Late Roman African Cookware of the Palatine East Excavations, Rome: A holistic approach 9781841715155, 9781407325460

Table of contents :
Cover
Title Page
Copyright
CONTENTS
LIST OF TABLES
LIST OF ILLUSTRATIONS
ACKNOWLEDGEMENTS
INTRODUCTION
FABRICS
FORMS
PRODUCTION
DISTRIBUTION AND CONSUMPTION
SUMMARY
BIBLIOGRAPHY
Plates
APPENDIX

Citation preview

BAR S1143 2003 IKÄHEIMO: LATE ROMAN AFRICAN COOKWARE OF THE PALATINE EAST EXCAVATIONS, ROME

B A R

Late Roman African Cookware of the Palatine East Excavations, Rome A holistic approach

Janne P. Ikäheimo

BAR International Series 1143 2003

Late Roman African Cookware of the Palatine East Excavations, Rome A holistic approach

J anne P. ll deposits

1.

40/50 - 80/90

2.

80/90 - 100/110

38 vessels

PEA horizon 1 > 200, 215; PEA horizon 2 > 177, 181, 182, 186, 187, 265; PED horizon 2 > 89, 98, 99, 185, 209, 219; PED horizon 3 > 65, 72, 80, 81, 83

3.

100/110 - 150/170

31 vessels

PEA horizon 3 > 173, 174, 175, 176, 180; PED horizon 4 > 85, 204.

4.

150/170 - 190/200

153 vessels

PEA horizon 4 > 171, 172, 194; PED horizon 5 > 43, 122, 131, 151, 165, 206

5.

220/230 - 270/290

12 vessels

6.

270/290 - 300/310

471 vessels

PEA horizon 6 > 93, 97, 104, 117, 118, 123, 124, 125, 126, 127, 130, 132, 135, 137, 140, 141, 144, 149, 150, 153, 157, 160, 161, 162, 163; PED horizon 6 > 53, 57

7.

300/310 - 310/320

411 vessels

PEA horizon 7 > 36, 37, 38, 39, 66, 92, 94, 96, 105, 113; PEB: 366

8.

310/320 - 320/325

609 vessels

PEA horizon 8 > 16, 33, 34, 48, 49, 62, 67

9.

320/325 - 350/360

251 vessels

PEA horizon 8* > 20, 22; PEB eastern barrel vault (ebv) horizon 1 > 352, 355, 357, 360, 361, 362, 363; PEB (ebv) horizon 2 > 351

10.

350/360 - 400/425

94 vessels

11.

400/425 - 525/550

310 vessels

12.

modern etc.

11 vessels

PED horizon 1 > 97, 137

PEA horizon 5 > 112

PEA horizon 9 > PEA: 12; PEB (ebv) horizon 3 > 350, 353 PEB central barrel vault horizon 1 > 180, 186, 206, 209, 222, 223, 228, 276, 281, 286; PEB (ebv) horizon 4 > 340, 341, 342, 348, 349 PEB (ebv) horizon 5 > 270, 271, 295, 309, 310. PEA: 45, 139, 179; PEB: 359, 364.

followed the pattern, where the initial peak is followed by a rapid or long-term decline, the date of a given form can obtained through an aggregate of deposits in which it occurs.82 This hopefully minimizes the effect of the residual component, as “the chances of finding and recognizing a sherd from a pot which had an extended life for its type are remote and the likely explanation is that the piece in question is residual.”83

sector and the number of the deposit (e.g. PEB 105, class 34 African cookware). All the bags belonging to the same deposit were stored together. Later on, the work was revised by the researcher responsible for further processing of that particular class. The excavations yielded high quantities of pottery body sherds, the abundance of which became a problem after only a few field seasons in respect of the storage capacity they filled up. Therefore, after the general class (e.g. African cookware) of the body sherds had been determined, they were counted, weighed and eventually discarded. The documentation of African cookware was accomplished by entering the data directly into a portable computer running Microsoft FoxPro 2.6 database software. At the early stage of documentation a separate database file was created for each vessel form by dividing the assemblage into the principal forms within every deposit. This system proved extremely inconvenient, both because the data was slow to enter and hardly manageable. Later on, all the separate files were united into one database including all the African cookware from the late Roman deposits of the Palatine East excavations. The arrangement proved itself very effective in data-processing, although the usefulness of a database is principally determined by the parameters it contains. Since the introduction of the parameters used in the present study might be of some general value to

1.3.3 Finds processing This chapter is intended to present the procedure to which African cookware finds were subjected after their recovery from the soil at the Palatine East excavations, as well as to discuss some issues related to the methodology of pottery studies. The former topic includes a description of those parameters that were inserted into a computer database, whereas the latter part of this section seeks to state arguments for their usefulness. As the method used to process the pottery loosely followed the standard set by the excavations of the British Mission at Carthage84, it is only summarized here. First, the pottery was separated from the other finds in the same deposit, washed with water and soft brushes and placed on drying racks set in the sunlight. This was followed by an initial sorting of the material into respective classes of pottery. Each class was then put in a separate plastic-bag together with a cardboard tag indicating the class of pottery, the

9

Chapter 1

the study of Roman pottery, the opportunity is taken here to describe and discuss them in some detail. In addition, some suggestions on the ways the recording of these parameters could be improved in the future are also offered. Twenty one variables were recorded from each African cookware vessel. As indicated in the following table (Table 2), most of the variables are based on estimations and only a few measurable variables (5-8) have been recorded. Variables indicating the location of each find (1-2) are self-explanatory, whereas number codes (3-4) based on the working typology are used to indicate an individual form and variant. The assemblage strengthened the picture of African cookware as a highly standardized product,85 because only a few diagnostic sherds were too small or worn down to be identified. At the final stage of the study, these codes were converted into the typology presented in the next section. The rim diameter (5) was determined with a standard diameter-ring chart, whereas the other two variables producing numerical data for the study – the height and width of the rim (6-7, see Figure 4) – were measured with calipers. Although these measurements have always formed an essential part of the standard documentation procedure of archaeological ceramics, one can hardly find an article, in which such data has been put to effective use. In this sense, the following pages will positively demonstrate that such measurements are not useless trivia we use to fill our note books with, but rather an essential aid in the examination of a typology of a vessel form or even several forms related to one another by their function. The last measurable variable is – despite its name – the estimated vessel equivalent (6), which was determined with a modified diameter-ring chart divided into sectors, each representing 1% of the total circumference of a vessel rim. As a rule, estimated vessel equivalents were measured from nucleated sherd families i.e. all the sherds belonging to the same vessel found in the same context. With more time and effort spent on the restoration of pottery, it would have been possible to group the material by extended sherd families, which include all the sherds from the same vessel irrespective of the deposit they were found in. However, in terms of the aims of this study it was

considered useless. A by-product of the chosen procedure was an estimate on the minimum number of vessels represented in each deposit. Therefore, the quantification of the study assemblage will be carried out with three methods (sherd count, minimum number of vessels and estimated vessel equivalents), the result of which is compared in section 5.1. The rest of the variables are mainly estimations. The majority is related to various stages of production, although some also indicate the use of African cookware. Pottery fabrics (9) were documented by breaking a fresh chip from each vessel and comparing it with a reference collection glued to index cards (which will be later on referred to as chip cards).86 In fact, the most effective method to register and control the variability of fabrics is to remove a small sample from every vessel. The procedure, which may sound ridiculous at first, is not so timeconsuming or exhausting, as it is generally believed to be. When the sampling is accomplished as a continuous task, it takes approximately an hour to take 50 samples and glue them to an index card, while with some experience ca. 250-300 samples can easily be fitted into a card measuring only 5x3 inches. For example, it took only nine index cards and ca. 40 hours of time to sample all the vessels examined in this study. The profit of the procedure is that the task of fabric determinations is moved from the field to laboratory conditions and the fabric of each vessel can be checked as many times as necessary, without laborious searches in storage rooms. The fabric classification presented in the next chapter, however, was established with the following procedure. A fresh fracture surface was exposed from each sherd or nucleated sherd family by removing circa 10x5 mm chip with pliers. Then, each chip was examined under a binocular microscope and attributed to a preliminary fabric class. A sufficient number of examples of each fabric class was glued to chip cards.87 A new chip card was started for every recognized fabric or fabric variant. The fabric classification was first accomplished separately for each vessel form (e.g. deep casseroles), but in the second phase of fabric studies they were united to form a master fabric classification.

1. sector

8. estimated vessel equivalent

15. patina cenerognola

2. context

9. fabric

16. rim soot

3. form

10. exterior surface finish

17. exterior surface soot

4. variant

11. interior surface finish

18. interior surface soot

5. diameter

12. exterior firing

19. wear marks

6. rim height

13. interior firing

20. number of fragments

7. rim width

14. core firing

21. other observations

10

Table 2. Variables recorded from each African cookware vessel in the study assemblage.

Introduction

vessels fired completely in oxidizing atmosphere, it became evident that a sample card with several examples of each firing pattern and running number or letter code would be also useful in this purpose. As many forms of African cookware are characterized by a black, soot-like surfacing (Plate 2h), which has been defined either as patina cenerognola or orlo annerito depending on the form it is found in, one entry (15) in the database was reserved to verify its presence as well as to indicate its color, the latter of which was recorded with a number code. From time to time, it was difficult to determine whether the black layer covering the vessel was to be interpreted as patina cenerognola or as a uniform layer of soot resulting from its intensive use. The advantages of use-alteration studies are obvious in case of cookwares, which frequently show two different forms of use-alteration. 89 Firstly, by studying the extent and density of soot-deposition on vessel surfaces, something can be deduced about the mode of cooking, and in favorable cases the substances processed can be identified. To note the presence of soot, the vessel surface was divided into three zones – the rim, the exterior and the interior (1618) – in which it was recorded separately with a fivestage scale including values from “absent” to “dense”. Although the documentation of this aspect had a greater importance in the original research plan, the registration of every possible soot pattern variant proved laborious if not impossible, partly because at the Palatine East excavations all the pottery had been cleansed with water and a brush. Another set of evidence results from the application of devices designed to modify the cooking atmosphere and temperature (e.g. lid), stir cooked substances (e.g. spatula) or clean the vessel after each heating cycle. Results of these tasks can be noted on vessel surfaces in form of patterned abrasion, scratching and other use-wear (Plate 2a-b, see also Figure 16). Their presence was also noted from the assemblage (19). However, these records turned out to be less useful than was initially expected, because the origin of a given wear type is often difficult to assign to a certain mode of use etc.90 As the assemblage was recorded by sherd families, one entry was reserved to indicate the number of sherds belonging to the same vessel (20). When used alone, the number of sherds is a somewhat inaccurate variable, especially regarding quantitative studies, sherd counts can be used with estimated vessel equivalents to determine the brokenness of the ceramics in a given archaeological unit. The last entry in every row of the database was reserved for additional observations (21).

h w

w 1.

4. h

w

w

h 5.

2.

w

h w

3.

6.

Figure 4. Points used in measuring rim height (h) and width (w). 1. A.I lid (Hayes 196); 2. B.I lid (Hayes 195); 3. F.II ring-handle; 4. O.II deep casserole (Hayes 197); 5. J.II shallow casserole (Hayes 23 B); 6. H.II pan-casserole (Hayes 26/181). Scale: 1:1. Some observations concerning various types of finishes detected both on the exterior and the interior surface were also made (10-11). The method of recording surface finishes with three-lettered codes, such as bsm (smoothed and burnished surface) showed itself fairly useful,88 although during the period of documentation work, the necessity of maintaining a reference collection with at least one example of each surface finish was noticed. The same could be said of a reference collection that would include all the variability resulting from differences in firing conditions. At present, the recording of this attribute was accomplished in an overly straightforward manner. The fracture surface of a sherd was simply divided into three imaginary zones: the exterior, the core and the interior (12-14) each of which was then assigned one of the following values: oxidized, partly reduced or reduced. Although the assemblage was dominated by

11

Chapter 1

CHAPTER 1 ENDNOTES 1

2 3 4 5 6 7 8

9 10 11

12 13

14 15

16 17

18 19 20 21 22

Tite 1999, 181-182; Sillar & Tite 2000, 14: Cumberpatch 2001, 270; Kingery 2001, 273; Tite & Kilikoglou 2002, 4. Millet 1987, 106-107. Tite 1999, 181-182; Sillar & Tite 2000, 3, 14. Kingery 2001, 273. Pavolini 1985, 201; Gandolfi 1994, 149. E.g. Panella 1991, 295, figs. 12-13. Fulford 1987, 59. Although a number of important studies touching this type of pottery has been published recently, they are centered either on source (e.g. Carthage and Leptiminus), trade routes (various shipwrecks) or intermediate stations (Ostia) rather than on one of its principal destinations, the city of Rome. In addition to a general lack of published 3rd and 4th century AD deposits from the capital of the empire (Pavolini 1996, 230), the only significant African cookware assemblage reported up to the present from Rome is the one from the church of San Sisto Vecchio, which cannot be dated accurately. Shorter reports have appeared on the excavations of S. Stefano Rotondo and the Caelian hill, see section 5.2.1. The term henceforth substituting "the Late Roman ceramic assemblage of the Palatine East excavations". On the definition of technological choice, see Sillar & Tite 2000. There were also other reasons for the negligence of commonwares, a group of ceramics which African cookware also belongs to. The lack of decoration is paralleled with lack of stamps and inscriptions, the existence of which would shed new light upon various aspects of this production. Furthermore, the vessel shapes of this class do not evolve rapidly, as was usually the case with decorated pottery. Moreover, the frequency of commonwares on most archaeological sites must have discouraged scholars from studying this abundant material, van der Werff 1982, 407; Panella 1996, 9-10. Finally, it was not until the 1970's that the fact that all the common wares did not necessarily represent local production was realized; rather they were also subject to exportation, sometimes even over considerable distances, Pavolini 1985, 201; Coletti & Pavolini 1996, 391. Skibo & Schiffer 1995, 80-82. Before the 1970's virtually all the progress in the study of ceramics produced in Roman Africa was made with other classes of pottery than cookware. For this reason, several paragraphs both outlining the development of African tablewares and pointing out their strong alikeness to cookwares of the same origin, have been included in this study, cf. Carandini 1983, 152. Rickman 1980, 118, 195-198; Fulford 1987, 67; Panella 1999, 188. The arguments concerning the questions why and how pottery should be used in the reconstruction of the Roman economy and social history have already been discussed in detail by Peacock (1977b, 21-25) and therefore I do not find it necessary to repeat them here. The present state of archaeological research on Roman Africa is reviewed in Mattingly & Hitchner 1995. The main reason for the late beginning can be attributed to the relatively modest appearance of African pottery. As even African Red Slip ware – a fineware also known as African sigillata – bears more often than not only monotonously repeated stamped decorations, African pottery did not manage to attract the attention of art-history oriented scholars of the early 20th century, see also Orton et al. 1993, 5-8. Falbe 1833; Gauckler 1897. Waagé 1933; Waagé 1948. Lamboglia 1941; 1958; 1963. The group initially identified as terra sigillata chiara B was later shown to have been produced in Gallia Narbonensis. The following summary describing the production of African Red Slip ware is based on: Salomonson 1968; Carandini 1970, Hayes 1972; Tortorella 1981b; 1987; 1995; Carandini 1983; Schuring 1988; Gandolfi 1994; Panella 1999. The production of tablewares was begun in Northern Tunisia, possibly in the area surrounding Carthage, with African Red Slip ware A (ARS A) from where it was first exported around AD 70. The date is confirmed by scattered examples found in the centers buried in the eruption of Vesuvius in AD 79. The forms belonging to early production were clearly influenced by Italian and Gallic Sigillata. However, exportation on a larger scale begun only after the reign of Domitian. From the Antonine era onwards, possibly up till ca. AD 300 ARS A virtually dominated tableware markets of the western Mediterranean. The production of African Red Slip ware C begun probably between AD 175-230 in Central Tunisia when the economic center of Africa Proconsularis shifted gradually from Carthage towards the south. Reynolds (1995, 12) dates the period of its dominance to AD 230-320. This production, distinguished from its predecessor by the quality of both fabric and slip, seems to have persisted until the mid-5th century AD. The success of this tableware both i n the markets of eastern and western Mediterranean was based on standardized, high quality mass products fabricated with molds. The production of African Red Slip ware D was started in the northern Tunisia around AD 275-325. In the absence of corroboratory archaeological evidence and because the forms are derived from ARS C and not from ARS A, this took very unlikely place in the same workshops that had manufactured ARS A. The mold-made ARS D vessels, the heyday of which is dated to AD 350-450, are frequently decorated with stamps. After the Vandal conquest of Roman Africa (AD 429-533), especially following the fall of Carthage in AD 439, ARS D was manufactured and exported more sporadically. However, the production surely did not persist the blow, which the invasion of Arabs (AD 647-698) gave to the last remains of late antiquity in North Africa. In addition to three major classes, several minor productions have been identified. These include: African Red Slip ware A/C that was produced in Central Tunisia from the early 3rd century AD to the AD 280's; African Red Slip

12

Introduction

23 24

25

26

27

28 29

30

31

32 33

34

35 36

ware A/D that was also produced in Central Tunisia, but from the late 2nd to the mid-3rd century AD; African Red Slip ware C/E, yet another Central Tunisian product that is dated AD 225-375 and African Red Slip ware E, the fabric variant that has been commonly regarded as the last Roman tableware fabric produced in North Africa. ARS E also shows how the focus of pottery production shifted between AD 450-550, for the second time, from the surroundings of Carthage towards the east coast of Central Tunisia. Salomonson 1962; 1968; 1969. The results of this campaign were published in four volumes Ostia I-IV. The first volume (Ostia I) presented material from two layers dated to the second quarter of the 3rd century AD (225-250 AD). The following volume (Ostia II) introduced only the finds dated between 80-100 AD (layers VA-VD) and 50-100 AD (Phase 1: layer IV), despite of the identification of three other phases (Phase 2: layer III, 115-180 AD; Phase 3: layer II, 180-230 AD and Phase 4: layer I, 220-250 AD). The volume of Ostia III includes material from three different contexts dated between the late Republican period and the mid-5th century AD, while the discussion in Ostia IV focused in two layers: I (ca. 250-400 AD) and II (225-275 AD). While the fifth volume of the Ostia excavations is expected to be published in the near future, further information has been offered in two preliminary reports, the purpose of which was to present a very large pottery dump, dating from the Flavian to the Antonine periods excavated just northeast of the baths, see Panella 1991 and Coletti & Pavolini 1996. Carandini, who can be indirectly credited for the establishment of these classes, doubted their existence from the very beginning (Ostia III, 410). For British scholars they are probably better known as “Black-Top ware” (Hayes 1976, 87), while other definitions include Spanish la ceramica cenicienta (e.g. Adroer 1963, 105-110) and German Schwarzrandware (e.g. Rüger & Niemeyer 1962, 112-113). Instead of this or the other definition, African sigillata A a strisce, used to describe this class, the term stripeburnished African cookware will be used throughout this study. As the custom to classify African cookware into three groups has been superceded only recently (cf. Tortorella 1981b, 208), the characteristics of each class are reviewed here in brief. The term ceramica ad orlo annerito sought to describe a blackened band that usually covers vessel rims in this group. It is believed to be a result of the way vessels were stacked on top of each other in kilns and the changes in the firing atmosphere (see section 4.3.1). The only product attributed to this class was the so-called lid-plate, the contested function of which will be extensively discussed in section 3.9.1. The class of ceramica a patina cenerognola – the name of which sought to describe an area of blackened surface usually covering the whole upper exterior surface – comprised all remaining African cookware forms. The most characteristic forms of this class are various shallow and deep casseroles as well as pans, some of which are also attested in ceramica da cucina polita a bande i.e. stripe-burnished African cookware. The study of African cookware has further been complicated by the persistent tradition to classify some of the most common forms – Hayes 23A, Hayes 23B and Hayes 181 in particular – as African Red Slip ware. Vegas 1973. Hayes 1972. A supplement adding new elements to the information presented in Late Roman Pottery was also published by Hayes (1980). Interestingly, Carandini (1983, 161) criticizes the common introduction of African table- and cookwares not only because of their different function, but also due to their presumed differentiated value, cf. section 5.3.3. An illustrative example of the attempts to study Roman common wares is Stephen Dyson's Cosa: The utilitarian pottery (Dyson 1976), which introduces eight deposits dating between the early 3rd century BC and the early 4th century AD. Although the original intention was presumably to present a chronology and a typology for local commonwares, several examples of African cookware were mistakenly included in the catalogue. Their presence, however, was recognized as "unusual types are the pieces related to Terra Sigillata Chiara that appear in the later deposits." Later on, this work was supplemented by a publication dealing with cookwares from the Roman port and fishery at Cosa (Oleson & Oleson 1987). Tortorella 1981b. Tortorella has also updated his work twice, see Tortorella 1987; 1995. The British excavations at Carthage yielded over 15 metric tons of pottery. The first report dealt with the excavations of Avenue du Président Habib Bourguiba that produced some 4 metric tons of pottery dated AD 400-650. The study of coarsewares included over 1000 diagnostic sherds and was one of the first reports in which necessary attention was paid to fabric descriptions, see Fulford & Peacock 1984. The second pottery report described the finds dated between the 2nd century BC and the mid-2nd century AD, which originate from the area north of the circular harbor, see Fulford & Peacock 1994. A brief report on the late 2nd/mid-3rd century AD finds from the area south of the circular harbor has also been published, see Tomber 1986. The six volumes introducing the University of Michigan excavations include three pottery reports, which present finds from deposits dating from the 1st to the 7th century AD (Hayes 1976; 1978; Riley 1981). In addition, the second Canadian expedition at Carthage has published some 200 kg's of pottery found in the mid-5th century (AD 425-475) deposit in 1978, see Neuru 1980. The excavations at the circus and the Byzantine cemetery of Carthage yielded three contexts rich in pottery (Tomber 1989, 437, 441-504). The largest of them was a re-deposited dump dated to the late 4th/early 5th century AD, which contained a comprehensive and informative selection of late African cookware forms. The two other deposits dated from the late 1st to the early 2nd century AD, and from the mid-5th t o the 7th century AD. The most recent pottery publication describes the finds related to the early Roman colony of Carthage, see Freed 1998. Ben Lazreg & Mattingly 1992; Mattingly et al. 2001; Striling et al. 2001. Dore 1988, 61-63. The UNESCO Libyan Valleys Survey (ULVS), for example, succeeded in providing new information on the distribution of Roman pottery based upon 43,000 recovered sherds from some 1100 sites. The results of this survey are summarized in Greene 1986, 127-132; 1992, 46. At Sabratha, the excavations of the British

13

Chapter 1

37

38

39

40

41 42

43 44 45

46 47 48

49 50 51 52 53

54

55 56

57

58 59

60 61 62 63 64

expedition yielded nearly 4000 coarse ware vessels, most of which date from the 3rd century BC to the 4th century AD (Dore 1989, 87), while the report of the Italian expedition (Pucci 1975) on the excavations of so-called mausoleum is another significant contribution to our knowledge on the distribution of cookwares in western Tripolitania. The report on the excavations of Berenice (Benghazi) introduces coarseware finds dated from the Republican period t o the early 6th century AD, see Riley 1979. After the Ostia reports, several important works, each including a section devoted to African cookware, have been published in Italy. They include pottery reports on the excavations of Luni (Luni II), villa of Settefinestre (Settefinestre II), Albintimilium (Olcese 1993) and the church of San Sisto Vecchio in Rome (Schuring 1988). Studies presenting the distribution of African cookware in Roman Spain, have also come out. They include works o n African pottery from Baetulo (Aquilué Abadías 1987) and Tarragona (Aquilué Abadías 1991), which have been later o n complemented by a comprehensive study on imported cookwares from ager Tarraconensis (Aguarod Otal 1991). Two recent compilations (Ibèrica 1995; Contextos 1997) devoted to the Late Imperial coarsewares on the Iberian peninsula also include several articles on African cookware. A full-length discussion on this topic can be found e.g. in: Hayes 1972, 296-297; Tortorella 1981b, 209; Carandini 1983, 145-146. Both African table- and cookware were also manufactured to meet the local consumption together with more elaborate and high quality products destined for exportation. It has been suggested that the term African Red Slip ware should be reserved to describe the fineware fabrics exported on a regular basis, see Hayes 1972 Tortorella 1981b, 209. The fabric of African cookware vessels is usually orange-reddish in color and somewhat coarser than any African tableware fabric due to the frequent presence of medium-sized quartz grains, which are present in the paste either as a natural component of clay or as a temper used to prepare clay otherwise poor in inclusions. Lavoie 1989; Peacock et al. 1989; 1990; Peacock & Tomber 1991; Mackensen 1993. Freed 1998, 21 (note 43), 31, 37. Hayes 1972, 17-18; 1978, 75; 1980, 518. The Early Cooking ware I has recently been identified as a product native to Carthage region, while the Early Cooking ware II that occurs at Carthage from the early 2nd century onwards was probably produced in Byzacena, Freed 1998, 21. On the production of cookwares in Byzacena during the Republican period, see van der Werff 1982, 110-112. Olcese 1993, 137. Freed 1998, 30-37. In Central Italy, evidence of the early 1st century AD African cookware imports comes from layer C2 in the House of Yellow Walls at Ostia (Zevi & Pohl 1970, 75-76) and deposit 22II of Cosa (Dyson 1976, 115-138, see also Hayes 1979, 75, note 2). While the few examples found in contemporary contexts at Settefinestre have been considered intrusive (Settefinestre II, 118), the plentiful evidence from Spain is likely enough to confirm the early 1st century AD date, see e.g. Aquilué Abadías 1985, 210-211; Beltran Lloris 1990, 136-138; Aquilué Abadías 1991, 970; Aguarod Otal 1991, 239. van der Werff 1982, 112; Freed 1998, 31, see also Hayes 1976, 93; 1978, 75; Tortorella 1981b, 211. Freed 1998, 24-25, 31. Tortorella 1981a, 358; Pavolini 1985, 201; Aquilué Abadías 1995, 71; Freed 1998, 37. Pavolini 1996, 226-227. An illustrative example of the status of African cookware in Pompeii can be pointed out from the House of Vestals (Regio VI, Insula I), which has been excavated since 1994 by the Anglo-American Project in Pompeii, see e.g. Bon et al. 1997. The careful exploration of the whole building complex has produced only six diagnostic sherds of African cookware, which the author has identified in person: two rims of an early Hayes 196 lid (see variant A.I), a fragment of a ring-handle (see form F), two casserole rims and a protruding carination of an unidentified shallow casserole form. The other examples of this date from the Bay of Naples area comprise fragment(s) of the Hayes 194 casserole and some sherds of uncertain provenance, see Carandini 1970, 108; 1977, 24. At the Palatine East excavations, for example, the quantities of African cookware found in deposits dated before the 3rd century AD are the following: AD 40/50-80/90, 4 vessels; AD 80/90-100/110, 38 vessels; AD 100/110150/170, 31 vessels; AD 150/170-190/200, 153 vessels; cf. Volpe 1998, 240-241. Tortorella 1981b, 211. Not surprisingly, two phases have also been distinguished in the production of Central Tunisian cookwares, of which the latter – characterized by products of somewhat higher quality – is dated after AD 175/200, van der Werff 1982, 110-111, 124. Hayes 1972, 423; Gandolfi 1994, 151. The production of Central Tunisian cookware for exportation may have already ceased in the early 4th century, while on local scale the production possibly continued to the 6th or even t o the 7th century AD, van der Werff 1982, 85-87, 124, 390. Tortorella 1983b, 130; cf. Tortorella 1981b, 208; Gandolfi 1994, 151. Riley 1981, 88-89; Tortorella 1986, 220-221. In Italy African cookware has been found in modest quantities even from the 6th and 7th century AD contexts, for example, at Portus (Paroli et al. 1993, 207; Coletti 1998, 409-415). However, at least the 7th century AD examples are in all likelihood residual. Hayes 1976, 95-100; 1978, 75; Tortorella 1981b, 211; Neuru 1986, 71-77; cf. Tortorella 1986, 220-221. For more profound introduction, see Hostetter et al. 1990; 1991; 1994; Hostetter & Brandt 1999. E.g. Peña 1999, 57-59. Arguments both for and against the usefulness of archaeological data have been presented by Pucci 1983. Cf. Peacock 1984a, 2-3; Fulford 1987, 58.

14

Introduction

65

66 67 68

69 70 71

72 73 74 75 76 77

78 79

80 81 82

83 84 85

86 87 88

89

90

Vegas 1973, 1, 157; Fulford 1980, 68; Carandini 1983, 158; Greene 1986, 10, 167; Wickham 1988, 184. Still, the greatest benefits are achieved with an integrated approach, as “the ceramic record can be used to provide important clues and additional data to those of epigraphic and documentary sources in an attempt to reconstruct trade links and shipping routes within the Roman Mediterranean through time”, Reynolds 1995, 139, see also Young 1981, 209; Morel 1983, 66. Millet 1987, 101. Crummy & Terry 1979, 54-57; Fulford 1987, 59-60; Tomber 1993, 148. In practice this is an artificial way to increase the size of a deposit, which has been proposed to indicate roughly its usefulness, as the possible changes in the relative proportions of various components are easier to demonstrate i n substantially large assemblages, rather than in smaller ones, Fulford 1980, 70. At the Palatine East excavations, the size of deposits was not necessarily a problem, as several of them contained pottery in hundreds of kilograms, Peña 1998, 9. Peña 1998, 6. Riley 1979a, 104-111; Tomber 1993, 150-158; Pavolini 1996, 238-239; Hayes 1998, 17. Hayes 1976, 48. As the concept of residuality is not unambiguous (see Peña 1998, 5), several definitions have been proposed for the term “residual”. Crummy and Terry (1979, 51) apply it to artifacts “which derive from occupation earlier than their respective context suggests”, while for Peña (1998, 6) residual is any find “initially discarded before the beginning of the formation of the context from which it was recovered.” See also Millet 1987, 104-105. Crummy & Terry 1979, 49-51. Crummy & Terry 1979, 57; Peacock 1982, 162; Fulford 1987, 61. Crummy & Terry 1979, 54, cf. Peña 1998, 11-13. De Rossi & Mandarini 1998, 55. Peña 1998, 12-13. Bradley & Fulford 1984, 43-47. The idea itself is in a way related to the determination of estimated vessel equivalents (see section 5.1). Peña 1998, 6. Crummy & Terry 1979, 54-59; Fulford 1983, 7. Unfortunately, the ideal of well-documented archaeological units i s more often than not contrasted by matters of secondary importance, such as lack of time, funds or trained employees. Zeggio & Rizzo 1998, 126-127, 146. Peña 1998, 7. Millet 1987, 99-101. The relationship of the present study and the chronology of African cookware is basically the same that was articulated by Dore (1988, cf. Dore 1989, 87) in connection with the coarseware finds of Tripolitania: “At the outset I did not consider that the coarseware could contribute in a major way to the refinement of dating of the site and thus my main concern was to transfer the established dating to the coarseware.” Crummy & Terry 1979, 50-51. Peacock 1984a, 2-4; see also Greene 1992, 40-41. The term “standard” will be occasionally used in the form catalogue (section 3.2.) to indicate the most abundant vessel variant of its group. On “standardization” as a process see section 4.5 and Rice 1996b, 176-182. On previous experiments with chip cards, see e.g. Riley 1979a, 103. Rye 1981, 50-51. The surface-finishes distinguished were: unfinished, rilled, wheel-ridged, turned, smoothed, self-slipped, slipped, burnished and their various combinations. So far, the best example of the results produced by this method is Skibo's monograph (Skibo 1992), which meticulously reports every form of use-alteration encountered in Kalinga pottery (Philippines). Tite 1999, 208.

