The economic connections of the Greek island of Crete under Roman rule have been a significant point of scholarly focus over the past few decades.Footnote 1 Of central importance in these discussions is the distribution of products packaged in amphoras manufactured on the island, which serve as the key archaeological indicator of Crete's export economy.Footnote 2 For the 1st–3rd c. CE, there are four amphora types manufactured on Crete, the Amphores Crétoise (AC) 1–4, relevant to discussions of economic connectivity. While each type is attested at various sites in the Mediterranean basin, the AC4 is the primary, and often the only, Cretan type found in Rome's northwestern provinces and along the Danube frontier. This paper examines the distinctive distribution patterns of AC4 amphoras within Roman trade networks through critical assessment of the morphological attributes of this amphora type compared to AC1–3 jars and through consideration of the mechanisms that underlie these patterns. Following an overview of the attributes and distribution of products packaged in AC4 amphoras, the discussion turns to river and overland transport and the benefits of the AC4 for this type of trade. A brief overview of Atlantic trade as a potential alternative means of distribution is also undertaken. This analysis, overall, builds on a growing number of studies that have focused on the design attributes of amphoras as important factors tied to their economic role.Footnote 3 It also demonstrates the importance of engaging in more nuanced and detailed investigations that question assumptions about amphora distribution within the Roman world.
What is distinct about the AC4?
The AC4 amphora, also known as the Dressel 43, is among the most recognizable Early Roman Cretan types (Fig. 1). John Hayes connected the AC4 with production on Crete following analysis of finds from the Villa Dionysos at Knossos, and manufacture is now documented at four sites on the island: Dermatos, Heraklion, Hierapytna, and Tsoutsouros-East.Footnote 4 In her study of amphora production on Crete, Antigone Marangou-Lerat identified three variants, which corresponded to different production sites (AC4a – Heraklion; AC4b – Dermatos; AC4c – Tsoutsouros-East).Footnote 5 At Hieraptyna, production may have included both the AC4a and AC4b. The AC4a has a button point, narrow ovoid body, vertical handles with prominent horns at the top that terminate above the rim, a swollen neck, and a folded, triangular rim. AC4b vessels have the same body shape, handles, and rim as the AC4a, but the neck is cylindrical. The AC4c has the same body as the above two variants, and a neck and rim comparable to the AC4b, but the handles have a smaller horn that tends to terminate below the rim.
Fig. 1. AC1–4 Amphoras. (After Marangou-Lerat 1995, fig. 29, fig. 30:A19, fig. 37:A37, fig. 42:A57, fig. 52, fig. 58:d, fig. 62:c, fig. 64, fig. 69, figs. 74–75.)
Marangou-Lerat argued that production of the AC4 began in the early 1st c. CE and continued into the first half of the 3rd c.Footnote 6 A later publication based on finds at the Cretan site of Gortyn suggested a more contracted chronology (mid-1st c. CE to mid-2nd c. CE), but there is additional data to support Marangou-Lerat's interpretation.Footnote 7 For the beginning of production, Marangou-Lerat cited finds of AC4 vessels at Vindonissa, Switzerland from a context of the first half of the 1st c. CE, and from two deposits excavated at the Meta Sudans in Rome, one dated to 20–30 CE and the other to 14–40 CE.Footnote 8 In addition, the assemblage of the La Chrétienne H shipwreck near Saint-Raphaël, France, dated to 15–20 CE, included one AC4 container.Footnote 9
Evidence for small-scale exports this early into the 1st c. CE may signal that production of the AC4 type began by the latter part of the 1st c. BCE, even earlier than suggested by Marangou-Lerat. The AC4 developed from the Rhodian amphora tradition, and in several publications these vessels are even mistaken as coming from Rhodes.Footnote 10 This is an example of a shared regional style, and Crete had strong economic connections with the southeastern Aegean, including Rhodes, throughout the Hellenistic period as the island helped to facilitate trade between that region and Egypt.Footnote 11 In eastern Crete, an amphora type described as East Crete type 2, datable from the late 2nd c. through 1st c. BCE, is interpreted as an imitation of Rhodian amphoras.Footnote 12 Petrographic analysis of the clay fabric of East Crete type amphoras indicates they were manufactured within the Hellenistic territory of Hierapytna, one of the documented locations of AC4 production.Footnote 13 Unfortunately, surviving examples of the East Crete type 2 are limited to two stamped handles found at the site of Mochlos in northeast Crete. Neither comes from a well-dated context, making it challenging to assess the overall morphology and the chronology of this type. This also restricts identification of when vessels specifically identifiable as AC4 amphoras appeared, but the above evidence suggests that this had occurred at least by the late 1st c. BCE/early 1st c. CE.
