.6.83 (British Archaeological Reports, International Series 257): 105–14. Oxford, British Archaeological Reports. Riley, J.A. (1974–5) Excavations of a kiln site at Tocra Libya in August 1974. Libyan Studies 6: 25–9. Riley, J.A. (1976–7 [1983]) Excavation of a kiln site at Tocra, Libya in August 1974. Libya Antiqua 13–14: 235–63. Riley, J.A. (1979) The coarse pottery from Berenice. In J.A. Lloyd (ed.), Excavations at Sidi Khrebish Benghazi (Berenice), vol. 2 (Supplements to Libya Antiqua 5): 91–467. Tripoli, Department of Antiquities. Riley, J.A. (1979–80) The excavation at Tocra 1974: additional observations in the light of the Berenice excavations. Libyan Studies 11: 53–64. Rives, J.B. (1995) Religion and Authority in Roman Carthage from Augustus to Constantine. Oxford, Clarendon Press. Romanelli, P. (1925) Leptis Magna (Africa Italiana 1). Rome, Società Editrice d’Arte Illustrata. Rottländer, R.C.A. (1990) Naturwissenschaftliche Untersuchungen zum Römischen Glas in Köln. Kölner Jahrbuch für Vor- und Frühgeschichte 23: 563–82. Ruegg, D. (1988) Minturnae: a Roman river seaport on the Garigliano river. In A. Raban (ed.), Archaeology of Coastal Changes (British Archaeological Reports, International Series 404): 209–28. Oxford, British Archaeological Reports. Schörle, K. (2011) Constructing port hierarchies: harbours of the central Tyrrhenian coast. In D. Robinson and A. Wilson (eds), Maritime Archaeology and Ancient Trade in the Mediterranean: 93–106. Oxford, Oxford Centre for Maritime Archaeology. Swift, K. (2006) Classical and Hellenistic Coarse Pottery from Euesperides (Benghazi, Libya): Archaeological and Petrological Approaches to Production and Inter-Regional Distribution. University of Oxford, D.Phil. thesis. Tatton-Brown, V. (1994) The glass. In H.R. Hurst (ed.), Excavations at Carthage. The British Mission II.1. The Circular Harbour, North Side (British Academy Monograph in Archaeology 5): 282–8. Oxford, Oxford University Press for the British Academy. Tchernia, A. (1997) Le commerce maritime dans la Méditerranée romaine. In P. Pomey (ed.), La navigation dans l’antiquité: 116–45. Aix-en-Provence, Édisud. Testaguzza, O. (1970) Portus. Illustrazione dei porti di Claudio e Traiano e della città di Porto a Fiumicino. Rome, Julia Editrice. Tomber, R. (1987) Evidence for long-distance commerce. Imported bricks and tiles at Carthage. Rei Cretariae Romanae Fautorum Acta 25–6: 161–74. Verità, M. (1999) Le sabbie e il vetro. In A. Ciarallo and E. De Carolis (eds), Homo Faber. Natura, scienza e tecnica nell’antica Pompei (Studi della Soprintendenza Archeologica di Pompei 6): 108–10. Milan, Electa. Votruba, G.F. (2007) Imported building materials of Sebastos Harbour, Israel. International Journal of Nautical Archaeology and Underwater Exploration 36: 325–35. Wedepohl, K. and Baumann, A. (2000) The use of marine molluskan shells for Roman glass and local raw glass production in the Eifel area (western Germany) Naturwissenschaften 87: 129–32. Wilson, A.I. (1999) Commerce and industry in Roman Sabratha. Libyan Studies 30: 29–52. Wilson, A.I. (2001a) Ti. Cl. Felix and the date of the second phase of the East Baths. In L.M. Stirling, D.J. Mattingly and N. Ben Lazreg (eds), Leptiminus (Lamta): a Roman Port City in Tunisia. Report no. 2 (Journal of Roman Archaeology, Supplementary Series 41): 25–8. Ann Arbor, Journal of Roman Archaeology. ROMAN PORTS AND MEDITERRANEAN CONNECTIVITY Wilson, A.I. (2001b) The water-mills on the Janiculum. Memoirs of the American Academy in Rome 45: 219–46. Wilson, A.I. (2002) Urban production in the Roman world: the view from north Africa. Papers of the British School at Rome 70: 231–73. Wilson, A.I. (2005) Une cité grecque de Libye: fouilles d’Euesperides (Benghazi). Comptes Rendus des Séances de l’Académie des Inscriptions et Belles-Lettres (nov.–déc. 2003): 1,648–75. Wilson, A.I. (forthcoming) Trading across the Syrtes: Euesperides and the Punic world. In J. Quinn and J. Prag (eds), The Hellenistic West. Cambridge, Cambridge University Press. Yorke, R.A. and Davidson, D.P. (1985) Survey of building techniques at the Roman harbours of Carthage and some other north African ports. In A. Raban (ed.), Harbour Archaeology: Proceedings of the First International Workshop on Ancient Mediterranean Harbours, Caesarea Maritima, 24–28.6.83 (British Archaeological Reports, International Series 257): 157–64. Oxford, British Archaeological Reports. 391