15

Fabrics

2 FABRICS 2.1 INTRODUCTION In his thorough work on the African Red Slip ware Hayes characterizes the fabric of African pottery as follows:1 “fairly coarse with granular appearance, color orange-red to dark red, commonest inclusion lime: either in small particles or occasional larger lumps, white or brownish fine quartz particles together with occasional black particles and rare small specks of silvery mica.” Although the study of the composition of archaeological ceramics has progressed considerably since the early 1970's, Hayes' fortunate statement (cf. section 1.2.1) is still a decent summary of the current knowledge on the petrology of African pottery. The reason why the study of African fabrics has not advanced at the same speed with other classes of Roman pottery is obvious: on macro- and even on microscopic level the majority of these fabrics are substantially uniform in their composition, and thus difficult to distinguish from one another.2 The compositional uniformity of African pottery fabrics arise from the geology of North Africa, characterized by substantially homogeneous quartz sand and limestone bearing tertiary sediments, from which precise source areas can rarely be defined3. As a result, the principal inclusions found in virtually all African pottery, including various types of African Red Slip ware, utilitarian ware and amphorae, are quartz and calcareous particles of heterogeneous origin. Small amounts of feldspar (plagioclase, orthoclase and microcline), quartzite and trace amounts of muscovite with infrequent biotite mica and rare pyroxenes, finegrained sandstone, foraminifera and quartz-bearing limestone have also been observed.4 A more profound discussion on the nature of different tempering materials will be presented in section 2.5, which attempts to evaluate their effect upon the production and use of these vessels. Although the preceding review on the state of the compositional analysis of African cookware may seem somewhat pessimistic, it must be stressed that the study of some other subcategories of African pottery has lately advanced considerably, thanks to the extensive utilization of various petrographic and chemical analyses.5 The most recent example on the advantages of chemical analysis has been published by Mackensen and Schneider, who used wavelengthdispersive x-ray fluorescence (WD-XRF) to define compositional differences between the African Red Slip ware products of several kiln-sites located in the present-day Tunisia.6 The results confirmed that at the very least the North Tunisian production can be distinguished from the Central Tunisian one,7 while even the identification of individual production sites

within a restricted geographical area seems to be possible to a certain extent.8 The following discussion will show that the distinction between North and Central Tunisian products can also be made in African cookware both chemically and mineralogically.9 The most apparent feature to be utilized to separate the two major productions from one another, is the appearance of Central Tunisian fabric, which customarily has a distinctive dark red color resulting from the firing process and contains regularly abundant limestone inclusions.10 The somewhat traditional definition of Carthaginian (i.e. North Tunisian) fabric, on the other hand, as buff colored and very sandy,11 is also accurate. Central Tunisian production has been further divided into three fabric variants, the identification of which is based on the variability in the amount of quartz and limestone temper.12 Still, a production center making use of several clay deposits of different geological origin may have produced several fabrics, which cannot necessarily be separated from the output of several production sites dispersed on a wide area but exploiting mineralogically similar raw material sources.13 African cookware classifies as a potential example of such case, because it has been attributed to only one group on the basis of mineralogy,14 while the results of chemical analysis evidenced the existence of two distinct groups.15 It will be shown, however, that these two groups correspond with the North and Central Tunisian components of African cookware defined in the present study. The fabric classification presented in the following pages is mainly based on visual observations accomplished with a high-powered stereomicroscope. In spite of recent advances in ceramic archaeometry, visual comparison is still a valid research method that can be successfully applied to the classification of archaeological ceramics. Frequently, the limit for the precision of fabric determinations is set by the number of finds. If the quantity of research material exceeds several thousand diagnostic sherds, visual inspection is in fact the only applicable method if the examination of every find in the assemblage is desired.16 The results are also accurate enough for most purposes, even if the assemblage to be examined is extremely homogeneous, like most African pottery.17 The next logical step in fabric studies, however, involves the testing of the samples from each of the categories based on visual inspection with some method of compositional analysis. Hence, laboratory analyses

17

Chapter 2

should not be used only when visual inspection has failed to answer the questions asked from the research material.18 In the present case, the visual identification of inclusions was strengthened with a selection of polished samples set in blocks of epoxy resin and studied with a combination of scanning electron microscope and energy dispersive spectrometer (SEM/EDS).19 It was doubted at first whether the use of this combination would produce any results, because previously the examination of African cookware samples with SEM had only confirmed their homogeneous composition.20 Contrary to this negative precedent, the use of SEM/EDS gave this time evidence on the nature of inclusions. The elemental composition of the clay was studied by analyzing the matrix between the temper particles with electron probe microanalyzer. Although the results thus produced are hardly more than semiquantitative, they are sufficient to illustrate the differences between the North and Central Tunisian production. In the near future, the work with the African cookware fabrics of the Palatine East excavations will be complemented with thin-section petrography.21 The use of methods, like x-ray fluorescence (XRF), which would probably have been the most efficient way to indicate differences in the bulk composition of this fairly uniform material, was excluded from the study. However, the method has previously been used to study the chemical composition of African cookware and these results22 will be used as a point of reference in the following discussion. The preliminary classification of the Palatine East fabrics was carried out with a high-powered stereomicroscope that aided to identify several fabric sub-groups i.e. fabric variants. The introduction of these sub-groups in the following section stems from a need to illustrate the variability included in a single fabric, although at a macroscopic level these fabric variants are strongly related to one another. The discussion on the differences in the chemical

composition of the two African cookware fabrics has been incorporated into section 2.4, whereupon attention will be turned to the advantages and disadvantages resulting from raw materials used in the making of African cookware. In the present study, a fabric is regarded as a multiphase clay ceramic body prepared using a distinct set of raw materials, including both base clay and temper, paste preparation and surface treatment practices.23 The variability among groups thus defined has been recognized as variants of a fabric. Each entry includes both a hand specimen description and a microscopic description. The latter entry has been accomplished with an ordinary binocular microscope (magnifications of 10-60x) and a high powered stereomicroscope equipped with photo capturing facilities (magnifications of 5-200x). As the consistency of macroscopic observations has been strongly stressed,24 the fabric descriptions to be presented are structured as indicated in the following table (Table 3). Ultimately, before the fabric classification is presented, the usefulness of high precision fabric descriptions must be questioned. In recent years archaeological science, the study of archaeological ceramics in particular, has become increasingly dependent on research methods intended to establish chemical or mineralogical composition of the examined subject. As expected, most of these methods are spin-offs of the triumph of the natural sciences and their advantages seem customarily evident. For this reason alternatives of compositional analysis are seldom, if ever, thought of. Recently, the weaknesses of these techniques have been summarized by Rice,25 who stresses the validity (i.e. usefulness) of the results. On the other hand, the exploitation of compositional data is encouraged, because it does not only contain important information on the origin of pottery, but also includes evident traces of human behavior. The choice to use certain tempering materials in the pottery production, for example, arises likely from a

Table 3. The structure of fabric descriptions 1) the name of fabric or fabric variant 2) hand specimen description (hsp) a) hardness: soft, slightly soft, hard, very hard b) break: sharp, regular, irregular c) fracture surface: slightly gritty, gritty, coarse d) color: Munsell numbers are used e) inclusions: concentration, size, condition and color 3) microscopic description (mic) a) general characterization of texture: fine-, medium or coarse-grained b) general characterization of matrix: calcareous, ferruginous c) inclusion concentration: absent, rare, sparse, frequent, abundant, very abundant d) inclusion size (mm): minute (1) e) inclusion type: grain, body, plate, reaction rim, void f) inclusion shape: angular, subangular, subround, round, irregular, elongated g) inclusion color/appearance: colorless, milky, white, red, brown, reddish brown etc.; glistening h) inclusion identification (question mark denotes highly conjectural identification)

18

Fabrics

mixture of experimentation and tradition, which are both indicators of a vigorous craft industry. To meet these challenges, the last two sections (2.4-5) of this chapter present an intentional attempt to exploit the data introduced in preceding descriptive sections. The discussion will focus on the main advantages and disadvantages resulting from the choice of particular clays and tempering materials, which the potters of Africa Proconsularis used as a part of their craft. Careful selection of raw materials is essential as far as the usefulness of high-fired cooking utensils is concerned. Functional cookwares are repeatedly subjected to thermal and mechanical shocks, both of which set different requirements for the products.26 For

this reason, the potters probably tried to improve their products by seeking – consciously or unconsciously – a balanced admixture of clay, water and tempering materials. The search for equilibrium, however, is a continuous process and the changes are usually detected only in a production that has lasted for several decades or centuries. Bearing this in mind, the intention is first to characterize the two African cookware clay pastes and then to evaluate the effect of clay matrix and different mineral non-plastics, either naturally present or deliberately added to the clay paste. The usefulness of African cookware with respect to the raw materials used in the production will also be discussed.

2.2 CENTRAL TUNISIAN FABRIC The Central Tunisian cookware fabric27 is distinguishable due to its dark red clay matrix (2.5 YR 4/4-8) and abundant calcareous temper consisting of yellowish or white bodies and reaction rims. The other inclusion present in significant quantities is quartz, which is chiefly monocrystalline, although rare occurrences of cryptocrystalline grains have also been reported. The quartz temper is dividable into two clusters by size, shape and color. The cluster of smallsized quartz, which includes predominantly colorless subangular grains from 0.1 to 0.25 mm in diameter, belongs very likely to detrital component of the clay. By contrast, the presence of translucent or in rare cases milky, subrounded quartz grains exceeding frequently 0.75 mm in diameter, gives evidence of the custom to manipulate the clay paste with a sand temper. Due to the scarcity of other inclusions, like potassium feldspars, augite or argillaceous rock fragments (herefore ARFs), together with their subround shape and a size equal to that of large quartz grains, they were probably introduced into the ceramic paste with the sand temper. The appearance of ARFs in particular, which are usually dark brown in color and fairly spherical in shape, distinguishes the group from the North Tunisian cookware fabric. In the present study, the Central Tunisian cookware fabric has been divided into four variants using the compactness of the matrix and the variation in the amount and the size of the inclusions as criteria. The principal characteristic of variant I.1 is the high density of calcareous inclusions (Plate 1b), whereas variant I.2 (Plate 1a) presents a fairly balanced mixture of quartz and calcite temper. The matrix of variant I.3 (Plate 1g) is more compact and also sparsely tempered in contrast to the other three fabric variants, whereas the use of sand temper is clearly seen in variant I.4, showing customarily high concentrations of large, subrounded quartz grains.

I.1 Highly calcareous Central Tunisian fabric variant HSP: hard to very hard with regular break and gritty fracture surface; section and surfaces uniformly dark red (2.5YR 4/4-6) in color, with occasional slip, selfslip or light salt scum on surfaces; salt scum often darkened in firing, the section shows sparse, small colorless grains and frequent small white bodies; MIC: medium-grained fabric, ferruginous matrix; frequent to abundant, small to medium, subangular to round, colorless or milky grains (quartz), frequent to abundant, minute to medium, subround to round, white or yellowish bodies and reaction rims (calcite), absent to rare, small to medium, subround to round, reddish or light brown bodies (ARFs), absent to rare, small to medium, angular to subangular, black grains (augite); rare, medium, subangular to subround, porous black bodies (?); sparse to abundant, small to medium, irregular and elongated voids. (Plate 1b) Comparanda: Peacock & Tomber 1991, 294-298, fabric 1. I.2 Standard Central Tunisian fabric variant HSP: hard to very hard with regular break and gritty

fracture surface; section and surfaces uniformly dark red (2.5YR 4/4) in color, rare overfired grayish-red (2.5YR 4/2) examples, slip or light salt scum occasionally on surfaces; salt scum often darkened in firing, the section shows sparse, small colorless grains and rare small white bodies; MIC: mediumgrained fabric, ferruginous matrix; frequent to abundant, small to medium, subangular to round, colorless or milky grains (quartz), sparse to frequent, minute to medium, subround to round, white bodies and reaction rims (calcite), absent to rare, small to medium, subround to round, reddish or dark brown bodies (ARFs), absent to rare, medium to large, subangular, black grains (augite), sparse to frequent,

19

Chapter 2

small to medium, irregular and elongated voids. (Plate 1a) Comparanda: Peacock & Tomber 1991, 298, fabric 2.

high firing temperature and the inferior quality of products with respect to the following phase. In the second phase, which lasted to the early 4th century AD, a fine sand temper low in calcite and somewhat lower firing temperature were used and the production is said to be of better quality than in the preceding phase. The possibility that these observations would correspond to the fabric classification presented above was examined by tabulating the distribution of Central Tunisian fabric variants according to chronological phase (Table 4).29 The starting hypothesis was that the highly calcareous (I.1) and quartz-rich (I.4) fabric variants would belong to the first production phase (A) determined at Uzita, while the standard (I.2) and compact (I.3) variants represent in all likelihood the latter phase (B). The results give only partial support to the idea about the two stages of production, as the decline in the number of products attributed to the phase A is evident only with the highly calcareous but not so readily observable with the quartz-rich variant. While the overall ratio between the two stages may equally be considered as confirmative evidence, excluding the proposed decline in the quality of these products in the early 4th century,30 the following discussion on forms (see chapter 3) will show that the occurrence of these fabric variants is also related to certain Central Tunisian vessel shapes. Notwithstanding the apparent similarity, the present classification is not comparable either to Dore's work with the quartz-limestone coarseware fabric variants of Sabratha that were reported to offer only a little evidence on their origin besides being generally North African.31 Contrary, the resemblance to the fabrics reported by Peacock & Tomber in connection with the British-Tunisian kiln site survey32 speak strongly in favor of Central Tunisian provenance. The variation in the amount and proportion of diverse tempering materials, however, cannot be attributed at present to differences in geographical location, even though the calcite-rich variant (I.1) is generally assumed to represent the production of workshops located in the coastal area of Central Tunisia.33

I.3 Compact Central Tunisian fabric variant HSP: hard to very hard with regular break and slightly gritty to gritty fracture surface; section and surfaces usually uniformly dark red (2.5YR 4/6-8) in color, rare examples showing a reduced core (2.5YR 3/0) with sharp margins, occasional slip or light salt scum on surfaces; salt scum usually darkened in firing, the section shows sparse, small colorless grains and sparse to frequent small white bodies; MIC: porphyritic/medium-grained fabric, ferruginous matrix; sparse, small to large, subangular to round, colorless or milky grains (quartz), rare to frequent, minute to medium, subround to round, white bodies and reaction rims (calcite), absent, small to medium, angular to subangular, light brown bodies (ARFs), rare to frequent, minute to small, irregular and elongated voids. (Plate 1g) I.4 Quartz-rich Central Tunisian fabric variant HSP: hard to very hard with regular break and gritty to coarse fracture surface; section and surfaces usually uniformly dark red (2.5YR 4/6) in color, occasional slip or light salt scum on surfaces; salt scum usually darkened in firing, the section shows sparse, small colorless grains and sparse small white bodies; MIC: medium-/coarse-grained fabric, ferruginous matrix; abundant to very abundant, small to large, subangular to round, colorless or milky grains (quartz), rare to sparse, small to medium, subround to round, white bodies and reaction rims (calcite), sparse to frequent, small to medium, irregular and elongated voids. Previously, Central Tunisian coarseware fabrics has been studied in detail at Uzita, where the production of both red Byzacenan coarseware (rood Byzaceens gebruiksaardewerk) and stripe-burnished Byzacenan pottery (gepolijst Byzaceens aardewerk) ware was divided into two chronologically distinct phases.28 The characteristics of the first phase, dated from the 3rd century BC to AD 175/200, were defined as the use of abundant coarse sand temper, relatively

Variant

I.1

I.2

I.3

I.4

A

B

N=

Phase 6

16.7

15.0

48.3

20.0

36.7

63.3

60

Phase 7

11.6

20.0

52.6

15.8

27.4

72.6

95

Phase 8

6.6

25.2

57.0

11.2

17.8

82.2

151

Phase 9

7.4

23.5

63.2

5.9

13.3

86.7

68

Phase 11

5.6

33.7

47.2

13.5

19.1

80.9

89

20

Table 4. Chronological distribution of Central Tunisian fabric variants (%).

Fabrics

2.3 NORTH TUNISIAN FABRIC The five characteristics distinguishing the North Tunisian cookware fabric34 from the Central Tunisian one are: a) the light tone and the orange color of clay (2.5 YR 5-6/6-8; 5 YR 6/6), b) the decreased presence of calcite, c) the evident increase in the amount of other inclusions than quartz, of which d) the occurrence of large lens-shaped yellowish ARFs and the strong presence of ARFs in general, and e) rare irregular, blackish-brown iron-rich patches consisting of a mixture of various minerals including quartz, must be stressed. The bimodal distribution of quartz temper, however, is very similar to the Central Tunisian fabric. As with Central Tunisian fabric, the differences in the compactness and temper were thought for a time to indicate the existence of several fabrics and fabric variants. Contrary to this, the three fabric variants introduced below were discovered to form a continuum ranging from examples enriched by natural non-plastics but lacking eolic quartz to those tempered with substantial amounts of quartz sand.35 Thus, much of the variation results very likely from the use of heterogeneous clay deposits, where the amount of minute quartz grains and other detritus varies by microstratigraphy. The fabric variant II.1 displays only a few grains of coarse quartz sand temper per sample, whereas the major presence of this mineral can be attributed to detrital component of the clay, which also contains minute ARF grains (Plate 1d). Many examples of the variant II.1 also show a section with a peculiar color transition from orange to pale brownish-yellow towards the exterior surface (Plate 1c), the cause of which will be discussed in detail in section 4.3.1. The second variant II.2 has been defined as the standard North Tunisian cookware fabric thanks to the homogeneous nature of its inclusions (Plate 1e), whereas the variant II.3 is characterized by a compact clay matrix enriched with coarse quartz sand temper (Plate 1f). The possibility that the distribution of these three fabric variants could result from chronological differences in North Tunisian cookware production was explored with negative results.

to frequent dark brown and light yellowish-brown bodies and absent to rare, small white bodies; MIC: medium-/coarse-grained fabric, ferruginous matrix; abundant to very abundant, minute to large, subangular to round, colorless or milky grains (quartz), absent to sparse, minute to large, subround to round, white bodies and reaction rims (calcite); absent to abundant, minute to large, subangular to round or irregular dark brown bodies (iron oxide), absent to frequent, small to very large, subround or lens-shaped, light yellowish-brown or brown grains (ARFs); absent to rare, minute to small glistening plates of mica (muscovite and biotite); sparse to frequent, minute to large, subround and irregular/elongated voids. (Plate 1c-d) Comparanda: Peacock 1984b, 15, fabric 2.3 Blacksurface ware; Schuring 1984, 166-168, fabric E (?).

II.1 Quartz-Rich North Tunisian fabric variant HSP: hard to very hard with regular break and coarse fracture surface; in most examples the section turns gradually from orange (2.5YR 5-6/6-8) to pale brownish yellow (5YR 6/6) in the last quarter towards the exterior surface, rare completely brownish-yellow examples, exterior surface often covered by grayishblack deposit (patina cenerognola), otherwise either slipped, self-slipped or unfinished surfaces; the section shows sparse to frequent, small colorless grains; rare

HSP: hard to very hard with regular break and slightly

II.2 Standard North Tunisian fabric variant HSP: hard to very hard with regular break and gritty to

coarse fracture surface; both the section and surfaces generally orange (2.5YR 5-6/6-8) in color; exterior surface often covered by grayish-black deposit (patina cenerognola), otherwise either slipped, self-slipped or unfinished surfaces; the section shows sparse to frequent, small colorless grains; absent dark brown bodies and absent to rare, small white bodies; MIC: medium-/coarse-grained fabric, ferruginous matrix; abundant to very abundant, minute to large, subangular to round, colorless or milky grains (quartz), absent to sparse, minute to large, subround to round, white bodies and reaction rims (calcite); absent to rare, minute to large, subangular to round or irregular dark brown bodies (iron oxide), absent, medium to very large, subround or lens-shaped, light yellowish-brown or brown grains (ARFs); absent to rare, minute to small glistening plates of mica (muscovite and biotite); sparse to frequent, minute to large, subround and irregular/elongated voids. (Plate 1e) Comparanda: Peacock 1984b, 15, fabric 2.1 African red-ware. II.3 Compact North Tunisian fabric variant gritty to gritty fracture surface; both the section and surfaces usually orange (2.5YR 5-6/6-8) in color; exterior surface often covered by grayish-black deposit (patina cenerognola), otherwise either slipped, selfslipped or unfinished surfaces; the section shows sparse to frequent, small colorless grains; absent dark brown bodies and absent, small white bodies; MIC: medium-grained fabric, ferruginous matrix; frequent to abundant, small to large, subround to round colorless and milky grains (quartz); rare to sparse, minute to

21

Chapter 2 yellowish or brownish fabric color.36 On the other hand, the Black Top ware – the fabric variant II.1 of the present study – has been occasionally considered to be a fabric of its own.37 It is also uncertain whether Hayes is referring to the internal variability of North Tunisian cookware when describing the two cookware fabrics, the common feature of which is the high number of vessels with gray-black exterior surface.38 Furthermore, the terminology has been recently complicated by a suggestion, according to which the ‘African blacktop coarse ware’ should be henceforth be called ‘Carthage/Nabeul blacktop cooking ware’.39

large, subround to round white bodies (calcite), absent, medium to very large, subround or lens-shaped, light yellowish-brown or brown grains (ARFs); absent to rare, minute to small glistening plates of mica (muscovite and biotite); absent to sparse, minute to medium, subround or irregular voids. (Plate 1f) A short survey on the terminology used in connection with the North Tunisian cookware leads to interesting observations. Besides the occasionally used “plain African Red Slip ware” the definition appearing more often as a synonym for this fabric is “Black Top ware”, the identification of which is aided by its

2.4 CLAY MATRIX The dissimilarity of North and Central Tunisian fabrics can be defined in terms of the non-plastic content, as well as by the composition of the base clay. The examination of the two fabrics with electron probe microanalyzer was accomplished by analyzing the area between non-plastic grains (Table 5).40 The results indicate concentrations of CaO significantly lower than 5% (wt.) in the clay matrix of both Central and North Tunisian fabric. Therefore, both fabrics may be characterized ferruginous,41 although Central Tunisian fabric shows consistently somewhat higher values of calcium. The likelihood that the discrepancy in the chemical composition would arise from differences in firing conditions is low, because in Roman Africa pottery was constantly fired in temperatures under 1100 °C, which do not normally alter the elemental composition of the clay.42 The possibility that the amount of calcite in pottery may point at the origin of the clay has previously been suggested in many studies, which also indicate that Byzacenan (i.e. Central Tunisian) fabric variants show generally higher concentrations of calcite than those of Zeugitanan (i.e. North Tunisian) origin.43 However, it should be asked whether the elevated presence of calcite in Central Tunisian fabric results from the use of calcareous clays, tempering materials or both. Previous analyses have indicated that the clay of African cookware is not particularly calcareous, if not calcareous at all,44 as they have indicated the use of both ferruginous and carbonateferruginous clay pastes in Roman Africa.45 The question can be examined by comparing the present results with the results of X-ray fluorescence analysis published by Olcese, which reflect the nonplastic component of the Central Tunisian fabric as the samples must be powdered for the analyses.46 The results show constantly higher (2-4 %) values of

calcium, indicating the use of calcite-bearing tempering materials in the preparation of the clay paste. Regarding the distribution of individual elements, the North Tunisian fabric can be observed to contain more aluminum and titanium, of which the increased presence of aluminum is the clearest difference to Central Tunisian products showing increased values of calcium, sodium and magnesium. These differences are also reflected in the appearance of the two fabrics. The hues of yellow and beige characterizing certain North Tunisian fabric variants most likely result from the combination of elevated Al2O3 and low CaO content,47 whereas the strong presence of iron in Central Tunisian cookware has been said to give the fabric its distinctive dark red color.48 The color is also a clear indication of the low calcium content of the clay, as lighter hues could be expected from an iron-bearing calcareous clay fired in high temperature.49 The low percentage of components decreasing the refractoriness50 of the clay, such as Na2O, CaO, FeO and MgO in North Tunisian pottery is reflected in the moderate to high fabric porosity observed in connection with previous studies.51 These fabrics do not show, however, any evidence on the use of organic tempering materials, which has occasionally been claimed to explain the presence of pores in African pottery.52 Increased porosity of cookware vessels may contribute positively to their performance characteristics as pores obstruct crack propagation and even the distribution of heat through the ceramic body. Both of these features are essential to prevent the tension damage caused by thermal stresses.53 The relatively strong presence of calcium and magnesium in Central Tunisian fabric is an indication

22

Fabrics

North Tunisian

Central Tunisian

Table 5. Electron probe microanalyses of African cookware Fabric

Form

Na2O

MgO

Al2O 3

I.1

C.II

0.689

2.635

I.1

G

0.465

I.1

G

I.3

SiO2

K2O

CaO

TiO2

V 2O 3

CrO 3

MnO

FeO

NiO

ZnO

19.374 50.126

2.493

3.288

0.575

0.020

0.012

0.055

4.140

0.008

0.018

2.868

21.544 51.458

2.970

2.823

0.704

0.026

0.049

0.013

4.622

-

0.086

0.573

2.854

22.004 54.580

2.733

4.012

0.741

0.011

-

0.042

8.018

0.007

-

J.II

0.235

1.719

24.151 51.057

2.679

0.615

0.816

-

0.025

0.013

3.534

-

-

I.3.

O.IV

0.133

1.81

26.551 49.382

2.244

0.675

0.549

0.029

0.016

0.037

4.408

0.052

0.084

I.4

G

0.666

2.431

19.697 48.297

2.437

1.601

0.648

0.048

0.020

0.052

5.023

0.007

0.080

II.1.

O.VI

0.291

1.353

22.858 53.379

2.385

0.571

1.266

0.055

0.062

0.048

4.224

-

-

II.2

A.II

0.217

1.759

26.673 50.041

2.714

0.614

0.890

-

0.012

-

4.506

0.004

0.039

II.2

H.II

0.472

1.502

28.271 53.522

2.291

0.611

0.671

0.038

-

0.109

6.144

0.017

0.008

II.2

I.1

0.170

1.652

23.023 48.105

1.963

1.072

0.689

0.055

0.042

0.050

3.822

0.020

0.012

II.2

O.VI

0.262

1.727

26.639 53.250

2.943

0.956

0.909

0.254

-

0.040

5.400

0.004

0.033

II.3

A.I

0.169

1.948

26.513 55.016

2.589

1.296

1.511

0.049

0.014

0.014

4.668

0.024

0.069

II.3

I.I

0.265

1.566

22.433 55.140

2.055

0.929

0.609

0.041

0.024

0.021

3.853

0.013

0.015

of low refractory clay, which tends to vitrify at substantially low temperatures as clay minerals react to form new compounds with the fluxing components.54 Therefore, the statement that the fabric of the earliest African table- and cookwares had a low resilience to overfiring was probably referring to Central Tunisian production.55 The vitrification structure of such clay is retained relatively unchanged at temperatures between 850-1050 °C, which is an advantage as the firing does not have to be as carefully monitored as in the case of calcite-poor clays.56 Experiments with calcareous clays have shown that at a temperature exceeding 900 °C clay minerals and calcite react to form calcium silicates and calcium aluminum silicates, and the resulting vessels are not only tough and resistant but also rigid.57 The compound effect of calcium and magnesium was apparent under scanning electron microscope, as the clay matrix of Central Tunisian fabric had been in several cases fused together, while individual particles in the matrix of North Tunisian fabric were still clearly detectable. The difference is also observable by the naked eye in African cookware, as the fabric of Central Tunisian cookware is generally harder than the North Tunisian fabric.58 The main disadvantage of tough fabrics is their poor modulus of elasticity, which increases the probability of such vessels being broken down by thermal and mechanical shocks. But cookwares that can well withstand both types of stress are rare in general. As cookware production cannot be based solely on the combination of calcite-rich clay

and high firing temperature, the presence of a component increasing fabric elasticity, like quartz sand, has been an important prerequisite for the making of useful products.59 Another potential explanation for the unexpectedly good performance characteristics of African cookware may be the presence of kaolinite, which is a fairly common mineral in secondary clay deposits.60 It is detectable both directly and through the presence of gehlenite, which is formed as calcite-bearing kaolinite clay is fired to 900-1000 °C.61 With regard to the mineral content of African pottery clays, several scholars have reported the rare occurrence of both muscovite and biotite mica.62 Although the moderate presence of biotite is one characteristic of Tripolitanian products,63 it has also been postulated that fabrics showing frequent mica cannot be of North African origin.64 This is not the case with the two African cookware fabrics, however, which display only rare muscovite and virtually absent biotite mica.65 The presence of mica has been explained by the transformation of clay in firing,66 but since micas are common detrital minerals they may have equally been present in the base clay.67 In addition, the examination of minute nonplastic grains with SEM/EDS showed increased concentrations of phosphate in association with calcium and silica. Phosphate is most likely present due to post-depositional contamination, which may occur as an amorphous coat over silicate grains or as small particles of calcium phosphate formed into the pores of the fabric.68

23

Chapter 2

2.5 INCLUSIONS The purpose of this section is to introduce the principal mineral and non-mineral inclusions either observed in the study assemblage or previously reported in various studies on African fabrics. An attempt has been made to present them in meaningful groups with a preference to discuss in detail those inclusions that likely provide additional information on the acquisition of raw materials, the manufacture of vessels, their use or post-depositional history. The results of various archaeological experiments that have only seldom been extrapolated to “real” archaeological assemblages have also been incorporated into this section. The group names of minerals and other inclusions are indicated in boldfaced letters in the title of each section. In the following parenthesis, the inclusions with major presence in African cookware are in italics, and inclusions observed in the study assemblage are underlined. At first, although a myriad of other factors than the search for optimal vessel performance can be offered to explain the use of particular temper in pottery production,69 the most important observation to be made regarding the non-plastic content of both African cookware fabrics is that it corresponds strikingly well to the ideal outlined by various archaeological experiments.70 The strong presence of natural non-plastics – particularly quartz sand and calcite – diminishing the need of paste preparation of North African clays, favored the production.71 Without a component increasing the elasticity of the vessel, in this case quartz sand, high-fired clay rich in calcite produces a tough and resistant fabric, which is too rigid for the production of cooking utensils.72 With the wide availability of suitable tempering materials, the potters of Roman Africa had an advantage over their competitors working in other parts of the Empire; they were able to utilize the same clay in the production of both table- and cookwares without significantly changing their paste preparation procedure.73 Because similar clays readily suitable for cookware production dot no exist in every corner of the Mediterranean, the areas possessing these resources are very likely to turn to the exportation of cookwares, as they can easily compete with the local production of mediocre use-related properties.74

components of plutonic rocks. Because of their origin, alkali feldspars should not be present in ceramics produced in a sedimentary environment. Still, moderate amounts of K-feldspars,76 especially sanidine, which is related to clinopyroxenes by its origin, have been reported in samples of African pottery. The two other main constituents of the alkali feldspar group, orthoclase and microcline, do not occur only in plutonic rocks but also in sandstones. In favorable conditions the alkali feldspar group may alter to sericite or kaolinite,77 which have also been detected in African pottery. Notwithstanding their somewhat obscure origin, the presence of alkali feldspars in trace amounts has hardly had effect on the performance of African cookware vessels. 2.5.2 Carbonates (calcite, chara [calcareous algae], foraminifers, limestone) Calcite is present in African cookware both in form of calcareous bodies and reaction rims (e.g. Plate 1b), the occurrence of which may be explained in a variety of ways. First, the clay employed in the pottery production probably contained calcareous elements, some of which subsequently dissolved either in firing or in post-depositional environment. As a result, some of the calcite in the African cookware is detectable only in the form of calcareous reaction rims.78 Hayes corroborates that the presence of lime in the African fabrics results from impurities in the clay, whereby he further deduces that potters preferred unwashed clay.79 In fact, at least some Central Tunisian products seem to contain primary calcite,80 formed chemically by marine microfauna.81 Calcite derived from calcareous algae or foraminifers, which are minute single-celled animals with calcite-bearing shells, often occurs both in sedimentary clays and beach sands.82 In the present case, both the size and the shape of these inclusions seem to point towards calcareous clays (marls) or, even more likely, to semicalcitic clays of sedimentary origin.83 On the other hand, there are known cases where potters have intentionally added lime or other calcite bearing material into the clay paste, possibly to both increase its workability and to improve the performance characteristics of the products.84 Although limestone is a dominant inclusion in African fabrics, the nature of this inclusion has been seldom discussed.85 The possibility that calcite was intentionally added to the clay paste in some areas of Central Tunisia has been suggested by Peacock and Tomber on the basis of the density and abundance of calcareous material observed in their samples.86 If calcite bearing material was added, it was probably in a powdered, rather than in crushed form, because

2.5.1 Alkali feldspars (microcline, orthoclase, sanidine) The presence of alkali (i.e. potassium- or K-) feldspars in African fabrics is unexpected, because they are igneous minerals. Some alkali feldspars like sanidine are primarily formed in extrusive igneous rocks,75 for example as hot magma cools down rapidly in a low pressure, while other k-feldspars are mostly

24

Fabrics

subrectangular calcareous inclusions that indicate crushed temper are absent from the study material.87 The last possibility, which at the moment seems to be the most improbable one, is that all calcareous bodies observed in African cookware were formed in a post-depositional environment. Generally speaking, the deposition of calcite into ceramic pores is accelerated in favorable post-depositional conditions.88 This has been documented in Monte Testaccio (Rome),89 where the post-depositional environment had been enriched with lime to neutralize the smell resulting from the tainted remains of oil in amphorae. Although the example is obviously an extreme one, the definitive exclusion of the possibility that some of the calcite observed in African cookware would have formed in post-depositional conditions is unreasonable. Regardless of whether the clay used in the pottery production contains natural calcite or deliberately added calcite temper, the presence of calcareous component in a clay paste can produce severe problems in firing. Although calcite-based temper may react with the clay paste and form stable calcium silicates and ferrosilicates,90 it is often accompanied by the decomposition of calcium carbonate, which usually takes place in temperatures around 650-1000 °C. The reaction results both in carbon dioxide and calcium oxide,91 of which the latter substance has an unfortunate quality to re-hydrate with water vapor in the air. As the resulting calcium hydroxide is clearly more voluminous than its parent material, it may cause “lime-blowing” that comprises structural damage ranging from small spalls to severe cracking and crumbling. Fortunately, the temperature in which calcium carbonate decomposes can be increased by adding sodium chloride, either in form of salt or salt water, to the clay paste.92 The proposed mechanism behind the risen reaction margin is that salt lowers the vitrification point of the ceramic body, which does not only strengthen the fabric against lime blowing but also seals off a significant fraction of its apparent porosity. Hence, the reaction between vapor and calcium oxide is prevented.93 The effect of sodium chloride as an agent lowering the vitrification temperature by accelerating the formation of calcium silicates and ferrosilicates was confirmed by the SEM/EDS scans of presumed carbonaceous matter, which showed the presence of silica and iron in such particles. At a macroscopic level, the effect is shown in the study assemblage by 17 overfired or vitrified vessels, of which 11 belong to Central Tunisian production. The presence of sodium chloride in calcite-rich clays produces some disadvantages, which can also be detected in African cookware. For their sake the modern brick industry has found the addition of salt both uneconomical and commercially impracticable.94

