Introduction. The glass production tanks and pottery kilns of BEY 015

Transcription

1 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Abstract Pottery from a late 1st century AD workshop and kilns located in Beirut (site BEY 015), in a previous programme of work chemically analysed and proven to be products of the workshop, are here presented in full (a complex range of at least five amphora types, including the Beirut amphora and the miniature carrot amphora/schöne- Mau XV, as well as kitchen wares and drains). Carrot amphorae and other Levantine forms found in Gaul and included in the analyses, some found to be Beirut products, others not, are also presented. The possible contents and the complex regional typologies of the amphorae of BEY 015 are also discussed. Keywords Pottery kilns, typology, amphorae, carrot amphora, dried fruit; chemical analyses Introduction The glass production tanks and pottery kilns of BEY 015 Excavations in Beirut in 1996 brought to light the location of an early Roman, 1st century AD industrial quarter just outside the most eastern section of the Hellenistic walls (the site had been defined as BEY 015 in the Beirut Central District archaeological plan designed for the international rescue excavations in the city following the Lebanese civil war) (Figure 2). A summary report of the excavations has recently been published (Kouwatli et al 2008). It was discovered that a sequence of large rectangular vats designed to produce raw glass for the subsequent manufacture of glass vessels and probably also window glass, dating to the mid 1st century AD, was replaced by four pottery kilns, associated with pottery and wasters of primarily late 1st to early 2nd century AD date (Figures 3a-b). Chemical analyses of samples from a wide selection of the ceramic finds established that several amphora forms, kitchen wares, primarily cooking pots, as well as drains and pot stands were produced on the site (Waksman et al 2003; Roumié et al 2004). More recent excavations nearby (site GEM 002), directed by Fady Beayno on behalf of the Directorate General of Antiquities of Lebanon in 2005 and 2006, discovered another area of pottery kilns a little further to the south, also outside the Hellenistic walls and overlying an earlier Hellenistic cemetery (Figure 2, for its location). The ceramics produced in this workshop were of the mid 1st century AD, in other words, contemporary with the glass-making to the north on BEY 015. Wasters indicate that the Beirut amphora (Reynolds Beirut 2) and thin-walled wares were made there. Some were identical to the beaker Figure 19.14, from BEY 006 (Reynolds is grateful to Fady Beayno and Abdullah Ala Eddine for showing me this material). The thin-walled wares were not attested on BEY 015, probably because of its later date. The scarcity of Beirut 2 on BEY 015 (eg Figures 5f and 6.1-3) with respect to later variants (Beirut 3: eg Figure 5g-h) is another indication of the later chronology of BEY 015. The proposed date range of the BEY 015 production site late 1st or early 2nd century AD is based primarily on the variants of the Beirut amphora present, as well as the range of local cooking pots, all of which are well paralleled and dated from the assemblages of BEY 006

2 72 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli (Anglo-Lebanese Beirut Souks Excavations). The chronologies of both GEM 002 and BEY 015 were thus relatively short, comprising spans of roughly 50 years each, or perhaps even a few decades in the case of BEY 015. The pottery With the aid of a CEDRE/CNRS grant 1 the assemblage recovered during the excavations on BEY 015 was examined and a selection of the principal forms of amphorae, kitchen wares (primarily cooking pots) and other pottery (a drain and pot stand) were sampled and analysed (see Roumié et al 2004 and Waksman et al 2003, for methods and archaeometric data). It should be said at the outset that there was no attempt to assess the finds according to the stratigraphic record. Given the size of the assemblage and the evident short date range of the material studied, the aim was to determine and chemically define the range of ceramics produced on the site. In this section the pottery from the BEY 015 production site will be described and discussed in full for the first time with respect to the results of the chemical analyses, local-regional ceramic typologies and other wider issues, such as Roman Phoenicia and the evidence for the exports of some of these Beirut products to Gaul and other Roman western provinces. There were two stages in the sampling process of the project aimed at, first, establishing the range of products of the kiln site and second, identifying amphorae from this Beirut source, particularly the small carrot amphora Schöne-Mau XV, found as imports on certain Gallic sites. First, samples from the BEY 015 kiln were taken in order to characterize the range of products of the Beirut workshop (65 samples). In two successive phases of analysis, samples of carrot amphorae from Gallic sites were added to the database for comparison (samples primarily from Lyon and Reims, and some from Bram) (Waksman et al 2003; Lemaître et al 2005). As part of the aforementioned work in France, Gallic samples were taken of four other amphora forms of likely north Palestinian origin found in Gaul and also added to the database in order to both characterize their clays and compare their chemical profiles against the Beirut samples (Figure 4b, for the full dendrogramme of Beirut and Gallic samples, from Lemaître et al 2005, Fig. 4, with some minor errors of the labeling corrected for the present publication): Kingsholm 117, a large module amphora of Phoenician- Palestinian tradition, with ring handles on the shoulder, with a fabric similar to that of Beirut products (Figure 16); three other north Palestinian forms, perhaps from Acre-Akko/Ptolemais that should not match the Beirut samples: Célestins 1A/Camulodunum-Colchester 105 (Figure 11) (note in this respect a complete vessel with rim like Célestins 1A/Colchester 105, but size, body and base like Augst 46 and 47: Figure 12.1, P 11638, from the Athenian Agora, fabric close to that of Beirut products); Augst 46 (Figure 12.2 and 6; Martin-Kilcher 1994); and Célestins 2A, related to Augst (Figure 12.4, from Lyon; a mid 3rd century Beirut example of this variant is also illustrated, Figure 12.5). Célestins 2A is also clearly related (rim, handles, fabric) to that of the partly contemporary, larger amphora AM 14, of early 3rd and 4th century date (Reynolds 2005, , Plate 14; Lemaître et al 2005, Fig. 11.4, for an example from Ostia). The complexity and variation within these forms is evident. No examples of Augst or Colchester 105 were found on BEY 015, confirming non-beirut origin of these products evident in the results of the chemical analyses (Lemaître et al 2005) (also argued in Reynolds 1 The project, directed by Yona Waksman ( Développement des analyses par faisceau d ions; application à l étude du patromoine archéologique du Liban ) involved the collaboration of the Ion Beam Analysis Laboratory of the Lebanese Atomic Energy Commission (Beirut) and the Laboratoire de Céramologie of the Maison de l Orient Méditerranéen (Lyon). The French part of the team comprised Yona Waksman (chemical analysis) and Séverine Lemaître (pottery), and the Beirut team comprised Paul Reynolds (pottery), Hans Curvers (director of the excavations of BEY 015), Mohamed Roumié and Bilal Nsouli (chemical analysis).

3 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology on the basis of typology and fabric). Only one possible example of these non-carrot amphora Gallic imports, and yet another variant, was identified in the BEY 015 assemblage, a hollow cone base, similar to that of Kingsholm 117, though this could equally be the base of a late Beirut 2 amphora (see below, for details) (Figure 17.6). For the BEY 015 material, the results indicate that the samples fell into two main Beirut groups (Beirut A and B), all being products of the workshop (Figure 4a, for the dendrogramme, with some minor errors of labeling corrected for the present publication: Waksman et al 2003, Fig. 2). A third group comprised a variety of products that should be considered marginals : a Beirut 2 amphora (Lib 3, Figure 6.1), carrot amphorae (Lib 58 and 60, Figure 7.9 and 19, respectively), one of the AM 72 variants (Amphora 2D: Lib 43, Figure 14.5) and a cooking pot (Lib 23, Figure 18.7). Some of these marginals could still be considered local to Beirut (see below). Some of the Gallic imports were also marginals. With one exception, perhaps, all were carrot amphorae (Lemaître et al 2005, Lib 68, 70, 71, 85: Figure 8.6-7). All of the sampled sherds, the majority of the BEY 015 examples being here illustrated, are identified with their sample numbers. A selection of other vessels of typological interest that were not sampled but can be assigned to the workshop on the basis of fabric are also illustrated. A typology and list of the sampled and non-sampled vessels is given in Appendix 1. Table 1 provides a list of the vessels in order of their appearance on the dendrogramme Figure 4b. 1. The amphorae The analyses indicate that three quite distinct types of amphorae were produced in Beirut on BEY 015: the Beirut amphora, the small carrot amphora generally known as Schöne-Mau XV and a complex group of interrelated amphorae that comprise more than one form (AM 72/Amphorae 1-3). These all fall within the two Beirut groups of the dendrogramme (Beirut A and Beirut B) (Figure 4a). Amphora 2D (Lib 43, Figure 14.5) was considered to be marginal, but could still be a BEY 015 product (the visual characteristics of the fabric do not differ from those of Amphora 1). A small hollow cone-mushroom base poses problems of classification: several possibilities are discussed below (Section 1d; Lib 7, Figure 17.6). a) The Beirut amphora (Figures 5-6) A typology for this form was published some years ago (Reynolds 1999, Appendix; Reynolds 2000; see also Reynolds 2005; see now also Ala Eddine 2005). 2 Probably for wine, it was the most long-lived of the amphorae produced from the late 2nd century BC, possibly, to the mid 7th century AD (production may have ended with the Umayyad occupation) (Figure 5). Production sites of the type have been located in Beirut itself (BEY 015, the evidence presented here), as well as other sites in its territory, at Khalde, some 15 km to the south of Beirut (personal observation of the AUB excavation material that includes wasters), and at Jiyeh, ancient Porphyreon, just to the north of Sidon, but presumably still within the very southern limits of the territory of Berytus (Figure 1; Wicenciak 2002, 2008b). We should not forget that some mid 1st century AD examples, so far only found in Beirut, were clearly identified as amphorae of the Roman colony of Beirut, being stamped COL[onia] BER[ytus] (or in the genitive: COL[oniae] BER[yti]) (Figure 5f). The range of additional fabrics encountered for the type is varied, however, suggesting that the Beirut amphora was produced at even more centres, presumably still within its territory, though probably not in northern Lebanon (see below, for north Lebanese fabrics). Rather pale (yellow) fabrics (southern Lebanese?) and 2 Ala Eddine published his typology of Beirut amphorae (2005) with no reference to the existing typologies and included forms that are not in the Beirut amphora series (namely his Type 8 = Reynolds [1999; 2005] AM 14 and the carrot amphora, his Type 5).