In her assessment of the end of production, Marangou-Lerat highlighted 38 fragments of AC4 vessels recovered from contexts dated between 230 and 250 CE at Ostia.Footnote 14 Additional finds from 3rd-c. deposits at other sites indicate that production may have continued into that century. This includes Rome, where small numbers of AC4 jars are attested in early 3rd-c. deposits.Footnote 15 At Viminacium (modern Kostolac) in Serbia, AC4 amphoras appear predominately in contexts datable from the second half of the 2nd c. to the early 3rd c. CE.Footnote 16 AC4 amphoras also account for 11 percent of eastern amphoras recovered at Lyon from deposits dated from 190 to 250 CE.Footnote 17 At Colchester, England, a single AC4 container was recovered from a deposit dated to 225–250 CE and a second from a deposit of 250–275 CE.Footnote 18 Additional evidence comes from Corinth, Greece, where AC4 jars account for 9.4 percent of amphoras by count and approximately 5.9 percent by weight from a dump dated between 250 and 300 CE.Footnote 19 While it is possible that some proportion of AC4 amphoras at any or all of these sites are residual in 3rd-c. deposits or represent vessels that were in use for extended periods of time, or were perhaps even in a state of re-use, a hypothesis that manufacture continued into the first half of the 3rd c. is not unreasonable.Footnote 20
The AC4, along with AC1–3 jars, is part of a tradition of ovoid amphoras that rose to prominence across the Mediterranean from the 2nd c. BCE through the 1st c. CE.Footnote 21 According to Enrique Garcίa Vargas and collaborators, the shape, chronological and geographical range, and historical significance of ovoid amphoras suggest that “they make up a kind of homogenous family.”Footnote 22 This “family” has different branches, including containers inspired by Koan vessels, such as the AC2.Footnote 23 Rhodian amphoras and their offshoots form another branch. Chronologically, there is some overlap between the various Early Roman Cretan types. The AC2 and AC3, analogous to the AC4, have connections to Hellenistic traditions on Crete and likely arose as independent types during the latter part of the 1st c. BCE or the early 1st c. CE.Footnote 24 Both continued to be manufactured through the 2nd c. CE.Footnote 25 The AC1 first appears by the mid-1st c. CE and was produced at least into the late 3rd c.Footnote 26
A comparison of AC4 versus AC1–3 amphoras shows several similarities along with several key differences (Table 1). All four types, for example, tend to be thin walled but well fired. This ensured a favorable capacity-to-empty-weight ratio and the possibility of packaging the greatest volume of liquid in the lightest possible container.Footnote 27 The overall morphology of the AC1–3 is also consistent with AC4 containers.Footnote 28 They typically comprise a rounded base, often with a button toe, an ovoid body, and a cylindrical neck in the case of AC1 and AC2 amphoras, or a swollen neck for AC3s. Rim profiles vary across each type, which contributes to the identification of particular sub-types. AC2b vessels also tend to have a distinct ledge on the shoulder. Average height is comparable across the four types.Footnote 29 A distinguishing feature between Early Roman Cretan amphoras, on the other hand, is the shape of handles. The AC1 has handles that are smaller than the other types and lack the more triangular outline of those characteristic of the AC2 and AC3. Handles on the AC1 and AC3 tend to be round in profile, while on the AC2 they are either pseudo-bifid or double-round. None of these types, however, has examples of the horns seen on AC4 handles.
Table 1. Comparison of characteristics of AC1–4 amphoras.
A significant point of distinction between the AC4 and the AC1–3 is their capacity. The AC4 is the smallest of the four Early Roman Cretan types with an average capacity of 10.9 L.Footnote 30 Marangou-Lerat also reports that two half-size AC4s have been documented on Crete, with a capacity of 5–6 L.Footnote 31 For the other three types, average capacity is more consistent and approximately double that of a standard AC4.Footnote 32 Some half-size examples of the AC3 have also been identified, one from the sea near the coastal site of Kaloi Limenes in south-central Crete and a second at Pompeii.Footnote 33 Smaller capacity meant, in addition, that the AC4 tends to have a narrower overall diameter than the other three types.Footnote 34
Where are AC4 amphoras found?
As excavations continue and classification of AC amphoras becomes more refined, distribution maps are regularly augmented. This section discusses the current state of the question. The distribution of products packaged in AC4 amphoras was more extensive than that of goods in AC1–3 containers. Each type is well attested at sites across the Mediterranean basin, in some cases in quite large quantities, but the AC4 is the only one to appear regularly along the Danube frontier and in Rome's northwestern provinces. This pattern was first revealed by Marangou-Lerat in her monograph on Cretan amphoras. An important element of her work was the creation of distribution maps that showed sites across the Roman Empire where Cretan amphoras had been attested.Footnote 35 She created an individual map for each of the Early Roman types and an additional map that combined their distributions. In a later study, she published an updated comprehensive map.Footnote 36 As datasets from different sites continue to be published, this has enabled further modification of distribution maps for Cretan amphoras.Footnote 37
While distribution maps like those created by Marangou-Lerat and later scholars can provide important aids for investigating economic patterns across the Roman world, they do have limitations. One issue is recognizing that these maps are never complete and require constant updating. Any conclusions must be taken with caution since new data could provide alternative perspectives. Most of the maps representing distribution of Cretan amphoras are also hindered by the fact that they only contain dots marking known findspots. Important variables like the chronology of the finds and the quantity of Cretan amphoras at a given site are not documented. This can lead to the false assumption that all dots are equal even though the number of vessels found at some sites may be far larger than at most of the sites represented on the map.
To populate these maps with findspots presents the additional challenge of reviewing significant numbers of published reports, looking for attested examples of specific types. Many publications pre-date the development of typologies for various amphoras, and some act as the source of a particular typology. This has led to several names being applied to individual types. The AC1, for instance, has been described in different publications as the Agora G197, Mid-Roman Amphora 2, Peacock and Williams class 41, Schöne-Mau 10, Cretan 1, etc. A search for findspots requires looking for examples under each potential name. Misidentification can be another issue, such as the challenge noted above where AC4 amphoras have been misattributed as Rhodian.Footnote 38 Detailed catalogue descriptions and illustrations are vital for assessing the identification of specific vessels. There is also the potential issue of including local or regional imitations of widely distributed types, which may only become apparent after archaeometric analysis. For instance, an amphora that morphologically appears to be an AC4 or very closely related type recovered from a villa in the area of Venice, Italy was assigned a local provenance following petrographic analysis.Footnote 39 None of these issues invalidate the benefits of creating distribution maps for amphora types, but they do point to the importance of working with multiple variables to tease out novel perspectives.