During the firing process the salt evaporated to the vessel surface in drying phase results in a whitish scum (Plate 2i),95 which characterizes both Central Tunisian cookwares and amphorae. The significance of this surfacing will be examined more thoroughly as the firing of African cookware is discussed (chapter 4.3). Experiments have indicated that sodium chloridecontaining bricks fire to a darker color than those without added salt.96 The presence of salt often produces a yellow reaction rim around calcareous particles, which consists presumably of glass formed by the fluxing action of the lime on the clay with the iron compound providing the color. In the study assemblage such formations are typical for the calcareous Central Tunisian fabric variant I.1. The use of clay containing natural or added calcite in pottery production would be more than absurd, unless it also produced some advantages. Archaeological experiments exploring the effect of limestone temper have yielded interesting results. In comparison with grit and grog, the addition of carbonaceous material facilitates the forming and enables the production of strong vessels with thin walls.97 The last mentioned property is advantageous, because when in use a thin-walled cookpot profits by its improved heat conductivity and decreased expansion differential between vessel surfaces.98 Calcium-rich limestone temper also makes the clay more workable by increasing flocculation.99 The strength of calcite-tempered vessels is indicated by the high peak load they can absorb,100 but due to the toughness of the fabric, calcite-tempered wares also have usually a low modulus of elasticity. This makes them less resistant to thermal shocks, even though thinner vessel walls and similar thermal expansion co-efficient of calcite to clay may have partially compensated this.101 The most severe disadvantage related to the mechanical properties of limestone-tempered pottery is its low post-peak response, in other words, such vessels tend to break very easily after the appearance of first cracks.102 2.5.3 Clinopyroxenes (augite, diopside) The presence of clinopyroxenes, like augite, in pottery products of Roman Africa has usually been taken as an enigmatic feature, because these igneous minerals do not normally occur in sedimentary environments. None the less, the occurrence of few grains of pale and green augite was reported by Peacock and his colleagues in almost every sample of their Central Tunisian fabric.103 Although the anomaly was already recognized in the mid-1980's, it still requires satisfactory explanation. One possibility to be taken into account is explosive volcanic activity that may transport tephra, the volcanic debris, thousands of kilometers away from the parent volcano.104 Therefore, a potential though not very probable origin for clinopyroxene –

25

Chapter 2

which is more common in autoclastics and hyaloclastites than as a pyroclastic heavy-mineral phenocryst of tephra105 – in African pottery could be Mt. Vesuvius located some 600 kilometers NE of Africa Proconsularis.106 The presence of clinopyroxenes in the Sahel region has also been explained by their occurrence in sandstone,107 as augite is “moderately common as detrital mineral” and sandstone often contains accessory minerals with an abundance of 1-2%.108 Just like the observation that volcanic minerals become more abundant in North Africa towards Tripolitania and the delta of the river Nile,109 these minor differences in the occurrence of certain marker minerals can hopefully be utilized in the future to identify the output of individual production sites. The other clinopyroxene observed in some African fabrics is diopside, which has the same main characteristics as augite. Regarding the origin of diopside, it should be stressed that in favorable conditions, which include a firing temperature around 1000-1100 °C, the combination of montmorillonite rich clay and dolomite inclusions may result in the formation of clinopyroxenes (especially diopside).110 Nevertheless, clinopyroxenes occur normally in trace amounts and they probably had no effect on the userelated properties of African cookware.

does not automatically imply that the larger fraction should be interpreted as temper,114 a closer examination of the fabric samples taken from the study assemblage produced a pattern supporting the idea. Had the clay been poor in natural quartz inclusions, the potter would have decreased its plasticity with the addition of sand-sized non-plastics, the major component of which was eolian quartz.115 The bimodal quartz temper is said to result in a fabric of high initial strength, as the smaller fraction consists mainly of angular grains.116 With respect to its widespread presence in ancient pottery, the study of the performance characteristics of quartz sand temper played only a minor part in experimental archaeology until very recently. For this reason, the few results and ideas based on the size and shape of quartz inclusions as well as on their expected firing performance, were able to persist for a long time. The alpha-beta inversion of quartz, which takes place at a temperature of 573±5 °C and is accompanied with 2% expansion in the volume of the temper, has always been the focal point of interest.117 Due to an unusually high thermal expansion coefficient, the use of quartz sand was long thought to produce highly porous and fragile fabric, even though fine-grained temper was supposed to diminish the formation of microcracks resulting from the mechanical stress created by the differential coefficient of thermalexpansion of clay minerals and quartz.118 Based upon this reasoning, fabrics with a fine quartz sand temper have been said to resist crack initiation fairly well, but opinions concerning the post-cracking performance of such fabrics have ranged from the rapid strength loss of ceramic body to the effective prevention of shattering, at least when contrasted to a coarse quartz sand temper.119 The presumed effect of grain size has even been expressed as a rule, according to which “the strength of the ceramic body increases as the size of temper decreases”.120 Recent experiments have shown that grain size does not have a significant effect on performance characteristics, as the maximum strength and toughness of quartz-tempered pottery is achieved when the non-plastic fraction accounts for 20% of the total volume of the clay paste.121 Experiments exploring the impact of the size and the quantity of quartz sand temper have also shown that the formation of microcrack networks as well as exclusive debonding resulting from different thermal expansion coefficients of the clay matrix and quartz temper is a desirable phenomenon that increases the energy dissipation, i.e. the toughness of the pottery.122 Because the resulting fabric has the same elastic behavior as material containing an equal volume of equal sized pores, the use of quartz temper also enhances crack propagation stability, which means that the possible breakage of the vessel can be anticipated.123

2.5.4 Quartz ( quartz sand) Quartz is the other of the two principal non-plastic constituents of African cookware. In contrast to calcite, quartz inclusions are also more abundant and they have been reported to occupy 15-20 % of the total volume of ceramic body.111 In most fabric variants, especially in Central Tunisian cookware, quartz inclusions can be separated into two fractions by grain size and shape. The larger grains measure 0.15-0.3 mm in diameter with their shape ranging from subrounded to rounded, whereas the smaller fraction is composed of subangular and angular quartz grains falling into the 0.03-0.05 mm size class. The difference in the size and the shape of these fractions is said to reflect their different origins. Small angular quartz, the presence of which is a characteristic of many African fabrics, most likely belongs to a non-plastic fraction of clay deposited fairly close to its parent material.112 The large rounded quartz grains on the contrary have most likely found their way into the clay paste as a sand temper (Plate 1f). In closer examination these large quartz grains show frequently frosted or pitted surfaces, which classify them as eolian (i.e. wind-deposited) sand usually formed in deserts.113 The presence of eolian quartz sand is a factor distinguishing African pottery from its imitations produced in the European continent, because such formations do not occur in accessible deposits on the northern side of the Mediterranean. Although a gap in the grain-size distribution

26

Fabrics

The toughness of quartz tempered ceramics is also reflected in improved wear-resistance, as the protruding quartz grains tend to protect ceramic matrix, which is more easily abraded.124 Wearresistance has been shown to depend on the volume, rather than on the size of the temper, even though wear-resistance does not increase significantly in fabrics containing over 20% of non-plastics.125

Tunisian ARFs are generally rich in iron and calcium while their North Tunisian counterparts are characterized by increased aluminum content. At least three explanations have been offered for the origin of ARFs in African pottery. First, they are said to consist of undissolved lumps of clay or conglomerations of argillaceous minerals formed during the firing process.128 The interpretation is untenable, as these argillaceous formations are shaped like lenses or pellets, they do not only have a clear boundary with the clay matrix, but also differ from it in color. Of the two remaining possibilities, the suggested use of grog i.e. crushed potsherd temper in African pottery,129 is also very dubious. Grog and ARFs are usually hard to distinguish from one another, unless the grog was prepared from distinctly different pottery fabric or such inclusions preserve traces of slip.130 It has recently been concluded that grog particles often have their internal structure aligned parallel to the vessel surface, they crack parallel to their greatest length and are usually evenly distributed.131 Because neither of the two African cookware fabrics include temper that would meet these characteristics, the presence of grog can be excluded. The size and the shape of these argillaceous pellets indicate that they were probably introduced into the clay with the sand-sized temper. Some examples of the North Tunisian fabric also include sparse reddish-brown grains, in which the iron content exceeds 15%. These fragments of iron-rich sedimentary rock or iron ore, which is their traditional definition,132 are most likely conglomerations of hematite coated detrital nuclei.133 Alternatively, the presence of hematite may result from goethitehematite transformation that takes place in temperatures around 250-280 °C.134 The presence of iron compounds, like hematite or limonite, in pottery is believed to reflect careless clay paste preparation.135 The sparse amounts of iron-rich rock fragments in African cookware was most likely introduced to the ceramic body with the sand temper, whereas more irregular dark brown iron-rich stains are probably original components of the clay.

2.5.5 Rock fragments ( argillaceous, iron-rich, quartzite) Both sedimentary and metamorphic rock fragments have been included in this group. The only example of the latter rock type is quartzite, which has been occasionally detected in African cookware.126 But as quartzite is a siliceous rock formed of metamorphosed quartz arenites,127 its behavior as a tempering material is practically identical to quartz. As quartzite occurs only rarely in the study material, its presence probably had no particular effect on the production or use of these vessels. The same can be said of argillaceous rock fragments (ARFs), which are, however, clearly more abundant in African cookware than quartzite. Although ARFs, which include occasional grains of minute microcrystalline quartz are present in both major African cookware fabrics, their appearance in these two productions is clearly different. In North Tunisian fabric variants ARFs are composed of somewhat finer particles with respect to Central Tunisian fabric, their color of ranges from light yellow to orange and the grain is more often than not lens-shaped. Yellowish lens-shaped ARFs are especially characteristic of fabric variant II.1 (Plate 1d), less common in other North Tunisian variants and completely absent in Central Tunisian pottery, in which ARFs are predominantly light brown to redbrown spherical bodies (Plate 1b). The examination of ARFs with SEM/EDS was successful enough to point out that in spite of their increased aluminum and potassium content indicating strong presence of clay minerals, the chemical composition of ARFs is very similar to the corresponding clay matrix. Therefore, Central

27

Chapter 2

CHAPTER 2 ENDNOTES 1

2

3 4 5 6 7

8

9

10 11 12 13 14 15 16 17 18 19

20

21 22 23 24 25 26 27

28

29

30 31 32 33 34 35 36 37 38 39 40

41

Hayes 1972, 13-14. However, some twenty years later Hayes (1994, 110) denied the presence of mica, and re-evalued the calcium content of African Red Slip ware. Taylor & Robinson 1996a, 231-232; cf. Dore 1988, 76; Aguarod Otal 1991, 404. Due to similar geology, difficulties have also been encountered in distinguishing Sicilian coarsewares and amphorae from North African products, see Alaimo et al. 1997, 47-49. Also the mineralogy of cookwares produced in the province of Cyrenaica resemble closely the products of Africa Proconsularis, but they can be sometimes identified by the presence of chert, which is a foreign inclusion to other African fabrics, see Riley 1979b, 40-41. Tite 1999, 196, see also Burollet 1967. Peacock 1984b, 14; Peacock and Tomber 1991, 292-293. Peacock et al. 1990, 61; Taylor & Robinson 1996a; cf. Taylor & Robinson 1996b. Mackensen & Schneider 2002. Mackensen & Schneider 2002, 141-142; see also Freed 1998, 18, 21 note 44. Tortorella (1995, 95) attributes the discovery of North and Central Tunisian fabrics to Dore (1989) and his work with the finds from Sabratha. Mackensen & Schneider 2002, 134-142. Already the results published by Taylor et al. (1997) supported the possibility to assign precise provenances for seemingly homogeneous productions. In their study on the find material recovered from the Plemmirio B shipwreck, the Africana 1 and 2A amphorae were shown to belong more likely to the output of a kiln site located at Sullecthum than the one of Leptiminus. The precision of these results i s reflected by the fact that both sites are located in coastal Central Tunisia, less than 50 kilometers apart from one another. In addition, altogether eighteen African cookware vessels – representing less than 1 % of the study assemblage – were discovered in more or less unique fabrics. Because of their heterogeneous nature and small number, the description of these fabrics has been omitted. Peacock 1984b, 14-18. Tomber 1986, 36. For fabric characterizations, cf. Peacock & Tomber 1991, 294-298. Peacock & Tomber 1991, 302. Peacock 1984b, 18; Olcese 1993, 87-88, 138; Aguarod Otal & Pérez Arantegui 1995, 145; Freed 1998, 21 note 44. Olcese 1993, 138-139, see also Tite 1999, 200. Peacock 1984b, 6; Tite 1999, 226. Peacock 1977b, 25; Schuring 1987, 64-65; Schuring 1988, 13, 26, 45, 51. Cf. Jones 1994, 15. The application of the analytical scanning electron microscopy in the study of archaeological ceramics has been summarized by Tite et al. 1982; Freestone & Middleton 1987; Tite 1992. On the use of SEM in the materials science, see Gabriel 1985. Schuring 1988, 50-51. All the studied examples were classified into one petrographic group characterized both b y the presence of varying amounts of silt in the clay matrix and a gap in the grain size distribution between naturally present silt and quartz sand temper. The advantages of thin-sectioning have been brought up e.g. in Williams 1979; 1990; Freestone 1995; Rice 1996b. Olcese 1993, 80-81, 138-139. Peña 1999, 60; Tite 1999, 195; Tite et al. 2001, 303. Jones 1994, 15-16. Rice 1996b, 168-173. Picon & Olcese 1995, 106; Tite et al. 2001, 302-316. The Central Tunisian pottery fabric was first recognized and described by Peacock (1984b, 17-18, fabric 2.6) See also Aguarod Otal 1991, 393-396 (ceramic paste no. 14); Freed 1998, 21 (Byzacenan blacktop red cooking ware); Peña 1999, 185 (fabric 2I). van der Werff 1982, 110-111, 122-124, 415-421. To my mind, the synchronous change of these two presumed wares, which was also observed by van der Werff (1982, 77, 124), confirms that they pertain to the same production. Due to insufficient number of examples (N=24) in Central Tunisian fabric, the phase 10 was excluded from the analysis. van der Werff 1982, 125. Dore 1989, 89-81. Peacock & Tomber 1991, 294-298. Peacock 1984b, 14. Peña 1999, 185 (fabric 2f). Cf. Peña 1999, 133. Hayes 1976, 94; Hayes 1980, 518; Freed 1998, 21 note 44. Peacock 1984b, 15, fabric 2.3. Black-surface ware. Hayes 1980, 518. Freed 1998, 21 (note 43). The samples were analyzed at the Institute of Electron Optics of Oulu University, using a Jeol Superprobe JCXA-733 electron probe microanalyzer with an automatic Link Lemas detector system. All analyses were made using a 15 kV accelerating voltage and 15 nA beam current. In order to analyze the area between temper particles, a beam size of 10µm was used. Freestone & Middleton 1987, 23. According to Tite (1992, 115) a calcareous clay contains over “5% of finetextured, well-dispersed lime plus magnesia”, whereas both Picon & Olcese (1995, 107) and Aguarod Otal & Pérez Arantegui (1995, 145) have set their limit to 7-8%. In his recent article, however, Tite (1999, 185, 216) has shifted

28

Fabrics

42 43 44 45 46

47 48 49 50

51 52 53 54 55 56 57 58 59 60 61

62

63 64 65 66 67 68 69 70

71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87

88 89 90 91 92

93

the limit to 15-25% of well-dispersed lime. Alternatively, the matrix of Central Tunisian fabric could perhaps be defined as semi-calcitic, cf. Hayes 1994, 109. Cogswell et al. 1996. Cf. Burragato et al. 1994b, 148-149. Aguarod Otal 1991, 45-46, 369; Aguarod Otal & Pérez Arantegui 1995, 146, fig. 2. Pacetti & Sfrecola 1989, 501-503. Olcese 1993, 80, 138-139. Compositional groups 7 & 8 with the raw data introduced in the appendix. The relative absence of calcite together with the compositional similarity to African Red Slip ware D identifies the group 7 as North Tunisian fabric, while the identification of group 8 as Central Tunisian fabric is based both on the increased calcium content and the frequent presence of a band of whitish scum on the rim of these vessels, see Olcese 1993, 80. Veniale 1994, 64. van der Werff 1982, 418. Heimann 1989, 137; Picon & Olcese 1995, 107-108; Tite 1999, 216. Refractory= “ceramic materials, usually high in alumina or silica, that can withstand high temperatures and are slow to melt.”, Rice 1987, 481. Schuring 1988, 21. Schuring 1986, 196; 1988, 38-39. Aguarod Otal 1995, 136. Freestone 1982, 100. Schuring 1988, 23. Tite et al. 1982, 119; Hayes 1997, 59; Tite 1999, 216. Heimann 1989, 136; Hoard et al. 1995, 829-30; Picon & Olcese 1995, 110. Peña 1999, 135. Picon & Olcese 1995, 106-110; Tite et al. 2001, 307-309. Heinrich 1965, 339; Press & Siever 1986, 103-104; Picon & Olcese 1995, 106-110. Heimann 1989, 136; Mitchell & Hart 1989, 148-150, tabb. I-II; cf. Veniale (1994, 60) who estimates the temperature range for the formation of gehlenite in oxidizing firing as 750-950 °C. Kaolinite has also been observed in samples taken from clay formations at Carthage, see Bullard 1985, 130-132. Aguarod Otal 1991, 393; Mackensen 1993, 170-172; Olcese 1993, 87; Freed 1994, 26-32; Sánchez Sánchez 1995, 279. Burragato et al. 1994b, 146. Freed 1995, 165. Peacock & Tomber 1991, 292-298. Sánchez Sánchez 1995, 279. Heinrich 1965, 281, 288; Rye 1981, 35; van der Werff 1982, 418; Rice 1987, 37, 410; Courty & Roux 1995, 26. Freestone et al. 1985, 164-175. Kilikoglou et al. 1998, 275-276. The results of various experiments as well as the concepts of strenght, toughness and thermal shock resistance are described in Tite et al. 2001 as well as in Tite & Kilikoglou 2002. Cf. Hayes 1997, 59; see also Peña 1992, 114-115. Picon & Olcese 1995, 110; Tite et al. 2001, 321-323. Schuring 1988, 43-44. Peña 1992, 117-119; Picon & Olcese 1995, 111-112. Klein & Hurlbut 1993, 541. van der Werff 1982, 418-419; Burragato et al. 1994b, 146. Heinrich 1965, 339; Burragato et al. 1994b, 146. Cf. Fabbri et al. 1997, 188. Hayes 1972, 14; see also van der Werff 1982, 419; Burragato et al. 1994b, 146. Burragato et al. 1994b, 144-146, see also Brown & Mattingly 2001, 392. Press & Siever 1986, 74-75; Pacetti & Sfrecola 1989, 501-503. Press & Siever 1986, 74-75; Rice 1987, 410. Rice 1987, 410. Tite & Kilikoglou 2002, 3. Peacock 1984b, 14-; Freed 1994, 26-32; Freed 1995, 165. Peacock & Tomber 1991, 292-298. Hoard et al. 1995, 825. On the other hand, Freed (1998, 24 note 58) ranks angular lime as a decisive factor i n distinguishing Central Tunisian products from North Tunisian ones. On the formation of secondary calcite in archaeological ceramics, see Cau Ontiveros et al. 2002. Burragato et al. 1994b, 144. Rye 1981, 33; Rice 1987, 98; Kilikoglou et al. 1998, 262. Arnold 1985, 26-28; Rice 1987, 97-98; Schuring 1988, 38; Feathers 1989, 580-587; Fabbri et al. 1997, 187-190. Rye 1976, 120-135; The prevention of lime blowing was studied extensively in Central Europe between 1880-1910. The results of these experiments are summarized in Butterworth 1956. Rye 1976, 131. Lime blowing can also be avoided by immersing the objects in cold water immediately after firing or by firing them in reducing atmosphere (see Laird & Worcester 1956, 555), but these methods were not used in Roman Africa.

29

Chapter 2

94

Laird & Worcester 1956, 559. The examination of a large clay preparation tank dug in sand aquifer at Leptiminus has raised the possibility that these structures were naturally supplied with slightly saline groundwater, see Brown & Mattingly 2001, 389-390. 95 Peacock 1984c, 263-264. 96 Butterworth 1956, 540; Laird & Worcester 1956, 549-560. 97 Hoard et al. 1995, 823-824; Fabbri et al. 1997, 190. 98 Hoard et al. 1995, 825; Tite et al. 2001, 313. 99 Peña 1992, 113; Hoard et al. 1995, 824-825, 831, cf. Fabbri et al. 1997, 186. 100 O'Brien et al. 1994, 284-285 (fig. 5); Hoard et al. 1995, 826-829. 101 O'Brien et al. 1994, 283-285; Hoard et al. 1995, 829-830; Tite et al. 2001, 316. 102 Hoard et al. 1995, 828-829. 103 Peacock et al. 1989, 182; Peacock & Tomber 1991, 292-298. 104 Keller et al. 1978, 591; Freestone 1982, 109. Tephra is “material of any size or composition ejected by volcanic explosions.” 105 Steen-McIntyre 1985, 268-269, 285; Prothero & Schwab 1996, 313-316. 106 While most tephras ejected by Italian volcanos contain both augite and sanidine, this combination is characteristic of a tephra that resulted from the eruption of Mt. Vesuvius ca. 1200 BC. This tephra is present in deep-sea drilling cores obtained from south of Crete, some 1200 kms southwest of the volcano. Still, the possibility that augite i n African pottery would derive from other, closely located volcano – with the exception of weakly explosive Mt. Etna in Sicily – cannot be excluded, see Keller et al. 1978, 595-602. 107 Williams 1984, 220; Peacock & Tomber 1991, 292-298, 301-302. 108 Heinrich 1965, 221; Prothero & Schwab 1996, 82. 109 Pacetti & Sfrecola 1989, 501-503; Pasquinucci & Menchelli 1996, 505, see also Burollet 1967, fig. 6 (fold-out map). 110 Heimann 1989, 136; Mitchell & Hart 1989, 153. 111 Schuring 1988, 50. 112 Rice 1987, 37; Mannoni & Sfrecola 1990, 315-318; Aguarod Otal 1991, 393; Burragato et al. 1994b, 146. 113 Prothero & Schwab 1996, 157-167. 114 Peña 1992, 118; Magetti 1994, 27-28. 115 While the difference between the temper and natural inclusions of clay is clearer in North Tunisian fabric, the recent research with the clay sources at Leptiminus (Central Tunisia) has produced evidence of clay deposits showing natural bimodal distribution in the size of quartz grains, see Brown & Mattingly 2001, 392: table 6.3. Peña (1999, 133), o n the other hand, has identified the coarser fraction of quartz sand in North Tunisian fabric as a natural component of clay. 116 Schuring 1986, 196. The bimodal distribution of quartz in African pottery has previously been observed also b y Freed (1994, 26-32) and Sánchez Sánchez (1995, 279). 117 Rice 1987, 95; Schuring (1986, 191) has reported an increase of 5 %. 118 Bronitsky & Hamer 1986, 98; Schuring 1986, 192, see also fig. 16 i, 191; Schuring 1988, 44; Peña 1992, 114. The idea has been doubted only lately (see O'Brien et al. 1994, 282-283). 119 Braun 1983, 123-124; Schuring 1986, 193, cf. fig. 24, 195; Feathers 1989, 585, fig. 1; Hoard et al. 1995, 825. 120 Bronitsky & Hamer 1986, 95-97. 121 Kilikoglou et al. 1995, 2963-2965; Kilikoglou et al. 1998, 267-268, 273-274. 122 Kilikoglou et al. 1995, 2962-2965; Kilikoglou et al. 1998, 272-273; Tite & Kilikoglou 2002, 2-3. 123 Kilikoglou et al. 1995, 2960-2962; Kilikoglou et al. 1998, 274. 124 Vekinis & Kilikoglou 1998, 286. 125 Vekinis & Kilikoglou 1998, 286-288. 126 E.g. Aguarod Otal 1991, 394. 127 Press & Siever 1986, 77. 128 Schuring 1984, 161, 166; Schuring 1988, 39, 50; Aguarod Otal 1991, 394. 129 E.g. Schuring 1988, 50; González Villaescusa 1990, 35-36 (Hayes 197: nos. 031, 034, 046, 047); Freed 1994, 2632; Freed 1998, passim.; Brown & Mattingly 2001, 392: table 6.3 130 Whitbread 1986, 79-84. 131 Cuomo di Caprio & Vaughan 1993, 29-39. 132 Mackensen 1993, 170-172, 437; Freed 1994, 26-32. 133 Heinrich 1965, 74; Prothero & Schwab 1996, 302. 134 Aguarod Otal 1991, 362. 135 Burragato et al. 1994b, 146.

30

Forms

3 FORMS

3.1 INTRODUCTION This section is designed to describe the methods and conventions used in the following pages, which offer a discussion on some twenty-one hundred African cookware vessels that have been divided into several forms and form variants. The material will be presented in six groups, of which the most consist of forms related to one another by the function. The most apparent exception is the group of Central Tunisian casseroles, which is presented separately from the North Tunisian production, because of its distinct form repertoire. By contrast, lids of Central Tunisian origin will be introduced with the rest of African cookware lids, as some of the forms are also common in North Tunisian production. In most cases, the identification of a new form has been based on the shape of the entire vessel, whereas conspicuous differences in the shape of a rim are generally considered to be the principal criteria for defining variants. Five of these classes, namely lids, pans, shallow casseroles and both deep casserole groups, are definitions commonly used with this particular class of ancient ceramics. The last group has been reserved for miscellaneous finds, a heterogeneous assemblage consisting of less than a dozen sherds. Most of these belong to bases, which were intentionally separated from ring-shaped handle attachments (form F). The latter form is shown to belong together with lid rims, which have been more often than not erroneously identified as lid/plates. Because of the evident relationship, they will be presented together with lid rims. The presentation of each group begins with an overall characterization, which applies to all the forms it includes. Although these forms are functionally related to one another, the function of the group as a whole is not necessarily fixed. For this reason, the function of each group will be discussed separately from the general introduction (see section 3.9).1 Only in cases when the study assemblage offers specific evidence on the use of particular form or variant, or the function may have differed from the group it has been traditionally attributed to, this question will be picked up at the point of its introduction. This theme will be examined here with a variety of methods. These include comparisons with potential associated forms, both by establishing correlation coefficients based on their distribution in horizons and charting relative cumulative frequency distributions from vessel diameters.2 Both the shape of a form and the characteristic surface finishes may also point to its use. The traces resulting from use-alteration, visible

in form of wear marks and soot patches, are yet another important source of information. The discussion on each group may include a description of several forms and variants. A form is defined here as “a distinct vessel shape attained by a specific set of primary forming operations”, whereas minor variations in vessel morphology or secondary forming operations have been used to distinguish (form) variants. These variants may bear witness to “workshop origin, a distinct function of method of use, or some diachronic morphological changes.”3 No particular set of rules was followed when defining the variants, although the variability detected in the shape of the rim was usually the most decisive criterion. The material was approached from "splitter's" point of view, meaning that it was first split to the smallest possible fractions. The establishment of such careful classification was assumed to provide additional information on both the typology and chronology of African cookware. However, much of the variability first thought to be chronologically significant, was both so inconsistent and randomly occurring that it probably resulted from "a potter's attempt to relieve the tedium of throwing so many pots each day."4 Each entry contains the following information. The distribution of a form or a variant in the study assemblage is summarized in a table, which also includes the principal vessel dimensions – such as rim diameter and width – and fabric distribution. As the criterion used to identify forms follows examples set by previous research, a form may be referred to with several names. The classification proposed in the pages of Late Roman Pottery5 is generally preferred. After a short characterization of the form, attention is turned to its chronology. Examples of the form or variant recovered from a known production site located in Africa Proconsularis will be mentioned separately, while the full discussion about this evidence is presented in section 4.4. A list of other published examples of the same form or variant (i.e. comparanda), which is intended to be a representative sample instead of complete list of known occurrences, has been incorporated in the find catalogue. The find catalogue (see Appendix 1) includes a detailed presentation of one to three indicative examples of each form or variant.6 Besides the description of a vessel form, some remarks on manufacturing techniques are also included, although they will receive a more thorough treatment in section 4.1. All the catalogued examples are illustrated, but because in many cases the material found in Late 31

Chapter 3

Roman horizons is very fragmentary, a better idea about the appearance of the form can be obtained by consulting the publications listed under the heading “comparanda”. The catalogue also includes numerical

data indicating the dimensions of a sherd(s) or vessel in question, its colors expressed as Munsell soil color chart (1975 edition) values and some remarks on its preservation.

3.2 LIDS observation,10 according to which the use of lids in cooking distinguished the Mediterranean from the northern provinces of the Roman Empire. The suggestion finds strong support in African cookware, for which vessels equipped with a lid seating are almost as typical as they are in the West-Central Italian cookware. The general variability within this group is fairly well known, as it consists of two or three principal forms (Hayes 182, 195 and 196), which can be divided to a limited number of variants. Moreover, the study assemblage also confirms Fulford's observation concerning the typology of this group, according to which vessels with plain rounded rims give place to profiled, upturned rims over the course of the centuries.11

This group consists of shallow open forms with a straight or slightly convex wall ending in either thickened or upturned rim. The center of the exterior surface is either plain or equipped with one of the three features indicating an additional step in the forming sequence after the vessel body had been thrown on the potter's wheel: a disk knob, turned7 ring-shaped ridge or a set of concentric grooves. This vessel part has received constant attention in the debate on the use of this group (see section 3.9.1). However, the one not leaving room for another identification than a lid, is the disk knob, which is not only an early but also a rare feature in African cookware (cf. form E). The number of fragments with a ring-shaped protrusion (cf. form F) also differs notably from the number of lid rim sherds, as their ratio is only 1:8. The discrepancy can be explained in a variety of ways. First, in some forms the exterior wall is known to curve uninterruptedly from one side of the vessel to another without a ring-shaped ridge or knob. This is likely the case with both the Hayes 195 series (group B) and Central Tunisian lids (group C). Secondly, differential breakage affects the recovery rate of various vessel parts. The center of a vessel, even when crowned with a ring-shaped protrusion, is obviously less resistant to postdepositional wearing, than a robust, thickened rim. The circumference of a rim exceeds often at least three or four times the circumference of the vessel center increasing equally the recovery probability of the former sherds, as the breakage certainly results in more rim than other diagnostic fragments. The shape of the rim can be used to give an approximate date for the forms in this group, even though the establishment of a comprehensive typology has turned out to be difficult.8 The similarity of the forms belonging to this group of ceramica ad orlo annerito9 was pronounced already by Lamboglia, who distinguished it from other African cookware. On the contrary, the aim of this study is to demonstrate that in use this group formed an inseparable entity with the rest of African cookware vessels. The possible relationship of a given lid form/variant to other African cookware forms will be examined under a respective entry, while a more profound discussion will be offered in section 3.9.1. As the attention is turned to the individual lid forms and variants, it is useful to mention Dore's

3.2.1 Plain or thickened North Tunisian lids The group to be touched here first is composed of North Tunisian lids with plain or thickened rim, the Hayes 196 series, that forms a typological continuum extending from the early 1st at least to the 5th century AD. The longevity of this form has caused some difficulties as practically every scholar has used a different set of parameters to define variants within this group. Aguarod Otal, for example, distinguishes two variants (Ostia I.261 & III.332) on the basis of the presence of a knob or ring-shaped protrusion.12 The application of this method may present severe problems, especially if the chronological coexistence of several variants is concerned. On every archaeological site, the main body of diagnostic African cookware lid fragments usually consists of rim sherds and fragmentary knobs or ring-shaped protrusions, whereas examples preserving both rim and knob are rare. As reliable links between different rim and knob forms remain to be established, the method must be regarded as insufficient for now. Still, the possibility that certain knob forms, ring-shaped protrusions and lid rims are related to one another will be touched upon in the following pages. The method applied in this study is simple and straightforward. According to the shape and thickness of the rim, three variants have been distinguished from the material. Variant A.I includes round rims, which are either of even width relative to the vessel body or slightly thicker. The two remaining groups consist of thickened angular rims, with every rim sherd less than 15 mm wide classified as variant A.II and those

32

Forms

Horizon

Phase 6

Phase 7

Phase 8

Phase 9

A.I

7.53

7.03

5.65

6.38

5.80

6.38

A.II

6.90

8.46

6.68

9.33

5.78

6.48

A.III

5.50

8.05

8.05

8.58

8.50

8.27

Table 6. Fragmentation of the A.I-III series by phases. Fragmentation rates for variants in each phase were determined as an average of estimated vessel equivalents.