4 74 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli occasionally examples in kaolinitic clays (not typical of Beirut or its environs, one would have thought) attest to the diversity of production sites (for the kaolinitic clay versions of the later 6th century, see Roumié et al 2005). 3 Though we know that the sanctuary and city of Baalbek/ Heliopolis lay within the territorium of Beirut, Baalbek produced its own idiosyncratic amphora type, a small table amphora with a ring-foot base and grooved, Koan-style handles that would suggest that it carried wine (eg Figure , from the Homs Survey; Hamel unpublished). The Beirut amphora type is the most common amphora form found in Beirut and is justly named, even though, as is clear from the analyses undertaken on the BEY 015 finds, this was not the only amphora produced in Beirut. All the other local amphorae we shall discuss, however, date solely to the early Imperial period, or, more correctly speaking, to the 1st to early/mid 3rd centuries. A pictorial summary of the development of the Beirut amphora is offered here (Figure 5). Ala Eddine (2005) has recently provided evidence for the earliest variants, pushing back the initial date previously proposed (Reynolds 2000: ca 75 BC) into the late 2nd century BC, as well as providing drawings of complete examples of other stages in the development of the type (some being incorporated into Figure 5). It is further proposed here, on typological grounds, that the earliest Beirut amphora (Figure 5b-c) derived from Figure 5a, a Sidonian imitation of a Greek Asia Minor shape with mushroom rim and hollow knob toe. It is suggested here that Beirut, newly emerging as an independent political and city entity (polis), breaking away from Sidon, 4 3 The fabrics are whitish to pinkish, rather porous, with surfaces often fired to a shiny pale orange, similar to the appearance of African Red Slip Ware. The surfaces can bear fine fissures or cracks. In addition to the late 6th to early 7th century kaolinitic examples of Beirut 8, the latest version of the Beirut amphora, mid 3rd and early 5th century examples in this fabric also occur on BEY 006. This is important evidence for the longevity of this source. It is still unclear if the production of these Beirut amphorae is related in any way to that of a certain class of cooking ware with similar characteristics, the forms of which follow those of Beirut very closely (Reynolds and Waksman 2007: CW 34 ). Large storage jars, perhaps for oil, that are found in the Beqaa, particularly in the south, may also be from a related source ( CW 26 ). It is perhaps no coincidence that the Hellenistic fine ware of Kamed al Loz has similar qualities. We are also reminded of the Ottoman and present day production of painted wares and oil jars at Rashaya al Fouhar, even further to the south. It may be that kaolinitic clays are more common than were previously thought, and that sources are concentrated in the mountains of southern Lebanon, and perhaps also further to the south, in the Jordan Valley-Lake Tiberias. It is surely nevertheless unlikely that the kaolinitic examples of the Beirut amphora were produced beyond the limits of the city s territory. For kaolinitic clay versions of the carrot amphora found in Gaul and included in the analyses, see here Figure (Catalogue nos ; Lemaître et al 2005, , with reference also to Rashaya al Fouhar and Beirut amphorae in kaolinitic fabrics). 4 For the political changes in the region during the later 2nd century BC, see Butcher (2003) and Sartre (2001). From the late 2nd century BC major cities on the Levantine coast gained independence from Seleucid control (Tyre in 126/125 BC, Tripoli in 112/111 BC, Sidon in 111 BC and Seleucia in 109/108 BC) (Butcher 2003, 29). It is perhaps in this context, particularly the independence of Sidon to which Berytus was closely linked, that we ought to interpret the appearance of the Beirut amphora (ie Figure 5b). Berytus and other Phoenician cities received the patronage of the Herodian dynasty based in Judaea, some parts of the lower Beqaa actually falling within their dominion. The Lower Beqaa, Chalcis, and the Ituraean principality of Arqa, were independent until their annexation by Rome to form part of the province of Syria in the 70s AD, a phase in the general reorganisation of Syria and Palaestina following the Jewish Revolt and the fortuitous deaths of key members of the Herodian and Ituraean dynasties (see Butcher 2003, 43-44, and especially Chapter 3; Reynolds 2003, and Jones Hall 2004). Reynolds has argued that the appearance of north Lebanese amphorae (here Figure 15) from the Flavian period onwards may reflect a new initiative of settlement and agricultural exploitation by Rome in these newly annexed lands (Reynolds 2003, with reference also to the survey and excavations in Yanouh [Qartaba]). A similar phase of settlement can be perceived in the Homs/Emesa region, to the north, perhaps linked to its annexation between AD 72 and 79 (Butcher 2003, 43; Reynolds forthcoming). Damascus too, formerly possibly under Nabatean control, had been brought under Roman rule a little earlier, by ca AD 63. The cult centre of Baalbek, located in the northern Beqaa, was technically part of the vast territory of Beirut when the colonia Berytus was founded. As the only Roman colonia in Phoenicia, in this case veteran the two legions are celebrated in the Augustan coinage Berytus gained new status and territory at the expense of the more established Phoenician cities of Sidon and Byblos, to the south and north, respectively.

5 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 75 created a new type at this time basing the shape of its amphora on the Sidonian Greek form. The distinctive toe, however, was not reproduced, but survived only as a small hollow. Indeed, recent excavations in Jiyeh, ancient Porphyreon, have rather unexpectedly produced large deposits with wasters not only of the Beirut 2 amphora, but also of the Hellenistic Sidonian amphora like Figure 5a (Domżalski et al 2005, especially Figures 6, 8 and 9; for further details, see Wicenciak 2008a). 5 This is surely no coincidence. Perhaps Jiyeh/ Porphyreon acted as a centralized workshop for both Beirut and its own agricultural hinterland. From its beginnings in the late 2nd century BC, the Beirut amphora proper evolved over the many centuries of its production. The body capacity varied accordingly, the largest vessels being those produced in the late 2nd to early 3rd centuries AD. BEY 015 confirms the late 1st to early 2nd century production of the amphora in Beirut on this site (Figure ), both from wasters of Beirut 3 and from the chemical analysis of the clays. As we have seen, earlier production in Beirut of the mid 1st century version (Beirut 2), as well as contemporary kitchen wares and thinwalled wares, is now attested by excavations on site GEM 002, located some 100 m to the south. Analyses of two bases of Beirut 2 found on BEY 015 confirm its shared source to the later examples more typical of BEY 015 (Figure 6.2-3, 5 In 2008, in order to better define the somewhat similar production of Beirut amphorae and other local products found in Jiyeh and Beirut, samples were taken of 1st century Beirut 2 amphorae and cooking wares produced at Jiyeh, for their chemical analysis and comparison with samples taken of Beirut 2 amphorae found in Beirut (primarily from the cistern deposit BEY /12237). This project aims to determine whether the Beirut 2 amphorae found in Beirut were produced at Jiyeh. The project is a joint venture undertaken by Ula Wicenciak (University of Warsaw), Carole Atallah (Directorate General of Antiquities [DGA] of Lebanon), Mohamed Roumié (Lebanese Atomic Energy Commission) and Paul Reynolds (University of Barcelona). Though very similar and difficult to distinguish both in terms of typology and fabric, there are some appreciable differences in the texture and colour of the two sample groups (Beirut and Jiyeh) and in the manufacture of the handles. There are certainly more obvious differences in the typological details of the kitchen wares. For some time Reynolds has distinguished Beirut city products as those Lib 16, Lib 32). However, a typical Beirut 2 rim found on BEY 015 appeared to be marginal on the dendrogrammes, perhaps because of its high lime content (Figure 6.1, Lib 3). A number of rim types of Beirut 3 are represented on BEY 015 (Figure ), all of which are well-paralleled in assemblages excavated in BEY 006 (see Reynolds 1999 and 2000; Reynolds in preparation). All have typical Beirut handles, with a central flat band and a concave moulding on either side. This handle type persisted through the production of the Beirut amphora. The vessel Figure 6.20 (Lib 31), with its centrally-grooved handle, is unusual. As such, it may be related to the complex typological range of Amphorae 1 and 2 (all with grooved handles), serving as a small module carrying the same goods. The rim type is closer to that of late 1st century Beirut amphorae, however (eg Figure 6.9, Lib 35). Another perhaps more relevant parallel for this vessel is the early Imperial amphora of Baalbek that characteristically has grooved handles and not Beirut-style handles (see above: Figure ). Baalbek products are generally in a very fine, lime-rich fabric, usually fired to a dark grey. So far Reynolds has found no Baalbek amphorae in Beirut, despite its known inclusion within the territory of Berytus (see above and note 4). No such small grooved handles in a local with a fairly rough surface, with variable lime content (very close to the oxidized vessels of the Crusader period produced only a few 100 m distant from BEY 015, at the Place des Martyres), and not the denser, often red-brown fabric of some products ( heavy Beirut fabric ). We should bear in mind here that the red fabric of the modern cooking ware and jugs produced at Ghaziyeh, on sale in Beirut in the late 1990s, is also close to that of these red-brown products. Some of the red clay(s) used at Ghaziyeh derive from the Chouf mountains to the south-east of Beirut. The clay of Beirut 2 amphorae and cooking wares produced at Khalde are fortunately more easily distinguishable (a bright orange fabric with fine lime that leaves traces of colour on one s hands). The Khalde products are, in contrast, absent in the Beirut Souks excavations. The fabric and typological details of the Jiyeh Hellenistic Sidonian amphora (Figure 5a) differ, fortunately, from that of true Sidonian products found as imports in Beirut (the latter in a yellow-ochre fabric with common lime and quartz).

6 76 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli fabric have occurred on BEY 006 and the piece is so far unique, being encountered solely on this production site. b) The carrot amphora/schöne-mau XV/ Camulodunum 189/Augst 44/Oberaden 85/ Peacock and Williams Class 12 (Figures 7-10) Form, distribution and contents Long suspected to be Levantine in origin (Peacock and Williams 1986, ), the production of at least some carrot amphorae/schöne-mau XV exported to western sites can now be attributed unequivocally to Beirut. The type a small narrow-bodied amphora of Phoenician tradition (ie short rim, like hole-mouth jars, ring handles on the shoulder) and cone-like, carrot-shaped body is well-known from exports to Gaul, the Rhine provinces, Britain (153 examples were found at Colchester in the excavations of the 1930s: Vipard 1995, 70), as well as Athens (Figure 8.20, a third century example possibly in Beirut clay, though it recalls equally some fine-grained north Palestinian amphorae) and Pompeii (ie the Schöne-Mau XV type piece) (Vipard 1995; Ehmig 2000; Carreras Monfort and Williams 2002; Reynolds 2005, 571) (Figures 9-10, for some complete examples found in the West). The western distribution is concentrated on Germany, Gaul, Pannonia and Britain, with more sporadic finds in Pompeii-Naples, Rome and Barcelona (Vipard 1995, 68-73, for a catalogue of western finds; also Carreras Monfort and Williams 2002, 135, Fig. 2 and note 14, for additional Spanish finds). The occurrence of 45 examples at Vieux, on the north coast of France, is notable (Vipard 1995, 52 and Appendix). The focus of exports on military sites is likely: note, for example, those on the somewhat isolated Roman forts of Petavonium in north-western Spain and Inchtuthil in Scotland. The examples from Lyon and Reims sampled for the purposes of this project need also to be added to the published distribution maps (Table 1; Lemaître et al 2005). The type is also found on military sites along the lower Danube (Bjelajac 1996, 28-9, Type 5; Dyczek 2001, 91-3, Type 7a). There is a strong case for the argument that the form carried locally-produced Syrian dates, as well as figs and damson plums. Dates, kouk[ai], are mentioned on a dipinto on a typical body sherd of this type found at Carlisle (Carreras Monfort and Williams 2002, 133 and especially 136, Fig. 3, where they are illustrated; the epigraphic evidence is fully presented by Vipard [1995, 65-68], who argues, in contra, that they carried salted fish, or perhaps shell fish or dye derived from the latter). Dried dates supplied to Rome were from two principal sources, referred to as caryotae and thebaicae. Whereas the latter were clearly Egyptian, caryotae, we are informed by Varro (On Farming, ), were from Palestine (eg Jericho) and Syria (Dalby 2000, 169, 174). Both varieties were served at the cena Trimalchionis (ie in the Neronian period) (Petronius, Satyricon, 40) and Domitian was pelted with these during one Saturnalia festivity (Statius, Silvae, 1.6; Dalby 2000, 231). A more common use for dates, of course, was in cooking, there being many of Apicius recipes containing this ingredient (eg Apicius, 3.4.3: Flower and Rosenbaum [1958]; Dalby [2000], 174: in this case caryotae are mentioned). Syrian dates were also on sale as snacks during theatre performances in 1st century Rome (Martial, Epigrams, 11.31: notae caryotides theatris, writing here during the reigns of Domitian to Trajan; Dalby 2000, 235). Dried Damascene and Syrian damson plums (ie prunes) ( damascena and syriaca pruna ) are however equally noted in the sources, Martial making a clear reference to these amphorae carrying this fruit: pointed jars of ancient damsons (Martial, Epigrams, 5.18: acuta testa ). Martial elsewhere (Epigrams, 13.29) offers a gift for Saturnalia of a jar of damsons ( vas Damascenorum ). Accept plums wrinkled with the decay of foreign old age: their use is to relieve the burden of a stuffed stomach ; Dalby 2000, 169). A remarkably accurate description of the amphora, a twisted cone ( torta meta ), carrying cottana (small figs) is also given by Martial (Epigrams, 13.28; 7.53).