Marangou-Lerat's original overarching distribution maps demonstrated the benefits of not simply placing generic dots on a map but instead focusing on particular amphora types. Her most informative map was one that showed the distribution of AC1–3 amphoras alongside the distribution of AC4 vessels.Footnote 40 This organizational strategy reflected the distinct morphology of the AC4 and helped to highlight its unique presence in certain regions compared to the other three types. Rita Auriemma and Elena Quiri, in a later study, also emphasized the distinction in distribution between the Early Roman Cretan types and produced maps to help illustrate this.Footnote 41 They did not elaborate on a reason behind this pattern, however. The number of sites where Cretan amphoras have been documented has grown substantially in the three decades since both Marangou-Lerat and Auriemma and Quiri's studies, and it is necessary to see if the addition of new data continues to emphasize the same dichotomy in distribution. A revised map for AC1–3 vessels (Fig. 2) shows a few more inland European sites, while distribution continues to focus around the Mediterranean. For AC4 amphoras (Fig. 3), along with a well-documented presence at Mediterranean sites, distribution of these containers and their contents throughout Rome's northwestern provinces and along the Danube frontier is even more apparent.
Fig. 2. Distribution of AC1–3 Amphoras, 1st to 3rd c. CE. Sites represented on the map: Abusir, Abyar el Njam, Alexandria, Alicante, Antikythera, Aquileia, Argos, Ashekelon, Athens, Barzan, Berenice, Boscoreale, Braives, Brescia, Brindisi, Butrint, Caesarea, Capomulini, Carunutum, Carthage, Cassandra, Chersonesos, Corinth, Cremona, Cyrene, Demetrias, Didyma, Dreamer's Bay, Durres, Ephesus, Eretria, Herculaneum, Isthmia, Istria, Leptis Magna, Lipari, Lyon, Magdalensberg, Mainz, Marina el-Alamein, Merida, Milan, Miletus, Naples, Nettuno, Nora, Novae, Novara, Olympia, Oplontis, Ostia, Padua, Paphos, Perissa, Pisa, Pompeii, Port-la-Nautique, Porto Recanati, Puteoli, Pyrgi, Rome, San Foca, Schedia, Settefinestre, Stabiae, Susa, Tebtynis, Tenos, Thasos, Thebes, Trieste, Troy, Turin, Verona, Veštar Port, Vicenza, Zadar. (Map by S. Gallimore.)
Fig. 3. Distribution of AC4 Amphoras, 1st to 3rd c. CE. Sites represented on the map: Adamclisi, Adony Fort, Aegyssus, Aenona, Ajdovšina, Alexandria, Altinum, Antikythera, Apollonia Pontica, Aquileia, Arras, Augst, Ausburg, Bavay, Berenice, Belgrad, Braives, Brescia, Brijuni, Brindisi, Caerleon, Caesarea, Capua, Carnuntum, Carthage, Celje, Chersonesos, Cioroiu Nou, Colchester, Corinth, Cremona, Delos, Dobruja, Dreamer's Bay, Emona, Ephesus, Fishbourne, Fos, Frejus, Gonio-Apsarus, Haltern, Herculaneum, Istres, Kempten, Kostol, Kostolac, Kurvingrad, Leptis Magna, Lipari, London, Luni, Lyon, Magdalensberg, Mainz, Mala Vrbica, Margherita di Savoie, Marina el-Alamein, Marseille, Mesola, Milan, Miletus, Mons Claudianus, Naples, Nijmegen, Novae, Novara, Nuragha Losa di Abbasanta, Oderzo, Olympia, Orange, Orlea, Ostia, Ottagono, Pola, Pompeii, Porto Recanati, Potenza, Ptuj, Rimini, Rome, Sabratha, Samos, San Foca, Settefinestre, Seville, Sexaginta Prista, Sopron, Sparta, Stabiae, Strasbourg, Sybaris, Sympherpol, Szombathely, Tebtynis, Thasos, Tongeren, Tortosa, Troesmis, Tulcea, Turin, Vada Volaterrana, Veliko Tarnovo, Verulamium, Vienna, Voorburg, Windisch, Xanten, Zadar, Zalalöv. (Map by S. Gallimore.)
Why is the distribution pattern of AC4 amphoras different than AC1–3 amphoras?
The distribution pattern for AC4 amphoras clearly differs from that of AC1–3 containers, and the number of sites involved confirms that this is not a random phenomenon. Instead, it must be the result of certain factors that are poorly defined at present and require further investigation. The distinct characteristics of AC4 vessels, particularly their smaller capacity and narrower profile, are important variables to consider as informing these patterns. Marangou-Lerat suggested that the smaller capacity of this type points to use as packaging for higher-quality wine.Footnote 42 Unfortunately, she did not provide data to support that hypothesis and available evidence is limited. An AC4a amphora found in a mid-1st c. CE deposit at Caerleon, Britain, bears a painted inscription on the shoulder that has been reconstructed as: III/AACII/(passum) perprimum.Footnote 43 The latter part may translate as “first-grade raisin wine.” This reading is not secure, however. Perprimum is otherwise unattested in Latin and interpreting it as an assessment of quality is tied to the fact that grading wine in tiers occurred within the Roman world.Footnote 44 Pliny the Elder (HN 14.82), for instance, described secundarium passum as a second-grade version of raisin wine where water was added to the grape skins after pressing. Even if perprimum does refer to quality, such a designation on an amphora could have been more of an attempt at marketing than an accurate reflection of its contents. Perhaps comparable is the use of the phrase Cret(icum vinum) Exc(ellens) in a painted inscription on an AC2 container found at Pompeii.Footnote 45 Even if AC4 amphoras did contain higher-quality vintages, which seems unlikely, that variable by itself does not explain their near monopoly among Cretan amphora types at sites outside of the Mediterranean basin.