Phase 10 Phase 11

exceeding this limit as variant A.III.13 The method may be criticized for being unnecessarily technical and subjective, as the limit between the variants is based on nothing else than intuition. Although this artificial and highly processualistic approach obviously does not relate to the way the people making use of these objects grasped them, it enables the examination of this group in smaller units. Moreover, the possibility that rim width would positively correlate with vessel diameter was excluded by the cross-tabulation of appropriate parameters. The fact that the average width of lid rims in the group A increases as a function of time is illustrated as their cumulative distribution (Figure 5) is broken down into chronological phases.14 Additional evidence supporting the division can be found by inserting the chronological dimension into the fragmentation of these variants. The cross-tabulation shows clearly that examples of the variant A.III are generally less fragmented than examples of the other two variants (Table 6). Although the discrepancy may partially result from differences in the vessel breakage, as a lid with thickened rim shatters to bigger pieces than one with indistinct rim, it may equally reflect the post-

depositional history of these vessels. Thus, the examples of variant A.I belonging to phases 8-11 are as likely residual as the presence of variant A.III in phase 6 has to be regarded as an intrusive component in the assemblage. The variation in the rim shape reflects minor changes in the vessel forming sequence, the main stages of which were not altered during the production. The lid was thrown rim side up and left to dry until it reached the leather-hard stage. Thereupon the lid was inverted, re-mounted to the wheel-head and turned from the upper exterior surface. In the case of Hayes 196 lids, the procedure seems to have included also the turning of ring-shaped protrusion on the vessel top. The firing conditions have also remained highly uniform as over 90% of the examples in each entry show completely oxidized fabric, excluding the utmost part of the rim, which is often covered from brown to black orlo annerito deposition. The following presentation of the Hayes 196 series (variants A.I-III, see Table 7) is concluded by the introduction of variant A.IV, which will be interpreted as an early example of Late Cookware production. Although the variant A.IV does not

100 %

80 %

60 %

40 %

20 %

4

6

Phase 6

8

10

Phase 7

12

14

Phase 8

16

18

Phase 9

20

0% 22 Thickness

(mm)

Phase 11

Figure 5. The cumulative rim thickness distribution of A.I-III lid series (combined) by chronological phases. 33

Chapter 3 Table 7. The size and the distribution of plain or thickened North Tunisian lids. Diameter

A

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

590

A.I

14

21/27

34

-

-

-

5

7/9

13

75

35

34

21

10

16

16

175

-

191

A.II

16

24/27

39

-

-

-

7

10/12

16

72

65

90

21

9

27

-

283

1

284

A.III

22

26/30

43

-

-

-

13

17/19

22

6

19

40

19

4

11

6

91

2

99

A.IV

14

19/23

25

-

-

-

6

8/10

11

-

3

1

3

-

9

2

14

-

16

strictly belong to the Hayes 196 series, it is clearly a late derivative of the same basic shape and such being the case, worth presenting in this context. The variant A.I (Plate 3:1-2), which has not only been regarded as the earliest African cookware lid but also the earliest vessel form ever produced in African cookware,15 has a round rim that does not differ significantly from the vessel wall in width. The variant represents the first stage in the typology of the Hayes 196 lid series, which is characterized by the thickening of the rim as a function of time. It is present in the study assemblage with nearly two hundred examples, practically all of which are of the North Tunisian fabric. The only feature in the production differing significantly from later variants is the high percentage of examples showing no orlo annerito deposition on the vessel rim.16 This may result from slight adjustments made to firing methods in the course of production.

Due to heterogeneous criteria used in the definition of the Hayes 196 variants, both the introductory and the closing date of variant A.I are fairly hard to fix. At Carthage examples of this variant have been recovered from deposits dated to the last decade of the 1st century BC.17 The date of introduction outside Africa Proconsularis, on the other hand, has traditionally been placed to the Flavian period,18 while Spanish scholars have recently been inclined to push it back to the Augustan-Tiberian epoch.19 In the Italian peninsula the earliest verified evidence on the presence of this variant comes from several early 1st century AD contexts, whereas in France similarly dated evidence has been reported from Fréjus.20 The presence of the variant A.I characterizes also the early deposits (dated ca. AD 50-90) at Carthage and the Terme del Nuotatore (Ostia), as well as the mid-2nd century AD deposits of Roselle.21 While the study assemblage can not add new elements

100 %

80 %

60 %

40 %

20 %

0% ø 14

16

18

20

A.I (early Hayes 196)

22

24

26

28

J.I (early Hayes 23b)

30

32

34

O.I (early Hayes 197)

Figure 6. The cumulative rim diameter distribution of three North Tunisian cookware form variants: A.I lid, J.I shallow casserole and O.I deep casserole.

34

Forms

100 %

80 %

60 %

40 %

20 %

0% ø 15

17

19

21

23

25

27

29

North Tunisian

31

33

35

37

39

Central Tunisian

Figure 7. The cumulative rim diameter distribution of lid variants A.II & C.II, a comparison of North and Central Tunisian production. to the discussion, as its horizons date from the late 3rd century AD onwards, the early Imperial deposits of the Palatine East excavations confirm the presence of variant A.I outside Africa Proconsularis by the early 1st century AD. The date of the transitional period between variants A.I and A.II is equally obscure. For example, while Aquilué Abadías dates the end of the production of variant A.I to the first quarter of the 2nd century AD, according to González Villaescusa it did not cease before the latter half of the 2nd century AD.22 It is also possible that both estimates compress the chronology of the variant A.I, as the production seems to have continued without any significant modifications at least until the early 3rd century AD, whereupon the rim becomes clearly more pronounced and is identifiable as variant A.II.23 In the study assemblage, this is indicated by the decreasing number of A.I variants towards the late 3rd/early 4th century AD horizons. Thus, the examples of variant A.I found in the 4th century AD contexts at Ventimiglia are most certainly residual,24 whereas the latest closing date of the mid 5th century AD proposed by Fulford, may actually refer to other variants in this group.25 When it comes to the vessel use, which will be extensively discussed in section 3.9.1, one of the few scholars to relate the variant A.I directly with a certain casserole type is González Villaescusa, who suggests that lids of the Hayes 196 series were used either with

the Hayes 23B shallow casserole or the Hayes 197 deep casserole.26 The interpretation is only partially supported by the study assemblage, as the comparison of the cumulative rim diameter distributions emphasizes only the relationship of early Hayes 196 lid and the early Hayes 197 variants (Figure 6). Conversely, the early Hayes 23B variants are generally a couple of centimeters smaller in diameter than the two other forms. The variant A.II (Plate 3:3-4), characterized by a slightly angular, thickened rim and a ring-shaped protrusion topping the upper exterior surface, represents the intermediate stage of the Hayes 196 typology and the definition “standard Hayes 196” describes its importance. As indicated above, the distinction to variant A.III is not unambiguous, but something created to facilitate the processing of the study assemblage. Neither is the provenance of this variant homogeneous as the fabric in circa 27% of nearly four hundred examples is clearly Central Tunisian (see variant C.II). For the sake of clearness, these productions will be treated separately, although even in closer examination the production sequence of North and Central Tunisian vessels does not show remarkable differences. The similarity is also observed in the comparison of cumulative vessel diameter distributions of both productions (Figure 7), although Central Tunisian lids are generally slightly larger in size than their North Tunisian parallels.

35

Chapter 3

100 %

80 %

60 %

40 %

20 %

0%

ø 14

16

18

A.II (Hayes 196)

20

22

24

26

28

J.II (Hayes 23B)

30

32

O.I (Hayes 197)

34

36

38

O.I (Hayes 197)

Figure 8. The cumulative rim diameter distribution of four North Tunisian cookware form variants: A.II lid, J.II shallow casserole and O.I & O.II deep casseroles. All the 1st century AD examples of the variant A.II27 can be safely regarded as intrusives, because the production in large scale seems only to have started in the 2nd century AD, although not later than the early 3rd century AD.28 Due to gradual development, which began in the early 2nd century AD,29 the validity of an alternative date from the early 2nd to the early 3rd century AD, is a question beyond the scope of this discussion.30 The evidence of variant A.II is abundant in the early/mid-3rd century AD contexts,31 but a closing date of the mid- or late 3rd century AD has also been proposed.32 The study assemblage, in which variant A.II occurs also in phases 10 and 11, does support this interpretation but rather suggests that the production continued at least to the late 4th or even the early 5th century AD. A late closing date is equally indicated by several late Roman contexts from Carthage,33 where the terminal date has been fixed to 450 AD.34 Although this is very likely the right date, it does not automatically imply that the examples found in the late 5th century AD deposit of S. Maria Capua Vetere35 are residuals. In all, it seems acceptable to date the production of the variant A.II from the 2nd to the 5th century AD.36 Carandini has postulated that production took place in the same workshops that also manufactured late African Red Slip ware A,37 while the whole Hayes 196 series is said to originate in the region of Carthage.38 As indicated above, the study assemblage does not fully support this view, because the fabric in

over one fourth of the examples is Central Tunisian. Thus, a part of the production must have also taken place in workshops other than those manufacturing African Red Slip ware A. Although the “standard Hayes 196” lid is one of the most abundant African cookware forms found almost on every archaeological site along the West Mediterranean coast after the 2nd century AD, its possible association with certain pan and casserole forms has not received much attention. This negligence results partly from the tradition of interpreting the Hayes 196 series as plates rather than lids. Hayes, who took this variant as an heir to his form 192, suggested that it was used to cover the Hayes 23 shallow casserole, even though smaller examples may have been used with deep casserole forms 197-198 of his classification.39 As indicated by the cumulative rim diameter distribution of several North Tunisian vessel forms and variants (Figure 8), the observation itself is fairly accurate as the Hayes 23B shallow casserole is related to the variant A.II rather than either of the Hayes 197 deep casserole reference groups. The variant A.III (Plate 3:5-6), with heavily thickened angular rim, represents the end point of the three-staged Hayes 196 typology,40 but only a few scholars have previously distinguished it from its predecessors.41 This category includes nearly a hundred vessels, of which the vast majority belongs to North Tunisian production. Thus, the production of Hayes

36

Forms

196 lids seems to have ceased in Central Tunisia some time before this variant was introduced by the workshops located in the northern parts of Africa Proconsularis. Regarding the vessel production sequence, the only novelty shown by the variant A.III is the increased presence of a shallow furrow on the interior surface immediately adjacent to rim. While this feature is present only in 3% of the examples of the variant A.I, in case of variant A.II the percentage is 8 and in group A.III 20%. This modification may have been necessary to retain the usefulness of a cooking set. With the addition of a furrow under the lid rim, the stability of the Hayes 196 lid on the top of a cooking vessel with a reduced or non-existent horizontal flange was maintained.42 Turning to the chronology of the variant A.III, the mid-2nd century AD is the earliest suggested date of introduction,43 while the date of the late 2nd/early 3rd century AD is supported by vast archaeological evidence.44 Although production seems to have begun in the late 2nd century AD, various excavations have shown that the heyday of variant A.III dates to the 3rd and 4th century AD.45 As always, there are few exceptions, like the room XIV of the Terme del Nuotatore (Ostia), which included a prominent mid-3rd century stratum without any rims that could have been classified as the variant A.III.46 In other contemporary contexts, like the one dated between 240-260 AD at Ebusus, A.II is still the dominant variant of the Hayes 196 series.47 Hence, it is not unreasonable to date

variant A.III after the 3rd century AD. This interpretation is further supported by Carandini's reference to sites, where the variant A.III has been found among kiln debris together with African Red Slip ware D, for which the earliest possible dating is the early 4th century AD.48 Interestingly, he has also identified potential production sites of the variant A.III from the surroundings of Tipasa and Caesarea (Algeria). The closing date for the production is not as hard to fix as the date of introduction, as the plentiful evidence from sites around the western Mediterranean, of which the deposits unearthed at Carthage are of primary importance, point unanimously towards the late 4th or early 5th century AD.49 Although the latter chronological terminus has recently gained more support, a suggestion that the production extended to the 6th century AD has also been put forward.50 But, for now it is reasonable to date the production from the late 3rd century to the mid 5th century AD,51 with the most intensive phase of production taking place in the 4th century AD. As far as the use of variant A.III is concerned, it seems natural that it continued the tradition of being used together with the Hayes 23B shallow casserole or the Hayes 197 deep casserole, as Marín Jordá has previously suggested.52 The interpretation is only partially supported by the study assemblage, as the cumulative rim diameter distribution of several late North Tunisian vessel forms (Figure 9) associates variant A.III with medium-sized pan-casserole H.II.

100 %

80 %

60 %

40 %

20 %

0%

ø 17

19

21

A.III (Hayes 196)

23

25

27

29

31

H.II (Hayes 26/181)

33

35

37

J.II (Hayes 23B)

39

41

43

O.VI (Hayes 197)

Figure 9. The cumulative rim diameter distribution of four North Tunisian cookware form variants: A.III lid, H.II pan-casserole, J.II shallow casserole and O.VI deep casserole.

37

Chapter 3

Like the section 3.3 will clearly demonstrate, the result is somewhat surprising, because also the Hayes 195 lid variants can be related to North African pancasseroles. Generally speaking, such a profound change in pairs consisting of a lid and a casserole is hard to imagine, but has likely taken place in this case towards the end of the Hayes 196 production sequence. Although the variant A.IV (Plate 4:7) clearly differs from the three preceding entries in certain details, it simultaneously bears a close resemblance to them. The presence of corrugated surfaces, indicating a decrease in time spent on the production of each vessel, is the most obvious difference. The changes in the production sequence are reflected, besides small rim diameter and rather thin rims, in the vessel surfacing; only a fourth of the examples were smoothed. The firing technique, however, does not differ from other variants in the group A, as most examples show a completely oxidized fracture surface. The ambivalent nature of the variant A.IV is also reflected by bipartite comparanda, which offers two dating possibilities. The first alternative is to assimilate the group with vessels found in contexts dated between the early 1st and the second half of the 2nd century AD at Sabratha (Tripolitania).53 Additional evidence on the early presence of a lid similar to variant A.IV comes from late 2nd/early 3rd century AD context examined at Carthage,54 where also several undated examples have been published.55 The nature of comparanda introduced above must be questioned, as the material from the late Roman deposits of Carthage and other sites56 offers means for an alternative interpretation. According to Tomber, the vessel introduced here as variant A.IV should be identified as a Late Roman Lid 1, the production of which flourished only in the late 5th century AD.57 The following arguments can be introduced to support this identification. Firstly, corrugated surfaces are characteristic of Late Roman cookware I, just like the shallow concavity at the junction of the rim and the interior wall also found in the examples included in the study assemblage. While most of them belong to phase 11, which has been dated preliminarily to AD 400/425-525/550, the variant A.IV is absent in the earliest phase (6) and the fragmentary nature of examples belonging to the following phase (7) suggests that they are residual.

Another feature separating the variant A.IV from other African cookware lids is the rarity of examples with blackened rim surface (orlo annerito), which is highly characteristic of the late Hayes 196 variant A.III. The fabric of these vessels, however, does not match with the description offered by Hayes on the fabric of Late Cookware I, which he relates to africano grande amphorae of Central Tunisian production.58 Therefore, the attribution of the variant A.IV to the Late Cookware I series must be regarded as tentative. 3.2.2 Everted, heavily enlarged North Tunisian lids This group consists of lids related to one another, in addition to their North Tunisian origin by a thickened vessel rim that folds almost to a straight angle with respect to the vessel body. Both the shape of the rim and the way it joins with the interior surface divide the group into four variants (B.I-IV, see Table 8). In addition, two somewhat peculiar versions of variant B.I have been catalogued to illustrate the variability that a single entry may include. In this case, both subvariants result more likely from the accidental slips the potter made while finishing the product than from the deliberate alteration of the vessel design. Another feature characterizing all the variants, is the total absence of protrusions, both ring- and diskshaped, that could be interpreted as lid knobs. Instead, the upper exterior surface frequently bears one to three shallow concentric grooves, which were produced intentionally after the turning process. It is unclear whether these grooves were designed to be functional or not, because alone they are too shallow to act as handles and a device that might have been attached to them to provide better grip is equally hard to imagine. Thus, it is imaginable that these grooves could result from an instrument used to turn the upper exterior surface of these vessels (see section 3.9.1). Generally speaking, the production sequence of vessels belonging to the group B does not differ significantly from the group A. The vessel was first thrown on the potter's wheel rim-side up, and when sufficiently dry inverted for further treatment. Hence, the upper exterior surface shows frequently traces of turning, while the area bordering the rim and the whole interior surface have customarily been left unfinished. The firing in nearly all the examples is

Table 8. The size and the distribution of everted, heavily enlarged North Tunisian lids. B

Diameter

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

174

B.I

21

26/33

43

7

10/12

15

12

15/18

23

16

14

38

9

2

16

3

92

-

95

B.II

20

28/31

38

5

7/9

12

11

13/15

17

-

2

1

-

1

5

2

7

-

9

B.III

27

24/34

44

8

10/12

14

12

15/18

22

1

2

17

5

4

8

-

37

-

37

B.IV

24

29/37

47

10

11/15

19

12

15/18

21

-

2

6

6

1

18

1

32

-

33

38

Forms

fully oxidized, excluding a momentarily subjection to reducing atmosphere long enough to create blackish orlo annerito deposition on the vessel rim. It would be no exaggeration to define the variant B.I (Plate 4:8-9) as the standard late Imperial lid with a hooked rim, because it is abundantly present in most deposits of the period. The variant can be recognized by a rim folded on the exterior surface to form a hook with a rounded tip (lip). Joints between the rim and both surfaces are nearly always sharp, and together with a slightly convex wall an impression of a domed vessel body is created. The form itself is in fact so common that it has long been divided into two variants, the Hayes 182 (C.I) and the Hayes 195 (B.I), which differ from one another both in terms of their fabric and surface finish. The latter variant, which is commonly considered to be a typical late Roman lid of North Tunisian production will be presented here, while the variant C.I, whose production appears to have taken place in Central Tunisia, will be introduced in the following section. With this distinction, nearly a hundred examples of variant B.I in the study assemblage form a very homogeneous group with respect to the production sequence used. For example, the vessel exterior is frequently turned and/or smoothed while the interior has been left in almost 80% of the cases unfinished.59 Regarding the chronology of variant B.I, it should be borne in mind that due to the similar appearance of Hayes 182 and 195 both variants have usually been discussed under the same heading. This mistake is obvious at least in those cases where the variant B.I has been introduced as a “Hayes 195 variant” instead of being defined as an actual Hayes 195.60 Due to the confusion it is often difficult to determine, whether the suggested dates are meant to apply to the Hayes 182 or Hayes 195, or possibly for them both. Hence, in the following discussion references to other sites are often based on my personal interpretation than on facts presented by other scholars. Because of the aforementioned problems, a precise date for the introduction of variant B.I is hard to establish. Evidence both from Carthage and several Spanish sites indicates the late 2nd century AD as the earliest possible date of appearance,61 but some of these examples may be identified as Hayes 182. In any event, the distribution in the study assemblage attests the 3rd and 4th centuries AD as the main period of its occurrence.62 It has also been stated that although variant B.I was already produced in the 3rd century AD, it was only extensively distributed in the 4th century AD.63 The evidence on the closing date of the production, including phase 11 of the study assemblage, indicates that it did diminish in the late 4th century AD, but continued into the early 5th century AD.64 Still, the evidence does not lend any

support to the suggestion that the production of other African cookware lids than the variant B.I had declined by the early 5th century AD.65 It is somewhat difficult to point out a matching casserole form for the variant B.I, even though a set formed by the Hayes 181 pan and the Hayes 195 lid has been postulated.66 Although the association is basically correct, it must be slightly revised since the group of African cookware pans can be divided into Central Tunisian Hayes 181 pans (form G) and North Tunisian Hayes 26/181 pan-casseroles (form H), of which the former is associated with the Hayes 182 and the latter with the Hayes 195 lid. The definition of these two cooking sets is based on similarities in fabric, surface finish and rim diameter, but the full argumentation will be presented only in section 3.9.2 discussing the use of pans and pan-casseroles. Nevertheless, it is worth pointing out that lid variant B.I is more likely related to the medium-sized than to the large pan-casserole like some other lid variants, the variant B.IV in particular. The following two malformed sub-variants are obviously derivatives of the variant B.I. They will be discussed here to illustrate variability that may at first appear significant, but actually results from secondary factors affecting the vessel production sequence, such as careless turning. For this reason, comparanda for such oddities is often laborious to find. Anyhow, their presence in Rome should lead to a question about the status of African cookware, because at least aesthetically they are potential second-rate products. The following explanation chain offers one potential answer. Consumers did not pay overly much attention to the appearance of cookwares, as they were looking for products with good performance characteristics. The distribution of African cookware was based both on good performance characteristics and the volume of trade; and the demand of African cookware rendered possible the sale of second-rate products. This model as well as alternative explanations will receive more profound treatment in section 5.3. The first sub-variant (Plate 4:10) is the otherwise normal Hayes 195 lid with a hooked rim, except that the usual rounded lip is replaced by a flat angular one. In closer examination it becomes evident that at least a part of the utmost section of the lip was cut away with a flat-edged tool, presumably while the potter was turning the upper exterior surface. The cut is in all likelihood unintentional, as the four examples in the study assemblage do not differ from the variant B.I in any other respect. The second Hayes 195 sub-variant is a lid with a hooked rim and a pointed tip (Plate 4:11). As hypothesized above, the presence of such oddities in Rome suggests that the production was not overly standardized as far as forms are concerned, or alternatively the demand was high enough to render possible the sale of products somewhat unusual in

39

Chapter 3

100 %

80 %

60 %

40 %

20 %

0% ø

23

25

27

29

31

33

35

37

B.IV (Hayes 195)

39

41

43

45

47

H.III (Hayes 26/181)

Figure 10. The cumulative rim diameter distribution of two North Tunisian cookware form variants: B.IV lid and H.III pan-casserole. shape. The two examples of this sub-variant have both whitish scum cover on the vessel rim (see Plate 2i) and parallels in the late 4th/early 5th century AD contexts of Carthage.67 The variant B.II (Plate 4:12), which is by far the most infrequent version of the Hayes 195 lid in the study assemblage, differs from the variant B.I in several details. The vessel body is straight instead of being slightly convex, the rim is triangular rather than hook-shaped and a slight protrusion, sometimes defined as a “hanging lip”, is often observed at the junction of the rim and the interior surface. The production sequence does not differ significantly from that for the standard Hayes 195, as most examples show turned and smoothed exterior surface paired regularly with unfinished vessel interior and fully oxidized fabric with the usual orlo annerito deposition on the vessel rim. Aquilué Abadías has dated this variant to AD 350-450,68 and the concentration of examples into phase 11 (AD 400/425-525/550) together with the late 4th/early 5th century AD parallels from Carthage, Otranto and Rome seem to confirm his idea. The interpretation is further supported by the existence of a similar vessel in African Red Slip ware D2,69 produced from the late 4th century AD onwards. The vague similarity between variants B.II and A.IV, which some scholars have classified as a Late Roman cookware lid form, is equally worth stressing. Due to a small number of finds, a matching cooking vessel cannot be found through the comparison of

cumulative rim diameter distributions. Still, it is improbable that the variant B.II would have been designed to go together with any other form than the Hayes 26/181 pan-casserole. The rim of variant B.III (Plate 4:13) is hooked and has often a pointed tip and a “hanging lip”, i.e. a shallow protrusion in the junction of the rim and interior surface. In respect of the previous variants, these modifications have hardly had significant effect on the vessel use. Still, the idea that such a protrusion increased the stability of a lid when placed over a cooking vessel is as plausible and hard to verify as the idea that a shallow groove cut on the interior surface near the rim on some African cookware lid variants (cf. variant A.III) had a similar purpose. The date of variant B.III cannot differ markedly from the one proposed for the standard Hayes 195 (variant B.I). Although the production may have begun in the late 2nd century AD,70 the large-scale distribution of variant B.III took place only from the late 3rd to the 4th century AD.71 Half of the occurrences in the study assemblage belong to the early 4th century AD phase 8, while the strong presence in phases 9-11 confirms the extension of the vessel chronology at least to the first quarter, if not to the second half of the 5th century AD.72 The identification of the late and robust variant B.IV (Plate 5:14) is fairly easy thanks to its heavily thickened rim with a hemispherical top and a sharp, protruding junction between the interior surface and the rim. With respect to the other variants in this

40

Forms

group, variant B.IV is not only thick-walled but also large. Still, the exceptional diameter does not necessarily correlate to the other dimensions of the rim, while the vessel production sequence has been identical to other variants in this group. Due to its robust appearance highlighted by a rim that is more hemispherical than hooked in shape, this variant has long been discussed separately in excavation reports.73 As a result, the evidence on the date of the variant B.IV is exceptionally homogeneous. With one exception,74 no examples from contexts that date earlier than the 4th century AD have been published. The idea, according to which the production ceased already in the early 5th century AD,75 is equally supported by the study assemblage, where the absence of variant B.IV in phase 6 is further highlighted by the clustering of examples in phase 11. The corresponding cooking vessel form is also fairly easy to point out. In addition to the fairly identical cumulative rim diameter distribution with the large Hayes 26/181 pan-casserole (variant H.III, Figure 10), their probable relationship is also suggested by their similar distribution in the study assemblage. Hence, it is possible to conclude this section by claiming a general interrelationship between the Hayes 195 lids (group B) and the Hayes 28/181 pan-casseroles (group H).

subsequently introduced as the Hayes 182 lid (C.I). On the contrary, the two other forms of this group (C.IIIIV) are found almost exclusively in a calcareous dark red Central Tunisian fabric (Table 9). It remains to be examined, however, whether they represent adjustments that potters are occasionally forced to make to meet ever-changing demand, or variations that are characteristic of pre-industrial production. The shape of the form C.I has already been described and illustrated in connection with the introduction of variant B.I (cf. Plate 4:8) and in various publications.77 As far as vessel dimensions or shape are concerned, the some twenty examples of the variant C.I included in the study assemblage do not differ much from variant B.I. Therefore, rim fragments of these two productions are difficult to distinguish from one another solely on a metrical basis. The same can be said about the vessel top, which is more likely equipped with one to three shallow concentric furrows than with a knob or other handle type. Thus, the most apparent difference between the Central and North Tunisian production is fabric (cf. sections 2.2-3), which is further reflected in rims belonging to the former group by a layer of whitishgray scum.78 Although the distribution of scum correlates often with blackish orlo annerito deposition, it resulted from the use of salt water with calcareous clay in the production, while the blackened rim zone was produced in the last moments of mainly oxidizing firing (see section 4.3.1). The careful finish given to the exterior surface is another feature distinguishing these vessels from their North Tunisian parallels in the study assemblage. In addition to the usual combination of turning and smoothing, nearly three fourths of the examples are slipped. While over a half of slipped vessels have also been burnished, the combination of smoothed and burnished exterior surface is not uncommon either. The burnishing was accomplished most likely while the vessel was rotating on the top of the potter's wheel, because it frequently occurs in form of concentric stripes or spirals. Due to small area covered, stripe-burnished surface has traditionally been thought to have hardly had any function besides improving the appearance of the vessel,79 but lately it has been proposed to increase the vessel strength (section 4.2.2). The attention paid to the finish of the

3.2.3 Central Tunisian lids The group C is composed of African cookware lids produced most likely near the coast in eastern Central Tunisia.76 While the fabric described in chapter 2.2 (fabric group I) clearly differs from the two groups introduced above, also the distribution of examples within the area of former Africa Proconsularis equally supports the idea of their Central Tunisian origin. In spite of geographic diversity, some forms to be introduced next have parallels in North Tunisian production, indicating either the movement of potters within the borders of Africa Proconsularis or the importance of intra-regional trade as a source of influence. The best known African cookware lid form simultaneously mass-produced both in North and Central Tunisian workshops is the mid-Imperial Hayes 196 (variants A.II & C.II). Another form of considerable importance is a lid with a hooked rim, the Central Tunisian version of which will be

Table 9. The size and the distribution of Central Tunisian lids. C

Diameter

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

avg

max

P6

P7

P8

P9 P10 P11 CT

Total

NT

?

217

C.I

22

28/34

43

8

9/11

18

11

14/16

19

5

5

4

3

1

1

17

2

-

19

C.II

20

25/28

39

-

-

-

8

10/13

15

6

21

37

16

4

22

106

-

-

106

C.III

20

25/30

35

8

12/14

18

8

10/12

18

10

7

11

5

3

3

33

6

-

39

C.IV

21

26/30

40

6

8/11

14

8

11/13

20

3

7

14

7

3

19

45

8

-

53

41

Chapter 3

100 %

80 %

60 %

40 %

20 %

0% ø 18

20

22

24

B.I (Hayes 195)

26

28

30

32

C.I (Hayes 182)

34

36

G.I (Hayes 181)

38

40

42

H.II (Hayes 26/181)

Figure 11. The cumulative rim diameter distribution of four African cookware form variants: B.I & C.I lids, G.I pan and H.II pan-casserole. by plentiful archaeological evidence.84 Particularly indicative are the finds made at Leptiminus (Leptis Minor), where the recovery of nearly a thousand examples as surface finds lead to the discovery of a kiln site involved in the production of this form.85 Other production site evidence includes comparanda from the El Mokaïda kiln site located several kilometers west of Sullecthum and the site of Oued el Akarit located 25 kilometers north of Gabès, where Hayes forms 181 and 185 were also produced.86 Yet another production district has to have been located in Pheradi Maius, in the northernmost part of Central Tunisia.87 The abundance of the Hayes 182 in the excavations of the Italian expedition at Sabratha shows equally from which direction the production areas of the form C.I should be primarily searched for.88 Scholars have voiced an unanimous opinion about the use of the form C.I, agreeing with Hayes, who first proposed that the Hayes 182 lid was more often than not used in conjunction with the Hayes 181 pan.89 Neither does the study assemblage offer possibilities to alter this idea, as the only cooking vessel form that has been customarily both slipped and burnished – though in most cases on the interior surface only – is the Hayes 181 pan. The correlation naturally strengthens the idea of a two-piece cooking set, while the discrepancy in the cumulative rim diameter distribution of these two vessel forms (Figure 11) may partially result from the small number of Hayes 182 rims. On the other hand, there

variant C.I is equally reflected by the vessel interior, which is smoothed in over a third of the cases. The appearance of Hayes 182 (C.I) in archaeological deposits considerably earlier than the Hayes 195 (B.I) probably reflects the chronological order of these two productions. The most extreme examples in this respect are perhaps the cities of Bararus (Rougga, Central Tunisia) and Sabratha (Tripolitania), where the earliest Hayes 182 rims have been found in the 1st century AD deposits.80 Still, these finds do not seriously question the idea according to which form C.I became a common item of exportation between the late 2nd and the early 3rd century AD.81 If the published evidence – where a difference is seldom made between the two productions – is interpreted in the strictest sense, the Hayes 182 lid might have been residual already in the late 3rd century AD and was certainly residual by the first half of the 4th century AD.82 On the other hand, by stressing the similarity of the chronology to that of the Hayes 195, the heyday of the 3rd and 4th centuries AD should have been followed by a rapid decline of production in the early 5th century AD.83 Due to the limited number of examples, the study assemblage does not offer sufficient means for the examination of this question, even though it is tempting to interpret the decline in the amount of examples towards phases 10 and 11 as a decrease in the production of form C.I. Hayes' straightforward statement concerning the Central Tunisian origin of form C.I is today supported

42

Forms

is no other Central Tunisian vessel form that would match better with the Hayes 182 lid than the Hayes 181 pan. Instead of illustrating another standard example of the Hayes 182/195 series, a somewhat modified variant (Plate 5:15) comparable to the sub-variants of Hayes 195 has been catalogued. As a result of exceptionally careless turning, the hook that was produced by pressing the folded rim against the exterior surface was reduced to a slight thickening running parallel with the interior wall. Due to its malformed appearance, it is no wonder that only a single example of this sub-variant, which strongly reminds of an undated rim Lund has identified as a bowl,90 is included in the study assemblage. As mentioned above, the Central Tunisian version of the Hayes 196 lid (C.II, cf. Plate 3:3-4)91 differs from its North Tunisian parallel only in minor details. For example, due to high concentrations of calcareous material in Central Tunisian fabric, the blackened rim zone is often covered by an additional layer of whitish scum, which is the most obvious indicator distinguishing these vessels from North Tunisian products. Regarding the vessel use (Figure 12), the Central Tunisian Hayes 196 variant is more closely associated to the local Hayes 23B shallow casserole variant (J.II) than to any deep casserole variant of the same origin. Neither can the Central Tunisian fraction of the Hayes 196 be related to the Hayes 197 deep casserole, because in the study assemblage their number in Central Tunisian fabric hardly exceeds ten.92

The third common Central Tunisian lid form C.III (Plate 5:16) is characterized by a sharp protruding lip, which results from the presence of a shallow furrow on the vessel exterior adjacent to the rim. Otherwise, the shape of the thickened rim with slightly angular edges closely resembles variant C.II of the Hayes 196 series. While the examples of this form in the study assemblage present the usual range of surface finishes – turned upper exterior surface paired with unfinished lower zone and vessel interior – the vessel firing shows somewhat different pattern from most African cookware forms. A third of the examples display a reduced vessel core (cf. Plate 1g) and the number of overfired sherds, the presence of which is indicated by vitrified surfaces, is also exceptionally high. Together these features hint at short duration firing in a high temperature. The increased presence of examples, in which the blackish orlo annerito deposition covers the whole vessel surface, may equally be a symptom of diverse firing conventions. Evidence from other archaeological sites is surprisingly scarce, possibly because form C.III has been commonly assimilated with the Central Tunisian Hayes 196 lid variant. Therefore, the comparanda to be pointed out from relevant literature is limited. The distribution of form C.III in the study assemblage matches fairly well with the 4th or 5th century AD example found in the Henchir ech Choggaf kiln site near the city of Acholla,93 although the production may have even begun in the Severan period, as