7 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 77 The date of exports of the type ranges from 10-1 BC (at the fortresses of Oberaden and Augst), perhaps peaking in the Flavian period, with finds occurring in the late Antonine period and in the early 3rd century (not noted by Carreras and Williams [2002] are the early 3rd century examples from Lyon, Place des Célestins, the vessels that were analysed with the BEY 015 examples: Lemaître et al 2005). Carrot amphorae are fairly scarce elsewhere in Beirut. Only examples were found on BEY 006, all occurring in a few contexts of the late 1st century AD (the majority are illustrated here, Figure ). 6 Their absence in a massive mid 1st century AD cistern deposit (12237/12300), as well as a major phase of construction deposits of AD (eg the pottery presented in Reynolds 1999) and in the early 3rd century AD deposits filling the natatio of the Imperial baths (BEY 045), is striking. We may conclude that locally or regionallygrown dates were transported in these containers primarily when bound for export. It should also be noted that other Levantine, south Phoenician, non-beirut, amphorae were used for the transport of dried fruit. An amphora found on the Tradelière Wreck contained dates (Pollino 1986; Parker 1992, 433, no. 1174). This appears to be the vessel illustrated by Lemaître et al (2005, Fig. 8.4) and classified as Kingsholm 117. The twisted handles and thick walls and rim of the amphora, however, suggest that it is, in fact, a Tyrian amphora, usually thought to have contained wine (Reynolds 2005). One might ask if the diminutive form Augst 46-47, produced in southern Phoenicia (on the basis of its fabric) and exported to the same western sites as the carrot amphora, did not normally carry dried fruit (Figure ). An example of the form Colchester 105/Peacock and Williams Class 65 (ie as Figure 11), found at Avenches, a larger, free-standing vessel with a ring base, contained carbonized dates. Another variant of Colchester 105, with a more pointed body, also found in the same context at Avenches contained carbonized olives (Peacock and Williams Class 65; Carreras Monfort and Williams 2002; Reusch 1970, Abb. 1.5). Did the small module vessel with the same rim type as Colchester 105 and the body of Augst found at Athens also carry dried fruit (Figure 12.1)? In contrast, the south Phoenician form Agora M 334, typologically related to the latter two amphorae with respect to their rim and handles, produced from the early 4th century onwards, probably carried wine (Reynolds 2005, 571-2, Figs : associated with wine presses in the territory to the north of Akko, right up to the present Lebanese border, and presumably also beyond it, in the attested Byzantine villages to the east and south-east of Tyre). The small module (south Phoenician) amphora Célestins 2A, found at Lyons and Beirut in the early to mid 3rd century (Figure ), has a similar rim type and handles to Agora M 334. Did this also carry dried fruit, the rim type being transferred to the larger form in the early Byzantine period when the same region shifted to the production of wine? The BEY 015 and Gallic carrot amphora samples: analyses and typology (Figures 7-10) The BEY 015 examples of carrot amphorae can be divided into two groups, a division equally evident in their details of manufacture and clays used. The first (Type A) have the typical thin walls and fabric of Beirut products, for example the Beirut 2 and Beirut 3 amphorae of the 1st to early 2nd centuries (Figure ). The ribbing is narrow and concave. The second group, in contrast, have much thicker walls, wide convex ribbing and a far more calcareous clay, the large 6 BEY (2 rim/wall/handle fragments, from 2 vessels; another rim; a handle/wall; a handle fragment and 24 walls); BEY (a rim); BEY (a rim); BEY (a wall, possibly a carrot amphora); BEY (a rim); BEY (handle/wall); BEY (handle/wall: a carrot amphora?). The examples from 10033, and are related to each other in phase and area of BEY 006. The six examples that for some reason occur together in BEY are part of a late 1st century assemblage that was redeposited in the 4th century (or the 1st century level was truncated in the 4th century and this truncation overlooked during excavation). One of the rare examples of a Beirut 2, presumably, bearing the stamp COL on its shoulder occurred also in this deposit: BEY

8 78 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli lumps of lime causing spalling of the surfaces in places (Figure ). The latter tend to be fired to a pale green colour. A complete vessel of this type is perhaps that illustrated from Saalburg by Vipard as his Type Ba (1995, 53, Fig. 1 and Fig. 6.3; here, Figure 10.7). The BEY 015 Type A rims can be classified into 2 groups (Carrot 1 and Carrot 2). One of these, with an indented rim top, and particularly thin walls and sharp ribbing, can be equated with Vipard s rim type 3a4, the type piece being an example from Verulamium (St. Albans: Figure 9b; Figure 10.1) (eg BEY 015, Fig , Lib 37 and one rim not sampled). The second group, with a band rim face and flattish or lightly concave top, formed by folding the shoulder outwards (Figure 7.3-5, Lib 52, Lib 50 and a rim not sampled) (the method of manufacture of the Beirut amphorae of the same period) can be equated, perhaps, with Vipard s rim types 2 and 3a3 (Figure 9b). Other rim types illustrated by Vipard are not present on BEY 015 or in the Beirut BEY 006 assemblage (eg Figure 9b, rim types 1, 3a1-2, 3b2, 3c1-2). The distinctive well-bevelled rim type 3b1 seems also to be absent (note that the Gallic marginal vessel Lib 68, Figure 8.6, has this rim type). The simple rim types of BEY 006, though similar, have the typical rounded inner face of Beirut 3 amphorae (Figure 8.12, 14, 15, 18). The rounded rim face of Vipard rim type 3a2 may be equivalent to that of the Athenian (non-beirut) product Figure Though many base fragments can be assigned to BEY 015 Type B (thick-walled, wide convex ribbing, calcareous fabric), only one rim is attributable to them, a vessel that was not in fact sampled, but clearly belongs to Type B (Figure 7.12). The rim type is not very different from Carrot 2 (Type A), being manufactured in the same manner. The BEY 015 lower body fragment Lib 60 (Figure 7.19, thick-walled, Type B) and Gallic Lib 68 (Figure 8.6) have high concentrations of Ca (ca 25%) and were classified as marginals to the main two groups of BEY 015 products (Beirut A and B). Though this is not matched by the other 015 samples, similar values are found for some Medieval Beirut products (produced nearby at Place des Martyres). For these a local, Beirut origin can hence still be suggested (Roumié et al 2004, 201). Beirut finds of carrot amphorae from site BEY 006 are shown on Figs , rims, handles and bases (for details, see note 6). The majority of these have a rather dense, red -brown clay, unlike the BEY 015 kiln products. They have simple, band rims, similar to those of Beirut 3 amphorae found on BEY 015 (eg Figure ; Reynolds 1999). Analyses of finds of the carrot amphorae found in Gaul prove that some of them derive from the Beirut 015 workshop (Lemaître et al 2005) (Figure 8.1-4, rims Lib 67, 69, 82 and the handle Lib 83). One of the Gallic samples, dated to ca AD 80, may be related to Vipard Type 1, the rim in this case being not flattened, but hooked. It was also classified as a Beirut product (Figure 8.5, LEV 595). This distinctive variant was, nevertheless, not found on BEY 015 or on BEY 006. Some examples were considered to be marginal Beirut products (Figure 8.6-7). A fair number of Gallic carrot amphorae sampled were classified as non- Beirut products (Figure ), Lib 72 (Figure 8.8) being the most distinctive in shape. Gallic examples Lib 70 and 71 are unlikely to have been made in Beirut with concentrations of K and Rb well above the values met in Beirut samples (Roumié et al 2004, 201). That these are body sherds, and hence, not unequivocally identifiable as carrot amphorae, is also noted (ie they could be Augst 46-47). In conclusion, though some exports of carrot amphorae were produced on BEY 015 and perhaps elsewhere but maybe also in Beirut, a fair percentage of exports to the West were not Beirut products, as is evident in their chemical and typological profiles.

9 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 79 c) Reynolds AM 72 and related forms. Amphora 1, 2A-D and Amphora 3 (Figures 13-14) Amphora 1 and 2 The amphora type AM 72/Amphora 1 was defined by Reynolds (1999: AM 72 ; 2003; 2005, 568, Plate 9, Figs 59-60) (Figure 13). The form was produced in Beirut (BEY 015), but has also been found in a dense, lime-rich fabric that indicates that the type was manufactured also in northern Lebanon (Reynolds 1999, BEY , in fabric FAM 43) (Figure 13.15). AM 72 was produced from the late 1st/early second century to the early 3rd century (many examples in the Imperial Baths, BEY 045, natatio deposits), but is notably absent in mid 3rd century contexts (eg the 220 kg deposit BEY ) (for the latest version, see Figure ). A large amphora, with a tall, cylindrical neck (later examples, more tronco-conical: Figure 13.12), wide shoulder, thicker-walled than the Beirut amphora, the deeply-scored handles and a rim with a deep concave lid seat, presumably for holding a lid, are distinctive features. The base probably ended with a long cylindrical or slightly tronco-conical hollow toe (no complete examples are known) (Figure 13.11, 13-14). Though the overall characteristics of the rim and handles are similar to those of Koan-style wine amphorae (eg Campanian examples of the 1st century), Reynolds has suggested, on the basis of the hollow toe, that the form is a fish sauce amphora (like those of Spanish garum amphora Beltrán IIB or Dressel 7/11: Sciallano and Sibella 1991). Koan amphorae and their imitations tend to have a long solid toe (but note that the Amrit amphora has a small mushroom toe in the early Imperial period: Reynolds 2005, Fig. 46). The pronounced concave rim, however, is unparalleled. This is the only version of this class Reynolds has encountered in Beirut deposits. However, here on BEY 015, there are other variants that, like those illustrated, also have the grooved handle of AM 72, but lack the concave rim face (Figure ). Form 2D (Lib 43, Figure 14.5: erroneously illustrated as a classic AM 72 in some of Reynolds previous work), recalls another fish sauce amphora, Dressel 7/11, whereas Amphorae 2A and 2B seem to be close imitations of a Dressel 2-4 wine amphora. In fact, Amphora 1 could equally be a Dressel 2-4, but the provision of a lid seat suggests that the content differed from that of Amphora 2A-B. The tall, rather triangular rim of Amphora 2C (Figure 14.4) recalls the shape of one of the more common amphora forms that Reynolds believed to be characteristic products of northern Lebanon, here Figure (see Reynolds 2005, AM 202 ). The latter type is pretty large. The north Lebanese form, however, has wide, solid handles. The wide neck could indicate that it carried fish or fish sauce, or, perhaps, olive oil (there are many presses in the mountains of central, northern Lebanon, Mtein, Gabriel 2010). A smaller module of AM 202 is also common (Figure 15.3). The amphora AM 52, with solid rounded handles is another distinctive type (Figure 15.4). Also illustrated is a small tableamphora, quite common in late 1st century and first half of the 2nd century AD contexts in Beirut, very probably carrying north Lebanese goods from the numerous settlements in the mountains (Figure 15.6). Amphora 3 (Figure ) To add to the complexity that seems characteristic of BEY 015, another shape shares the concave rim of the classic AM 72, but bears the triple banded handles of the Beirut amphora type and not the scored handle of AM 72 (Figure , Amphora 3). The size of the rim, neck and handles of this type is notably smaller than that of Amphora 1 and 2. The vessels of Amphora 3 are well-fired, bearing sharp ribbing. Apart from these BEY 015 type pieces, Reynolds knows of no other examples in Beirut. An amphora found at Chhîm, on the other hand, could well be this form (CHM 1409, E XVII) (Reynolds thanks Ursula Wicenciak for showing him the finds of the site).