The attraction of Cretan products north of Mediterranean
While this study focuses to a large extent on the morphology of AC4 amphoras as a reason behind their presence north of the Mediterranean, there must always be concern in amphora studies that these were packaging containers for particular contents. Available evidence indicates that raisin wine (passum in Latin; γλυκύς in Greek) was the primary contents of Cretan amphoras.Footnote 46 Cretan passum was distributed widely across the Roman Empire and appears in large quantities, based on amphora evidence, at sites like Rome, as will be discussed below. There were several reasons behind this popularity, including the low cost of the product, its stability (i.e., resistance to transforming into vinegar), its use in cooking (as passum is a common ingredient in many Roman recipes), and its medicinal properties.Footnote 47 Such attributes had broad appeal and help to explain the presence of AC4 containers at both civilian and military settlements in the northern provinces. It is also likely that Cretan passum was consumed primarily by members of the lower classes as a product that was low cost and could be used in multiple ways.Footnote 48
The draw of Cretan products like passum to Rome's northern provinces, analogous to other goods from the eastern Mediterranean, was tied to the growth of economic networks in the Late Republic and Early Empire. In part, the development of such networks was driven by the importance of ensuring a steady supply of products to the thousands of soldiers stationed along the Rhine and Danube frontiers, or in outlying regions like Britannia. Greg Woolf argues that this helped contribute to a “consumer revolution” in this region.Footnote 49 The rapid appearance of various types of Roman material culture both before and after conquest was, in part, connected to this consumer revolution. Early in this process, merchants relied heavily on Mediterranean products, particularly those packaged in amphoras.Footnote 50 Diana Dobreva has noted the same phenomenon for the provinces of Moesia Inferior and Thrace during the Late Hellenistic and Early Roman periods.Footnote 51 Production and distribution of these goods had developed during the Hellenistic period, and they were readily available to mobilize to different locations. Reliance on commodities imported from the Mediterranean shifted over time as local and regional markets at both military and civilian sites stimulated trade in equivalent regional goods.Footnote 52
Contextualizing the quantity of AC4 amphoras north of the Mediterranean
Distribution maps illustrate that the AC4 is much more common at sites north of the Mediterranean than the other Early Roman Cretan types. An important consideration, however, is that the overall number of attested vessels is low despite the number of identified sites being high. For many of these sites, a single AC4 container has been attested in the archaeological record. In Pannonia, as an example, the AC4 is considered the rarest imported amphora type.Footnote 53 When the number of AC4s identified at each site is considered (Fig. 4), some interesting patterns become clear, including a concentration of these amphoras around Italy. Using Rome as an example, during the mid-1st c. CE, Cretan amphoras account for approximately 5.0 percent of overall amphora finds, with the AC4 the most commonly attested type.Footnote 54 Cretan amphoras found at Rome reach their highest quantities by the mid-2nd c., when they account for approximately 11.83 percent of overall amphoras.Footnote 55 Given the size of the market at Rome for amphora-borne goods, this points to significant numbers of Cretan containers being brought into the city.Footnote 56
Fig. 4. Quantity of AC4 Amphoras attested by site, 1st to 3rd c. CE. (Map by S. Gallimore.)
The quantity of AC4s and other Cretan amphora types found at Rome helps to illustrate the most important distribution network for the island's products during the Early Roman period. André Tchernia has argued that the Egyptian grain trade was an important stimulus behind the movement of Cretan goods to Italy.Footnote 57 Cretan containers are among the most common imported types at sites along the Bay of Naples in the 1st c. CE, when Puteoli served as the primary harbor of Rome.Footnote 58 At Herculaneum, for instance, 35 AC4 amphoras were found in a shop in Regio IV.14.Footnote 59 When Ostia and Portus took over as the main destination for Egyptian grain by the late 1st/early 2nd c. CE, this corresponded with the increase in finds of Cretan amphoras at Rome as illustrated in Giorgio Rizzo's study mentioned above. This suggests that Cretan products were piggyback commodities along east–west trade networks, with the island's harbors likely functioning as important stopover ports for ships sailing from Egypt and other eastern regions.Footnote 60 The draw of products to Rome also supports the distribution of AC4 amphoras across Italy and to neighboring coastal regions like the north Adriatic and the southern coast of France, as cargoes would have dispersed and goods packaged in these containers would have gained a foothold within many networks connecting east and west.