100 %

80 %

60 %

40 %

20 %

0%

ø 16

18

20

22

C.II (Hayes 196)

24

26

28

30

32

J.II (standard Hayes 23B)

34

36

38

S.I (Hayes 183)

Figure 12. The cumulative rim diameter distribution of three Central Tunisian cookware form variants: C.II lid, J.II shallow casserole and S.I deep casserole. 43

Chapter 3

100 %

80 %

60 %

40 %

20 %

0% ø 15

17

C.III lid

19

21

C.IV lid

23

25

27

29

G.I (Hayes 181)

31

33

35

J.II (Hayes 23B)

37

39

S.I (Hayes 183)

Figure 13. The cumulative rim diameter distribution of five Central Tunisian cookware form variants: C.III & C.IV lids, G.I pan, J.II shallow casserole and S.I deep casserole. suggested by Aquilué Abadías on the basis of the finds from Baetulo.94 As attention is turned to the related cookpot form (Figure 13), the cooking vessel the C.III lid was most likely used with turns out to be the Central Tunisian version of the standard Hayes 23B shallow casserole. Also their distribution in the study assemblage, as indicated by Pearson's correlation coefficient (0.568), speaks in favor of this relation in comparison to e.g. Hayes 181 pan (0.136, form G). This comes hardly as a surprise, as the study of corresponding material of North Tunisian provenance produced similar results. Thus, the claim that shallow casseroles of the Hayes 23 series were predominantly used without a lid certainly requires further attention in the discussion concerning the use of this common African cookware form (cf. sections 3.4 & 3.9.3).95 The form C.IV (Plate 5:17) is one of the few examples to be regarded as a native Central Tunisian vessel form, though in rare cases the fabric is clearly North Tunisian. The form can be distinguished by its segment-shaped rim with a pointed tip that results from a production sequence consisting of folding, turning and smoothing. In general, the examples in the study assemblage form a fairly homogeneous group. The turned upper exterior surface is frequently paired with unfinished lower zone and vessel interior surface, while the completely oxidized fabric is topped with usual combination of whitish-gray scum and a blackish orlo annerito cover on the vessel rim. The occasional presence of concentric furrows on the upper

exterior surface hints that form C.IV was not customarily equipped with a knob.96 In half dozen cases, a similar furrow was found on the interior surface adjacent to the rim, where its probable function was to increase the stability of the vessel. As it seems that most Central Tunisian cookware forms were not subject to such large scale exportation as their North Tunisian counterparts, the date of the form C.IV has to be based on relatively scarce evidence. In spite of the early desultory examples found in the 2nd century AD deposits,97 one can only speak of the continuous presence of this form in the 3rd century AD. The production may extend at least into the 4th century AD, if not into the early 5th century AD.98 The number of examples belonging to phases dated after the early 4th century AD in the study assemblage, however, seems to contradict the latter possibility. Recent archaeological research carried out in Byzacena and Tripolitania has produced new evidence supporting the idea about the origin of this form. Three Central Tunisian kiln sites: the Catacombs of Sullecthum, Henchir ech Choggaf and Henchir Mbarek have yielded examples of this form.99 The form is also abundant at Uzita and Sabratha, where it is often associated to deep casserole variants S.I and S.II.100 However, by comparing the cumulative rim diameter distribution to other Central Tunisian forms present in sufficient numbers in the study assemblage (Figure 13), form C.IV appears to have been used more likely together with the Hayes 181 pan than

44

Forms

with the most frequent Central Tunisian shallow or deep casserole variants. Several explanations for the association of two lid forms (C.I & C.IV) with a single cooking vessel form may be suggested. Firstly, it remains to be examined whether the group of Hayes 181 pans is dividable into two fractions that would correspond to these lids. If this question can be answered positively, further study is needed to clarify whether such a multiplicity rises from chronological differences or from the existence of several contemporary workshops producing the same pan form with a different lid. The latter explanation must be favored, at least for the time being, because the distribution of these lids in the study assemblage is fairly even.

well-represented in contexts dating from the mid-2nd century onwards. Thereupon, the production – that may have taken place from the mid-4th century AD onwards at the El Mahrine kiln site101 – seems to have continued at least to 475-500 AD.102 The other parallels from Luni and ager Tarraconensis do not help much, because of their residual character or the insecure date of the context they were found in.103 The search for a corresponding casserole form is equally problematic, but if a similar rim diameter range is emphasized, the standard Hayes 193 deep casserole (form L) qualifies as the most potential candidate. The form D.II includes two variants of different sizes, which resemble closely one another in form characterized by an everted, horizontal rim forming a short flange with somewhat angular edges. In the examples of the smaller variant (Plate 5:19) both the total absence of soot deposition and the prevalence of carefully smoothed interior surfaces are factors speaking against the current identification. The fabric, on the other hand, is in all examples at least similar to if not identical with the North Tunisian cookware fabric. As all the examples in the study assemblage derive from the 5th/6th century AD phase 11, the form itself – for which one can easily find parallels in other cookware fabrics – seems to be fairly late. In case of African cookware, comparanda from other sites is practically non-existent, as the example dated to 525-600 AD from Carthage is in all likelihood residual.104 The second variant of the form D.II (Plate 6:20) is similar to the one introduced above, except being apparently more robust and larger in size. But in this case, the combination of turned exterior and smoothed or unfinished interior surface together with the presence of blackened orlo annerito surface on the rim of several examples, makes the identification as a lid fairly easy. At Carthage the form, of which the most examples in the study assemblage belong strangely to Central Tunisian production has been dated between the 4th century AD and AD 425-450 by Fulford; while Tomber has reached a similar conclusion with the material from the excavations near the Circus of Carthage.105 Apart from an isolated late 5th century AD example from Carthage,106 a pure 5th century AD date is supported only by the study assemblage.

3.2.4 Miscellaneous lids The last group of African cookware lid rims includes all the miscellaneous forms present in the study assemblage with less than ten examples (Table 10). Due to their rarity these finds play understandably only a minor role when the intention is to establish an overall view on the production of African cookware. On the other hand, the presence of rare forms may imply that the group includes more variability than has, perhaps, usually been thought. Due to irregularities in vessel fabric and shape it is equally possible that some of the “forms” introduced here in a random order are neither lids nor African cookware. In many cases, however, the selection of vessel surfacings together with the presence of orlo annerito deposition supports the current identification. The examples of the form D.I (Plate 5:18) in the study assemblage are characterized by a slightly angular, small protrusion forming a thickened rim on the interior surface. All the examples have unfinished surfaces and a completely oxidized fabric, while the only somewhat unusual feature is the distinctive brownish tinge of the orlo annerito deposition. Its presence, however, is only one of the arguments supporting the identification of this form as a lid, as in half of the examples the rim is covered by patches of soot and vessel surfaces are predominantly unfinished. As often, the dating is based on scarce evidence. While the earliest occurrence of form D.I is reported in the late 1st century AD deposit at Carthage, it is only

Table 10. The size and the distribution of miscellaneous lids. Diameter

D

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

avg

max

P6

P7

P8

P9 P10 P11 CT

Total

NT

?

16

D.I

21

22

24

-

6/7

-

7

8

9

1

-

1

2

-

-

-

4

-

4

D.II

21

26/30

37

-

13

-

7

8/10

12

-

-

-

-

-

8

4

4

-

8

D.III

-

17

-

-

-

-

-

6

-

-

-

-

-

-

1

-

1

-

1

D.IV

-

22/30

-

-

-

-

2

-

-

1

-

-

1

2

-

3

9/10

45

Chapter 3 Table 11. The size and the distribution of lid knobs Diameter

E

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

avg

max

P6

P7

P8

Fabric

P9 P10 P11 CT

Total

NT

?

8

E.I

-

2

-

-

21

-

-

23

-

-

-

1

-

-

-

-

1

-

1

E.II

2

2/3

5

5

9/12

16

19

22

23

1

2

2

1

-

1

3

4

-

7

The identification of previously unpublished lid D.III (Plate 6:21), in which the convex wall of a small open form ends in a thinned rim with rounded lip, is based on the fabric resembling more closely African cookware than African Red Slip or African utilitarian ware. The form in question is relatively small and characterized by a domed, strongly convex body turning to a thinned rim with a rounded lip. In the total absence of comparanda, the unique find from phase 11 dates the form to the 5th century AD. The peculiar diagonal rim of the lid form D.IV (Plate 6:22) was produced as the excess clay was turned off from the exterior surface. Although the group of three examples in the study assemblage is fairly homogeneous, especially with respect to surfacing and firing, the form itself could possibly be interpreted as a badly cut version of some better known lid form, most likely either of the Hayes 182 or the Hayes 195.

The method has already been heavily criticized by Aquilué Abadías, who stresses the importance of rims rather than knobs in the classification of African cookware lids.111 This does not imply, however, that knobs and other potential handles would be completely useless in the study of African cookware. Notwithstanding one complete example preserved in the study assemblage, it is uncertain whether the rare hollowed knob variant E.I (Plate 6:23) should be merged with African cookware. The most important feature supporting positive identification is a hole pierced through the knob top. Similar holes, which were obviously designed to reduce the risk of overboiling, are often found among West-Central Italian cookware lids. On the other hand, a slipped exterior surface that does not show a trace of soot is highly unusual combination for cookware vessels, just like the exceptionally fine fabric of this example. Thus, it is not possible to point out the cooking-vessel form this variant was designed to go with. The number of disk knobs (variant E.II, Plate 6:24-25) in the study assemblage is ridiculously low in contrast to the quantity of preserved lid rims. Postdepositional conditions that would favor the preservation of lid rims probably should not be held accountable for this difference, because the material includes five completely preserved disk knobs, of which two examples have been catalogued to indicate the internal variability of the group. The small size of a disk knob offers perhaps a better explanation, as these round objects are certainly not so easily distinguished from gravel and dirt as the larger and more angular rim sherds. The third and the most probable explanation for the lack of lid knobs is that these vessels were not usually equipped with a knob (see section 3.9.1). Due to absolute rarity of disk knobs, they must be dated on the basis of few parallels. Aguarod Otal grouped all the lids with a disk knob under a single entry, which is dated from the late 1st century AD to the first half of the 2nd century AD.112 If this dating is accepted, all the examples in the study assemblage – in spite of their even chronological distribution – are very likely residual. However, an example found in Carthage113 shows that the convention of equipping lids with a disk knob was picked up in the late 5th/early 6th century AD, when the so-called Late Cookwares were chiefly directed to intra-regional trade instead of being exported on a large scale.

3.2.5 Knobs In spite of the fact that lids are virtually the most abundant kind of form in African cookware, indicated by nearly one thousand examples in the study assemblage, attention has seldom been paid to other vessel parts than rims. The rarity of disk- or buttonshaped knobs (Table 11), especially the ones that can be associated with a certain rim form, explains partially this custom.107 While many lid forms were equipped with a shallow ring-shaped protrusion instead of a knob; other forms do not seem to have had a knob or other handle device at all. The explanation for the absence of lid knobs is obvious. The addition of a knob would not have only slowed down the production sequence, but in transportation these lids would have also been more difficult to stack and thus easily broken. Hence, it is hardly surprising that knobs are characteristic of the African cookware lid production only in the 1st and 2nd century AD,108 whereas the ring-shaped protrusion i.e. ring-handle or grip is the dominant feature of the two subsequent centuries. Bearing this in mind, it is vexatious that the presence of a solid knob is sometimes regarded as the only unambiguous feature identifying a particular rim form as a lid.109 Recently, the classification of early African cookware lids has been revised by Aguarod Otal, who used the presence and absence of knobs and other handle devices as the principal criteria of classification.110

46

Forms

3.2.6 Ring handles Ring handles have long formed one of the most enigmatic groups of African cookware, even though their association with certain lid rims was recognized a long time ago. The discussion in section 3.9.1 includes arguments identifying the rims with which these diagnostic fragments join as lids; thus the appropriate term to be used here is a “handle”, in spite of the fact that ring handles may have not significantly improved the handling of these vessels. The aim of the following brief presentation is to introduce the variability included in this group. Previously, the topic has only been examined by Aguarod Otal, who proposes that it more likely results from the output of several contemporary potters or workshops rather than from chronological differences.114 The study assemblage supports this view, while it will be shown that much of the variability in this group arises from minor differences, such as the finish given to the top of the vessel. In spite of the introduction of a new classification system based on the occurrence of diverse handle types, Aguarod Otal was forced to admit that in many cases the rim is the most positively identifiable vessel part; and that handles are extremely difficult to classify without well-established links to respective rim forms.115 Nevertheless, thanks to recently published evidence, lid rims can occasionally be paired with handles. For example, Cabrera III and Trincere shipwrecks suggest almost exclusive association of ring handles with lid rims of the Hayes 196 series.116 They are also good examples of ring handles that are illustrated in publications only if they form a part of a complete lid profile, because while complete lids are easy to date fragmentary ring handles are not. For this reason, parallels for the ring handle variants presented below were sought only incidentally. The examination of the study assemblage showed that ring handles form a homogeneous group, in which North Tunisian examples clearly outnumber Central Tunisian ones (Table 12). Both productions are also evenly distributed in the study material instead of being associated with a certain form, variant or chronological period. Neither the vessel dimensions nor the firing show much variance, as virtually all the examples were fired in an oxidizing atmosphere and

the group includes only two occurrences of a patina cenerognola surface. The complete absence of soot from the exterior surface is another characteristic feature of this vessel part. Ring handles do not either show a trace of use-wear that would normally be found at least on the edge of the protrusion, if these vessels were plates or bowls (cf. section 3.9.1). Three variants – diagonal, angular and rounded ring handle – have been distinguished on the basis of differences in the shape of a protrusion. In addition, an example of an incomplete ring handle showing a protrusion that has not been turned out of the exterior surface,117 will also be introduced. Generally speaking, diagonal ring handles show often turned exteriors paired with unfinished interior surfaces, whereas the number smoothed or even slipped vessel surfaces is considerably high in the remaining two groups. Because the variation cannot be shown to result from chronological differences, it may actually stand for the output of several contemporary potters or workshops. The longevity of ring handles is indicated by their chronological distribution, which extends from the scattered examples in the 1st century AD to the abundance of the 3rd and 4th centuries AD and finally to more infrequent occurrences in the early 5th century AD.118 The first ring handle variant F.I (Plate 6:26) represents somewhat more careless production compared to the other two variants. As the ring handle was formed on the upper exterior surface, the protrusion was not only turned out from the vessel body, but also flattened from the top to form a diagonal lip tilted towards the center. For this reason, protrusions of the group F.I are generally more shallow and perhaps aesthetically less pleasant than in the two other groups. None the less, their even distribution in the study assemblage shows that such variation is insignificant, as the variant cannot be associated with any particular African cookware lid rim or chronological period. The ring handle variant F.II (Plate 6:27-28) with an angular protrusion belongs to somewhat more careful production, as the handle has been patiently turned from both sides to form a rectangular protrusion with pronounced edges. Still, neither the chronology nor the association with lid rims seem to differ from the other variants. The sole peculiarity is

Table 12. The size and the distribution of ring handles Diameter

F

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

124

F.I

5

7/9

11

1

3/4

6

4

6/8

12

10

7

9

9

2

4

3

38

-

41

F.II

5

7/8

11

2

4/5

8

5

6/7

8

17

13

15

5

4

3

4

53

-

57

F.III

6

7/8

11

2

4/5

6

3

5/6

14

5

5

7

5

1

2

2

23

-

25

F.IV

-

8

-

-

2

-

-

-

-

-

-

-

-

-

1

-

1

-

1

47

Chapter 3

the latter catalogued example with a possible inscription (graffito) consisting of a single letter scratched inside the handle after firing. As indicated in section 5.2.2, inscriptions and decoration are both extremely rare in African cookware. The production of the variant F.III (Plate 6:29) included an additional stage in the forming process, as the edges of a protrusion were smoothed after turning. However, as it does not seem likely that this additional investment in the forming sequence improved vessel performance characteristics, for example by offering a better grip, it was probably an

intentional act designed to improve the attractiveness of the products. The increased number of examples with smoothed or even slipped surfaces in this group may also reflect similar endeavors.119 The chronology and the associated forms of this ring handle are probably the same with the other variants. An example of unfinished ring handle F.IV (Plate 6:30) with flat, slightly hollowed lid top is unique in the study assemblage. It has been inserted here to indicate the final process used by North African potters to form lids, which involved the turning of a ring handle.

3.3 SHALLOW PANS AND PAN-CASSEROLES the early 2nd century AD.121 The influence of IRSC production is clearly seen in the shape of this wide, flat-based pan, in which the height of convex upper wall rarely exceeds 5 cm. The vessel wall with its strongly inward inclined upper section is, in fact, one of the features besides fabric that distinguishes this form from the Hayes 26/181 pan-casserole. The surfacing of the Hayes 181 pan is very distinctive. The exterior surface has nearly always been smoothed, while the interior surface regularly displays one of the following finishes or a combination of them: smoothing, burnishing or slipping. The burnishing of slipped or smoothed interior surface has produced lustrous stripes, which are found in over two thirds of the material. These differences in surfacing have been utilized to distinguish two Hayes 181 pan variants: slipped and stripe-burnished. This classification will not be used in the present study, mainly because these surfacings are frequently encountered both in Central and North Tunisian production. However, attention should be paid to the observation, according to which the rim profile of the Hayes 181 pans straightened in the course of production.122 It reveals that the study material, on which the preceding conclusions are based, has also included examples of the Hayes 26/181 pan-casserole. The burnished stripes on these vessels were once enough for the introduction of stripe-burnished ware (ceramica cucina polita a bande, cf. section 1.2.1) as a new sub-group of African cookware. Today, the use of this definition can reasonably be abandoned, because the group it seeks to describe is extremely heterogeneous by nature. Stripe-burnishing is not tightly associated with any particular production or vessel form. Additional factors underlining the similarity of the Hayes 181 pan to the rest of African cookware products are firing in a fully oxidizing atmosphere and regular presence of patina cenerognola deposition on the exterior surface. The Central Tunisian origin of this form is often indicated by a

This class includes two open pan-like vessel forms characterized by a wide body with a shallow convex or slightly outward inclined wall ending in an even or thinned rim with a rounded lip. In contrast to more restricted shapes, the forming of these vessels had to be based on the use of slowly rotating potter's wheel (see section 4.1), which is a precondition for the production of all wide dish-like forms. Moreover, the clay first centered and flattened on the wheel-head must consecutively be pulled outwards, while in the shaping of other cookware forms, such as deep casseroles, the clay is principally pulled upwards. But just as with other forms, at the leather hard stage the vessel was inverted and re-attached to the wheel top for turning. In many cases this produced closely spaced turning marks on the lower exterior wall. More often than not the turning of the vessel was followed by slipping or burnishing of either surface. Although the form itself has been long recognized as a pan, its typology, use, as well as various morphological aspects are not particularly well known. Even the existence of two forms – the Hayes 181 pan (form G) and the Hayes 26/181 pancasserole (form H) – was not fully realized until recently in spite of apparent differences both in their origin and in the production sequence.120 The Hayes 181 is a flat-based pan of Central Tunisian origin, the exportation of which can be dated between the late 2nd and early 4th century AD. The three Hayes 26/181 pan-casserole variants, on the other hand, are North Tunisian products that were exported between the end of the 3rd and the early 5th century AD. The principal aim of the following discussion is to provide a firm basis for their identification. 3.3.1 Hayes 181 pans The Hayes 181 pan (form G, Plate 7:31-32) has rightly been regarded as a late descendant of Internal Red Slip cookware (IRSC, a.k.a. Pompeian red ware) pan produced in several regions around the Mediterranean from the late 3rd century BC at least to

48

Forms Table 13. The size and the distribution of the Hayes 181 pans. G

Diameter

Variant min G

19

avg 24/29

Rim height

max min 31

-

avg -

Rim width

max min -

5

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

77

6/7

8

18

14

29

7

12

4

77

layer of whitish scum topping the blackened surface, although examples showing either of these covers are also fairly abundant. The chronology of the Hayes 181 pan has been related to the chronology of preceding Internal Red Slip cookware pans. Although the form was already produced in Central Tunisia during the 1st century BC,123 the excavators of Terme del Nuotatore (Ostia) have suggested that the Hayes 181, which occupied the place of the IRSC pan variant Goudineau 28 in the Roman batterie de cuisine, was introduced outside Roman Africa sometime between AD 90-120.124 Although its factual predecessor, namely the form CVI:1 of Atlante I, was probably among the first African cookware forms exported in the 1st century AD,125 the regular distribution of African pans seems to have begun only in the late 2nd/early 3rd century AD.126 The rapidity in which the Hayes 181 took over the western Mediterranean markets is indicated by the late Antonine deposits of Terme del Nuotatore, where African production makes up ca. 65% of shallow pans.127 However, neither the distribution of African pans nor the distribution of African cookware in general was geographically homogeneous. For example, at Otranto the Hayes 181 pan has been found only in contexts dated after 330-350 AD.128 The closing date for the production is substantially hard to fix, because many of the late examples published to date belong in fact to the Hayes 26/181 pan-casserole series. For this reason, the end of the production has often been dated to the early 5th century AD,129 but the drastic drop in the distribution of this form between phases 9-11 in the study assemblage (Table 13) does not give support to such interpretation. Rather it suggests, just as the evidence from ager Tarraconensis,130 that at least the production of exported pan variants ceased in Central Tunisia sometime around the mid-4th century AD. The Central Tunisian origin of this form was already suggested by Hayes,131 who introduced all the forms of this provenance under numbers 181-185 in his classification. Today, his idea is supported by a large body of evidence. Firstly, a kiln site involved in the production of this form has been recently excavated at Leptiminus (Leptis Minor),132 but examples of early Hayes 181 variants have also been found in abundance in Tripolitania.133 By the late 2nd/early 3rd century AD the production even seems to have reached such remote places as Kasserine (ancient Cillium) in south-western Central Tunisia, or at least the survey of the city and its surroundings has

1

8

61

been reported to yield local imitations (?) of this form.134 The form is also abundant in the 3rd century AD deposits at Sabratha.135 Therefore, it is puzzling how some scholars have been able to locate the production of Hayes 181 pans in Northern Tunisia, particularly in the surroundings of Carthage,136 unless they actually referred to the Hayes 26/181 pancasseroles. 3.3.2 Hayes 26/181 pan-casseroles Because the group of Hayes 26/181 pan-casseroles was long considered only a late variant of the Hayes 181 pan, its characteristics are still poorly known.137 While a fragmentary example of this pan-casserole was already published in Late Roman Pottery as African Red Slip ware form 26, its close resemblance to the Hayes 181 shallow pan was realized only later.138 Lately, the definition “Hayes 26” has been used to describe the North Tunisian parallel of Hayes 181.139 In the following pages the close relationship of the two forms will be highlighted by defining the examples of North Tunisian production as the Hayes 26/181 pan-casserole. As indicated by the definition “tarda versione con pareti rettilinee della forma Lamboglia 9A=Hayes 181”,140 the two productions may first appear strikingly similar. Still, the study assemblage includes abundant examples of a shallow pan, the shape of which does not meet the description of the Hayes 181. The most obvious difference is in the shape of the upper wall, which, in the case of the Hayes 26/181 pan-casserole, tends to thicken and straighten in the course of production. These changes were probably necessary both to increase the vessel durability and to facilitate their stacking both in kiln and ship's cargo hold. The Hayes 26/181 is usually also more robust and larger in size than the Hayes 181. Moreover, these two pan forms may have also been used differently from one another, at least this is indicated by numerous examples of the Hayes 26/181 with rather sagging than flat base. As sagging base implies that in use some of these vessels were placed like a casserole on a cooking stand set over a fire, pancasserole is actually more accurate definition than a pan.141 Still, it is generally believed – perhaps rightly due to the absence of a handle device – that the form was used in an oven, just like the Hayes 181 pan.142 It is evident that the existence of two pan forms in two fabrics cannot reflect the changes taking place in the production of some workshops, but a shift of

49

Chapter 3

the whole production from Central to North Tunisia. The transition probably took place either simultaneously with or somewhat earlier than the beginning of the African Red Slip ware D production. At least, such pattern is suggested by cookwares accompanying both the production of African Red Slip ware A and C.143 The rapport between the two productions during the period of transition is indicated by the late 3rd century AD examples published from Ebusus,144 which are not difficult to interpret as North Tunisian derivatives of the Hayes 181 pan. Chronological succession is further evidenced by many archaeological sites, where the distribution of the Hayes 26/181 pan-casserole is almost exclusively limited to late 4th and early 5th century AD contexts.145 Here, the group of Hayes 26/181 pan-casseroles (Table 14) has been divided into small (H.I), medium (H.II) and large (H.III) variant to emphasize the differences between the Central and North Tunisian production. Although these variants may, at first glance, seem to form an inseparable entity, it will be shown that the variation in vessel form and size can be chronologically significant. There may also be some variation in the fabric, as Neuru has identified two fabric variants: a) standard African fabric, medium red in color and b) darker, brownish-red fabric with fine inclusions.146 It is equally possible, however, that these definitions refer to North and Central Tunisian fabrics, rather than to two North Tunisian fabric variants. As the Hayes 26/181 pan-casserole belongs to the last phase of African cookware production, its distribution within the western Mediterranean is markedly more limited than the distribution of Hayes 181 pan.147 The small Hayes 26/181 (H.I, Plate 7:33) is by far the rarest of the three shallow pan-casserole variants. One potential explanation for the rarity is that due to similar rim diameter these sherds are easily misidentified as Hayes 23A shallow casserole. However, the upper wall of the small Hayes 26/181 is relatively straight and ends to a slightly thinned rim with a pointed lip, whereas the Hayes 23A shallow casserole is characterized by a convex upper wall turning to a slightly thickened rim with rounded lip. The small vessel size of the variant H.I is indicated by the catalogued example with only 2.6 cm high upper wall.

The surfacing of the small pan-casserole does not differ significantly from the other variants in this group, but is different from many other African cookware forms, as both vessel surfaces have customarily been smoothed. In many cases the careful smoothing of the vessel interior has also produced a self-slip, which is difficult to distinguish from a proper slip. Moreover, in nearly half of the examples the interior surface is covered with burnished stripes. All the vessels have been fired in a fully oxidizing atmosphere, and many carry also a band of patina cenerognola deposition on the upper exterior surface (Plate 2h). One of the characteristics of the whole Hayes 26/181 series is that this layer is more often brown or blackish-brown than black in color. Previously, the small Hayes 26/181 has only been discussed by Aguarod Otal, who identified an example (labeled Ostia IV.1) from Tarracona as a Hayes 181 pan variant and dated it to the second quarter of the 5th century AD.148 In the study assemblage, however, the distribution of the small pan-casserole is almost exclusively limited to the early 4th century AD contexts. As the excavations of the German expedition at Carthage have yielded comparanda from the late 4th century AD context, the most plausible date to be suggested for the small Hayes 26/181 pan-casserole ranges from the 4th to the first half of the 5th century AD. Evidence on the exact location of production sites of shallow pan-casseroles has been gained through chemical analyses, the results of which point at the El Mahrine kiln site in the River Medjerda valley.149 The medium-sized Hayes 26/181 pan-casserole (H.II, Plate 7:34-5) is the most abundant variant in the group. The variant is characterized by a straight, vertical or slightly outward inclined upper wall (height 4-5 cm) ending in an even or thinned rim with a pointed or rounded lip. Due to the similar appearance and size the medium-sized pan-casseroles have customarily been identified as Hayes 181 pans. The unfortunate convention was started by Lamboglia, whose pioneering articles on terra sigillata chiara (African Red Slip ware) A and D introduced the variant, characterized by a slipped interior surface, as form 9A.150 The mistaken identification is understandable, however, since the form of the earliest examples of the Hayes 26/181 pan-casserole, which date back to the late 3rd/early 4th century AD, seem to

Table 14. The size and the distribution of the Hayes 26/181 pan-casseroles. H

Diameter

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

171

H.I

18

19/26

28

-

-

-

4

5/6

7

3

4

6

1

-

-

1

13

-

14

H.II

20

26/31

40

-

-

-

5

6/7

7

14

19

30

15

12

11

4

97

-

101

H.III

25

29/37

46

-

-

-

6

7/8

10

-

3

7

2

4

40

8

48

-

56

50

Forms cease until the late 5th/early 6th century AD.154 Unlike the two other variants, the large Hayes 26/181 pan-casserole (H.III, Plate 7:36) can hardly be identified as any other form due to its distinctive, slightly outward inclined upper wall, which is not only tall (up to 6.5 cm) but also equipped with a thick rim. The shape of the bottom varies from flat to slightly sagging, although flat examples clearly outnumber the sagging ones. The size of the large pan-casseroles, however, did not diminish the attention that potters paid to the surfacing of these vessels. Every example in the study assemblage has a smoothed exterior surface and a half also show traces of stripe-burnishing. Although in some cases the interior surface has been treated similarly, its most common surfacing is smoothing combined with occasional patches of self-slip. By contrast, slipped vessel interiors do not seem to occur in this group frequently.155 As usual, all the vessels were fired in a fully oxidizing atmosphere and show a band of patina cenerognola deposition, often brown or brownish-black in color, on the exterior surface underneath the rim. In terms of chronology, the large Hayes 26/181 pan-casserole represents an endpoint in a typological sequence that had begun with the production of Internal Red Slip cookware pans and included a period of prosperity in the 2nd and 3rd centuries AD in the form of the manufacture of Central Tunisian Hayes 181 pans. Hence, most scholars are willing to date the variant H.III from the late 4th to the early 5th century AD,156 although examples of large pan-casserole are also reported to have been found in contexts dating between the mid-5th and the mid-6th century AD.157 A late dating is likewise suggested by the study assemblage with over two thirds of the examples clustered in the 5th/6th century AD phase 11.

imitate Central Tunisian Hayes 181 pans. For example, both the Hayes 181 pan and the Hayes 26/181 pan-casserole show frequently one to three concentric grooves at the junction of the base and the wall on the interior surface. When it comes to surfacing, it has been rightly observed that a wash or a slip coating the interior of the medium Hayes 26/181 is occasionally topped by burnished stripes.151 This surfacing was detected in one fourth of the examples, whereas less than a quarter showed a slipped interior surface. In many cases, however, the latter surfacing has more likely resulted from efficient smoothing rather than from the application of slip. Besides, vessel interior smoothing is also the prevalent finish of exterior surfaces, one third of which are also covered with burnished bands. All the examples were fired in a fully oxidizing atmosphere, although a wide band of patina cenerognola deposition, usually black or brown in color, is frequently found below the rim on the upper exterior surface. The date of the medium-sized Hayes 26/181 pancasserole should not significantly differ from the other variants in this group, though as a direct successor to the Hayes 181 pan, the first examples could at least theoretically precede the two other variants. This possibility is indicated by a set of late 3rd century AD examples published as Lamboglia 9A (Hayes 181) pans, both the form and the fabric of which identify them as pan-casseroles.152 Still, the presence of medium-sized Hayes 26/181 pan-casserole in all chronological horizons of the Palatine East excavations dated between the late 3rd and early 5th century AD, dates the main production period of the variant H.II back to the 4th and the early 5th century AD.153 Its strong presence in phase 11 (AD 400/425-525/550) even supports a proposition, according to which the production did not

3.4 SHALLOW CASSEROLES The group of shallow casseroles includes two forms, the Hayes 23A and the Hayes 23B, which share the same basic shape characterized by a shallow convex upper wall turning to a thickened rim with a rounded lip. The rim of the Hayes 23A joins smoothly with the wall, whereas in the Hayes 23B the rim has not only been folded over and pressed against the interior surface, but also separated from the wall with a cut. The junction of the upper wall and the base is always angular and frequently takes the form of a protrusion with angular edges on the exterior surface. It is possible that the protrusion marks the joint of two clay rings, at least carinated vessels are frequently made in several parts.158 The protrusion is also believed to have facilitated the placement of these

vessels on a cooking stand. Both shallow casserole forms were normally equipped with a sagging base, which frequently bears concentric corrugations on the exterior surface (Plate 2e). Experimental archaeology has recently shown that a surfacing of this kind was intended to increase the thermal shock resistance of cooking vessels rather than to improve their thermal conductivity.159 Both shallow casserole forms were subjected to large-scale exportation throughout the western Mediterranean. The Hayes 23B shallow casserole is, in fact, one of three forms, in addition to the Hayes 196 lid and the Hayes 197 deep casserole, present virtually on every archaeological site where African cookware has been found. 51

Chapter 3 Table 15. The size and the distribution of the Hayes 23A shallow casseroles. I