10 80 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli d) Unclassified base: Late Beirut 2? (Figure 17.6) A single example of a short, hollow cone base was also produced on BEY 015 (Figure 17.6, Lib 7). The classification of this piece is not straightforward. It was originally classified as Kingsholm 117, a type that remains poorly defined, especially given that the type piece lacks both its rim and base (Peacock and Williams 1986, 217, Class 66). This base does not correspond to that of the solid toe of a complete vessel classified as Kingsholm 117, found on the Dramont D wreck (Figure 16.3; Lemaître et al 2005, Fig. 1.6). Three rims, found at Lyon, Reims and Bram, two of them with pronounced ribbing on the upper shoulder (Figure ), hence close to that of the aforementioned Dramont D vessel, were analysed and, with base Lib 7, were found to be non-homogeneous in fabric. It was suggested, however, that the rim in Figure 16.2 could be grouped with base Lib 7 (Lemaître et al 2005, 521). Another complete vessel classified as Kingsholm 117 by Lemaître et al (2005, Fig. 8.4, with reference to Pollino [1986, 179, 24]), from the La Tradelière wreck, has a solid toe and is thick-walled, a point noted in the discussion of the piece (Lemaître et al 2005, 528). The wall thickness, band rim and the twisted handle (this can be appreciated from the drawing provided) make its identification as a (1st century) Tyrian amphora far more likely (for parallels, see Reynolds 2005, 570, Figs 89-91). An Augustan period Kingsholm 117 similis amphora from Lyon (La Favorite), included in the analyses (Lib 73), is also illustrated (Figure 16.4; Lemaître et al 2005, Fig. 13; Becker et al 1986). The base Figure 16.5 is probably from the same type. The BEY 015 base, however, also bears a close resemblance to those of several other amphorae: the Chhîm amphora, though the latter is clearly much larger (Figure 17.1; Reynolds 2004; 2005, 570, Figs 92-3); a roughly cylindrical amphora with collared rim and pinched rim top, and ring handles, is not an uncommon find in Beirut (eg Figure , from the cistern deposit BEY /12237). It has a compact, Beirut-looking fabric, as do the ribbed, hollow cone bases also illustrated here (Figure ). This base type is clearly also identical in shape to that of the Chhîm amphora. Reynolds has long wondered, nevertheless, if the latter are not from the cylindrical collared rim type (they are not in the lime-rich fabric of Chhîm products). The BEY 015 base (Lib 7) is similarly thin-walled but is not as large; a final, more plausible possibility, bearing in mind the rather small size of the base and its relatively thin floor, is that it is a late 1st century AD Beirut 2 amphora base (also suggested by Lemaître et al [2005, 521, note 11]). These tend to be larger than their mid 1st century predecessors (eg Figure 6.3). 2. Coarse and kitchen wares (Figures 18-19) The analyses of BEY 015 are important confirmation of the Beirut production of certain kitchen wares defined in previous work (Reynolds 1999): Figure , cooking pots with a tall collar rim, either plain or with the inner rim face thickened (mid to late 1st century AD); Figure , their successor (with a short, everted collar rim), 7 Figure 19.1, a lid, Figure 19.2, a carinated bowl (a Hellenistic shape that was equally produced in Augustan Carthage); 8 Figure 19.3, a large jar, Figure 19.4, a mortar and Figure , a drain. There were no thin-walled vessels (mugs-cups, like Figure ), jugs or flagons found on BEY 015. A large number of pot stands in the same fabric as the sampled material (ie produced on the site) may well have served as stands on which the potters rested vessels, the amphorae probably, during the stages of manufacture (eg Figure ). This is also a feature of the Jiyeh production site. 7 For comment on the typology of early Roman cooking pots in Beirut, see Reynolds (1999) and Reynolds and Waksman (2007). 8 Fulford and Peacock (1995, 56, Bowl 19).

11 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 81 Conclusions The range of amphorae and their variants, including those of the Beirut amphora, is far greater on BEY 015 than that of sites in the city itself. The definition of AM 72, as previously published, is here not so straightforward. The classic type (with a concave lid seated rim), regularly found in Beirut contexts, is Amphora 1. Whether Amphora 2A-D, that share the grooved handle of Amphora 1, represent distinct types is not yet clear: they appear not to have been for the Beirut market or were short-lived. Amphora 3, with a short neck and Beirut amphora type handles, is so far attested solely on BEY 015. It may be typologically related to north Lebanese amphorae such as Figure 15.3 and 4 (small AM 202 and AM 52, respectively), or perhaps served a similar purpose. The source of some of the Gallic imports of carrot amphorae, long thought to have originated in the Levant, and even in Egypt, is now proven to be Beirut. It is also clear that the type was produced at other locations. Carrot amphorae would have contained dried fruit. It is a point of interest that this type of amphora is scarcely found elsewhere in Beirut, but occurs almost exclusively as an export. The chemical analyses have also confirmed that forms such as Augst and Colchester 105 are linked in terms of source (also evident in their typologies) and that the source was not Beirut. Reynolds would suggest that the chemical comparison of their fabrics with those of Medieval vessels produced in Akko/Acre might help to locate or eliminate Akko/Ptolemais as the source of some of these. Others, Reynolds suspects, have the typical fine quartz pale yellow-orange fabrics of some examples of Agora M 334, probably those produced to the north of Akko and in southern Lebanon, what was southern Roman Phoenicia.

12 82 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Catalogue In the following Catalogue the order is as follows: Catalogue number; Lib number (sample number, if sampled); Excavation catalogue number (layer). Figure number, if illustrated; Type of sherd (rim, base, etc); Rim diameter (whether measurement taken from outer, or inner rim face); Form (eg Beirut 2, Beirut 3); Fabric description. Order of sampled sherd on the dendrogramme Figure 4b (for the order of other vessels analysed, but not illustrated here, see Table 1). Note that the pottery from BEY 015 and other sites in Beirut (BEY 006 and 045), as well as elsewhere in the region (eg Homs, Chhîm), is listed here in order of illustration. This was preferred to presenting all the BEY 015 material in a single block, with parallels following. Abbreviations: Rim, handle, etc: R = rim; H = handle; B = base; N = neck; Sh = Shoulder; W = wall, fr = fragment. a) Beirut amphorae BEY 015 Beirut 2 1. Lib 3. Figure 6.1. Rim/neck. Diameter: 9 cm inner rim. Beirut 2. Well-fired orange fabric with reduced core. (Order 76). 2. Lib 30. D Rim. Beirut 2. Occupation, not associated with kiln use? With a Hellenistic fish plate base. (Order 49). Beirut 2 base? 3. Lib 16. F Figure 6.2. Beirut 2 base. (Order 47). 4. Lib 32. D Figure 6.3. Fired orange-brown throughout. Beirut 2 base. (Order 37). Beirut 3 5. Lib 9. E Figure 6.4. R/N/H. Rim obscured. Diameter: ca 10 cm inner rim. (Order 3). 6. Lib 25. E Figure 6.5. R/W/H stump. Diameter: ca 12 cm outer rim. Beirut 3A. Flat band rim, folded with projection on inner face. Well known variant for late 1st and early 2nd century AD. Rather large handle. Coat of yellow mortar on the inner wall. Reused. (Order 12). 7.. Figure 6.6. No. 4. Diameter: ca 14 cm inner rim. Beirut 3A. Band rim with slight thickening on top inner face (folded). 8. Lib 24. F Figure 6.7. R/Nfr. Diameter: ca 13 cm outer rim. Beirut 3B. Band rim, light concave top. Outer rim face uneven, due to folding. Note that this piece is coupled with the unusual, grooved handle Lib 31 (Figure 6.20). (Order 45). 9.. Area C4. Figure 6.8. R/N. Diameter: 13 cm outer rim. Beirut 3B. Outer rim face folded, hence uneven, as drawn. Flat top. 10. Lib 35. E Figure 6.9. R/N/Hfr. Diameter: 10 cm inner rim. Beirut 3C. Thick, band rim. Orange-red fabric and cream-orange surfaces. (Order 14). 11. Lib 4. O Figure R/N/Hfr. Diameter: ca 9.6 cm inner rim. Beirut 3C. Pale salmon-orange fabric throughout. Rather pimply surfaces. (Order 61). 12. Lib 10. Niveau 13. Couche 354. Room 3. Figure R/N/Hfr. Diameter: 9 cm inside. Beirut 3C. Reduced grey fabric with creambrown surfaces. (Order 5) No. 5. Figure R/N/Hfr. Diameter: 12 cm outer rim. Beirut 3C. Plain inner face. Band rim. 14. Lib 19. F Figure R/N. Diameter: 10 cm inner rim. Beirut 3D. Square band band rim, bevelled inner face, concave top. Dark grey fabric with pale greenish- brown surfaces. (Order 80). 15. Lib 15. F Figure R/N. Diameter: 13 cm outer rim. Beirut 3D. Square band rim, flat top. Rough, reduced ware. (Order 40). 16. Lib 17. Z Figure R/N/Hfr. Diameter: 10 cm inner rim. Beirut 3E. Rolled rim. Beirut 3E. Close to Beirut 2, but such rims are found in 2nd century contexts. (Order 48) E Figure Rim. Beirut 3. Over-fired and rather warped. Perhaps a waster. 18. Lib 29. Figure (Rim now missing, taken as sample)/hfr/nfr. Beirut 3. (Order 54). 19. Lib 28. F R/Hfr. Beirut 3, classic type, as early 2nd century examples. Rim taken as sample. (Order 1).

13 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology Lib 27. Area C4. (Rim). Beirut 3. (Order 34). Beirut 3 Bases 21. Lib 47. Square E Figure Base. Beirut 3. Grey brown local fabric with greengrey surfaces. Spiral ribbing. (Order 22). 22. Lib 49. E2. Figure Base. Beirut 3. Reduced grey. (Order 24). 23. Lib 22. E Beirut 3 base. (Order 6). 24. Lib 21. E Beirut 3 base. (Order 31). Hybrid Beirut 3/Amphora 1/cf. Baalbek amphora Hybrid Beirut 3/AM 72 grooved handle. Plain band rim, like some Beirut 3 amphorae in this respect. Handle bearing a deep groove down the centre. Note that Baalbek amphorae are of this size and have the same handle type. 25. Lib 31. Z Figure R/N/H. Diameter: 10 cm inside. Fired orange-red brown throughout. Same fabric as AM 72/Amphora 1A and Amphora 2C. Here, it is thinner-walled, the core is not reduced. (Order 46). b) Baalbek amphora Homs Survey 26. SHR 758, CAT Figure R/N/Hfr. Baalbek amphora. Grooved handle. Hard, fine, black fabric with common lime. 27. SHR 758, CAT Figure Hfr/Nfr. Baalbek amphora. Same fabric. c) Carrot amphorae BEY 015 Type A: Thin-walled vessels. Light, concave ribbing. Type B: Thick-walled vessels with wide or narrow convex ribbing. Fabric B. Some examples are lime rich. Type A 28. Lib 37. D Figure 7.1. R/H/W. Diameter: 9 cm outer rim, 7 cm inner rim. Type A. Carrot 1. Large part of vessel. Folded band rim, marked indent on rim top. Pimply, local fabric (ie quartz grains under wet smoothed surfaces). Dark brown core and dark orange edges. Brownish-orange surfaces. (Order 59) Figure 7.2. R/N. Type A. Carrot 1. Folded band rim, marked indent on rim top. Compact fabric with abundant fine lime. Greencream surfaces. 30. Lib 52. Figure 7.3. R/W/H. Diameter: 7.4 cm inner rim. Type A. Carrot 2. Rusty orange edges and reduced core. Identical ware to AM 72 vessels. (Order 35) Area C 4. Figure 7.4. Found with other carrot amphorae and a Beirut 3 rim. R/N/Hfr. Diameter: 7 cm inside. Type A. Rough, hard, reduced throughout. Small, square band rim, rim top almost flat. Rather wide handle. 32. Lib 50. F Figure 7.5. R/N and H/W of same vessel. Type A. Carrot 2. Handle arched. Pinched-concave rim top. Slightly concave band face. Well-ribbed inner and outer surfaces. Fired deep orange. (Order 7). 33. Lib 40. D Rim. Type not noted, but presumably Type A. (Order 56). 34. Lib 41. D Rim. Type not noted, but presumably Type A. (Order 64). 35. Lib 39. D Rim. (Order 27). 36. Lib 53. E Figure 7.6. Lower section of wall. Orange-red brown fabric. Orange brown outer surface and dark grey brown inside. Thinwalled. Type A. (Order 33). 37. Lib 38. D Figure 7.7. Lower half of vessel. Type A. (Order 38). 38. Lib 57. D Figure 7.8. Rough local red brown fabric, with reduced, grey-brown surfaces. Type A. (Order 44). 39. Lib 58. D Figure 7.9. Lower wall. Thin-walled. Dark red-brown outer half and dark brown-grey inner half. Type A. (Order 53). 40. Lib 45. F Figure Base. Pale red brown fabric and pale brown surfaces. Type A. (Order 32). 41. Lib 61. D Figure Base/Wall. Type A. (Order 51). Type B 42.. Area D4. Figure R/N/H. Diameter: 7 cm. Type B. Carrot 3. Concave rim top. Wide handle, narrowing to base. Rather thick-walled. Fabric B. Over-fired dark brown, porous fabric with common fine 0.5 mm lime,