North–south trade between the Aegean basin, including Greece and western Anatolia, and North Africa was also important for the dissemination of Cretan products, including those packaged in AC4 amphoras, during the Early Roman period. AC4 containers are not particularly common across this region but are widespread, with examples attested within Greece and the Aegean (Antikythera, Corinth, Delos, Sparta), Asia Minor (Ephesos, Limyra, Miletos, Samos), Cyrenaica (Berenice), and Egypt (Alexandria, Marina el-Alamein, Mons Claudianus, and Tebtynis).Footnote 61 Crete was an important node for supporting trade between these different regions. Berenice, where several types of Cretan amphoras are well documented from the 1st through 3rd c. CE, helps to illustrate this network.Footnote 62 Along with Cretan amphoras, products that are well attested at Berenice include amphoras from the Black Sea, Pontic Sigillata, and Eastern Sigillata C (also known as Çandarli Ware).Footnote 63 Black Sea amphoras, along with Pontic Sigillata and Eastern Sigillata C, are also attested on various Cretan sites, including Knossos and Gortyn.Footnote 64 This highlights a pattern where ships carrying products along north–south networks passed through Crete, which in turn spurred the export of Cretan goods to the associated regions.Footnote 65
For AC4 amphoras north of the Mediterranean, the low overall numbers have been noted previously. According to Joost van den Berg in a study of amphoras found at Kops Plateau, Nijmegen:
In the western part of the Empire Cretan amphorae occur mostly in the forts and settlements along the Rhone-Rhine trade route. Given the small numbers it seems likely that Cretan wine reaches the people in the North as a lucrative form of side business, perhaps by hitchhiking on the trade vessels transporting the large quantities of Baetican olive oil or South Gaulish wine.Footnote 66
Publication of amphora data from Rome's northern provinces tends to be quite robust. A focus on amphoras as key archaeological data in the region is apparent from several early 20th-c. publications.Footnote 67 In a recent study of amphoras from the eastern Mediterranean found in the Rhône-Rhine area, Tyler Franconi identified at least 65 sites with quantified amphora assemblages.Footnote 68 This amount of data has also permitted larger economic studies of the region and consideration of economic networks that spread throughout northern Europe.Footnote 69 This robustness helps to substantiate that the widespread identification of small numbers of AC4 vessels in this region is a distinct pattern. Important amphora studies are also available from the Danube frontier, including those by Dobreva and by Ljiljana Bjelajac.Footnote 70
There are some exceptions to the overall low quantity of AC4 containers north of the Mediterranean, however, especially when distribution hubs like Lyon, Aquileia, Scarbantia (modern Sopron), and Turin are considered. Lyon, located approximately 270 km north of the Mediterranean coast at the confluence of the Rhône and Saône Rivers, was one of the most important distribution hubs in the northwestern Roman Empire. Not surprisingly, a high concentration of individuals with business interests are attested at the site. Of the 23 inscriptions found in the province of Gallia Lugdunensis that record businessmen, 22 have been identified at Lyon.Footnote 71 Similar names are attested at other coastal distribution hubs like Aquileia, demonstrating the importance of these sites within Roman exchange networks.Footnote 72 AC4 amphoras are well documented at Lyon. Séverine Lemaître, in a deposit dated between 190 and 250 CE, records 34 AC4 amphoras, accounting for approximately 11 percent of overall amphora finds from the eastern Mediterranean.Footnote 73
For Aquileia, Strabo (5.1.8) discussed its importance as an emporion that supported the overland trade of goods brought to the site by sea. Founded in the early 2nd c. BCE, the site served as a gateway from the Adriatic into several European provinces. Of key importance for Aquileia as a distribution hub was its connection to overland routes, particularly the Amber Road, an important route for both trade and military supply. That route is represented by a clear line of sites where AC4 amphoras are attested, stretching from the coastal sites of Aquileia and Trieste toward Vindobona (modern Vienna) and Carnuntum.Footnote 74 At Aquileia, AC4 containers have been attested in several excavations and are noted as being much more common than other Cretan types.Footnote 75
Scarbantia is significant for being one of the final settlements before reaching the Danube along the Amber Road and for being one of the earliest Roman settlements along this route.Footnote 76 It was a civic rather than military site and may have acted as a distribution hub for the large military centers at Vindobona and Carnuntum on the Danube, along with the smaller, surrounding forts. Recent excavations at Scarbantia, in a sanctuary complex comprising three temples, recovered approximately 50 fragments of AC4 containers from levels dating from 35/40 CE to 70 CE.Footnote 77
Turin, which was founded as Augusta Taurinorum in the late 1st c. BCE, served as a key distribution hub within the Po River valley, helping to connect trade moving inland from the northern Adriatic to land and river routes, including those that crossed into transalpine regions.Footnote 78 All Early Roman Cretan amphora types are attested at the site, but the AC4 is by far the most common. During excavations between 1991 and 2000 in two parts of the site, the Piazzo Castello and around the Royal Gardens, with layers datable between the 1st and 4th c. CE, approximately 2,453 fragments of eastern amphoras were recovered. AC4 amphoras account for 184 of those fragments.Footnote 79 At the site of Novara approximately 100 km to the west of Turin, there is another concentration of AC4 amphoras, perhaps because this site also served as a hub connecting sites along the Po and one its tributaries, the Ticino, along with various overland routes.Footnote 80
These distribution hubs help to contextualize the small number of finds of amphora types like the AC4 at most sites in northern Europe since cargoes would have been broken up at those places to accommodate different supply needs. Auriemma and Quiri describe this as a form of capillary trade, with the increasing dispersal of products leading to smaller quantities at individual sites as you continue further along the economic network.Footnote 81 As part of this capillary distribution, products packaged in AC4 amphoras could be purchased or perhaps, in the context of the Roman military, requisitioned. An AC4 amphora found in a mid-1st c. CE deposit at Caerleon, Britain may provide an example of supplies being purchased or requisitioned by the army (Fig. 5).Footnote 82 This amphora, which was identified in the publication as Rhodian but instead can be classified as an AC4a container, was discussed above with respect to a painted inscription on its upper shoulder, written in cursive Greek, that describes the contents as passum, or raisin wine. A second painted inscription is visible along the neck and part of one of the handles. Written in block letters in Latin, it reads: Leg(ionis) II Aug(ustae) “Property of the Second Legion Augustan.”Footnote 83 The reference to the legion appears to be a later addition and may indicate that this amphora and its contents were requisitioned or purchased by a military supplier trying to support forces stationed at Caerleon. Similar processes may lie behind the presence of AC4 amphoras attested at other sites across northern Europe and the Danube frontier.
Fig. 5. AC4a Amphora found at Caerleon, Britain (Burnham et al. Reference Burnham, Keppie, Esmonde Cleary, Hassall and Tomlin1994, 311, fig. 9.)