Diameter

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

37

I.I

16

19/22

26

-

-

-

4

6/7

8

8

3

9

1

1

5

4

19

3

26

I.II

19

20/22

23

-

-

-

-

6/7

-

1

-

-

1

-

2

1

3

-

4

I.III

21

26/27

29

-

-

-

-

6/7

-

1

-

1

1

-

4

2

5

-

7

3.4.1 Hayes 23A shallow casseroles The first form to be discussed here is the Hayes 23A shallow casserole,160 of which the following pages will introduce three variants (see Table 15). While variant I.I is defined as the "standard" Hayes 23A shallow casserole, the other two are somewhat more unusual examples. The common denominator for them all is the lack of rim with pronounced thickening on the interior surface, which is a highly characteristic feature of the chronologically successive Hayes 23B shallow casserole.161 In terms of chronology, the Hayes 23A shallow casserole is almost certainly a residual element in the study assemblage – in spite of nearly forty examples – as it is one of the earliest forms ever produced in African cookware. The abundant examples in the Flavian deposits at Ostia, as well as the late 1st century AD finds from Aix-en-Provence demonstrate its rapid spread throughout the western Mediterranean.162 The evidence from Celsa and Cesaraugusta, however, indicates that the production had already begun during the Claudian age.163 Still, the examples of the 1st century AD represent a minority as the main body of evidence comes from contexts dated to the early 2nd century AD.164 The closing date of the Hayes 23A shallow casserole is a problematic issue, because the earliest term suggested, the late 2nd century AD, has also been promulgated as a date of introduction.165 The Hayes 23A has been found in abundance in the early3rd century AD contexts at Porto Torres,166 while the finds from Baetulo, Tarragona, Valencia and Uzita suggest that it became residual only after the mid-3rd century AD.167 It is also indicative that at Luni the form has been found only in the pre-4th century AD contexts.168 The terminus ante quem of the late 3rd century AD is confirmed by both the excavations of the Italian expedition at Sabratha, where the ratio of Hayes 23 A and B forms was 1:5, and the excavations of Ebusus, where the ratio was reduced to 0:18.169 The 1:9 ratio in the study assemblage does not leave much room for alternative suggestions. Thus, the examples of Hayes 23A found in the late 4th and early 5th century AD contexts at Carthage and Ostia are certainly residual. 170 The typology of Hayes 23A is also poorly known, although a shallow and convex upper wall may be a feature distinguishing some early variants.171

The current knowledge on the location of production sites of Hayes 23A as well as the use of shallow casseroles in general is also inadequate. The idea that the production of Hayes 23A was centered in northern Tunisia, with Carthage as the principal port of distribution,172 is supported by the study assemblage, where the examples in North Tunisian fabric clearly outnumber the Central Tunisian ones. The standard Hayes 23A shallow casserole (I.I, Plate 8:37) is a medium-sized cooking vessel with slightly convex, shallow upper wall (height 3-4 cm) ending in a thickened rim with a rounded lip. In most of the examples in the study assemblage both surfaces were smoothed, often efficiently enough to produce patches of self-slip on the vessel interior, while the use of slip is surprisingly rare. Half of the examples have stripe-burnished exterior surface. The firing of the standard Hayes 23A took always place in a fully oxidizing atmosphere, but a band of blackish patina cenerognola surface only rarely covers the upper exterior wall. The absence of this feature is likely another symptom of the early date, because patina cenerognola seems to become more common towards the end of production. The study assemblage includes only rare examples of Hayes 23A shallow casserole variant I.II (Plate 8:38), characterized by a fairly straight upper wall with a thinned rim and pointed lip. The evidence from Ostia and Carthage shows that it was produced in the late 1st/early 2nd century AD.173 The finds from the Guadalquivir valley (Spain) have led to a suggestion, according to which the variant should actually be regarded as a predecessor of the standard Hayes 23A and the Hayes 23B shallow casseroles.174 The identification of the peculiar and uncommon variant I.III (Plate 8:39) as a Hayes 23A shallow casserole is not secure, because none of the examples preserve the junction of the upper wall and base. Thus, the variant could be equally interpreted as a lid, unless several factors supported the former identification. Firstly, both the vessel diameter and the shape of the upper exterior wall with thinned rim and pointed lip are very similar to the latest secure Hayes 23A variants.175 The dominance of smoothed surfaces with some stripe-burnished examples is also highly unusual for North Tunisian lids. The shallow corrugations covering the exterior surface in many cases, however, remind of the Late Cookware lid

52

Forms vessel of the Severan age.183 Generally, the end of the production has been dated, alike many other African cookware forms, to the late 4th/early 5th century AD.184 Only Aquilué Abadías has proposed a slightly later date, the second half of the 5th century AD.185 The starting hypothesis, according to which the heterogeneity observed in the rim form of the Hayes 23B shallow casseroles results from chronological differences, was proven partially wrong, as most of the variability included in the group showed itself insignificant regarding the form typology.186 Still, the earliest variant of Hayes 23B shallow casserole dated to the mid-2nd century AD, can be distinguished from the rest of the material by its relatively short, thickened rim and unarticulated transition between the rim and the vessel body.187 This idea finds support in the study assemblage, where most examples sharing these characteristics (variant J.I) date to the late 3rd / early 4th century AD. Another serious attempt to use the variation in rim shape to refine the typology of African shallow casseroles has been made by Villedieu,188 whose classification overlaps to a certain extent with the one to be introduced next. Contrary, the study assemblage does not lend support to a proposition, according to which the Hayes 23B shallow casserole occurs in two size classes (17-22 and 22-34 cm)189 and it is equally dubious whether such disparity could reflect functional or chronological differences. In spite of these minor differences in interpretation, it is safe to deduce that the earliest variant is characterized by a short, thickened and somewhat rounded rim, whereas the vessels belonging to the 3rd or 4th century AD production tend to have a tall and well-articulated rim. Thereafter, the rim of the Hayes 23B becomes more elongated, and the transition between the rim and the vessel wall gets less pronounced,190 while the rim itself frequently bears one to three shallow furrows (variants J.IV and J.V) on the interior surface. Their origin likely derives from the custom to press the folded clay against the interior surface while the potter's wheel was in motion. Several factors, including reduced time, production costs as well as enhanced vessel strength may have favored this typological change, but for the

variant A.IV, especially when the examples in the study assemblage are concentrated in the 5th/6th century AD phase 11. Thus, if the variant is not a lid, it is certainly one of the latest shallow casseroles produced in African cookware. 3.4.2 Hayes 23B shallow casseroles In addition to somewhat larger size,176 the Hayes 23B shallow casserole differs from its predecessor only in details. The form is still characterized by a protruding carination separating a shallow, outward inclined upper wall from sagging base with corrugated exterior surface. The vessel rim, however, has been formed by folding the uppermost section of clay against the interior surface. In all variants but J.III, an articulated rim was produced by separating it from the vessel body with a notch, often taking the form of a shallow dent between the two vessel parts. Due to this rather complicated forming sequence, which was modified several times in the course of production, the examples of the Hayes 23B shallow casserole form a fairly heterogeneous group. The following discussion presents an attempt to utilize the heterogeneousness to refine the typology of this form (Table 16). The introduction date of the Hayes 23B has been greatly disputed, as the evidence ranges from the JulioClaudian or early Flavian age to the late 1st/early 2nd century AD and even to the mid-2nd century AD.177 The last date corresponds to the results of two archaeological expeditions at Carthage, where the transition from the Hayes 23A to the Hayes 23B shallow casserole seems to have begun during the first half of the 2nd century AD.178 A further confirmation is the dominance of the Hayes 23B in the late 2nd/mid-3rd century AD deposit unearthed on the south side of the circular harbor at Carthage.179 Although additional evidence may still be pointed out,180 archaeological data can equally be used to date the event to the late 2nd century AD.181 Still, the heyday of the Hayes 23B shallow casserole is unanimously dated to the 3rd and early 4th centuries AD.182 The only discordant note has been played by Hayes, who placed the end of the production to the early 3rd century AD, though he also noted that the Hayes 23B was the leading African cookware

Table 16. The size and the distribution of the Hayes 23B shallow casseroles. J

Diameter

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

350

J.I

18

22/26

34

5

8/10

12

6

6/8

9

49

17

20

4

2

1

3

90

-

93

J.II

20

24/27

37

9

13/16

20

6

8/9

13

25

46

55

33

6

22

72

115

-

187

J.III

23

26/28

31

12

15/17

20

6

8

9

3

6

4

1

1

1

7

9

-

16

J.IV

22

25/27

31

14

16/17

19

7

8/9

10

4

15

15

7

1

7

29

20

-

49

J.V

26

27

28

18

20

23

-

8/9

-

-

2

3

-

-

-

4

1

-

5

53

Chapter 3

time being the development cannot be explained unequivocally. Another production related detail, which has been brought up, is that these vases were slipped with a brush instead of dipping them into a diluted suspension of clay.191 Evidence for production sites of the Hayes 23B shallow casserole is scarce, although Pheradi Maius has been proposed as one possible production district.192 Tortorella places the production in North Tunisia, especially the region surrounding the city of Carthage.193 Accordingly, the early Hayes 23B example published by Taylor and Robinson, which clearly belongs to the earliest phase of the Hayes 23B production, has been attributed to the El Mahrine kiln site located in the River Medjerda valley in North Tunisia.194 However, following the pattern of other forms of African cookware, the production of shallow casseroles shifted from northern to central Africa Proconsularis sometime during the early 3rd century AD. The assumption is supported by the distribution pattern of late Hayes 23B variants, which are virtually absent from Carthage, but occur in abundance in Mauretania Caesariensis and other parts of the Western Mediterranean.195 The sites producing cookwares for exportation in Central Tunisia were presumably located near the coast, as their products do not occur too often in the midst of Central Tunisia. In the first pottery report of the Kasserine survey Neuru actually noted the absence of this form, and added only later on that the Hayes 23B was imported from the region of Carthage during the 1st and the 2nd centuries AD.196 My short visits to the archaeological sites of Sufetula and Cillium in February 1998, on the other hand, produced evidenceof local shallow casserole production characterized by a heavily thickened rim with a wide hemispherical protrusion, of which no examples are included in the study assemblage.197 The Hayes 23B shallow casserole is usually thought to have been used with the Hayes 196 lid variants,198 and a comparison of cumulative rim diameter distributions in the study assemblage confirms this idea (see the discussion on the lid variant A.II). The earliest Hayes 23B shallow casserole variant J.I (Plate 8:40), the production of which took exclusively place in North Tunisia, is slightly smaller than the other variants in the group. The upper wall, for example, is only 4.0-5.5 cm high,199 whereas in later variants it commonly measures 5.5-8.0 cm. Although the small size is not necessarily reflected by the rim diameter, the reduced dimensions of the rim itself facilitate the identification. With respect to vessel surfacing, the examples of the variant J.I in the study assemblage form a fairly homogeneous group. The exterior surface was usually smoothed, sometimes efficiently enough to produce a self-slip, the combination of which is also the dominant finish on

the interior surface, while only a few examples were actually slipped. Another common feature, characterizing nearly half of the examples, is the presence of burnished stripes on the exterior surface. Although three quarters of the material show a band of patina cenerognola, generally 1.5-3.0 cm in height and from brown to black in color, on the exterior surface underside the rim, all the examples were fired in a fully oxidizing atmosphere. Soot deposition is substantially rare, and its distribution is limited to the vessel exterior suggesting a use over a fire. Evidence on the use of the Hayes 23B shallow casserole with a lid is provided by a third of the diagnostic sherds, in which use-wear is limited to the tip of the rim. The early date of the variant J.I200 is confirmed by the early Imperial deposits at the Palatine East excavations, where it is practically the only variant of the Hayes 23B series present. Elsewhere, the earliest secure example of the variant J.I has been found at the excavations of a Roman villa at Roselle in a deposit dated to AD 130-150.201 Other contemporary examples were recovered from deposit E49 at Baetulo, whereas the introduction date of the mid-2nd century AD is also suggested by the excavations of la calle del Templo (Can Boada), with examples found in a deposit dated to the third quarter of the 2nd century AD.202 As usual, the end of production is difficult to date, mainly because the group of Hayes 23B shallow casseroles has usually been treated as an inseparable entity. However, the abundance of these rims in phases 6-8 of the study assemblage, which is further highlighted by their scarcity in phases 9-11, suggests that the production of variant J.I ceased sometime in the late 3rd/early 4th century AD. Therefore, when found in the late 4th/early 5th century AD contexts the examples of this variant are most certainly residual.203 The adjective “standard” is attached to the Hayes 23B variant J.II (Plate 8:41-42), not only because of its abundance, but also for its production in both North and Central Tunisia. In the study assemblage the ratio is ca. 3:2 in favor of North Tunisian production. Still, in terms of the vessel size these productions do not differ significantly from one another. The standard Hayes 23B shallow casserole is always a medium-sized form. The only difference is that the upper wall of North Tunisian vessels is markedly more shallow (ca. 5.5-7.0 cm) than in their Central Tunisian parallels (ca. 6.5-8.0 cm). The possible functional or chronological implications of this difference remain to be demonstrated. As attention is turned to the surfacing of the standard Hayes 23B shallow casserole, the differences between North and Central Tunisian production become more apparent. The treatment of the exterior surface is fairly uniform with smoothing and

54

Forms

Central Tunisian Ext / Int

SLI

SMS SMO

North Tunisian UFI

SLI

SMS SMO

UFI

SLI

-

-

-

-

1

1

-

-

SMS

2

15

1

-

5

18

-

-

SMB

-

37

-

-

15

37

2

-

SMO

-

13

3

1

7

23

5

1

burnishing qualifying as the most common methods of finishing in both productions (Table 17). As usual, burnishing occurs in stripes and bands rather than in uniform patches covering large portions of the vessel surface. In Central Tunisian products the interior surface is predominantly smoothed and occasionally the procedure has produced a semi-lustrous self-slip surface, the quality of which cannot be by any means compared to the slipped interior surfaces of North Tunisian vessels. But as the presence of a slip is easier to identify on the light brown North Tunisian than on the dark red Central Tunisian fabric, the notion may be biased. Otherwise, two distinct production sequences were used contemporaneously in different parts of Africa Proconsularis for the manufacture of the same cookware form. Although the vast majority of the standard Hayes 23B shallow casseroles in the study assemblage were fired in an oxidizing atmosphere, the number of vessels with reduced core differs considerably as twenty-five out of thirty-three examples belong to Central Tunisian production. Due to very sharp core margin (Plate 1g), these products seem to have been fired at a higher temperature but for a shorter time than their North Tunisian parallels. It is unclear, however, whether the difference resulted from the fuel used to fire pottery or from different kiln types used in pottery production. The former explanation is perhaps more plausible, although agricultural by-products, the nature of which will be discussed more profoundly in section 4.3.3, seem to have been equally available for potters in North and Central Tunisia. Another indication of the high firing temperature used in Central Tunisia is the tough fabric reflected by the general absence of use-wear. In cases where usewear is present, it is strictly limited to the vessel lip. In North Tunisian products with softer fabric, attrition marks occasionally extend to both the interior and exterior surface. The distribution of use-wear also indicates that the variant J.II was frequently used with a lid, while the presence of soot on the exterior surface in over quarter of the material demonstrates that these vessels were used directly over a fire. Both productions, 85% of the Central Tunisian and 95% of the North Tunisian, also show a 2-3 cm wide band of patina cenerognola underneath the rim on the upper exterior surface. A layer of whitish scum, the origin of which lies in the combined use of calcareous clay

Key: SLI= slipped, SMS= smoothed with self-slip, SMB= smoothed and burnished, SMO= smoothed, UFI= unfinished.

Table 17. Surfacing of the standard Hayes 23B shallow casserole.

and salt water (see sections 2.2 and 4.3.1), is topping this surface on nearly half of the Central Tunisian examples. The dating of the standard Hayes 23B shallow casserole presents the usual problems, though its absence from the Early Imperial deposits of the Palatine East excavations points strongly towards a date after the mid-2nd century AD. The abundant finds from the early 3rd century AD shipwreck of Trincere and the mid-3rd century AD wreck of Cabrera indicate that it is by no means intrusive in the earliest horizons of the study assemblage,204 even though at Otranto variant J.II has only been found in contexts dating after 330-350 AD.205 The dominance of North Tunisian production in the earlier phases, especially in phase 6, suggests that it partially precedes the Central Tunisian one. The possibility that both productions had persisted at least to the early 5th century AD is suggested by nearly an equal number of examples included in phase 11. The variant J.III (Plate 9:43) differs from the standard Hayes 23B shallow casserole only due to its indistinct rim that was formed by folding the clay against the interior surface without cutting it separate from the interior wall. As a result, these rims tend to be somewhat taller and narrower than those in the preceding groups, while the rim diameter range is pretty much the same. The group includes examples of both productions, which are distinguished from one another only by the fabric, as the selection of vessel surfacings they show is the same. The most common technique, smoothing, is often accompanied by burnished stripes or bands on the exterior and patches of self-slip on the interior surface. In rare cases, all of which belong to the North Tunisian production, the interior surface seems to bear a slip. Nearly all the examples show a band of patina cenerognola cover on the exterior wall beneath the rim with its color ranging from brown to black in North Tunisian and from black to whitish-gray in Central Tunisian products. While the former production was fired in a fully oxidizing atmosphere, nearly all Central Tunisian examples show a reduced core with sharp margins indicating higher firing temperature and shorter firing time (cf. variant J.II). The evidence concerning the vessel use is fairly uniform. Use-wear is once again more abundant among North Tunisian examples, the abraded area

55

Chapter 3

only covers the vessel lip and the distribution of soot is limited to the exterior surface. In the absence of comparanda, the close resemblance to the standard Hayes 23B shallow casserole complemented by identical distribution of these variants in the study assemblage, the production can only be dated from the late 2nd to the early 5th century AD. The variant J.IV (Plate 9:44) should be regarded as another derivative of the standard Hayes 23B shallow casserole. The only factual difference is the presence of a shallow depression on the interior rim formed when enough force was used to press the topmost section of the clay against the interior surface. The close resemblance to the other Hayes 23B variants is underlined by nearly identical rim diameter range and rim shape, even though rims are generally somewhat higher in this group. The upper wall may also be slightly taller compared to the preceding variants, at least in the only well-preserved example in the study assemblage it is over 8 cm high. Nearly two thirds of the examples in the study assemblage are of Central Tunisian production. Their distribution is virtually identical to the standard Hayes 23B shallow casserole implying that the North Tunisian production began somewhat earlier than the Central Tunisian one. Both seem to have persisted at least till the first half of the 5th century AD. Unlike in other variants, these two productions do not differ significantly with respect to vessel surfacing. Predominantly smoothed exterior surface may also carry burnished stripes and bands or a patchy self-slip, the presence of which is also characteristic of the smoothed interior surface.

Firing in a fully oxidizing atmosphere characterizes both productions excluding a third of Central Tunisian examples, which has a reduced core with sharp margins. A 2-3 cm wide band of patina cenerognola deposition covering frequently the upper exterior surface ranges from brownish to black in color in North Tunisian production, whereas Central Tunisian production displays often a blackened vessel surface with a whitish scum. In addition to vessels with soot-covered exterior, examples in which sootdeposition extends to the rim or interior surface are also present. The distribution of use-wear is usually limited to the tip of the rim, though it may extend to the vessel interior or exterior. In all, the variant J.IV was very unlikely used differently from other Hayes 23B variants. The variant J.V (Plate 9:45) is distinguishable from the rest of the group by the peculiar interior rim bearing two or more shallow grooves. As indicated above, they probably resulted from the potter's efforts to press the folded rim tightly against the interior surface. The height is the only metric variable distinguishing the variant J.V from the previous entries. All but one example belong to Central Tunisian production characterized by smoothed vessel surfaces with occasional burnished areas on the exterior and scarce patches of self-slip on the interior surface. As both use-wear evidence and chronological distribution are practically identical with the previous entries, neither the function nor the date of the variant J.V could have differed significantly from other shallow casserole variants.

3.5 NORTH TUNISIAN DEEP CASSEROLES The group of North Tunisian deep casseroles is fairly heterogeneous, mainly because the basic vessel form was modified several times in the course of production that lasted at least from the mid-1st to the mid-5th century AD. None the less, practically all the North Tunisian deep casseroles have been formed in the following way. First, the vessel body was thrown right side up, whereupon the rim was either folded over to form a flange or pressed against the exterior surface to produce a protruding thickening. In many cases, the lid seating was either grooved or cut on the top of the rim, and the rim was distinguished from the exterior wall with a cut. Especially in the case of Hayes 197 derivatives

(groups M-O) the throwing of the vessel body was executed with potter's wheel rotating at a high speed producing a rilled surface on the vessel interior. At the leather hard stage, the vessel was attached to the potter's wheel upside down and the bottom was turned with a tool that often produced both a corrugated vessel surface and a sharp carination between the wall and the bottom (Plate 2e). Probably at this stage, the vessel rim was also given finishing touches by removing the excess clay with a flat-edged tool that produced the faceting frequently seen on the exterior rim. The result of this process was relatively deep, cylindrical casserole with somewhat indistinct rim and corrugated sagging bottom.

Table 18. The size and the distribution of the Hayes 194 deep casseroles. Diameter

K

Variant min K

-

avg 23/26

Rim height

max min -

-

avg 21/25

Rim width

max min -

-

Distribution

avg

max

P6

P7

P8

6/9

-

1

-

1

56

Fabric

P9 P10 P11 CT -

-

-

-

Total

NT

?

2

2

-

2

Forms

3.5.1 Hayes 194 deep casseroles The Hayes 194 (form K, Plate 9:46) is a deep casserole with a relatively tall vertical upper wall ending in a profiled, thickened vertical rim with pronounced lid seating at the inner edge. As indicated by particularly well-preserved example from Early Imperial deposits of the Palatine East excavations (Figure 14), the form has a slightly sagging bottom separated from the upper wall by an angular joint. In rare cases, the Hayes 194 was equipped with a pair of horizontal handles attached underneath the rim,206 but the study assemblage includes only two residual rim sherds.

the finds made at Ostia, Cosa, Fréjus (France) and on various Spanish sites212 point towards the early 1st century AD, the presence of Hayes 194 is also characteristic of the early deposits of Terme del Nuotatore (Ostia), which date to AD 50-90.213 In fact, the beginning of the production has been commonly dated to the late 1st century AD and the following century regarded as the main period of its distribution.214 The latter idea, however, has been contradicted by scholars who date the end of the production either to the first half,215 or to the mid/late 2nd century AD.216 The dominance of the Hayes 194 deep casserole in the late 1st and 2nd centuries AD is reflected by some twenty examples included in Early Imperial deposits of the Palatine East excavations. The fact that the terminal date of the Hayes 194 has been rarely placed to the early 3rd century AD,217 indicates also shortlived production. Although the two sherds of Hayes 194 in the study assemblage (Table 18) are certainly residual, they are otherwise informative. Firstly, their fabric supports Hayes' observation of the North Tunisian origin of this form.218 Another point of interest is the brown or brownish-black color of patina cenerognola deposition on the upper exterior surface, because for some reason these colors are characteristic of vessels belonging either to the earliest or the latest production phase of African cookware. The most interesting aspect, however, is the way the rim shape of this early form can be related to the typology of North Tunisian deep casserole rims. A pronounced inner lid seating is a risk regarding the efficiency of production, because not only the casserole but also its lid may warp during the drying stage, and it is therefore very difficult to get a proper fitting for the lid.219 Hence, the absence of African cookware casserole forms with a flanged rim or a pronounced lid seating from the early 3rd century AD onwards is understandable due to requirements of production and transportation.

Figure 14. A complete Hayes 194 deep casserole from the late 1st century deposit (PED 83) of the Palatine East excavations. The fact that the Hayes 194 belongs to the earliest phase of African deep casserole production is shown by the vessel form, borrowed through the influence of Early Byzacenan cookware from a Greek/Punic cooking vessel, lopada, the production of which had begun as early as the 5th century BC.207 The form of the Hayes 194 has been discussed in more profound manner in the third volume of the excavations of Terme del Nuotatore at Ostia,208 where the first examples were found in contexts preceding the introduction of African Red Slip ware A in the Italian peninsula.209 Thus, it is not a surprise that the Hayes 194 has been considered to be a parallel to a similar African Red Slip ware A1 form Hayes 19.210 The production of Hayes 194 in North Tunisian cookware might have begun already by the late 1st century BC,211 but the precise date of introduction outside Roman Africa is fairly difficult to fix. While

3.5.2 Hayes 193 deep casseroles The Hayes 193 deep casserole is a form characterized by a tall convex wall separated from a slightly sagging bottom by an angular joint. The inward inclined upper wall customarily ends in an even or slightly thickened rim with a pointed lip. The form is

Table 19. The size and the distribution of the Hayes 193 deep casseroles. Diameter

L

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

29

L.I

17

19/20

23

-

-

-

5

6/7

7

-

7

9

3

-

1

-

20

-

20

L.II

15

19

20

-

-

-

5

5/6

7

-

-

4

1

-

1

-

6

-

6

L.III

-

19/23

-

-

-

-

6

7

8

-

1

-

1

-

1

-

3

-

3

57

Chapter 3

fairly rare, even though the rarity may partially result from the fact that rim fragments of the Hayes 193 are easily misidentified as Hayes 181 pan or Hayes 26/181 pan-casserole. Examples included in the study assemblage have been divided into three variants (Table 19), the most abundant of which has been defined as the standard Hayes 193 deep casserole. The rim sherds of the standard Hayes 193 deep casserole (variant L.I, Plate 9:47; 10:48) form one of the most homogeneous groups in the study assemblage. The rim equipped with a pointed lip is usually fairly thin, though compared to the upper wall it was slightly thickened. In the few well-preserved examples the convex upper wall reaches a height of 68 cm. The exterior surface is regularly smoothed with nearly a half of the vessels showing additional traces of burnishing. Smoothing is the most common finish also on the interior surface, but the assemblage also includes examples with unfinished or self-slipped interior surface. In addition, examples of Hayes 193 with a thin slip cover on the interior surface have been reported.220 In three fourths of the examples the fully oxidized fabric bears also a band of patina cenerognola deposition, usually brown or black in color, on the exterior surface. Compared to previously discussed forms, the band is exceptionally wide with its lower limit generally located only some millimeters above the joint of the wall and base. As the extent of patina cenerognola cover is clearly form depended, it was more likely produced in the firing stage than during the use of the vessel (cf. section 4.3.1). The standard Hayes 193 is said to be one of the earliest African cookware forms with its locally distributed predecessors dating back to the 1st century BC.221 The earliest, but also most dubious, occurrence outside of Roman Africa comes from a burial dated to the Tiberian age at Ballesta (Spain).222 Otherwise, the introductory date of Hayes 193 could be fixed to the late 2nd / early 3rd century AD,223 when the form was produced in the same workshops as late forms of African Red Slip ware A.224 This is explicitly shown by a similar, but also substantially rare form found in African Red Slip ware A2 and A/D.225 The main period of production has been commonly dated to the 3rd century AD,226 when this form was exported in moderate amounts to all parts of the western Mediterranean. The clustering of the examples in the study assemblage into phases 7-8, suggests that the production did not cease before the early 4th century AD. Later examples, just like the late 4th century AD finds at Luni227 are contrasted by the late 4th/5th century AD deposit (stratum I) of room XIV at the Terme del Nuotatore (Ostia), which did not yield any examples of the standard Hayes 193.228 Its absence in such an important context supports the idea about the residual nature of the material found in the mid-5th century AD context at Porto Torres (Sardinia).229

Evidence for the use of the variant L.I is not unambiguous. Firstly, unlike in many other deep casseroles, the exterior bottom is frequently smoothed and rarely corrugated. In addition, the absence of examples with sooted exterior surface conflicts with the presence of sagging bottom, which implies that the vessel would have been placed on a cooking stand set over a fire. As wear-marks were detected from a few examples only and in most cases they cover the upper exterior surface besides the lip, it is not clear either whether the Hayes 193 was used with a lid and alternative explanations for a mechanism producing such wear pattern must also be sought. In the variant L.II (Plate 10:49) the upper wall is less convex than in the standard Hayes 193 and it shows often a wide but shallow furrow right above the junction of the wall and the base. Instead of being thickened, the rim is often either thinned or of the same thickness as the upper wall. As indicated both by the rim diameter and the height of the upper wall (5.0-5.5 cm) the variant is also considerably smaller than the standard Hayes 193.230 All examples in the study assemblage show smoothed, but often also burnished exterior surfaces. The vessel interior is either smoothed or unfinished, of which the latter surfacing often correlates with the smoothed exterior. All the vessels were fired in a fully oxidizing atmosphere and two thirds of them also show brown or blackish patina cenerognola cover on the exterior surface. Unlike with the standard Hayes 193, most examples display patches of soot on the exterior surface, suggesting a use directly over a fire, whereas the distribution of wear-marks is similar to the preceding variant. The production of the variant L.II probably persisted longer than the production of the standard Hayes 193 deep casserole, both of which have been found together in the Cabrera III shipwreck dated at AD 257.231 In the study assemblage most examples of the variant L.II are concentrated into the early 4th century AD phase, while the other evidence together with a parallel from Carthage may indicate the continuity of the production at least to the late 4th/early 5th century AD.232 The infrequent Hayes 193 variant L.III (Plate 10:50) is the most robust of the three entries in this group. It has a peculiar, relatively straight and strongly inward inclined upper wall that ends to a rim with a slightly thinned lip. On the other hand, neither the surfacing nor the firing differ from the rest of the group, excluding the absence of patina cenerognola covered exterior surfaces. Sooted exterior surfaces indicate use over a fire, while the absence of use-wear points towards cooking without a lid. These finds together with the presence of a similar vessel form in Late Cooking ware I are sufficient to date the variant L.III from the early 4th to the late 4th/early 5th century AD.233

58

Forms Table 20. The size and the distribution of the proto-Hayes 197 deep casseroles. M

Diameter

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

avg

max

P6

P7

P8

Fabric

P9 P10 P11 CT

Total

NT

?

9

M.I

21

23/25

34

10

11

12

11

12

13

4

-

1

-

-

1

-

6

-

6

M.II

18

21

23

6

7

8

-

13/14

-

1

-

1

-

1

-

-

3

-

3

3.5.3 Proto Hayes 197 deep casseroles The idea that the "proto Hayes 197 deep casserole" is an early version of a deep casserole that later on developed into a form known as the Hayes 197 – and represents an intermediate stage in a typological sequence of North Tunisian deep casseroles – has previously been brought up by various authors.234 The residuality of this form in the study assemblage (Table 20) is suggested both by the fragmentary condition of respective examples and the evidence from other sites dating the end of the production to the second half of the 2nd century AD.235 The date of introduction, which is occasionally placed as early as the 1st century AD,236 is more often presented with a broad margin from the late 1st to the early 2nd century AD.237 The form is characterized by a vertical or slightly inward inclined upper wall ending in a short flanged or thickened rim. Both rim forms, the dissimilarity of which has been used to distinguish two variants, were most probably designed to provide a sufficient seating for the lid. In a flanged rim the upmost section of the clay was folded and trimmed at the leather hard stage, while a protruding thickening was formed by folding the clay against the exterior surface. The other vessel parts, including the vertical upper wall and a sagging base with a corrugated exterior surface (Plate 2g), do not differ significantly from other North Tunisian deep casserole forms. The most common surfacing of this form is smoothing, but some examples show also traces of burnishing on the exterior. All vessels have been fired in a fully oxidizing atmosphere, excluding the regular band of a black or brownish black patina cenerognola deposition on the exterior surface. From the typological point of view it is interesting, however, that most examples of the form M have a smoothed interior surface, whereas in the Hayes 197 deep casserole it is predominantly covered by a shallow rilling. For the time being, it is uncertain whether the difference is other than chronological. A protruding rim thickened on the exterior surface is the main characteristic of the variant M.I

(Plate 10:51-52). The rim has a grooved lid seating at the inner edge and a notch on the underside separating it from the exterior wall. In more complete examples (Plate 10:52), an angular joint separates the vertical upper wall from a sagging base. On the basis of the shape, the variant M.I was evidently used over a fire, even though the study assemblage only includes a few examples with sooted exterior surface. Similarly, the absence of use-wear from rim surfaces does not challenge the idea about the use with a lid. The evidence suggesting the residuality of this variant in the study assemblage is plentiful. Not only the late 1st and early 2nd century AD contexts at Carthage,238 but also the early 2nd century AD deposits of other sites are well-attested with this variant.239 The variant M.II (Plate 11:53-54) is characterized by a short flanged rim with a shallow furrow on the rim top forming a lid seating. The presence of lid seating together with wear-marks on the rim top point towards use with a lid, while direct contact with fire is evidenced by an example with a sooted exterior surface. A presupposition for flawless typology would be that variant M.II precedes variant M.I in chronology, but the archaeological evidence does not support this idea. In fact, though earliest examples of the variant M.II date to the late 1st century AD, it seems to become more common only in the early/mid- 2nd century AD.240 The first rather than the second half of the 2nd century date is further supported by the deposits excavated on the south side of the circular harbor at Carthage.241 3.5.4 Hayes 197 minor deep casseroles The small North Tunisian deep casserole (form N, Plate 11:55-56) is introduced separately from the rest of the Hayes 197 deep casserole family due to probable differences in use. The existence of two size classes was previously observed by Sánchez Sánchez,242 whose definition (ø 12-20 cm) for the small Hayes 197's i.e. Hayes 197 minor correlates fairly well with the size of the examples included in the study assemblage (Table 21). For the sake of comparison, it should be mentioned that the

Table 21. The size and the distribution of the Hayes 197 minor deep casseroles. N

Diameter

Variant min N

14

avg 15/18

Rim height

max min 19

7

avg 8/10

Rim width

max min 11

6

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

30

7/8

10

16

6

4

3

29

-

30

59

-

1

1

Chapter 3

maximum rim diameter of a full-sized Hayes 197 deep casserole may be as much as 34 cm. The upper wall of Hayes 197 minor, on the other hand, is only 3-4 cm high and separated from the slightly sagging bottom by an angular joint. Excluding the metric differences, the form is an exact reproduction of the “standard” Hayes 197 on a miniature scale with a shallow, vertical upper wall ending in a rounded or almond-shaped rim thickened on the exterior surface and bearing a notched lid seating. In many cases the lid seating is so faint that it can be reasonably asked whether it had a practical function at all. The evidence from the necropolis of Porta di Cesarea (Tipasa) indicates that the Hayes 197 minor could have been used with a lid.243 Hence, it is imaginable that as the same workshops produced both Hayes 197 minor and full-sized Hayes 197, the potters automatically notched the top of the rim The vessel interior has customarily rilled surface that was produced by the potter's wheel rotating at a high speed, while the exterior wall is always smoothed and shows in a few examples darker stripes that are very close to stripe-burnishing. Practically all the examples in the study assemblage come in fully oxidized fabric, but in contrast to many other forms, only a third shows patina cenerognola deposition on the exterior surface. One potential, but yet hypothetical explanation is that this small and relatively fragile form was customarily placed with African tableware vessels at the top of kiln, where the atmosphere remained oxidized throughout the firing (cf. section 4.3). The date of Hayes 197 minor can reasonably be presumed to correspond with the rest of Hayes 197 series. Therefore, it is highly improbable that an example found in the early 1st century AD deposit at Cosa244 would date the beginning of production, especially when the form is absent from other contemporary deposits.245 The production began more likely only during the 2nd century AD, of which both the first246 and the second247 half have been

promulgated as the precise date of introduction. The distribution of the Hayes 197 minor in the study assemblage supports the idea about the continuity of production in the early 4th century AD,248 but contrasts its persistence to the early 5th century AD.249 In addition, the form is said to be more common than other Hayes 197 variants in Algeria,250 which may reflect differences in the distribution mechanisms or culinary habits. 3.5.5 Hayes 197 deep casseroles The Hayes 197 deep casserole, characterized by a high vertical wall ending in a thickened rim with a notched lid seating on the top, is one of the three African cookware forms found in abundance around the western Mediterranean. Still, statements stressing the irrelevance of the variation shown by this form are by far more numerous than attempts to elaborate its typology. For example, instead of concerning the possibility that the vessel typology could reflect conscious changes in the craft, Carandini has comprehended the production of Hayes 197 as a fortuitous development without any historical significance.251 The following discussion will demonstrate, however, that at least some of the variability detected in the shape of Hayes 197 rims could result from chronological differences or from the existence of several contemporary productions. To test this possibility, the shape of the rim was used as a criterion to divide the material into seven variants (Table 22). The method has previously been used by Leveau to explore similar material, but the results do not necessarily correspond to one another.252 In an alternative classification the form is divided into two variants (A and B) on the basis of the presence of an incision underneath the exterior rim, while a third variant – dated from the 3rd to the mid-5th century AD – is introduced as the late Roman casserole form Ostia III.108.253 In the following discussion this late form corresponds to variants O.V and O.VI.