14 84 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli occasional 2 mm lime. Outer surface fired green. Probably rim type for Type B bases. 43. Lib 56. E Figure Wall. Middle wall section. Type B. Fabric B. Some large lime lumps (2-5 mm). Reduced grey throughout. (Order 16). 44. Lib 2. D4 427 (or O4 427). Figure Lower wall. Thick-walled. Type B. Fabric B. Abundant lime. Pale rusty orange. (Order 23). 45. Lib 1. D Figure Base. Type B. Pronounced convex spiral ribbing. Fabric B. Coarse, lime-rich orange fabric. Orange-brown outer surface. (Order 19). 46. Lib 62. Figure Base. Type B. Orangebrown fabric, inner edge reduced dark-grey/ brown. (Order 55). 47. Lib 63. Figure Base. Type B. Wide ribbed. Fabric B. Reduced with lime. Pitting due to lime. (Order 57). 48. Lib 64. D Figure Narrow base. Type B. Fabric B. N.B.: the right side has abundant lime, up to 1.5 mm, whereas the left side has 0.5 mm lime inclusions, closer to a normal Beirut fabric. Usual convex ribbing of Type 2/Fabric B carrot amphorae. (Order 58). 49. Lib 60. D Figure Lower half, missing toe. Type B. Thick-walled with wide, stepped, spiral ribbing. Fabric B. Granular, green fabric with lime. Marginal. Analyses indicate that this is chemically different. It is much more calcareous. It nevertheless compares well with some of the Crusader Medieval products of the kilns of nearby Place des Martyrs (turquoise glaze vessels). (Order 84). 50. Lib 59. D Figure Wall fragment. Type B. Thick-walled with wide convex ribbing. Darkish orange surfaces and outer edge. Pale grey-buff fabric. (Order 67). Gallic examples 51. Lib 67. Lyon. Place des Célestins, US (Lemaître et al 2005, Fig. 7.1.). Figure 8.1. R/N. AD Beirut product. (Order 28). 52. Lib 69. Lyon. Place des Célestins, US (Lemaître et al 2005, Fig. 7.2). Figure 8.2. R/N/ Hfr. 2nd century AD. Beirut product. (Order 30). 53. Lib 82. Lyon. Place des Célestins, US (Lemaître et al 2005, Fig. 7.3). Figure 8.3. R/N. 2nd century AD. Beirut product. Small rounded rim. (Order 65). 54. Lib 83. Lyon. Place des Célestins, US (Lemaître et al 2005, Fig. 7.4). Figure 8.4. Handle fr. Unstratified. Beirut product. (Order 29). 55. LEV 595. Reims. Rue de l Équerre, 2002.A (Lemaître et al 2005, Fig. 7.5). Figure 8.5. R/ Nfr. Folded, hooked rim. Dated: AD 80. Beirut product. (Order 8). 56. Lib 68. Lyon. Place des Célestins, US (Lemaître et al 2005, Fig. 7.8). Figure 8.6. R/N/H. Inner rim face bevelled flat. AD Close to BEY 015 (marginal). (Order 85). 57. Lib 85. Lyon. Place des Célestins, US (Lemaître et al 2005, Fig. 7.9). Figure 8.7. Base. Thick-walled (Base B). Unstratified. Close to BEY 015 (marginal). (Order 13). 58. Lib 72. Lyon. Rue des Farges, C Figure 8.8. R/N/H. Not a Beirut product. Even, convex upper wall, with a small, rounded rim. Late 1st century AD. (Order 77). 59. Lib 86. Lyon. Place des Célestins, US (Lemaître et al 2005, Fig. 7.11). Figure 8.9. R/N. AD Not a Beirut product. (Order 79). 60. LEV 591. Reims. Crédit Agricole, F (Lemaître et al 2005, Fig. 7.12). Figure Rim. AD Not a Beirut product. (Could be Augst 47?). (Order 98). 61. LEV 589. Reims. Crédit Agricole, F (Lemaître et al 2005, Fig. 7.13). Figure R/ Nfr. AD Not a Beirut product. (Order 97). Beirut: BEY 006 examples Few carrot amphorae have been found in the Beirut Souks excavations (BEY 006, 007, 045), an indication that these amphorae were primarily for containing dried fruit for export. Examples are predominantly in a dense, well-fired dark reddish-brown fabric, local, but perhaps not from the BEY 015 production site (see note 6). It was not possible to include these examples in the analyses, but they are illustrated here for their typological interest. They all have simple, band rims and thin-walled upper sections. 62. BEY Figure R/N/H scar. Diameter: 8 cm inner rim top. Thin-walled. Heavy, dense fabric. 63. BEY Figure H/W. Thickwalled. Handle narrowing to base. Could be the

15 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 85 handle of Figure Heavy dense fabric. 64. BEY Figure R/Nfr. Diameter: 8 cm top. 65. BEY Figure R/Nfr. Diameter: 7 cm top. 66. BEY Figure R/Nfr. 67. BEY Figure R/N. Thinwalled. 68. BEY Figure Base. Thinwalled, light wide ribbing. Base Type A. 69. BEY Figure Hfr. Athens (Agora excavations) 70. Agora P 994. Figure Complete vessel. Thin-walled, light weight. Simple, rather convex band rim. Well ribbed, short ovular-shaped body. The rim, body (and fabric?) may link this piece with the Kingsholm 117 similis vessel Figure 16.4 (from Lyon, La Favorite), here catalogued under no Gallic and other western parallels (in addition to those illustrated on Figure 9) 71. Figure Verulamium/St. Albans, (Vipard 1995, Fig. 4.6). 72. Figure Straubing. (Vipard 1995, Fig. 4.8). 73. Figure Ambrussum. (Vipard 1995, Fig. 5.3). 74. Figure Knossos. (Vipard 1995, Fig. 4.7). 75. Figure London. (Vipard 1995, Fig. 3.3). 76. Figure Naples. (Vipard 1995, Fig. 3.8.). 77. Figure Saalburg. (Vipard 1995, Fig. 6.3). Type B. d) Célestins 1A/Colchester 105/Peacock and Williams Class 65 Lyon 78. Place des Célestins, Figure 11a. R/2H/W.. (Lemaître 2000, Fig. 9.6; see also photographs Lemaître et al 2005, Fig ). 79. Lib 89. Place des Célestins, Figure 11b. B/W. (Lemaître 2000, Fig. 9.8; Lemaître et al 2005, Fig. 9.3). (Order 72). Augst and similis Athens 80. Agora P Figure Complete vessel. Thin-walled. Cupped, band rim, similar to that of Célestins 1A/Colchester 105. Small hollow toe and body like Augst 46. Augst 81. Augst 46 type piece (reconstruction drawing?). Figure (Martin-Kilcher 1994, Fig ). Rolled rim, almost cylindrical neck. Small hollow toe. 82. Augst 47 type piece. Figure (Martin- Kilcher 1994, Fig. 237, 5436). Lyon 83. Lib 76. Place des Célestins, Figure Célestins 2A. (Lemaître et al 2005, Fig. 11.1). AD (Order 89). Beirut 84. BEY and 207. Figure Cf. Célestins 2A. R/Hfr and Base. Mid 3rd century. Lyon 85. Lib 74. Place des Célestins, Figure Augst 46. (Lemaître et al 2005, Fig. 10.3). (Order 78). e) AM 72/Amphora 1-3 AM 72/Amphora 1 Form 1. Classic type. Concave inner rim, convex or flat rim face. Tall, cylindrical neck and corresponding long, grooved handle (Figure 13.1). Some vessels with grooved band on upper neck. Probably similar grooved band on the outer shoulder. Shoulder/wall probably sharply carinated, probably with a tronco-conical body ending with a hollow cylindrical or slightly tronco-conical toe (Figure 13.11). Thicker walled than Beirut amphorae. Note that one example has a wider diameter than the rest (Figure 13.3). Later examples have a more tronco-conical flaring neck and the toe is ribbed (Figure ).

16 86 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli BEY Lib 44. Figure R/N/Shfr/2Hfr. Diameter: 13 cm outer face cm inner face. 100% of rim. Amphora 1. Triple convex groove-moulding on upper neck, below rim. Pronounced concave lid seat. Outer face well bevelled. Step on lower neck. (Order 62). 87. Lib 11. D Figure R/Nfr. Diameter: 14 cm outer rim. Amphora 1. (Order 43) E Figure R/N fragment. Diameter: 14.6 cm inside. Amphora 1. Lower rim face beveled and undercut. Pronounced concave inner face. Fired green throughout. BEY 015 AM 72/Amphora 1 or Amphora 2, hollow cylindrical bases 89. Lib 55. Figure Base. Fired dark brown throughout. Orange-brown surfaces. Square E (Order 60). 90. Lib 46. Area F2, surface find. Figure Base. Dark, rusty orange-brown surfaces and reduced grey fabric. (Order 20). 91. Lib 51. G Figure Base. (Order 9). Beirut BEY 006 Amphora BEY Figure Rim. Diameter: 15 cm outer rim. AD (Reynolds 1999, Fig. 32, Cat 22, for the context). 93. BEY Figure R/Nfr/H scar. Diameter: 12.5 cm top. Mid 2nd century AD. 94. BEY Figure R/N/Hfr. Diameter: 11.5 cm top inner rim. Double grooved band on upper neck. Mid 2nd century AD. 95. BEY Figure Handle fr. (Reynolds 1999, Fig. 31, Cat 133, for the context). AD BEY Figure Base. AD ? 97. BEY Figure R/N/2Hfr. Diameter: 11 cm top. Mid 3rd century AD. Late version of AM 72/Amphora 1, final phase of production. Small convex rim with indented lid seat. Well-arched handles, bearing a narrow groove (folded). Typical Beirut fabric. 98. BEY Figure Base. Early 3rd century AD. Base of Late Amphora 1. Ribbed. 99. BEY Figure Base. Early 3rd century AD. Base of Late Amphora 1. Ribbed BEY Figure R/N/Hfr. Diameter: 12 cm top. AD FAM 43, ie North Lebanese. Coarse brown fabric. Abundant fine-0.5 mm hard white lime (with even fracture). Common fine brown quartz. Moderate very fine black volcanic inclusions. Fairly hackly break with fine voids from missing inclusions. Moderate mm rounded argillaceous inclusions (Reynolds 1999, Fig. 33, Cat 23). BEY 015 Amphora 2A Grooved handle, upper section raised. Tall, cylindrical neck. Step on lower neck/upper shoulder, as Amphora 1. Bevelled triangular rim face. No concave inner face. Local version of Koan-style Dressel 2-4? 101. Lib 8. E Figure R/N/2Hfr. 100% of rim-neck surviving. Diameter: 11.6 cm inner rim. (Order 17). Amphora 2B Convex band rim face. No concave inner face. Grooved handle. Local Dressel 2-4? 102. Lib 12. E2, Nivel 13, Couche 954. Figure R/N/Hfr. Diameter: 10 cm inner rim. Amphora 2B. Light concave indent on inner rim. Rounded band rim. Handle raised. Central groove. Handle section is oval, not rectangular (as Amphora 1 and 2A). Brown core and orangebrown edges. Handle reduced greenish-brown. (Order 15) Square E2, Niveau 13, Couche 954. Figure R/N/Hfr. Diameter: 10 cm inner rim. Amphora 2B. Neck is a little troncoconical. An oval handle section. Reduced core and orange-brown edges-surfaces. Same ware as Lib 14 (Figure 14.4). Amphora 2C Rather elongated band rim, thickened on the top inner edge. Thick grooved handle. Note the