River and land transport
Even if the overall number of AC4 amphoras north of the Mediterranean is low, it is still necessary to explain the mechanisms behind their distribution. For these regions, transport of amphora-borne commodities was primarily by river or road.Footnote 84 Both overland and fluvial transport may have been necessary in many cases since, as Christoph Schäfer observes for the northern provinces, “there existed no unbroken river route.”Footnote 85 In western Europe, the Rhône-Rhine axis, which comprised a series of riverways and overland routes, may have been “the most important commercial, military and civil route linking the Mediterranean and the Gallic and Germanic territories.”Footnote 86 Ships from across the Mediterranean could access this economic network by first sailing to Marseille, with direct access to the mouth of the Rhône established by construction from 104 to 102 BCE of the fossa mariana, a canal that bypassed the coastal delta where accumulated sediment made accessing the river difficult. Goods were then brought up the river to Lyon for dispersal throughout the region. The third quarter of the 1st c. BCE through the mid-1st c. CE saw significant Roman investment in Gaul and along the Rhine frontier aimed at infrastructure for river and land transport.Footnote 87 Colonization in this region, including at sites like Arles, Lyon, Nîmes, Orange, Valence, and Vienne, was an important stimulus, as was the growing presence of Roman soldiers who required reliable supply lines. These efforts solidified Lyon as the central node within these networks. A combination of river and land transport was also important for reaching regions to the north of the Adriatic Sea and along the Danube frontier.
Transport of amphoras by river was not a simple matter of transferring whatever containers arrived by sea to smaller barges. The largest Roman river boats may have reached 34 metric tons, but most were smaller (Fig. 6).Footnote 88 Even a small amount of breakage would be amplified by their reduced cargo capacity. Attributes of amphora design that were conducive to large seafaring ships were not always beneficial for smaller river barges. This led to special considerations, as Jennifer Muslin describes:
the appearance of…more compact forms marks an important shift, since it suggests that seafaring ships were not the only watercraft that carried amphorae; flatbottomed and half-capacity amphorae are often found in first and second century C.E. river boat wrecks, suggesting that their spiked counterparts were as not [sic] well-equipped for the wide, shallow hulls of river boats, which were only one or two meters in height.Footnote 89
Fig. 6. Example of a river barge, the Arles-Rhône 3 wreck (mid-1st c. CE). From Wikimedia Commons; image by L. Brighton (CC-by-SA-4.0.)
Amphora manufacturing traditions in Gaul provide a clear example in support of this argument. Types manufactured in Gaul from the 1st c. BCE to 2nd c. CE are typically flat-bottomed and relatively short. The Gauloise 1, for instance ranged in height from 47 to 53 cm, while taller forms like the Gauloise 4 may have reached up to 69 cm.Footnote 90 The flat bases provided more stability during transport by river boat, and the height of these jars prevented the cargo from being top-heavy. Not surprisingly, while Gauloise-type amphoras are documented in different Mediterranean regions, they appear most frequently in the Rhône-Rhine axis.Footnote 91 Production of amphoras with flat bottoms, such as the Dressel 28 and Oberaden 74 during the Early Roman period, is also attested in the Guadalquiver and Ebro River valleys in Spain.Footnote 92 Such types were suitable for river transport and developed when Roman trade along river networks was increasing in volume and frequency during the 1st c. BCE and 1st c. CE.Footnote 93
Dressel 2–4 amphoras, a common Early Roman type manufactured in numerous regions of the Mediterranean, offer a contrasting example. As Muslin observes for Dressel 2–4 containers manufactured in the Campanian region of Italy, they were nearly as tall (ca. 95 cm) as a standard river boat hull, meaning the ship would be top heavy, making the containers more likely to jostle during transport.Footnote 94 The long spikes of amphora types like the Dressel 2–4 also made them less stable in these types of ships and could have contributed to breakage. Despite these limitations, some spike-bottomed Dressel 2–4s did find their way onto river barges.Footnote 95 As noted above, development of trade as part of a Roman commercial revolution along the Rhine and Danube frontiers was reliant primarily on Mediterranean products prior to the rise of more local and regional markets. This likely spurred the import of a variety of products to civilian and military sites in these regions, including wine from Italy. Eastern versions of the Dressel 2–4 types also appear on numerous sites north of the Mediterranean, particularly in the first half of the 1st c. CE.Footnote 96 Some scholars, in addition, have argued that a flat-bottomed version of the Dressel 2–4 manufactured in the black-sand fabric characteristic of the Bay of Naples region was developed to accommodate river transport.Footnote 97
Cretan types like the AC1–3 had heights (ca. 60 cm) more in line with Gauloise-type containers, but their small button toes and wider diameter may not have been ideal for river boats. They lacked stability and stackability, and likely were not popular choices for transport of goods by river. This is evident in the small number of AC1–3 jars identified at sites in northern provinces. The shape of AC4 amphoras also does not, at first glance, appear advantageous for transport by river barge, but this type does have certain advantages including its narrower profile and smaller capacity. These attributes may have made these containers more stable and stackable as part of a river-barge cargo, increasing their appeal to merchants seeking cargo for boats heading into the Rhône-Rhine axis and other river routes. For this reason, perhaps it is also not surprising that Camulodunum 184 amphoras from Rhodes, which are very similar in morphology to AC4 amphoras, are the most common eastern type attested north of the Mediterranean in the Early Roman period.Footnote 98 In other regions where river travel may also have been important, such as sites along the northern Adriatic, AC4 vessels are often found together with flat-bottomed amphoras like the Forlimpopoli type.Footnote 99
Amphoras in the Rhodian tradition may have had a reputation for traveling well as part of shipboard cargoes. A recent study by Anno Hein and Vassilis Kilikoglou analyzed the mechanical performance of Rhodian and Koan amphoras of Hellenistic date through virtual simulation using three-dimensional models.Footnote 100 Particular points of focus were the vertical load on the shoulder when amphoras were stacked as part of a cargo and the horizontal load on the body because jars were positioned side-by-side. Both the external surface at specific contact zones and the internal surface opposite those contact zones were examined. The Rhodian amphoras under consideration included types datable from the 4th through 2nd c. B.C. According to Hein and Kilikoglou, these simulations demonstrate that later Rhodian amphoras showed a decrease in documented vertical load stresses compared to earlier types for both the external and internal surfaces. They argue that this represents “an improvement and effectual technological development after the 4th century BC.”Footnote 101 Koan amphora types from the 4th to 1st c. BCE, on the other hand, did not show any improvement in documented vertical stresses on the external surface but had some improvement to stresses on the internal surfaces. Horizontal stresses did not change for either amphora type.Footnote 102 For Rhodian amphoras, morphological development continued after the 2nd c. BCE, and AC4 amphoras more closely resemble the form that originated by the 1st c. BCE. If vertical load continued to be a concern during development of these types, that could indicate they did not face the same amount of breakage and damage as other amphora forms, making them good choices for cargoes, including on river barges. This does not mean that Rhodian amphoras and associated types like the AC4 were developed initially with river transport in mind but shows that improvement of attributes that made these containers suitable for overseas transport were also applicable and advantageous for fluvial transport.