Table 22. The size and the distribution of the Hayes 197 deep casseroles. Diameter

O

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

Total

avg

max

P6

P7

P8

P9 P10 P11 CT

NT

?

246

O.I

18

22/27

34

11

13/17

20

8

10/12

24

33

22

20

6

6

8

2

93

-

95

O.II

15

19/24

30

8

11/13

16

8

10/11

13

28

9

11

3

2

1

4

50

-

54

O.III

20

20/23

27

10

12/13

18

9

10/11

14

3

1

1

2

-

-

-

7

-

7

O.IV

19

23/26

31

10

15/17

20

9

10/12

15

8

3

13

2

4

2

3

29

-

32

O.V

21

24/28

29

14

14/16

17

10

12/13

13

-

1

6

1

-

1

1

8

-

9

O.VI

18

24/27

33

12

13/16

19

9

10/12

14

2

9

15

11

1

5

1

42

-

43

O.VII

20

23

27

16

17

18

-

10/11

-

1

-

4

-

-

-

-

5

-

5

O.VIII

-

27

-

-

16

-

-

-

1

-

-

-

-

1

-

1

11

60

Forms 3rd and 4th centuries AD,264 although the 3rd century AD has also been suggested to be both the introductory and closing date.265 By the beginning of the 5th century AD, the shape of the Hayes 197 rim had markedly changed, probably as a consequence of an attempt to facilitate the production sequence. Now, the rim was simply formed by folding the clay over to exterior and pressing it against the wall without further turning or cutting. Thus, the rim and wall are often hard to distinguish from one another in the late variants,266 the production of which ceased likely in the late 4th/early 5th century AD.267 As a result, Hayes 197 has perhaps been rightly regarded as the end point of a typological sequence that led from deep casseroles with a distinct, wide flanged rims to thickened rims acting both as a lifting flange and a lid locator (i.e. lid seating).268 The fabric and the distribution identify the Hayes 197 as a typical North Tunisian product.269 Although the form has been found also in southern Africa Proconsularis and even in Tripolitania (Sabratha), importation seems to have been on a very modest scale and customarily eclipsed by locally produced deep casserole forms.270 The North Tunisian provenance is also supported by chemical analyses performed by Taylor and Robinson, the results of which pointed out the Oudna kiln site as one potential source for the production.271 The rarity of this form in Central Tunisian fabric is striking, especially when compared to the frequency the other two African cookware mass products, the Hayes 196 lid and the Hayes 23B shallow casserole, are attested in this fabric.272 One potential explanation is that the strong local tradition favored the production of the Hayes 183 (see form S), but the possibility that the differentiated properties between North and Central Tunisian clays would have induced this pattern cannot be ruled out completely. The element identifying the first early Hayes 197 variant O.I (Plate 11:57-58) is a hemispherical rim with a grooved lid seating near the inner edge of the rim top. The vessel dimensions do not differ significantly from the other variants and the same can be said of the surfacing and firing. In spite of a few burnished exceptions, the exterior surface was generally smoothed and the interior surface rilled. Only 15% of the interior surfaces have been smoothed. The chronology of the variant O.I is not unambiguous, as the dates of comparanda cover the entire period suggested for the Hayes 197 production from the late 2nd to the first half of the 5th century AD.273 Therefore, it is indicative that in the study assemblage the majority of examples concentrate among the late 3rd/early 4th century AD phases. As virtually all the examples belong to the North Tunisian production, the fabric of which is softer than the Central Tunisian one, the presence of wear-marks in nearly two thirds of the examples is not

Regarding the production sequence, the suggestion that the rim of this form would have been formed by applying a strip of clay to the upper exterior wall is doubtful,254 as the making of the Hayes 197 hardly differed significantly from the preceding forms. The deep rilling (Plate 2f), which resulted from the throwing of the vessel body with the potter's wheel rotating at a high speed, however, is highly characteristic on the interior surfaces of all Hayes 197 variants. Otherwise, the interior surface has customarily been left unfinished, as examples with smoothed interior are not frequent – only 15% of the cases – and do not correlate with particular Hayes 197 variant, vessel size or chronological phase. Unfortunately, the study assemblage did not include enough sufficiently preserved examples to confirm whether the floor of Hayes 197 is covered by a thin slip as it has sometimes been suggested.255 Over 93% of the Hayes 197's in the study assemblage show fully oxidized fabric paired with brown or black patina cenerognola deposition on the exterior surface. They are not matched in abundance by examples carrying traces of soot deposition on the exterior surface, the number of which is limited to couple of dozens. The low number results partially from the fragmentary nature of most diagnostic sherds, which often preserve only the topmost part of the wall and rarely the sooted bottom. After browsing through some body sherd material, however, I have no reason to believe that this form would not have been used on a metal tripod (tripes) set directly over a fire.256 Similarly, the idea that the Hayes 197 was often topped with variants of the Hayes 196 lid finds support in archaeological evidence.257 For example, in the late 2nd/early 3rd century AD context excavated at Marseilles the ratio between these two forms matched to such a perfection that Moliner had no other alternative than to stress their probable relationship.258 The origins of the Hayes 197 deep casserole may be traced down to the late Punic period,259 as the Hayes 191 flanged casserole, one of the earliest products in African cookware, has been interpreted as its actual predecessor. Generally speaking, the typology of forms preceding the Hayes 197 is substantially hard to follow, because all deep casseroles with rim thickened on the exterior surface have usually been lumped together in discussion. As a result, one may easily get the impression that the production of Hayes 197 began already in the late 1st century AD,260 though the period was marked only by the introduction of preceding deep casserole forms, such as the form M (“proto Hayes 197”). Even if the early Hayes 197 variants – characterized by a sharp cut separating the rim from the exterior wall261 – may have been produced already in the early 2nd century AD,262 it started to gain its dominance only in the latter half of the century.263 The main period of production has been dated to the

61

Chapter 3

necessarily exceptional. Their distribution, however, shows considerable variation by extending frequently to the exterior rim and the upper exterior wall. For the time being, the mechanism behind such an anomalous distribution cannot be identified, as the type of usewear does not seem to correlate with any other variable. Still, the prevalence of examples with a sooted exterior among sherds showing soot deposition, which make up slightly more than 10% of the material, points towards use directly over a fire. The second early Hayes 197 variant O.II (Plate 12:59) is very similar to the previous entry, though in this case the notch separating the rim from the exterior surface is clearly more pronounced. The closest parallel to be pointed out from the literature is Aguarod Otal's variant A, the main characteristic of which is said to be the undercut rim.274 However, both the vessel dimensions and the prevalence of smoothed exterior and rilled interior surfaces would have presented a good case to introduce these two variants together. Even the distribution of use-alteration, which include rare sooted examples and some anomalous patterns of use-wear, link these two variants. The early date of the variant O.II is perhaps slightly exaggerated by parallels, which are reported to have been found in the late 1st and early 2nd century AD contexts.275 The study assemblage with nearly a half of the examples belonging to the late 3rd century AD phase 6, does not only support the early date, but also suggests that the production of variant O.II had probably ceased by the early 4th century AD. The somewhat unusual and uncommon variant O.III (Plate 12:60) is characterized by an undercut rim with a profiled exterior. Like the variants introduced above, this medium-sized deep casserole shows frequently smoothed exterior surface paired with rilled vessel interior, fully oxidized fabric and blackish patina cenerognola cover on the exterior. Neither does the infrequent presence of use-alteration marks, usewear or soot, offer possibilities for further discussion. In fact, chronological distribution qualifies the variant O.III as an oddity of the 4th century AD. The appearance of the variant O.IV (Plate 12:61) is very similar to the early Hayes 197 variants (O.III), thanks to protruding rim with a rounded thickening on the exterior. Only the way in which the vessel rim was finished is different; instead of grooving the lid seating was notched on the rim top. Other variables do not distinguish the some thirty examples in the study assemblage from the rest of Hayes 197 variants. The distribution of use-wear, however, is more often limited to the rim top than extended to the upper exterior surface, possibly indicating that a notched lid seating retained the lid on the vessel top better than the grooved one. The chronology of the variant O.IV extends probably from the late 3rd to the early 5th century AD, but on the

basis of its distribution in the study assemblage, the possibility that production had already ceased by the late 4th century AD must also be considered. Attributes distinguishing the late Hayes variant O.V (Plate 12:62) from previous entries are a robust rim – formed by pressing the folded clay against the exterior surface, but without cutting it separate from the exterior wall – and a wide, grooved lid seating on the rim top. Otherwise, the evidence concerning the production sequence and vessel use does not differ markedly from the preceding entries. The mid-4th century AD introductory date of the variant O.V was previously based on the finds from Luni and Carthage.276 From now on, both its absence from the late 3rd century AD phase 6 and the strong presence in phase 8 dated to the first quarter of the 4th century AD may be counted in as supporting evidence. The production of the variant O.V very likely persisted to the 5th century AD,277 but the example found in the late 5th/early 6th century AD deposit at Carthage may already be residual.278 The second late Hayes 197 variant O.VI (Plate 12:63) has a short, profiled rim thickened on the vessel exterior and a wide but shallow lid seating grooved on the top of the rim, which was rarely cut apart from the upper exterior wall. Features observed in some forty examples of the variant O.VI – smoothed vessel exterior paired with rilled interior surface, fully oxidized fabric and black patina cenerognola cover on the upper exterior wall – are common for all Hayes 197 variants. The comment also applies to the use-alteration evidence consisting of few examples with a sooted exterior surface or worn rim top. The main production phase of variant O.VI dates to the late 4th and early 5th century AD,279 though it may have already been introduced in the late 3rd century AD.280 In the study assemblage most examples date to the 4th century AD, but the strong presence in phase 11 indicates the continuity of production at least into the early 5th century AD. As the two sherds belonging to phase 6 are probably intrusive, the date of introduction for the variant O.VI should be fixed in the early 4th century AD. A notched lid seating with a triangular section and a somewhat elongated rim profile are the main characteristics of variant O.VII (Plate 13:64). Due to its scarce presence in the study assemblage, it is uncertain whether the seemingly increased association of smoothed interior surfaces with this variant is an actual characteristic of the production or just a statistical anomaly. In either case, measurements of the variant O.VII do not differ significantly from the rest of the group and the same can be said of vessel surfacing and firing. The concentration of the examples into the early 4th century AD phase 8 raises the possibility that the variant reflects very short-lived production of an individual potter or a workshop.

62

Forms Table 23. The size and the distribution of late Roman deep casseroles. Diameter

P

Variant min P

avg

Rim height

max min

1 8 21/22 2 7

avg

Rim width

max min

1 3 19/21 2 4

9

avg

Distribution

max

13/14 1 7

Fabric

P6

P7

P8

P9 P10 P11 CT

-

2

4

1

-

1

4

Total

NT

?

8

-

4

8

The introduction of the Hayes 197 deep casseroles is concluded with a rare variant O.VIII (Plate 13:65) characterized by the usual slightly everted, protruding rim thickened on the exterior and a grooved lid seating on the rim top. The strongly inward inclined upper body that could be used as a criterion to identify the vessel shape as a cookpot, on the contrary, is highly unusual in this context.281 The rarity of the variant O.VIII is reflected both by the presence of only one example in the study assemblage and scarce comparanda. The finds from Ostia and Valencia point towards a production that had begun by the early 3rd century AD.282 The example from the Palatine East excavations, which is hardly residual due to 20% of the rim circumference preserved, signals for the continuity of the production at least in the early 4th century AD.

Compared to the previous entries, an increased number of examples bear other than a fully oxidized fabric, while patina cenerognola deposition, usually either brown or black in color, was detected on half of the exterior surfaces. The comparanda that can be appointed for the Late Roman deep casserole is both scarce and chronologically problematic, as similar material has been only published from three other sites, Luni, Villa dels Hospitals (El Morel) and Tarracona. While the examples found in Luni belonged to the late 6th/7th century AD context,283 the finds from ager Tarraconensis were not dated any more precisely than "late Imperial".284 Thus, as most examples in the study assemblage belong to early 4th century AD, just like the find from Villa dels Hospitals (El Morel),285 it is the most reliable date yet available.

3.5.6 Late Roman deep casseroles A deep casserole (form P, Plate 13:66-67), characterized by a tall, thickened rim with a projecting lid seating at the inner edge, can be interpreted as a late successor both to the Hayes 194 and the Hayes 197 due to peculiar shape of the rim. Two examples have been inserted into the find catalogue to illustrate both extremes in vessel size (Table 23). In difference to Hayes 197, form P has somewhat more robust rim with a lid seating formed by a shallow, stepped protrusion near the inner edge instead of a groove or a notch. Another difference can be pointed out in terms of the vessel surfacing, where the increased number of smoothed interior surfaces differs markedly from the Hayes 197 characterized by rilled or unfinished vessel interior. In most cases also the exterior surface of the form P was smoothed, though it may occasionally show burnished bands or stripes. The evidence concerning the use of the Late Roman deep casserole is fairly abundant as a third of the examples have a patchy soot cover on the rim and exterior surface, which points towards a use directly over a fire. The pronounced lid seating as well as a few worn rim tops suggests its use with a lid.

3.5.7 Hayes 199 deep casseroles The last form to be introduced in this group has usually been referred to as the Hayes 199 deep casserole (Table 24), but because it was not illustrated in the pages of the Late Roman Pottery, the use of other definitions is by no means exceptional in the archaeological literature. The form could reasonably be regarded as an odd Hayes 197 variant, the main characteristics of which are an angular rim thickened on the exterior and a notched lid seating on the vessel top, if the upper wall was not rather tall and the exterior surface covered with shallow ribbing. In this case the ribbing is probably an intentional surfacing as opposed to an accidental by-product of rapidly rotating potter's wheel. The similarity to the Hayes 197 has led many scholars to group the Hayes 199 together with other forms representing the intermediate phase in the North African deep casserole typology,286 like the protoHayes 197 (form M). On the other hand, Hayes did not only interpret the form as a contemporary product to the Hayes 197, but also as a successor the Hayes 198 deep casserole of the 1st century AD.287

Table 24. The size and the distribution of the Hayes 199 deep casseroles. Diameter

Q

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

avg

max

P6

P7

P8

Fabric

P9 P10 P11 CT

Total

NT

?

10

Q.I

-

24

-

-

13

-

-

12

-

1

-

-

-

-

-

-

1

-

1

Q.II

16

20/22

24

9

10

12

10

11/13

14

3

3

1

-

-

-

-

7

-

7

Q.III

-

19/24

-

-

13/16

-

-

8/12

-

1

1

-

-

-

-

-

2

-

2

63

Chapter 3

Due to heterogeneous nature of the ten examples in the study assemblage, the material has been divided into three variants, of which the standard Hayes 199 deep casserole (Q.I) is present with only one example. The status of the other two variants is not well established, but their angular rims and the sharp angle between the interior wall and the rim.288 make a good case for their introduction in this context.289 The only scholar to conclude anything on the use of Hayes 199 was Aguarod Otal, according to whom it served for stewing vegetables and/or meat in abundant amount of water,290 just like the other forms belonging to the third group of her classification (see section 3.9.3). Having been recovered in phase 6 (AD 270-330), the only example of the standard Hayes 199 (variant Q.I, Plate 13:68) in the study assemblage strengthens the idea about the probable cessation of the production by the late 3rd century AD.291 On the other hand, the first examples of this tall deep casserole form characterized by a notched lid seating on thickened, angular rim and light ribbing on the slightly convex exterior wall have been found in mid/late 1st century AD and Hadrianic contexts.292 The form occurs more frequently only in contexts dating from the mid-2nd century AD onwards.293 The latest and also almost certainly residual example was recovered from a mid4th century AD context at Luni.294 The examples of the variant Q.II (Plate 14:69) belong to a small deep casserole with slightly convex, vertical upper wall ending in a protruding rim with triangular section. In addition to completely oxidized fabric and smoothed surfaces, the vessel exterior frequently bears both shallow ribbing underneath the rim and brown to black, often patchy patina cenerognola deposition. In spite of the absence of soot deposition and wear-marks, there is no direct evidence questioning the interpretation on the use of this form. The variant Q.II has commonly been interpreted as a derivative of the Hayes 199 deep casserole,295 though the definition Ostia I-270 can be offered as another point of reference.296 Aquilué Abadías, on the other hand, stressed its similarity to the Hayes 197 and dated the variant between the late 2nd and the end of the 4th century AD.297 The evidence from both Valencia, where the variant Q.II was subjected to a minor scale distribution in the mid-3rd century AD,298

and the Palatine East excavations, where the majority of examples concentrate into phases 6-8, speak on the behalf of the late 3rd/early 4th century AD date. The extension of production to the late 4th century seems probable in light of the material found in the excavations of Carrer del Gasòmetre 32 (Tarragona).299 The identification and date of the variant Q.III (Plate 14:70) are both problematic. On the one hand, the two examples in the study assemblage are not very close parallels to the standard Hayes 199, though in respect of the production sequence or vessel use they do not differ much from it. The early 3rd century AD casserole forms 58 and 59 Dore has introduced in Central Tunisian fabric300 would actually be a natural identification, if the fabric of the examples in the study assemblage was not North Tunisian and their date of the late 3rd/early 4th century AD. Additional support for the 4th century AD date can be pointed out from Luni, where several examples of the variant Q.III have been found in such contexts.301 The end of the production may be shifted at least to the early 5th century AD, unless the example found in the Vandal period deposit at Carthage is residual.302 3.5.8 Miscellaneous North Tunisian casseroles In contrast to the previous sections, which have introduced similar African cookware forms and variants, the four North Tunisian casserole variants of this group are related to each other only by their provenance and scarce presence in the study assemblage. Most of them have also been encountered rarely on other archaeological sites of the western Mediterranean. The introduction of these miscellaneous North Tunisian casseroles is still important, as they will undoubtedly enrich the overall picture on the African cookware production. A short, everted rim with a rounded lip and a grooved lid seating on the vessel top are the sole diagnostic features preserved by the examples of the variant R.I (Plate 14:71). Due to their fragmentary condition, the identification of the variant as a casserole rather than a cookpot cannot be ascertained, especially when the latter alternative is suggested by the small rim diameter. It should be equally pointed out, however, that the variant R.I has also been

Table 25. The size and the distribution of miscellaneous North Tunisian casseroles. Diameter

R

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

avg

max

P6

P7

P8

Fabric

P9 P10 P11 CT

Total

NT

?

6

R.I

-

17/18

-

-

11

-

-

8/9

-

-

-

1

-

-

1

1

1

-

2

R.II

-

21

-

-

7/8

-

-

6/7

-

-

-

-

2

-

-

-

2

-

2

R.III

-

24

-

-

24

-

-

8

-

-

-

1

-

-

-

-

-

1

1

R.IV

-

28

-

-

11

-

-

17

-

1

-

-

-

-

-

-

1

-

1

64

Forms

deemed the typological end point of the Hayes 197 series.303 The presence of a patina cenerognola deposition and soot cover on the exterior surface are the two bases for the classification of variant R.I as African cookware. Turning to the variant chronology, the two finds in the study assemblage cover a period from the early 4th to the late 5th century AD, while comparanda from Carthage suggests that the production may have persisted even to the early 6th century AD.304 The only characteristic feature of the variant R.II (Plate 14:72-73) is a short everted rim with rounded lip and shallow furrows profiling the exterior. The examples in the study assemblage have the same rim diameter, smoothed vessel surfaces and fully oxidized fabric, but do not show a trace of patina cenerognola deposition. The soot seen on the exterior surface of both examples has most likely resulted from the direct use over a fire, whereas the absence of use-wear may be explained either by a small sample size or use without a lid. The date of the variant R.II is probably fairly late. Production may have begun only in the late 3rd/4th century AD and did not cease before the 5th century AD.305 The mid-4th century AD examples in the study assemblage fit fairly well into this picture, while a late closing date is supported by multiple examples found in the 5th century AD deposit at VilaRoma,306 though other evidence would seem to push it even to the late 6th century AD.307 As the fabric in both Palatine East examples is practically identical to North Tunisian fineware fabric, one could also relate them with finds described by Dore in the connection of the UNESCO Libyan Valleys survey as “finely

made and finished with finely tooled surfaces and rim, ... outer surfaces have crisply delineated horizontal grooves.”308 As decorated vessels are extremely rare in any cookware production, the variant R.III (Plate 14:74) with a simple zigzag line incised underneath a short everted rim with somewhat angular lip is an item of particular interest, which does not have any straight parallels. The fabric of the only example in the study assemblage, which dates to the early 4th century, is neither the usual Central nor North Tunisian fabric. Therefore, it is dubious whether a casserole found in the mid-5th century AD deposit at Carthage309 represents the same form and fabric. Another possible identification could be the Hayes 200 small olla used both for storage and cooking, even though the production of this form began already in the mid-2nd century AD.310 The last miscellaneous variant R.IV (Plate 14:75) – characterized by a slightly convex upper wall ending in a thickened, everted rim with rounded lip – is rare both in the study assemblage and elsewhere. The late 3rd century AD date of the only example from the Palatine East excavations supports the chronology proposed by Fulford, who dates the production between the 3rd and 4th centuries AD.311 The example recovered from the El Mahrine kiln site also fits well within this date range,312 while other potential parallels, none of which is exactly alike the example in the study assemblage, date to AD 360440.313 The nature of another parallel recovered in the kiln site near the Catacombs of Sullecthum314 is dubious, as it very likely belongs to the Central Tunisian production.

3.6 CENTRAL TUNISIAN CASSEROLES This section introduces a group of casseroles, the production of which most likely took place in Central Tunisia. Although the evidence available today suggests Central Tunisia as the main production area, the impression may well be biased, as the number of archaeological surveys and excavations carried out in Africa Proconsularis far outweighs the number of projects taken place in Tripolitania (modern Libya). The thirty-one examples in the study assemblage have been divided into two groups, of which the group of deep casseroles includes seven variants and is perhaps more interesting due to homogeneous appearance and relative abundance on other sites around the western Mediterranean. The nature of the three rim sherds classified as miscellaneous casserole or cookpot forms is more uncertain, as even their Central Tunisian origin can reasonably be questioned. Still, they may be important indicators of the standardization of the production.

3.6.1 Central Tunisian deep casseroles The definitions commonly used for this vessel form are the Hayes 183, tajine and marmitta.315 If the occurrence outside of Central Tunisia is used as a criterion, the deep casserole qualifies as the third most important cooking pot export after the Hayes 181 pan and the Hayes 23B shallow casserole. But unlike the other two forms, these vessels are not easy to mistake for contemporary North Tunisian production, because differences are manifold. Inward inclined, slightly convex upper wall ending in an everted rim with an undercut thickening on the exterior and a broad, shallow furrow on the interior rim forming a lid seating are both good marks of identification. But the most distinctive feature, however, is light to thick layer of whitish-gray salt scum covering frequently the exterior surface.316 The appearance of this surfacing is often very similar to patina cenerognola deposition on North Tunisian products,317 while the mechanism producing the effect was different (cf. section 4.3). 65

Chapter 3 Table 26. The size and the distribution of Central Tunisian deep casseroles. Diameter

S

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

Fabric

avg

max

P6

P7

P8

P9 P10 P11 CT

Total

NT

?

28

S.I

17

21/27

30

8

9/11

20

11

12/14

15

8

4

2

1

-

1

16

-

-

16

S.II

22

26

31

11

12

16

10

11

12

-

3

-

-

-

-

3

-

-

3

S.III

-

27

-

-

11

-

-

20

-

1

-

-

-

-

-

1

-

-

1

S.IV

18

18/19

22

8

9/12

15

9

10/11

15

-

-

1

1

1

2

4

1

-

5

S.V

-

19

-

-

8

-

-

14

-

1

-

-

-

-

-

1

-

-

1

S.VI

-

26

-

-

8

-

-

14

-

-

-

-

-

-

1

1

-

-

1

S.VII

-

25

-

-

10

-

-

7

-

-

-

-

-

1

-

1

-

-

1

The production sequence of deep casseroles hardly differed from the one used in North Tunisia. First, a cylindrical vessel body was formed by throwing it the right side up, whereupon the uppermost section of it was folded over, pressed against the exterior surface and tilted slightly outward to form a rim. After the vessel had dried enough, the rim was separated from the exterior wall with a cut and a lid seating was either pressed or cut on the interior surface. Often, the top section of the vessel exterior was also turned at this stage.318 After reaching the leather-hard stage, the vessel was inverted, re-mounted on the potter's wheel and the excess clay was turned off from the exterior bottom to produce a slightly sagging shape.319 Vessel surfaces were then smoothed, sometimes efficiently enough to produce burnished stripes and bands on the exterior. In spite of few vessels showing a reduced exterior surface similar to patina cenerognola deposition, Central Tunisian deep casseroles were generally fired in a fully oxidizing atmosphere. More often, the salt water used to prevent lime-blowing in a clay rich in calcareous inclusions resulted in a whitish scum on the exterior wall and the rim during firing.320 The absence of scum-covered base sherds together with more complete examples equipped with a sagging base from other sites confirm the identification of the form as a deep casserole rather than a cookpot. The idea about the Central Tunisian origin of this cookware form was first brought up by Hayes, while only a decade later van der Werff was ready to define eastern Byzacena as the specific area of production.321 Other scholars have been more conservative and localized the manufacture of this form to Central/South Tunisia or Tripolitania on the basis of its rare occurrence in Carthage.322 The recent finds made in several Central Tunisian and Tripolitanian kiln sites323 have lent further support to this idea. Another clear indication of the Central Tunisian origin is the Roman cemetery of Raqqada (Henchir el Ouiba), where the form was extensively utilized as a cremation urn.324 Otherwise, the normal use of Central Tunisian deep casseroles has very

unlikely differed from their North Tunisian parallels, especially because also this form has a sagging base and broad lid seating on the interior rim. Unfortunately, the abundance of deep casseroles on several sites in Central Tunisia has not resulted in a better knowledge of their typology, as all the abovementioned kiln sites have been assigned an approximate date between the late 2nd and the early 5th century AD. Furthermore, the variability of rim forms encountered at Raqqada is reported to have prevented further classification, although the material covers a substantially long chronological period between the 1st and the 4th century AD.325 Still, the shape of the rim is here used to distinguish seven variants (Table 26). For example, the variant S.II has a less pronounced lid seating and less sharply undercut thickening on the exterior rim than the variant S.I, its probable predecessor.326 In general, as the rim of the form S becomes more angular, elongated and outsplayed with time,327 the development is accompanied by minor changes in the fabric, which seems to get darker, harder and finer. Regarding the absolute chronology, the form (mainly variants S.I and S.II) has been found in abundance at Sabratha in deposits dating from the second half of the 2nd century AD onwards, although the possibility that some variants were already in production in the late 1st century AD cannot be excluded.328 The form is also characteristic of 3rd century AD deposits and its exportation seems to have continued at least to the late 4th century AD.329 The most common variant S.I (Plate 15:76) is characterized by an everted, slightly thickened rim with a pointed lip and a broad but shallow groove forming a lid seating on the interior. The rim diameter of the examples in the study assemblage is similar to the North Tunisian Hayes 197 deep casserole series. The production sequence used does not differ significantly from the one described in the general introduction of the form. The surface deposition on the vessel exterior can be divided into two common types: a) blackish patina cenerognola covered by a thin

66

Forms should be of a somewhat later date.332 The 3rd century AD introductory date proposed by Hayes333 is supported by contemporary material found in Ostia and Sabratha.334 This form has been identified both in “Byzacenan cookware” (form 12) and “Stripe-burnished ware” (form 10) at Uzita, where the end of the production is dated to the 3rd century AD.335 However, the clustering of the examples into phase 7 in the study assemblage, may indicate the continuity of production to the early 4th century AD. A strongly outward inclined rim with an undercut thickening on the exterior and rounded lip are the main characteristics of the fairly rare variant S.III (Plate 15:78), the presence of which has been documented at several other sites.336 A complete example published from Uzita shows that the rim belongs to a deep casserole.337 The dates assigned to contexts in which similar examples were found are very heterogeneous. So far the earliest occurrences have been reported in the 1st century AD,338 and the early 2nd century AD deposits,339 while exportation, even on a moderate scale, did not begin before the late 2nd/early 3rd century AD.340 The excavations of Uzita, where this form has been identified both in Byzacenan cookware (form 9) and stripe-burnished ware (form 4), suggests a date from the late 1st to the late 2nd century AD.341 At Luni, variant S.III has been found in contexts dated between the second half of the 3rd century and the early 5th century AD. Thus, the late 3rd century AD example in the study assemblage fits well into the prevalent chronology. A cluster of sherds including a

whitish scum, the combination of which is present in a third of the examples and, b) light to thick whitish scum cover. The evidence from the excavations of Uzita – where this form has been identified both in “Byzacenan cookware” (form 10) and “Stripe-burnished ware” (form 5) – suggests that the production had begun by the late 1st century AD, but the variant was not really exported before the late 2nd century AD.330 Thus, it is hardly a chance that the main period of the distribution, the 3rd century AD, coincides in part with the boom in the exportation of other Central Tunisian ceramic products, the African Red Slip ware C in particular. The distribution of the variant S.I in the study assemblage favors also the idea about the continuity of production at least to the late 3rd or even to the first half of the 4th century AD.331 With appropriate distribution chart (Figure 15) it can be observed that rim diameters of the most common Central Tunisian lids do not match particularly well with the variant S.I, although all the other forms except the Hayes 182 (C.I) may have served this purpose with equal probability. The rim of the second variant S.II (Plate 15:77) is straighter and more elongated than that of the preceding variant, while the lip is more often rounded than pointed. Otherwise, the evidence related to the vessel forming, firing and use repeats the pattern described above. Although the chronology of variant S.II has seldom been discussed profoundly, as it is regularly assimilated with the preceding variant, it

100 %

80 %

60 %

40 %

20 %

0% ø 16

18

20

22

C.I (Hayes 195)

24

26

28

30

C.II (Hayes 196)

32

34

C.III lid

36

38

C.IV lid

40

42

S.I (Hayes 183)

Figure 15. The cumulative rim diameter distribution of five Central Tunisian cookware form variants: C.I-IV lids and S.I deep casserole. 67

Chapter 3 Table 27. The size and the distribution of miscellaneous Central Tunisian cooking vessels. Diameter

T

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

avg

max

P6

P7

P8

Fabric

P9 P10 P11 CT

Total

NT

?