17 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 87 similarity of the rim type to that of north Lebanese amphorae Figure (AM 202) Lib 14. G Figure R/N/Hfr. Diameter: 12 cm inside. Amphora 2C. Large, thick, grooved handle. Rim is unique on BEY 015. Dark (grey) brown fabric and pale brown surfaces. (Order 36). Amphora 2D Wide, flat band rim, bent outwards, plain inner face. Handles raised and grooved, as usual for AM 72/Amphora 1-2 series. Unique Lib 43. Figure R/N/2H/Sh. Complete. Diameter: 14 cm top. Amphora 2D. Grooved handle. Orange red fabric with reduced core. Chemical analyses indicate this to be marginal within the range of BEY 015. (Order 82). Amphora 3 Short, tronco-conical neck and everted rim with inner edge bevelled and thickened to a point (to serve as a lid seat?). The pronounced ribbing on the outer neck is a distinctive feature. Unlike Amphora 1-2, the form has a Beirut-style handle and is much smaller (cf. AM 52, Figure 15.4) Cleaning Square D4. Figure R/N/H/Sh. Diameter: 10 cm inner rim top. Amphora 3. Reduced grey core and bright orange surfaces-edges. Moderate 0.5 mm lime Lib 5. Figure R/N/Hfr. Diameter: 10 cm top. Amphora 3. Fired orange -brown throughout. (Order 52) Figure R/N/Hfr. Diameter: 10 cm rim top. Amphora 3. Double groove on upper neck. Orange-brown with thin, dark-brown core. f) North Lebanese amphorae AM 202 A bevelled band rim, double groove on the upper neck, below the rim, with plain, wide, rectangular handles (atypical of Beirut products). The neck is large and tronco-conical. Probably with a sharply carinated shoulder-upper wall, and tronco-conical body (cf. a vessel of this type observed on BEY 004, courtesy of John Hayes). Rare on BEY 006. Beirut BEY BEY Figure R/N/H/ Sh. Diameter: 15 cm outer rim. Hard, well-fired and dense. Greyish pale brown fabric, with pale orange-brown outer surface. Abundant rounded fine 0.5 mm lime/limestone and quartz, with some fine fossil shell (or grey limestone). Moderate mm rounded red brown iron oxide pellets (or rounded basalt) (Reynolds 1999, fabric FAM 43 ). North Lebanese product. Upper level of a cistern deposit, ca AD BEY Figure R/N/Hfr. Diameter: 16 cm inside. Unique. Fabric seems to be local, but the rim type and large diameter suggest this is AM 202. (Reynolds 1999, Fig. 28, cat 130, for context and description). Red brown surface and dark yellow-brown core. Not a typical FAM 43 fabric. Closer to Beirut wares. AD BEY Figure R/N. Diameter: 10 cm inside. Small module of AM 202? Or a variant of AM 52 (Figure 15.4). This is a relatively common form on BEY 006 and 045. Other examples have a solid, rounded plain handle, like that of AM 52. Hard, over-fired. Rusty brown red surface with abundant lime dots. Granular, dark brown fabric with angular lime. Some fossil shell present. Fabric FAM 43C. AD AM 52 Thick-walled, tronco-conical neck, with narrow convex ribbing (cf. treatment of base Figure 15.7). Convex (sometimes bevelled) rim face. The handles have a plain rounded oval section and are attached to the rim face. It is possible that the form is related typologically to amphorae like Figure The majority of examples have a lime-rich fabric, but the best-preserved example, here illustrated, has less obvious lime and a quartz-rich fabric similar to that of Beirut amphorae, but it is rather dense, the surfaces being wet smoothed. Note that some non-beirut, north Lebanese amphorae and kitchen wares do occur without the characteristic lime (fabric FAM 43B). Perhaps this has been burnt out during firing.

18 88 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli 112. BEY Figure R/N/H/Sh. AM 52. Diameter: 9 cm inside. Pale grey brown core and pale brown orange surfaces. Compact, pimply surfaces. Quite dense. Occasional fine black volcanic glass. FAM 43B. Early 3rd century. Flagon-transport amphora 113. BEY Figure R/Nfr/H. Diameter: 20 cm. Medium-small module transport amphora. Flanged rim. Wide Beirut type handle attached to rim face. Hard. Pale red edges and dark yellow ochre core. Pale rusty orange to orange buff surfaces. Abundant fine- 0.5 mm angular lime (with some rounded grey fossil shell?) (fabric FAM 43C). North Lebanese, probably coastal (cf. shell). AD BEY Figure Complete vessel. Diameter: 9 cm. Small flagon-transport amphora. Abundant fine lime. North Lebanese. Lower cistern fill. Late 1st century AD. Yanouh 115. Large amphora base. Long solid toe. Thickwalled with narrow convex ribbing (as Figure 15.4). North Lebanese fabric with abundant lime. (Gatier et al 2001, Plate 1.1). g) Kingsholm 117 or similis The type is poorly defined. For comments on Kingsholm 117/Peacock and Williams Class 66, see Text. For comment and other illustrated examples of the type, generally with an ovoid body, see Pieri (2005, 115, 116, Fig. 75). Most illustrated examples lack the base. Lyon 116. Lib 65. Lyons, Place des Célestins, Figure Classified as Kingsholm 117 (Lemaître et al 2005, Fig. 8.1). (Order 92). Reims 117. LEV 597. Reims. Rue de l Équerre, Figure Classified as Kingsholm 117 (Lemaître et al 2005, Fig. 8.2). (Order 42). Dramont D wreck 118. Dramont D wreck. Complete vessel, classified as Kingsholm 117. Figure (Lemaître et al 2005, 521, Fig. 1.6, after Pollino 1983). Lyon 119. Lib 73. Lyons, La Favorite, Figure Kingsholm 117 similis. (Lemaître et al 2005, , Fig. 13; Becker et al 1986). Complete vessel. Dated to AD The vessel has a well-ribbed body from its base to its rim. The rim is convex and the toe is a hollow cone. A base to shoulder of the same type was found in a context of similar date in Beirut (BEY ). The fabric was close to that of local, Beirut products, in this case fired dark brown. The overall characteristics of this type, both form and fabric, resemble those of the wide-bodied carrot amphora Figure Lib 73 lay within the group of amphorae considered to be marginal within the BEY 015 analyses. The typological similarity of the rim of this piece and Figure should be noted. The Kingsholm type piece (its base is missing), like the latter, however, has a more cylindrical body, close to those of Gazan amphorae. Figure 16.4 has a more bag-shaped body that would link it to north Palestinian Late Roman Amphora 5 and their early Imperial predecessors (Pieri 2005, Fig. 76). (Order 83) BEY Figure Bfr. Hollow cone toe. Form as Figure 16.4 likely. Augustan. Reduced grey-brown surfaces, fairly rough break with common 0.5 mm lime (not visible on surface). g) Chhîm amphora 121. Chhîm. Complete amphora from a burial chamber (Reynolds 2004). Reynolds dated the amphora to the Byzantine period because it seems to be associated with the Byzantine presses on the site, but an early Imperial date for the type is also possible. The base type is distinctive, being a hollow, ribbed cone. The orange, lime-rich fabric is typical of Chhîm products. h) Collar-rim cylindrical amphora For comments on the type and bases illustrated, see the Text.

19 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 89 Beirut 122. BEY Figure R/2H/W. Diameter: 13.4 cm inner rim top. The base is unfortunately missing. The ware is compact, red brown, not very different from some Beirut products. Its chemical comparison with examples of Beirut 2 amphorae made at Jiyeh, macroscopically indistinguishable from Beirut products, alongside BEY 015 and other Beirut 2 amphorae is planned in order to resolve the identification of its source. Mid to later 1st century AD BEY Figure R/Shfr. Diameter: 12 cm inner rim. AD BEY Figure Base. Hollow, ribbed cone base. Same base type as the Chhîm amphora, but the ware is not that of Chhîm. It is close to that of Beirut products and may be the base type for vessels like Figure AD BEY Figure Base. Hollow, ribbed cone base. See comments on Fig AD BEY 015 Late Beirut 2 amphora? See the Text for discussion of the various possible attributions of this base Lib 7. Figure Small hollow cupped base with pronounced spiral ribbing. Over-fired dark brown fabric and green outer surface. Fabric as local. Common 0.5 mm lime. Chemically, this falls into the normal range of Beirut fabrics of BEY Cooking pots Cooking pot Lib 26. E Figure R/W/H. CP 1.1. Rim thickened and a little rolled. Dark reddishbrown fabric. Brown-grey surfaces. (Order 18) Figure CP 1.2. Waster. Diameter: ca 12 cm inside. Bubbles on handle surface due to misfiring. Light convex inner rim face, not as pronounced as CP Lib 20. E Figure R/W/H. Diameter: ca 13 cm inner rim. CP 1.2. Plain vertical rim. Beirut handle. Fired dark brown with grey-brown surfaces. Rough-local ware. (Order 21) Lib 13. E2 954, Niveau 13. Figure R/H/W. Diameter: 14 cm inside. Slight thickening on inner rim face. Beirut handle. Form 1.3A. (Order 25) Lib 18. F Figure R/H/W. Diameter: ca 13 cm inside. Form 1.3A. Tall rim, inner face thickened-folded. Beirut handle, light mouldings. Fired green, with a reduced fabric. (Order 26) E Figure R/W. Small diameter: 12 cm out. CP 1.3A. Rim thickened a little inside Lib 23. D Figure R/W/H. Diameter: 12 cm inner rim. CP 1.3A. Light Beirut mouldings on handle. Inner face foldedthickened. (Order 81) E Figure R/W. Diameter: 15 cm inner rim. Form 1.3B. Inner rim well bevelled. Plain light convex outer face. Light brown. Cooking pot Lib 48. E2 surface. Figure R/W/H. Diameter: 17 inside. CP 2. Round section handle. (Order 2) Lib 6. O Figure R/H/W. Diameter: ca 18 cm inner rim. CP 2. Beirut handle, shallow mouldings. Local with common fine lime. Could confuse with north Lebanese lime-rich fabrics. Dark brown/grey inside and pale red brown outside. (Order 63) Lib 34. Figure R/W/H. Diameter: 19 cm outer rim. CP 2. Round sectioned handle, not usual Beirut type. Very thin-walled. Fabric is normal. Orange brown fabric and pale brown surfaces. (Order 66) E Figure R/W. Diameter: 16 cm. CP 2. Cooking pot Lib 54. E Figure R/Shoulder. Diameter: 12 cm top. CP 3. Pale orange-red fabric with reduced core. Short slightly sloping collar neck with folded small band rim, concave top. Wide spaced ribbing on upper shoulder. The sloping collar is a feature of cooking pots of both the 1st century BC and the Flavian period. Note

20 90 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli that this piece was originally classified as an example of Kingsholm 117 similis or the collarrim amphora (cf. Figure ). (Order 39). Lid 140. Lib 36. Figure E R/W. Diameter: 26 cm inside. Order 11. Bowl E Figure R/W. Diameter: 31 cm. Reduced grey throughout. Jar Figure R/W. Diameter: 22 cm. Well bell-shaped rim with a step to the upper wall. Classic Beirut fabric. Reduced. Mortar Area D 4. Fig R/W. Diameter: ca 37 cm. Piece of clay attached to outer face. Double groove on hooked rim face. Well-known mortar form produced locally and probably also in the southern Beqaa. See Reynolds 1999 (Figs ). Here fired pale greenish yellow. Drain F5. Figure R/Flange/W. Diameter: 12 cm inner edge. Large piece, 100% of upper section, missing base. Dark grey-brown. Typical Beirut city, matt surfaces, reduced Lib 33. D Figure R/W. Diameter: 9 cm outer rim. Rough, reduced local fabric. (Order 10). Stands 146. Lib 42. D Figure R/W, start of base. Diameter: 13 cm outside. Folded band rim. Sampled, but failed. Not on dendrogramme. Over-fired, greenish fabric Figure R/W. Diameter: 13 cm top. Fired orange-brown Figure R/W. Diameter: 11 cm top. Orange brown fabric with a reduced core Figure R/W. Diameter: 13 cm top. Wide convex (folded) band rim. Fired salmon-orange Figure B/W. Diameter: 16 cm. Reduced fabric Figure B/W. Diameter: 17 cm. Reduced black surface and dark brown fabric with a black core Figure B/W. Diameter: 15 cm. Ware as oxidized AM 72. Dark brown fabric with bright orange edges. This piece was found with a carrot amphora handle/shoulder, an early 3rd century Beirut 4 amphora rim and a 3rd century AD cooking pot. BEY 006 Thin-walled wares Ware and forms as found on the nearby Beirut pottery production site GEM BEY / Figure R/W/B. Diameter: 8 cm. Local. Cylindrical beaker BEY / Figure R/W/H. Diameter: 8 cm. Local. Cup BEY / Figure R/W/H. Diameter: 9 cm. Local. Jar.