Underwater excavations in the Rhône River as it passes through Arles, France help to corroborate the presence of AC4 amphoras on river barges. Along with over a dozen wrecks that have been discovered in this part of the Rhône, excavations have revealed several large underwater pottery dumps. One of these dumps is situated near the right bank and overlays the Arles-Rhône 3 and Arles-Rhône 4 shipwrecks.Footnote 103 Excavation of this deposit prior to the discovery and subsequent investigation of the Arles-Rhône 3 wreck recovered several thousand ceramic vessels, including a large number of amphoras, all dating from the mid-1st c. CE through the mid-2nd c. Over half of these containers were Gauloise-type amphoras, but examples from many other regions were also attested.Footnote 104 Among the finds from the eastern Mediterranean were a small number of AC4 containers and a couple of AC1 jars.Footnote 105 The presence of AC1 and types like the traditional spike-bottomed Dressel 2–4 in this deposit shows that even amphoras that were not well suited to transport by river barge could still be carried on these ships. For Cretan vessels, however, a larger number of AC4 amphoras were recovered than AC1 containers, suggesting this was a preferred type for transport aboard river barges.
Finds of AC4 containers at sites to the north of the Adriatic likely also arrived first at coastal ports and were then transported by river barges and/or overland caravans.Footnote 106 A shipwreck discovered off the coast of Croatia in Koromašna Bay, along the northern side of Žirje island, may point to cargoes of goods packaged in these containers being transported to the northern coast of the Adriatic.Footnote 107 This wreck, datable to the 1st c. CE, had over 100 AC4s as part of its cargo along with some AC2s and Rhodian amphoras. While a shipwreck's final destination can only ever be a tentative hypothesis, the presence of AC4 amphoras at harbor sites in the northern Adriatic and at sites in regions to the north of the Adriatic could indicate the ship was destined for a harbor like Aquileia, as noted above.Footnote 108 Another option could be Altinum, an important port site just to the east of Venice. Nearly two dozen AC4 amphoras are noted in excavations there.Footnote 109 AC2s also appear in this region occasionally, including at Trieste on the Adriatic coast and inland at Magdalensberg.Footnote 110 This trade, as discussed above, likely followed overland routes, including the Amber Road, as one mechanism of distributing products arriving at the coast across a wider region. The northern Adriatic was also tied to the fluvial and overland routes of the Po River valley.
AC4s are attested at a number of sites along the Danube to the east of Vindobona and Carnuntum.Footnote 111 Transport of these containers to these sites was likely by river and could have come from two directions. Some AC4 amphoras may have been shipped to these other centers after arriving at Vindobona or Carnuntum and being placed on river barges at those sites.Footnote 112 There are also finds of AC4 jars at sites located at or near the mouth of the Danube, including Aegyssus (modern Tulcea).Footnote 113 Interestingly, at Histria, which is located just to the south of the mouth of the Danube along the coast of the Black Sea, single examples of both the AC1 and the AC2 amphora have been found.Footnote 114 These containers made it close to the Danube but rarely appear to have traveled down the river as part of river barge cargoes. Other sites where AC4 amphoras have been identified in this region, like Troesmis in Romania, are located on tributaries of the Danube, and likely were supplied by river barges.Footnote 115 For some locations that were not positioned along a river, such as Tropaeum Traiani (modern Adamclisi) to the south of the Danube or Dobruja (modern Dobrogea) to the north of the Danube, overland transport would have provided an essential means of supply.Footnote 116 Dobreva has observed the elevated number of sites where AC4 amphoras are attested along the Danube frontier, especially in comparison to other Cretan types.Footnote 117 She also notes that Popilian's type I amphora from his 1976 publication, which comprises containers in the Rhodian tradition, may include the AC4, raising its profile in the region even higher.Footnote 118 Analogous to the Rhône-Rhine axis, the AC4 appears to have been ideal for overland and river transport along the Danube frontier as part of Early Roman trade networks.
Alternative routes? A consideration of Atlantic trade
We can identify the final resting place of an amphora, but its journey to that location, and potential stops along that route, are incredibly challenging if not impossible to discern. While the above discussion focuses primarily on the dissemination of AC4 amphoras throughout Rome's northern provinces via river and land routes that originated along the coast of the Mediterranean, consideration must also be given to trade routes along the western Atlantic. The importance of Atlantic sea trade between Spain, western France, and Britain has seen increased attention over the past few decades.Footnote 119 It was a vital economic network, particularly for the dissemination of olive oil from the Baetican region of Spain to military forces stationed along the Rhine frontier and in Britain. Perhaps some AC4 amphoras that have been documented in these same regions traveled aboard ships that plied the Atlantic.