21 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 91 Bibliography Ala Eddine A 2005 The development of Beirut amphorae. A general approach. BAAL, Hors Série II: Becker et al 1986 Becker C, Constantin C, Desbat A, Jacquin L and Lascoux J P Le dépôt d amphores augustéen de la rue de la Favorite à Lyon. Figlina 7: Bjelajac L 1996 Amfore gornjomezijskog Podunavlja. Belgrade. Butcher K 2003 Roman Syria and the Near East. London. Carreras Monfort C and Williams D F 2002 Carrot amphoras: a Syrian or Palestinian connection? In The Roman and Byzantine Near East, 3 (Journal of Roman Archaeology Supplementary Series, 49): Dalby A 2000 Empire of Pleasures. Luxury and indulgence in the Roman world. London and New York. Domżalski et al 2005 Domżalski K, Wicenciak U, El-Tayeb M and Waliszewski T Late Hellenistic and early Roman pottery production center at Jiyeh. Recent excavations, In Polish Archaeology in the Mediterranean 16 (Reports 2004). Warsaw, Dyczek P 2001 Roman amphorae of the 1st-3rd centuries AD found on the Lower Danube. Typology. Warsaw. Ehmig U 2000 Cottana ermittelt: Syrische Feigen und andere Warenimporte. Augsburger Beiträge zur Archäologie 3: Étienne R and Mayet F 2002 Salaison et sauces de poisson hispaniques. Paris. Flower B and Rosenbaum E 1958 The Roman cookery book. A critical translation of the Art of Cooking, by Apicius. London: Harrap. Fulford M G and Peacock D P S 1995 Excavations at Carthage, The British Mission, Volume II, 2. The Circular Harbour, North Side: The Pottery (British Academy Monographs in Archaeology 5). Oxford. Gabriel Z unpublished Les pressoirs à raisin au casa du Metn-Nord au Liban (étude ethnoarchéologique), projet de diplôme d études supérieures. MA Thesis, Lebanese University, Dekwane, 2010 Gatier et al 2001 Gatier P L, Charpentier G, Harfouche R, Maqdissi M, Mercier F, Nordiguian L, Pieri D, Poupet P and Rousset M O Mission de Yanouh et de la haute vallée du Nahr Ibrahim. Rapport Préliminaire BAAL 5: Hamel, H unpublished Die Reste der Feste. Datierung und Funktion eines spätantiken Fundkomplexes aus Baalbek (Libanon). MA Thesis. Institut für Klassische Archäologie, Freie Universität Berlin Hayes J W and Ala Eddine A A transitional Byzantine-Umayyad pottery group BAAL 3: Jones Hall, L 2004 Roman Berytus. Beirut in Late Antiquity. New York. Kouwatli et al 2008 Kouwatli I, Curvers H H, Stuart B, Sablerolles Y, Henderson J and Reynolds P A pottery and glass production site in Beirut (BEY 015). BAAL 10 (2006): Lemaître S 2000a Les importations d amphores de Méditerranée orientale à Lyon au III e siècle ap. J.-C. In Rei Cretariae Fautorum Acta 36 (Selçuk-Ephesus 1998). Abingdon, Lemaître et al 2005 Lemaître S, Waksman S Y, Reynolds P, Roumié M and Nsouli B, A propos de l origine levantine de plusieurs types d amphores importés en Gaule á l époque impériale. Actes du Congrès de Blois, 5-8 mai SFECAG, Martin-Kilcher S 1994 Die Römischen Amphoren aus Augst und Kaiseraugst. (Forschungen in Augst 7). Augst. Parker A J 1992 Ancient shipwrecks of the Mediterranean and Roman provinces. (BAR International Series 580). Oxford. Peacock D P S and Williams D F 1986 Amphorae and the Roman economy. An introductory guide. Avon.

22 92 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Pieri D 2005 Le commerce du vin oriental à l époque byzantine (V e -VII e siècles). Le témoignage des amphores en Gaule. Beirut: IFPO, BAH 174. Pollino A 1983 Dix ans d archéologie sous-marine en Provence orientale en corse. Catalogue d exposition au Musée d Histoire et d archéologie. Bastion Saint- André. Antibes. Pollino A 1986 L Épave de la Tradelière In Exploitation de la mer. La mer, moyen d échange et de communication (Antibes), Reusch W 1970 Kleine, spitzkonische Amphoren. Ein Beitrag zur römischen Schwerkeramik. Saalburg Jahrbuch 27: Reynolds P 1999 Pottery production and economic exchange in second century Berytus: Some preliminary observations of ceramic trends from quantified ceramic deposits from the Souks excavations in Beirut. Berytus 43 ( ): Reynolds P 2000 The Beirut amphora type, 1st century BC-7th century AD: an outline of its formal development and some preliminary observations of regional economic trends. Rei Cretariae Romanae Fautorum Acta (Ephesus, 1998) 36: Reynolds P 2003 Amphorae in Roman Lebanon: 50 BC to AD 250. Archaeology and History in Lebanon (formerly National Museum News) 17: Reynolds P 2004 The Pottery in R Ortali and B Stuart Two rock-cut Roman tombs in Chhîm. BAAL 6 (2002): Reynolds P 2005 Levantine amphorae from Cilicia to Gaza: a typology and analysis of regional production trends from the 1st to 6th centuries. In: J M Gurt i Esparraguera, J Buxeda i Garrigós and M A Cau Ontiveros (eds), LRCW I, Late Roman coarse wares, cooking wares and amphorae in the Mediterranean: Archaeology and Archaeometry (Barcelona, March 2002) (BAR International Series 1340): Oxford, Reynolds P 2006 The pottery finds. In Kouwatli et al (2008): Reynolds P and Waksman S Y 2007 Beirut cooking wares, 2nd to 7th centuries: local forms and north Palestinian imports. Berytus 50: Reynolds P forthcoming The Classical and early Islamic pottery, for the final publication of the project Settlement and landscape development in the Homs region. (Durham University: Director, Graham Philip). Reynolds P in preparation The Classical Pottery. (The Archaeology of the Beirut Souks 4 Berytus). Roumié et al 2004 Roumié M, Waksman S Y, Nsouli B, Reynolds P and Lemaître S Use of PIXE Analysis technique for the study of Beirut amphora production in the Roman Period. Nuclear Instruments in Physics Methods B, 215: Roumié et al 2005 Roumié M, Nsouli B, Atallah C and Waksman S Y Application of PIXE using AI funny filter for cluster analysis of Byzantine amphorae from Beirut. Nuclear Instruments in Physics Methods B, 227: Sartre M 2001 D Alexandre à Zénobie. Histoire du Levant antique, IV e siècle av. J.-C. III e siècle ap. J.-C. Paris-Beirut. Schöne R and Mau A 1909 Corpus Inscriptionum Latinorum, Supplement 2, Pars III. Berlin. Sciallano M and Sibella P 1991 Amphores. Comment les identifier? Aix-en-Provence. Vipard P 1995 Les amphores carottes (Form Schöne- Mau XV). État de la question, SFECAG. Actes du Congrès de Rouen (1995). Marseille, Waksman et al 2003 Waksman S Y, Roumié M, Lemaître S, Nsouli B and Reynolds P Une production d amphores «carottes» à Beyrouth à l époque romaine? Revue d Archéometrie 27 (2003 published 2005): Waliszewski T et al 2008 Jiyeh (Porphyreon). Hellenistic, Roman and Byzantine settlement on the southern coast of Lebanon. Preliminary report on the 1997 and seasons. BAAL 10 (2006): Wicenciak U 2002 Preliminary information on the local pottery from Chhîm. BAAL 6: Wicenciak U 2008a Coarse pottery production at Jiyeh in Sector B1-B3. In T Waliszewski et al, 2008:

23 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 93 Wicenciak U 2008b Local coarse pottery. Material from Well B4 at Jiyeh. In T. Waliszewski et al 2008: Table 1. Order of sampled sherds on dendrogramme Fig. 4b All samples are Lib numbers, except where indicated. Samples taken in Gaul (Lemaître et al 2005) are indicated as Gallic. Analysis Lib Location Location and Form Figure Order sample 1 28 BEY 015 BEY 015. Beirut 3.1 rim 2 48 BEY 015 CP 2. Round section handle BEY 015 BEY 015. Beirut LEV 596 Gallic Carrot rim. Reims. Rue de l Équerre, 2002.B 5 10 BEY 015 BEY 015. Beirut 3 rim BEY 015 Beirut 3 base 7 50 BEY 015 Carrot rim. Type LEV 595 Gallic Carrot, hooked-folded rim. Reims. Rue de l Équerre, A 9 51 BEY 015 AM 72/Amphora 1-2 base BEY 015 BEY 015. Drain BEY 015 Lid BEY 015 BEY 015. Beirut Gallic Carrot base. Base Type B. Lyon. Place des Célestins, US Unstratified BEY 015 Beirut 3 rim BEY 015 Amphora 2B rim BEY 015 Carrot wall. Base Type B BEY 015 Amphora 2A BEY 015 CP BEY 015 Carrot base. Type B BEY 015 AM 72/Amphora 1-2 base BEY 015 CP BEY 015 Beirut 3 base BEY 015 Carrot wall. Type B BEY 015 Beirut 3 base BEY 015 CP 1.3A BEY 015 CP 1.3A BEY 015 Carrot rim Gallic Carrot wall. Lyon. Place des Célestins, US AD Gallic Carrot handle. Lyon. Place des Célestins. US Date uncertain

24 94 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Gallic Carrot wall. Lyon. Place des Célestins, US nd century AD BEY 015 Beirut 3 base BEY 015 Carrot base. Type A BEY 015 Carrot wall. Type BEY 015 Beirut 3 (rim) BEY 015 Carrot rim. Type BEY 015 Amphora 2C rim BEY 015 Beirut 2 base. BEY BEY 015 Carrot wall. Base Type A BEY 015 Cooking pot BEY 015 Beirut 3 rim LEV 594 Gallic Carrot handle. Reims. Rue de l Équerre, 2002.C 42 LEV 597 Gallic Reims. Rue de l Équerre, 2002 (Lemaître et al 2005, Fig ). Date: ca AD BEY 015 Amphora BEY 015 Carrot base. Type A BEY 015 Beirut BEY 015 Hybrid Beirut 3/grooved handle BEY 015 Beirut 2 base BEY 015 Beirut 2 or 3 variant BEY 015 Beirut 2 rim 50 7 BEY 015 Fig Hollow cupped amphora base BEY 015 Carrot base. Type A BEY 015 Amphora 3 rim BEY 015 Carrot wall. Base Type A. Marginal BEY 015 Beirut 3 rim BEY 015 Carrot base. Type B BEY 015 Carrot rim. Type A BEY 015 Carrot base. Type B BEY 015 Carrot base. Type B BEY 015 Carrot rim. Type A BEY 015 AM 72/Amphora 1-2 base BEY 015 Beirut 3 rim BEY 015 Amphora BEY 015 CP 2. Beirut handle. Common lime BEY 015 Carrot rim (type not noted: Type A?) Gallic Carrot rim. Lyon. Place des Célestins, US nd 8.3 century AD BEY 015 CP 2. Normal fabric BEY 015 Carrot wall. Base Type B Gallic Amphora Célestins Type 1A rim. Lyon. Place des Célestins, US (Lemaître et al 2005, Fig. 9.2). Early 3rd century Gallic Amphora Augst 47 handle. Lyon. Place des Célestins, US Early 3rd century 8.2

25 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology Gallic Amphora Célestins Type 1A rim. Lyon. Place des Célestins, US Early 3rd century Gallic Amphora Augst 47 wall. Lyon. Place des Célestins, US Early 3rd century Gallic Amphora Célestins 1A base. Lyon. Place des Célestins, US Early 3rd century 73 LEV 593 Gallic Amphora Célestins 1A base. Reims. Rue de l Équerre, (Lemaître et al 2005, Fig. 9.4) Gallic Carrot handle. Lyon. Place des Célestins, US Unstratified (Lemaître et al 2005, Fig. 10.3). Marginal Gallic Carrot rim. Place des Célestins, US AD BEY 015 Beirut 2 rim. Marginal Gallic Carrot wall. Lyon. Farges 75, C Late 1st century 8.8 AD. Marginal Gallic Augst 46 rim. Place des Célestins, US Residual 12.6 in Medieval context Gallic Carrot rim. Lyon. Place des Célestins, US AD Marginal BEY 015 Fig Beirut 3 rim. Marginal BEY 015 CP 1.3A. Marginal BEY 015 Amphora 2D. Marginal Gallic Complete amphora. Kingsholm 117 similis. Lyon. La 16.4 Favorite, Augustan (Lemaître et al 2005, 523). Marginal BEY 015 Carrot wall. Base Type B. Marginal Gallic Carrot wall. Lyon. Place des Célestins, US AD Marginal 86 LEV 590 Gallic Carrot wall. Reims, Carrot, BC-AD 15. Marginal Gallic Amphora Célestins 2A wall. Place des Célestins, US Early 3rd century Gallic Amphora Célestins 2A wall. Lyon. Avenue Alexandre Max, US nd to 3rd century AD Gallic Amphora Célestins 2A rim. Place des Célestins, US (Lemaître et al 2005, Fig. 11.1). AD Gallic Amphora Célestins 2A wall. Lyon. Place des Célestins, US Early 3rd century Gallic Amphora Célestins 2A wall. Lyon. Palais St Jean, US AD Gallic Amphora Kingsholm 117 rim. Lyon. Place des Célestins, US Early 3rd century (Lemaître 2000; Lemaître et al 2005, Fig. 8.1) Gallic Kingsholm 117 wall. Bram. Augustan (Lemaître et al 2005, 521) 94 LEV 587 Gallic Carrot rim. Reims, Crédit Agricole, F AD (Lemaître et al 2005, 518, Table) 95 LEV 588 Gallic Carrot handle. Reims, Crédit Agricole, F AD (Lemaître et al 2005, 518, Table) 11b

26 96 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Gallic Carrot or Kingsholm 117 wall. Lyon. Odeon 94. C st-2nd century (Lemaître et al 2005, 518, Table). Marginal 97 LEV 589 Gallic Carrot rim (kaolinitic clay). Reims, Crédit Agricole, F AD (Lemaître et al 2005, Fig. 7.13) 98 LEV 591 Gallic Carrot rim (kaolinitic clay). Reims, Crédit Agricole, F AD (Lemaître et al 2005, Fig. 7.12) 99 LEV 592 Gallic Carrot wall (kaolinitic clay). Reims, Crédit Agricole, F AD (Lemaître et al 2005, 518, Table)

27 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 97 Ras al Basit Latakia/ Laodicea Beirut/Berytus Khalde Jiyeh/Porphyreon Sidon Tyre Tartus Amrit Tripoli Byblos Chhîm Apamea Hama Homs Baalbek Damascus Akko Caesarea Figure 1 Map of principal coastal and other Levantine sites mentioned in the text (from a base map provided by Yona Waksman). Figure 2 Excavated sites referred to in the text.

28 98 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Figure 3a Bey 015 tank furnace complex phase 2.

29 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 99 Figure 3b Bey 015 kiln complex schematic.

30 100 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli références de Beyrouth amphores carottes autres amphores Figure 4a amphores Beirut 2 amphores Beirut 3 céramiques communes Beyrouth Gaule Célestins 1A Célestins 2A Augst 47 AM 72 Figure 2 : Classification ascendante hiérarchique des échantillons de référence de Beyrouth et des différents types d'amphores considérés. Classification basée sur les concentrations centrées et réduites de 11 éléments (cf. texte). Les principaux groupes de composition chimique sont soulignés. amphores carottes Beyrouth Gaule autres amphores K Célestins 1A Célestins 2A Augst 47 Kingsholm 117 LIB 28 LIB 48 LIB 9 LEV596 LIB 10 LIB 22 LIB 50 LEV595 LIB 51 LIB 33 LIB 36 LIB 25 LIB 85 LIB 35 LIB 12 LIB 56 LIB 8 LIB 26 LIB 1 LIB 46 LIB 20 LIB 47 LIB 2 LIB 49 LIB 13 LIB 18 LIB 39 LIB 67 LIB 83 LIB 69 LIB 21 LIB 45 LIB 53 LIB 27 LIB 52 LIB 14 LIB 32 LIB 38 LIB 54 LIB 15 LEV594 LEV597 LIB 11 LIB 57 LIB 24 LIB 31 LIB 16 LIB 17 LIB 30 LIB 7 LIB 61 LIB 5 LIB 58 LIB 29 LIB 62 LIB 40 LIB 63 LIB 64 LIB 37 LIB 55 LIB 4 LIB 44 LIB 6 LIB 41 LIB 82 LIB 34 LIB 59 LIB 81 LIB 90 LIB 88 LIB 87 LIB 89 LEV593 LIB 84 LIB 75 LIB 3 LIB 72 LIB 74 LIB 86 LIB 19 LIB 23 LIB 43 LIB 73 LIB 60 LIB 68 LEV590 LIB 77 LIB 79 LIB 76 LIB 80 LIB 78 LIB 65 LIB 66 LEV587 LEV588 LIB 70 LEV589 LEV591 LEV592 LIB 28 LIB 48 LIB 9 LIB 10 LIB 22 LIB 50 LIB 51 LIB 12 LIB 25 LIB 33 LIB 36 LIB 26 LIB 56 LIB 8 LIB 21 LIB 46 LIB 35 LIB 45 LIB 53 LIB 13 LIB 18 LIB 20 LIB 47 LIB 1 LIB 2 LIB 49 LIB 32 LIB 67 LIB 83 LIB 69 LIB 27 LIB 52 LIB 14 LIB 38 LIB 54 LIB 85 LIB 11 LIB 57 LIB 24 LIB 44 LIB 16 LIB 17 LIB 30 LIB 82 LIB 37 LIB 55 LIB 41 LIB 5 LIB 58 LIB 29 LIB 62 LIB 61 LIB 40 LIB 34 LIB 4 LIB 6 LIB 63 LIB 64 LIB 39 LIB 59 LIB 81 LIB 90 LIB 88 LIB 87 LIB 89 LIB 84 LIB 75 LIB 3 LIB 72 LIB 23 LIB 43 LIB 77 LIB 79 LIB 78 LIB 80 LIB 76 LIB 60 LIB 68 LIB 74 LIB 86 LIB 70 LIB Beyrouth I Beyrouth II 1 2 Figure 4b K K K KK Beyrouth 1 2 kaol. Figure 4a-b Dendrogrammes of chemical analyses of Beirut kiln (BEY 015) and Gallic samples (from Waksman et al. 2003; Lemaître et al. 2005).

31 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 101 a b c d e Sidonian 'Greco-Italic' ca 250 BC -150/125 BC Beirut 1b (Ala Eddine 2005, fig. 6: roughly to scale) Beirut 1d Early Augustan ca BC ca BC (AUB Museum Cat 483) Beirut 1a (Ala Eddine 2005, fig. 4) Beirut 1c (Ala Eddine 2005, fig. 6) h Beirut 3.1b ca f g i Beirut 3.2 ca j Beirut 3.3 ca 150 Mid 1st century AD 0 10 cm / Beirut 2 l k m Beirut 4a Late 2nd century Beirut 4b and c Early 3rd century q r s Late 4th century Beirut 5 u BEY / 30/033 Beirut 3.1a v Late 1st century AD x n o e-m: Early 2nd century to mid 3rd century: BEY 006 and 045 deposits p Mid 3rd century t and Beirut 8.1a AD y AD Beirut 5 (Ala Eddine 2005, fig. 21 roughly to scale) Beirut 7 ca AD 410 w Beirut 8.1b AD Beirut 8.1b AD 551 BEY 004 (Hayes and Ala Eddine 2000, Fig. 6) Beirut 8.2 ca AD 640/ 650 Figure 5 The development of the Beirut amphora, from its late Hellenistic late 3rd to 2nd century Sidonian predecessor (a) and its first appearance (in Beirut fabric) in the late 2nd century/100 BC (b), to the latest examples of the 7th century AD.

32 102 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli BEY BEY 015 Lib 16 Lib 3 Beirut 2 3 Lib 32 Lib 9 Beirut Lib 24 Lib Lib 35 Lib Lib 10 Lib Lib Lib Waster? 17 Lib Lib 47 Lib BEY 015 Baalbek amphora Homs Survey Lib 31 SHR 758, CAT Figure 6 Beirut amphorae, BEY 015 (1-20) and Baalbek amphora (21-22).

33 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 103 Carrot 1 3 Carrot 2 12 Carrot 3 Lib 52 1 Lib Lib 50 Lib Lib Lib Lib 1 Lib Lib Lib 57 Lib Lib 63 Lib 64 Lib 45 Lib 38 Thin-walled Type A Lib 61 Lib 60 Lib 59 Thick-walled wide, convex ribs Type B Figure 7 BEY 015 carrot amphorae. Type A (1-11) and Type B (12-20).

34 104 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Lib Lib Close to Beirut BEY 015 (marginals) 4 5 Lib 82 Lib 83 7 Not Beirut products 5 LEV 595 Beirut products LEV Lib 68 Lib Lib 72 Lib LEV 591 Carrot amphorae sampled from Gaul (From Lemaître et al. 2005, Fig ) LEV BEY BEY BEY BEY BEY BEY BEY BEY Athens Agora P 994 BEY Figure 8 Carrot amphorae: (1-11) BEY 015 and Gallic sampled sherds, (12-19) BEY 006 examples; (20. Agora P994).

35 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 105 Camoludunum 189 Type piece Figure 9a. Forms Figure 9b. Rim types Figure 9a-b Vipard (1995, fig. 1). Carrot amphora typology.

36 106 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Verulamium (Vipard 1995 fig. 4.6) 4 Straubing (Vipard 1995 fig. 4.8) Knossos (Vipard 1995 fig. 4.7) 6 Saalburg (Vipard 1995 fig. 6.3) 5 Ambrussum (Vipard 1995 fig. 5.3) Naples (Vipard 1995 fig. 3.8) London (Vipard 1995 Fig. 3.3) Figure 10 Carrot amphorae from western sites (from Vipard 1995): Beirut products?

37 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology Lib 89 Figure 11a & b Lyon. Célestins 1A / Colchester 105/ Peacock and Williams Class 65 ca AD (Lemaître 2000, Fig. 9.6 and 8).

38 108 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Athens P Augst 46-47/ Célestins 1A 'hybrid' Augst. Augst 46 (Martin-Kilcher 1994, fig 196.4) Augst.Augst 47 (Martin-Kilcher 1994 fig. 237, 5436) Lib 76 Lyon. Célestins 2A (Lemaître et al. 2005, fig. 11.1) AD Beirut. As 'Célestins 2A' BEY and 207 Mid 3rd century Lib 74 Lyon. 'Augst 46' c. AD (Lemaître et al.2005, fig. 10.3) Figure 12 Athens, Augst, Lyon and Beirut. Small module amphorae related to Augst (not Beirut products: south Phoenician, from Akko/Ptolemais?).

39 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology Amphora 1/'AM 72' BEY Lib Lib 55 Lib 46 Lib 44 Lib 51 Amphora 1/'AM 72' BEY BEY BEY Mid 2nd century AD BEY Mid 2nd century AD BEY AD BEY AD ? 12 BEY Mid 3rd century AD Late Amphora 1/'AM 72' BEY BEY ca Amphora 1/AM 72 North Lebanese version Late Amphora 1/ AM 72 bases BEY and BEY Early 3rd century Figure 13 Amphora 1/AM 72, local (1-14) and north Lebanese (15).

40 110 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Form 2A Form 2C BEY 015 Form 2 4 Lib 14 1 Form 2B Lib 8 Form 2D 2 Lib Lib 43 BEY Lib 5 Form 3 6 Figure 14 BEY 015. Amphora 2A-D and Amphora 3. 8

41 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology BEY 006 BEY AD AM 202 BEY ca. AD Yanouh small AM 202 BEY c Gatier et al. 2001, PL AM 52 BEY Early 3rd century 5 BEY ca BEY Late 1st century AD Figure 15 Beirut and Yannouh. North Lebanese transport amphorae, small to large modules.

42 112 Paul Reynolds, Sylvie Yona Waksman, Séverine Lemaître, Hans Curvers, Mohamed Roumié and Bilal Nsouli Lyon 1 2 Lib 65 LEV 597 Reims Kingsholm 117 from Lyon and Reims Lemaître et al. (2005, fig ) 3 4 Lyon. La Favorite Kingsholm 117 similis. 1st century AD Lemaître et al. (2005, fig. 13) Lib 73 5 Beirut. Augustan Kingsholm 117? BEY Dramont D wreck (after Pollino 1983) Lemaître et al. (2005, fig. 1.6) Figure 16 Kingsholm 117 or similis: sampled Gallic examples (1-2) complete examples from the Dramont D wreck (3) and Lyon (4), and a base (Beirut, BEY 007) (5).

43 An early Imperial Roman pottery production site in Beirut (BEY 015): chemical analyses and a ceramic typology 113 Chhîm Beirut. 'Collared rim' amphorae Local? 2 3 AD BEY Chhîm amphora Mid 1st century AD BEY / Lib 7 BEY BEY BEY Beirut. Hollow cone bases AD Local? Figure 17 Chhîm. Chhîm amphora type (1); Beirut (BEY 006) collared rim amphorae: local? (2-3); Beirut (BEY 006), hollow cone bases (4-5); BEY 015, hollow cone base (Lib 7) (6).