Much of the focus on Atlantic trade has been on its role in the distribution of Dressel 20 amphoras, the primary container for packaging and transporting olive oil from Baetica.Footnote 120 The Dressel 20, in many ways, offers an interesting contrast to the AC4. Morphologically, the Dressel 20 has a rounded point at the base, a thick-walled, wide, globular body, a short, vertical neck, and an outward-thickened, rounded rim. Large handles, round in section, attach to the neck and upper shoulder. The average capacity of the Dressel 20 amphora is approximately 75 L, while its height ranges from 70 to 97 cm.Footnote 121 Its size and base type do not appear, at first glance, to be conducive to travel by river barge, but this amphora type is attested in substantial quantities at numerous sites in northern Europe.Footnote 122 The importance of a steady supply of products like olive oil to the army was likely a key driver behind this, with the need to mobilize large-scale supplies from a region capable of producing it at the time. Distribution of Baetican olive oil appears to have been organized through the Roman state beginning at least by the 1st c. CE.Footnote 123 By building on earlier assessments and additional amphora assemblage data, Schäfer estimates that olive oil equivalent to the contents of approximately 23,000 amphoras was required per annum in the German provinces.Footnote 124 Much of this trade, as discussed below, may have been tied to ships plying the Atlantic as opposed to relying primarily on river barges.
There has been much debate concerning the routes by which olive oil packaged in Dressel 20 amphoras reached these northern sites. Some scholars have argued that ships from Spain sailed across the Mediterranean to southern France, where these containers were transferred onto river barges for transport into Rome's northern provinces.Footnote 125 Others have argued that Atlantic trade routes were more important for distribution of Dressel 20 amphoras.Footnote 126 Schäfer, more recently, has examined variables like meteorological conditions, distance, cost of sea versus river versus land travel, and estimates of travel time along different routes, and argued that these all support Atlantic trade as being more favorable for distribution of Baetican products to northern provinces.Footnote 127 Other scholars have posed related arguments, and the current consensus does seem to favor Atlantic trades as central to distribution patterns for Dressel 20 containers.Footnote 128 The sheer mass of olive oil transported in these amphoras along this trade route likely had a significant influence on other products as well.
Did AC4 amphoras travel along Atlantic trade routes? Evidence supporting this is limited at present. There are a small number of sites in Spain where AC4 amphoras are documented, suggesting these vessels and their contents reached that province intermittently but never in large quantities.Footnote 129 AC4 amphoras are also essentially unattested at sites along the Atlantic coasts of Spain and France. One possible exception is a single fragment identified at Setúbal that may be from an AC4 amphora.Footnote 130 The excavators suggest this interpretation quite tentatively and observe that other interpretations are also feasible. Along the Atlantic coast between Spain and western France, the only attested Cretan amphoras are two fragments of AC3 jars at Barzan where the Monards Channel meets the sea.Footnote 131 A cluster of sites in Belgium and the Netherlands with attested AC4 amphoras could perhaps suggest some degree of Atlantic trade reached those sites, but it is equally feasible to hypothesize that those containers arrived in that region via the overland and river routes of the Rhône-Rhine axis.
In contrast, AC4 containers are documented in underwater assemblages and at several terrestrial sites around the mouth of the Rhône River in southern France, which led to Lyon where AC4 amphoras are well documented.Footnote 132 The evident concentration of AC4 jars along the northern Adriatic, particularly the western side that led into the Po River valley, with routes connecting to transalpine regions, also provides indications of how AC4 amphoras and their contents would regularly make their way into northern Europe. The massive draw of Cretan goods to Italy and Rome during the Early Roman period may have facilitated trade more regularly to the coastal regions of southern France and the northern Adriatic, emphasizing the importance of these networks, as opposed to Atlantic trade, for disseminating Cretan products within the Rhône-Rhine axis.
Conclusions
The distribution of AC4 amphoras and their contents is distinct from AC1–3 containers during the 1st to 2nd c. C.E. for their presence in Rome's northwestern provinces, along the Amber Road, and across the Danube frontier. Critical assessment of the mechanisms that contributed to this pattern suggests that morphological attributes of AC4 containers, including their smaller capacity and slimmer shape, may have been significant factors. Cretan wine, the presumed contents of these amphora types, had great appeal, but the AC4 was most practical for river and overland transport to the regions in question. The implication of this is significant because the AC4, as part of the Rhodian family of amphoras, was likely designed originally with the intention of overseas shipment in cargo ships to destinations within the Mediterranean. Its selection by merchants over other Cretan forms for transport of goods north of the Mediterranean along road and river routes shows an important modification to its intended function. It is possible that this modification influenced subsequent use and distribution of these containers over time, perhaps prompting stronger connections between Cretan sites and trade centers along the coast of the northern Adriatic and southern France that fed into networks north of the Mediterranean.
This assessment builds on an increasing body of work that focuses on design attributes of amphoras and the importance these played in their use and distribution. It demonstrates that such attributes were critical points of consideration for merchants and consumers who needed to ensure a steady supply of different commodities to sites whose supply chains extended beyond the coast of the Mediterranean. While focus on the distribution of pottery types like amphoras is a vital part of economic analysis in the Roman world, continued emphasis on additional variables like design is necessary to develop more detailed interpretations of the structures within economic networks and past decision-making that contributed to patterns evident in the archaeological record.
Open access