3

T.I

-

20

-

-

10

-

-

6

-

1

-

-

-

-

-

1

-

-

1

T.II

-

16

-

-

11/13

-

-

9

-

-

-

-

-

-

2

2

-

-

2

waster and indicating the presence of a kiln site involved with the production of deep casseroles, has been spotted in the area southeast of the amphitheater at Leptis Minor.342 As these finds confirm the Central Tunisian provenance of variant S.III, they also exclude its typological interdependence with the Hayes 197 deep casserole, which has also been suggested.343 The examples of the rim variant S.IV (Plate 15:79) with a rectangular thickening on the exterior and a flat rim top, form the most homogeneous unit in this group of Central Tunisian deep casseroles. Otherwise, they do not differ from other variants. With regard to vessel usage, examples with worn rim top and sooted exterior surface are somewhat more frequent than in other entries. The evidence from ager Tarraconensis, Luni and Sabratha indicates that the production of variant S.IV had begun by the 2nd century AD, whereupon it persisted until the late 4th/early 5th century AD.344 The proposed closing date coincides fairly well with the study assemblage, where variant S.IV is present in contexts dated from the mid-4th century AD onwards. Evidence from production sites is scarce, although the kiln site located near the Catacombs of Sullecthum may be reasonably suggested as one potential source.345 The variant S.V (Plate 15:80), in which a slightly convex upper wall ends in an everted, lozenge-shaped rim with angular lip and undercut thickening on the exterior, is very likely the same vessel, which Aguarod Otal has defined as the VilaRoma form 5.32.346 If so, the variant has a very long production history extending from the 1st to the 5th century AD. The earliest examples have been unearthed at Sabratha, though for the most part the parallels the site has produced pertain to the 2nd century AD or later levels.347 Hence, the example belonging to the late 3rd century AD phase 6 of the study assemblage nicely fills the gap between the Sabratha material and other parallels belonging to the late 4th/early the 5th century AD contexts.348 The variant S.VI (Plate 15:81) has a slightly convex wall turning through an angular transition to an everted rim with sharply undercut thickening on the exterior and somewhat angular lip. This variant, of which the study assemblage includes only one example, is dated from the late 4th to mid-5th century AD by all the occurrences,349 except the 3rd century AD material found at Sabratha.350 As the rim diameter of the Sabratha finds is markedly smaller than in the other examples, the possibility that they do not

belong to the same production at all must also be taken into consideration. The example of the variant S.VII (Plate 16:82) with slightly inward inclined, convex upper wall turning to an even, slightly everted rim bearing somewhat angular lip is unique in the study assemblage, but has some parallels elsewhere. On the basis of the finds made at Carthage, Fulford has dated the production between the 3rd and 4th centuries AD.351 It is also the date to be suggested for the example in the study assemblage, whereas another parallel from Carthage classified as the Late Roman cooking pot 3, comes from the late 4th/early 5th century AD context. 3.6.2 Miscellaneous Central Tunisian cooking vessels The two forms forming the group of miscellaneous Central Tunisian cooking vessels are related to one another only by their rarity in the study assemblage (Table 27). In fact, the fewness of these rim sherds confirms for its part the dominant position of North Tunisian cookware products in overseas markets, a phenomenon for which the reasons will be sought later on in this study. The form of the variant T.I (Plate 16:83), characterized by small everted rim with undercut thickening on the exterior and a spherical lip, vaguely resembles the form of the Central Tunisian deep casserole. Still, the example in question can hardly belong to that production, because it is too small and lacks the whitish scum on the exterior surface. An additional factor negating such identification is the absence of comparanda. Aquilué Abadías has published a somewhat similar form as Atlante CVIII.10 and dated it to AD 300-450/475.352 The other potential parallel, identified by Dore as a small jar or a beaker, has been dated to the second half of the 4th century AD,353 whereas the example in the study assemblage dates more likely to the late 3rd century AD. The identification of a small cookpot (variant T.II, Plate 16:84), characterized by a thickened rim with a diagonal cut forming the lid seating on the rim top is uncertain as no parallels to this form with smoothed surfaces and fully oxidized fabric, have been found. The rim diameter in both of the examples is 16 cm, the vessel surfaces are smoothed and the firing took place in a fully oxidizing atmosphere. Thus, the two examples in the phase 11 of the study assemblage date the form to the 5th century AD.

68

Forms

3.7 MISCELLANEOUS FORMS The last section of the find catalogue introduces an omnium-gatherum of African cookware vessel forms and parts, none of which is present in the study assemblage with more than a few examples. Cooking basins and pitchers certainly do not belong to the most usual repertoire, and only rare African cookware vessels were equipped with a handle or a disk base. Nevertheless, the presentation of these oddities is justified by the need to illustrate the variation in the African cookware production.354

3.7.2 Trefoil-mouthed pitchers The African cookware pitcher with a trefoil-shaped mouth (form V, Plate 16:87) has commonly been regarded as an unusual item of export, because closed vessel forms are believed to have been more vulnerable to breakage during transportation than the open ones.360 The exportation of open vessel forms, such as lids or shallow casseroles was generally preferred, as they were easy to stack on top of each other, saved cargo-space and thus provided more profit with reduced risks. But as indicated by numerous African commonware jugs found in increasing quantities from the late 2nd century AD onwards at Ostia,361 this was not necessarily the case with all closed forms. Imported cooking pitchers, on the other hand, are extremely rare. The examples in African cookware known thus far come from Central Tunisia, ager Tarraconensis and the Rome-Ostia region,362 to which the one badly preserved sherd in the study assemblage can now be added. In spite of its rarity, the African cookware pitcher has already been a subject of a special study. González Villaescusa introduces in his profound article two variants, which can be distinguished from one another on the basis of both form and fabric. In his opinion, the production of the pitcher variant Uzita 48.1, which the example from Palatine East has been identified as by rim shape, took place in Central Tunisia.363 The attribution of provenance, however, does not match with this example, the fabric of which is clearly North Tunisian. More likely the fabric is the yellowish-orange (creamy) fabric with sparse limestone inclusions, rare biotite and coarse-grained quartz described by González Villaescusa in the introduction of his example no. 3,364 in spite of minor differences in vessel form, such as the rapidly expanding neck. At present the chronology of African cookware pitcher is unclear. The form is said to have a possible predecessor in Punic cookware, which dates between the 4th and 2nd century BC, as well as 6th century AD successors in other Carthaginian fabrics.365 Still, in spite of the early appearances in the late 1st/early 2nd century AD contexts in Spain,366 the main period of production dates very likely from the late 2nd to the 3rd century AD.367 If not residual due to poor state of preservation, the late 3rd century AD example in the

3.7.1 Cooking basins Had the fabric of the cooking basin variant U.I (Plate 16:85) been slightly different and the vessel itself lacked the patina cenerognola cover on the upper exterior surface, its exclusion from the study assemblage would not have presented a problem. However, both the above-mentioned features as well as thinly sooted exterior surface identify it as African cookware. Although this open form with outward inclined, almost diagonal wall ending in a flanged rim with a pointed lip is previously unknown, the evidence points more towards an identification as a flanged cooking basin than a lid. A possibly related vessel, published as a shallow casserole/pan, was found at Sabratha in a deposit dated to the early 1st century AD.355 An alternative and perhaps closer parallel is the late Roman basin form 6 of the late 4th/early 5th century AD,356 as the example in the study assemblage belongs to the 5th/6th century AD phase 11 (Table 28). The identification is further supported by an example, which Leveau published as a commonware vessel without offering a further discussion on the form itself.357 At first, the other cooking basin variant U.II (Plate 16:86) appeared to be related to African cookware lids by fabric and surface finish, but the examples in study assemblage did not show a trace of soot or orlo annerito deposition on vessel surfaces. In fact, this peculiar commonware form is more often encountered in African Red Slip ware (Hayes 10B bowl), while the examples in cookware (coarseware) fabric represent the late exportation of the 4th century AD.358 The occurrence of this variant in late phases 10 and 11 of the study assemblage corresponds remarkably well in date to numerous examples found in the late 4th/early 5th century AD deposits at Carthage.359 Table 28. The size and the distribution of cooking basins. Diameter

U

Variant min

avg

Rim height

max min

avg

Rim width

max min

U.I

-

21

-

-

8

-

U.II

-

16

-

-

13

-

-

Distribution

Fabric

avg

max

P6

P7

P8

22

-

-

-

-

-

-

1

-

-

-

-

1

1

12/15

69

P9 P10 P11 CT

Total

NT

?

3

-

-

1

1

-

2

-

2

Chapter 3 Table 29. The size and the distribution of trefoil-mouthed pitchers. Diameter

V

Variant min V

-

avg

Rim height

max min

5

-

-

avg

12

Rim width

max min

-

-

Distribution

avg

max

P6

P7

P8

7

-

1

-

-

study assemblage (Table 29) fits well into this picture, while the longevity of production could be indicated by the two early 5th century AD parallels from Carthage.368 Notwithstanding its seemingly unusual upper part, the cooking pitcher has a slightly convex sagging bottom, implying that it was used on a tripod set over a fire. Cooking pitchers are hypothesized to have been used to warm up liquids, certainly water, but possibly also wine.369 It is known that in the Roman world wine was also drunk warm or even boiled to produce concentrated defrutum (must).370 The connection with water, however, is supported by evidence from ager Tarraconensis, Ebusus (Ibiza) and Jávea (Alicante), where all but one example found showed a thin layer of calcium carbonate deposited on the interior surface.371

Fabric

P9 P10 P11 CT

-

-

-

-

Total

NT

?

1

1

-

1

vertical variant. Vertical handles are known to belong to various African cookpot forms, most of which have been illustrated by Aguarod Otal together with a couple of African cooking jar forms (Hayes 131 & 200) and the African cooking pitcher introduced above.373 However, neither the loop handle of jars nor the broad, grooved pitcher handle resemble the Palatine East example very closely.374 Therefore, it probably belongs to one of the many cooking pot forms, although the fabric of this sherd differs slightly from the ordinary North Tunisian cookware fabric. The parent vessel of the horizontal handle is equally difficult to identify. One, although seemingly distant, possibility is that it belongs to the Hayes 194 casserole, which sometimes has a pair of horizontal handles attached underneath the rim (see 3.5.1). If identified as such, the example found in the 5th century AD horizon is certainly residual, as the production of the Hayes 194 deep casserole only lasted from the late 1st to the late 2nd/early 3rd century AD. For the time being, other potential identifications for this handle are practically non-existent.

3.7.3 Handles Unlike Aegean and West-Central Italian cookwares,372 only a few African cookware forms were equipped with handles (form W, Plate 16:88) attached separately to the vessel body. In fact, apart from lids, handle devices of any kind are extremely rare in this class and the vessels they go with are poorly known. Still, such forms and variants do exist, although their importance with respect to the total volume of African cookware production has been marginal. The scarcity of handles results very likely from the efforts to optimize both the vessel production and transportation. In a production sequence, a handle decreases the volume of production by increasing the time put into the fabrication of a single vessel, whereas the transportation of cookware vessels with protruding handles probably suffered from the same disadvantages as the exportation of closed vessel forms. In fact, many closed cookware forms have handles and this twofold impact was probably often enough to discourage their exportation. The study assemblage includes only two examples African cookware handles (Table 30), one vertical and one horizontal. Their orientation was determined from wheel-throwing marks, as both examples had luckily preserved a small portion of the wall attached to the handles. Only the horizontal one will be catalogued here, as it is clearly rarer than the

3.7.4 Bases The preceding discussion on forms and variants has hopefully succeeded to underline that bases, at least the ones maintaining the stability of the vessel, are fairly uncommon in African cookware. The only exception to this rule is the group formed by the Hayes 181 pan and the Hayes 26/181 pan-casserole, though – due to diameter generally exceeding 20 cm – their fragments are easily distinguished from the rest of the material. Conversely, the diameter of all the examples to be introduced here is less than 10 cm. Of the four examples included in the study assemblage, three base variants have been distinguished (Table 31): a flat base, a turned flat base and a steep disk base. Strangely, the first two variants do not seem to have any direct parallels, whereas the third example can be interpreted without hesitation as a base fragment of the Hayes 131 cooking cup. The fabric in all but one of the examples is clearly North Tunisian. The two examples of the variant X.I (Plate 16:89), a flat base joining with a strongly outward expanding lower body do not offer much room for

Table 30. The size and the distribution of handles. Diameter

W

Variant min W

-

avg -

Rim height

max min -

-

avg 9/15

Rim width

max min -

-

Distribution

avg

max

P6

P7

P8

18/23

-

-

-

-

70

Fabric

P9 P10 P11 CT 1

-

1

-

Total

NT

?

2

1

1

2

Forms Table 31. The size and the distribution of bases. Diameter

X

Variant min

avg

Rim height

max min

avg

Rim width

max min

Distribution

avg

max

P6

P7

P8

Fabric

P9 P10 P11 CT

Total

NT

?

4

X.I

-

4/7

-

-

-

-

-

-

-

-

1

-

1

-

-

-

1

1

2

X.II

-

5

-

-

6

-

-

-

-

-

-

-

-

-

1

-

1

-

1

X.III

-

4

-

-

13

-

-

7

-

-

-

1

-

-

-

-

1

-

1

drinking.375 Still, both the fabric and the presence of patina cenerognola deposition on the vessel exterior confirm the present classification, especially when the Hayes 131 cooking cup could have been used to heat liquids directly on hot coals or on a cooking stand set over a fire.376 Its first appearance has been dated to the second half of the 2nd century AD, whereupon the production continued at least through the 3rd century AD.377 The late 3rd century AD closing date is supported by the finds from ager Tarraconensis, where the introduction of the Hayes 131 may have already taken place in the Neronian era.378 Although the single example from phase 8 may not be enough to push the closing date to the early 4th century AD, together with comparanda deriving from different parts of the western Mediterranean, it will necessitate the re-evaluation of Hayes's statement, according to which this form rarely occurs outside Tunisia.379

discussion. Both have a smoothed exterior paired with an unfinished interior surface, which is also covered by faint traces of wheel ridging. In the absence of comparanda, their identification as African cookware is strengthened only by the presence of deep black patina cenerognola deposition and patches of soot on the exterior surface, while the form to which they belong to is uncertain. The same can be said of the variant X.II (Plate 16:90), a flat base with turned exterior surface, as one isolated example from the 5th/6th century AD phase 11 can hardly be regarded as reliable evidence. The hollowed disk base, which turns smoothly to strongly outward expanding, convex wall (variant X.III, Plate 16:91) is the only example of the Hayes 131 cooking cup in the study assemblage, as no rims pertaining to this form have been identified. This small closed form, which has occasionally been classified as tableware, was most likely used for

3.8 LONG-TERM DEVELOPMENTS 3.8.1 Chronology The detailed presentation of the study assemblage has raised several issues of general interest, some of which will be touched upon in this section. The first and most obvious of them is the nature of formation processes behind this African cookware omniumgatherum. As previously pointed out, the assemblage can be divided into two parts according to the number of examples in each entry. It is unsurprising that the main body of the material is formed by few cookware forms and variants: a lid (Hayes 195-196), a pan (Hayes 26/181 and 181), a shallow casserole (Hayes 23 series) and a deep casserole (Hayes 197 series), which are also found in abundance on other sites. The dominance of few forms has been said to indicate standardized production,380 but, on the other hand, the number of forms and variants represented in the Form / Phase

assemblage by less than 10 vessels is equally impressive. In spite of the fact that the dominance of certain forms in African cookware has long been acknowledged, attention has seldom, if ever, been paid to the possible changes in their ratio against chronological background. To examine this question, the study assemblage was divided into four major forms by excluding both irrelevant vessel parts, such as knobs and bases, and vessels of uncertain identification. Moreover, since attention was only paid to forms, the two productions – North and Central Tunisian – were in this case counted as one. The relative share of these forms in each Late Roman phase of the Palatine East excavations is the following (Table 32). The table raises some interesting points, the

P6

P7

P8

P9

P 10

P 11

Lids

45.1

47.5

51.2

51.5

49.4

55.7

Pans & pan-casseroles

8.0

10.4

12.5

10.9

19.5

19.9

Shallow casseroles

20.8

23.2

18.6

21.0

12.6

14.2

Deep casseroles

26.1

18.8

17.6

16.6

18.4

10.1

71

Table 32. The distribution of the principal African cookware forms (%)

Chapter 3

validity of which should be tested with similar material derived from elsewhere than the city of Rome. First, the only group that maintains its share in a relatively stable manner from the late 3rd to an indeterminate moment in the 5th/6th century AD consists of lids. At first glance, this would seem to support the idea that lids were equally used with all three principal vessel forms of African cookware. In closer examination, however, it becomes evident that the increase in the number of lids is accompanied by a similar increase in the number of African cookware pans and pan-casseroles. The decrease in the number of African casseroles is more apparent with deep than shallow forms. This may indicate that the production of deep casseroles ceased somewhat earlier than the production of the shallow ones. On the other hand, the correlation could equally be used to promulgate the idea that some forms, the Hayes 23B shallow and the Hayes 197 deep casserole in particular, were used as a set. The evidence to be introduced in section 3.9, including the discrepancy in the vessel size and the lack of the Hayes 197 in Central Tunisian fabric, makes such association very improbable. Moreover, the drop in the number of African shallow casseroles has a strong negative correlation with the introduction of similar form in West-Central Italian cookware (cf. Table 34). It remains to be examined, however, whether the introduction of imitations resulted in the disappearance of shallow casseroles or did they just occupy the open markets after the importation of cookwares from African provinces had ceased. Conversely, in spite of locally produced copies, the introduction of the Hayes 26/181 pan-casserole seems to have seen an increase in the exportation of African pans, especially from the late 4th century AD onwards. One potential explanation for the development is that African products gradually replaced Internal Red Slip ware pans, the production of which had ended in the late 3rd century AD. The increase in the amount of pans may also be connected with the decreasing amount of shallow casseroles, which together maintain a constant share of 29-35 % of all cookwares. As these vessels were used for different purposes, the mechanism behind is difficult to explain, unless it is related to changes in the nutrition of the average Roman (see section 3.9.2). 381 When attention is turned to forms and variants present in the study assemblage with less than 10 examples, at least five factors can be introduced to explain their scarce presence. Firstly, the production of some forms or variants was short-lived, which may show up as a cluster in chronological distribution. The most apparent examples of such forms are the large Hayes 26/181 pan-casserole variant H.III (late 4th/5th century AD) and the Late Roman deep casserole form P (early 4th century AD). In terms of the elaboration of vessel chronology, these are

naturally the most ideal cases, which can be used to establish rudimentary dates for deposits etc.382 Secondly, certain African cookware forms were either rarely exported or their exportation had ceased by the late 3rd century AD. The most obvious example of the former case is the African cookware pitcher, the known examples of which outside Tunisia number less than ten, whereas open forms of the same origin were customarily subjected to exportation on a large scale.383 The evidence concerning pottery production after the mid-6th century AD at Leptiminus is highly indicative in this respect. As the output was increasingly directed to the local consumption, the majority of new forms are restricted and equipped with handles.384 On the other hand, the cessation of the production may be the explanation for the small number of Central Tunisian deep casseroles in the study assemblage, even if the absolute volume of Central Tunisian cookware exports had always been inferior to the North Tunisian one (cf. section 5.1.2). Certain forms and variants of African cookware are also easier to misidentify than others. The most apparent pair, in this sense, consists of the Hayes 26/181 pan-casserole and the Hayes 193 deep casserole. In fragmentary condition, the rims of these two forms are practically identical to one another. Similar problems have been encountered with other African cookware pan and casserole forms.385 These difficulties can be overcome by simply paying more attention to the identification of the study material. It is also inevitable that some forms and variants in the study material are residual, as several forms introduced in the find catalogue were probably not produced after the late 2nd century AD. Beyond doubt, both the Hayes 194 and the proto-Hayes 197 deep casseroles should be identified as such. The identification of residual elements from among the examples of principal African cookware forms, like the Hayes 23B shallow or the Hayes 197 deep casserole, is notably more difficult. However, as evidenced by more readily identifiable material, even some of these finds are out of phase. The most likely candidates, in spite of the fairly high number of examples present in the study assemblage, are the first variants of every major form, in other words A.I (Hayes 196), J.I (Hayes 23B) and O.I (Hayes 197). The identification of intrusive components, which are obviously present in the study assemblage,386 is markedly more difficult in comparison to residual material, because the question is related to the closing date of African cookware production. In light of the Palatine East evidence, the commonly proposed closing date in the early 5th century AD is not entirely acceptable. At least the set formed by the late Hayes 195 lid (B.IV) and the late Hayes 26/181 pan-casserole (H.III) should be regarded as a later production. But even when found in the post mid-5th century contexts AD,387 these forms do not

72

Forms

necessarily witness continuing production, because African cookware vessels may have been exported for some time after the cessation of production. In addition, the use-life of robust and thick-walled vessels, like the forms mentioned above, may have been longer than that of many other cookpot forms. The sudden appearance of local copies of the Hayes 26/181 pan-casserole in the 5th century AD phase 11 (cf. Table 34), however, coincides perfectly with the introduction of these African cookware imitations in the hinterland of Pesaro.388 Thus, the possibility that the production, exportation or the use of certain African cookware forms persisted to the late 5th or even to the early 6th century AD cannot be excluded.389

Another feature related to the overall typology of African cookware is the increase in vessel size towards the end of the production.392 As this topic is examined by counting an average rim diameter for selected North Tunisian forms by chronological phase (Table 33), the growth is interestingly more evident in cases in which the vessel form is one of those that dominate the last exportation phase of African cookware. On the other hand, the pattern is not so clear with products of declining importance, like the Hayes 23B shallow casserole or the Hayes 196 lid. For the time being, no satisfactory explanation can be offered for this phenomenon, which is possibly related to the simplification of African cookware forms towards the end of the production. In any event, rather than witnessing the period of degeneration that is often thought to precede the cessation of any ceramic production,393 such pattern may indicate the adaptation of workshops to the changing demands of production and consumption. Still, more evidence is obviously needed to link this development with possible changes in demography or eating and serving habits. 394 The find catalogue also sheds some light on the similarities and differences between the cookware production of North and Central Tunisia. As previously pointed out, the production of certain forms is characteristic of both of these productions. Previously, the best-known examples of such were the Hayes forms 181-185, which are rarely found in North Tunisia. Now, forms like the Hayes 26/181 pancasserole and the Hayes 197 deep casserole can be added to this list of characteristic North Tunisian products. It has also been shown that many forms, of which the Hayes 23 shallow casserole series is probably the best example, were produced in both North and Central Tunisia. The question whether these productions are contemporaneous remains to be resolved, even if a part of the evidence hints at a North/Central/North shift, similar to the pattern observed in the exportation of African Red Slip ware. This does not imply that the production of cookwares in North Tunisia had ceased for an indeterminate period, especially when they outnumber Central Tunisian products (cf. section 5.1.2) in every horizon of the Palatine East excavations.

3.8.2 Typology Another aspect the preceding catalogue has hopefully managed to underline is the chronological significance of the changes in the typology of certain African cookware forms. Previously, common thought maintained that rather than evolving typologically, certain African cookware forms were from time to time replaced by new variants.390 Today, not only can the typology of individual forms and variants be outlined, but there are also features characteristic of the whole assemblage. A good example of such is the overall thickening of vessel rims, which can be observed in all the main groups excluding the Hayes 197 deep casserole. Thickened vessel rims may have been advantageous for several reasons. Firstly, due to the fairly simple forming sequence, in which enough clay is either retained in shaping or the utmost section of the clay is folded over to form the thickening, these rims are easier to adapt to mass production than everted or hooked ones. Due to fairly complicated shape, the forming of the Hayes 195 rim (B.I), for example, has certainly been more time-consuming in comparison to any of Hayes 196 variants (A.I-IV). Secondly, a thickening may also have been a convenient way to amend the impact resistance of the vessel rim against chipping and abrasion.391 Thus, the vessel was more likely to be exported as the probability of breakage during the transportation had been reduced.

Table 33. The average rim diameter of selected North Tunisian forms Form / Horizon

P6

P7

P8

P9

P10

P11

Hayes 196 lid (A.I-III)

25.21

25.70

26.92

26.63

26.75

26.60

Hayes 195 lid (B.I-IV)

30.13

31.95

29.93

28.80

28.14

32.28

Hayes 26/181 pan casserole (H.I-III)

26.88

28.38

29.29

29.47

27.53

31.90

Hayes 23B shallow casserole (J.I-IV)

24.64

25.60

24.46

25.90

25.67

25.46

Hayes 197 deep casserole (O.I-VIII)

23.25

24.93

24.89

24.38

24.08

25.53

Key: light shading= minimum, heavy shading= maximum

73

Chapter 3

Generally speaking, the existence of identical vessel forms is not the only similarity between African cookware and Red Slip ware, as both productions are characterized by open forms that were easy to stack inside one another for transportation. The production of the same form in two different fabrics, however, would have been a waste of resources, unless they were used differently.

The significance of imitations has been interpreted differently. Usually, their existence has been taken as an indication of the popularity of African cookware, which local potters tried to exploit with their own products of inferior quality.403 The direct implication of this interpretation is that exported cookwares were superior in some fashion, most likely due to their improved technical qualities.404 Hence, in the late 3rd/early 4th century AD African pottery was such a point of reference that it was well worth copying.405 The idea that the largescale importation of African pottery directed to the use of the upper class would have developed a taste and markets for low-cost copies, which were economically within the reach of anyone,406 cannot be accepted as long as there is no evidence regarding the market value of African cookware (see section 5.3.3). Due to differences both in the available raw materials and production methods, imitations of African cookware (and tableware) often lacked certain features that are characteristic of the originals.407 On the other hand, the importation of African cookware probably forced local potters to specialize in the production of few indigenous forms, of which the West-Central Italian flanged casserole is an illustrative example.408 Generally speaking, while the African form repertory is characterized by pans, the percentage of casseroles and ollae is somewhat higher in the WestCentral Italian production. This may result both from differences in the customs of acquisition and the output of pottery production centers.409 In addition, indigenous vessel forms may equally reflect locally available food resources and therefore they may also have functioned differently from imports.410 Therefore, it is difficult to comprehend the production of imitations as a sign of cultural osmosis in the western Mediterranean.411

3.8.3 The impact of imitations The steadily increasing presence of vessels produced in local fabrics with forms apparently seeking to reproduce the African table- and cookware vessels,395 is another aspect to be taken into account when the importance of pottery production that took place in Roman Africa is evaluated. When speaking of imitations, it should be pointed out first that besides the tradition established between the mid-2nd century BC and the age of Augustus by Carthaginian cookwares,396 the pottery production in Africa Proconsularis had been strongly influenced by vessels manufactured in Greece and West-Central Italy.397 With regard to cookwares, imitations of African cookware, as they should probably be defined, are present in small to moderate quantities on archaeological sites around the western Mediterranean from the mid-Imperial period onwards. At the latest their production began during the 2nd century AD.398 Most of the evidence published so far comes from Spain and France,399 while Sicily and Britain may be considered as geographic extremes of the production.400 The most imitated forms are usually the ones that were the most popular outside Roman Africa, such as the Hayes 181 pan and the Hayes 197 deep casserole,401 at least if their frequency is used as an index. But imitations of forms of moderate distribution, such as the trefoil-mouthed pitcher, are also known.402

Form / Phase

P6

P7

P8

P9

P10

P11

TOTAL

A. Hayes 196 lid

-

6

8

2

1

14

31

G. Hayes 181 pan

2

1

2

5

7

5

22

H. Hayes 26/181 pan-casserole

-

2

1

1

-

15

19

I. Hayes 23A shallow casserole

1

7

8

3

-

11

30

J. Hayes 23B shallow casserole

3

10

22

36

15

93

179

L. Hayes 193 deep casserole

-

2

3

2

-

2

9

N. Hayes 197 minor

9

8

9

5

2

1

34

O. Hayes 197 deep casserole

14

16

17

19

12

10

88

P. Late Roman deep casserole

-

1

1

1

-

-

3

S. Central Tunisian deep casserole (S.III)

1

-

1

-

-

-

2

V. Cooking pitcher

1

5

6

7

2

5

26

TOTAL

31

58

78

81

39

156

443

74

Table 34. WestCentral Italian imitations of African cookware

Forms

Pottery type / Phase

P6

P7

P8

P9

P10

P11

West-Central Italian cookware vessels

532

596

848

574

403

2575

5.6

9.7

9.2

13.9

9.7

6.1

471

411

609

251

94

310

5.3

12.3

11.3

24.2

29.3

33.4

proportion of imitations (%) African cookware vessels proportion of imitations (%)

A fair amount of kiln sites involved in the production of African cookware imitations have been located. The evidence from western Languedoc (South France) includes several 2nd/early 3rd century AD workshops, the production of which included imitations of the three most common and popular African cookware forms: the Hayes 23, 196 and 197.412 Both the Hayes 23 shallow casserole and the Hayes 197 deep casserole were also the most frequently copied forms produced by kiln sites located in the Roman province of Baetica, Spain.413 A good example of the impact these imitations could have had on local markets is the recently published mid2nd/mid-3rd century AD deposit from Colonia Patricia Corduba, where African cooking vessels account for 16% and their imitations 27% of all the fine- and cookware finds.414 This is a good case to presume that the introduction of local imitations had affected the market share of African cookware imports. To test the impact and the chronological relation of imitations of African cookware production, these vessels were identified from among 5528 examples of the West-Central Italian cookware included in the Late Roman ceramic assemblage of the Palatine East excavations. Although the data concerning these imitations is highly preliminary by nature (Tables 89),415 it deserves further comment. When the attention is first turned to the imitated vessel forms (Table 34), the most striking point is the very low number of cookware lids. To a certain extent, the absence may result from the difficulties of identifying West-Central lid forms as African copies, but alternative explanations can also be offered. Firstly, the reproduction of African cookware lids was perhaps useless, because in contrast to other cookware forms, the performance characteristics of lids did not

Table 35. The share of West-Central Italian imitations in the study assemblage

require much attention. Thus, as indicated by the study assemblage, it was more profitable to concentrate on producing copies of African cookware forms of recognized performance characteristics. Secondly, the absence of imitations may well indicate that the design of African cookware lids was not very user-friendly, as the ease of manufacture and transportation had forced the producers to exclude the addition of a proper knob. Still, had these forms been plates or bowls of distinguished performance characteristics (see section 3.9.1), their copies in local fabrics should be found in abundance with other imitated forms. By contrasting the quantity of imitations both to African cookware and West-Central Italian production as a whole (Table 35), two substantially diverse patterns emerge. The first point to be made is that in respect of African cookware the share of imitations grows steadily towards the end of the chronological sequence under examination.416 While these vessels play only a minor role in the late 3rd century AD phase 6, the same cannot be said of the situation in the 5th century AD. By then, a third of the material, which would have once been considered typical examples of African cookware, actually belongs to West-Central Italian cookware production. In regard to the total volume of West-Central Italian cookwares, however, imitations of African cookware clearly pass out of sight, in spite of slight increase in their quantity towards the 5th century AD. The question of, whether the strategy behind the success of these products was based on rapid conquest or slow occupation of the markets, is also of considerable interest. The latter alternative is perhaps more probable, given the inferior appearance and finish of imitations

3.9 VESSEL USE 3.9.1 Lids The present study is also designed to produce new information on the use of different African cookware forms as well as to identify functional sets, i.e. vessels used together for a certain purpose. Data for the examination of the former question was gathered by paying special attention to the use-alteration marks – both use-wear and soot – visible on vessel surfaces.

In spite of laborious efforts, the registration of usealteration marks with a standardized procedure that would be of use in data processing turned out to be quite impossible. In addition, the fragmentary nature of the study assemblage, in which most sherds are rims preserving only a small fraction of the upper wall, does not offer many possibilities for further analysis. As a result, the presence of the use-wear on

75

Chapter 3

the rim top has been taken as evidence on the use with a lid, while soot deposition implies a direct contact with fire. Although such observations are of very generic nature, they will be used in the following discussion to show that the group of vessels traditionally classified as ceramica ad orlo annerito consists of lids. The discussion concerning the function of this group in the archaeological literature can be justly characterized as puzzling. Recently these vessels have been regarded as either lids or plates, although most scholars have avoided taking a firm stand by using a diplomatic definition “lid-plates”. Lamboglia obviously assumed these vessels were plates, as his self-explanatory definition piatti a bordo annerito (i.e. plates with blackened rim) shows. Similarly, in Ostia I these vessels are classified as plates or soup-plates.417 Although this identification was generally accepted in the early 1970's, the possible double function of these forms was also brought up regularly418 and some scholars even expressed their disagreement by identifying them as lids rather than plates.419 In addition, these vessels have been claimed as baking covers.420 Therefore, a serious attempt will be made here to introduce new elements to the debate concerning the vessel function, which has so far provided a fertile ground for speculation and more or less acceptable hypotheses. Generally speaking, the assumptions on the use of this group have frequently been based on a single argument, referring either to presumed performance characteristics or other properties, which seemed to be of considerable importance at the time. Attempts to review the question in an all-inclusive manner, which is the main aim of this section, are rare, if not nonexistent. Hence, the discussion will begin with aspects related to the vessel shape and surfacing, after which the focus turns to the two principal forms of use-alteration: sooting patterns and attrition marks. Thereafter, the material will be reviewed in light of modern expectations of vessel performance characteristics.421 Each property will be evaluated independently against every proposed use of these vessels, (lid, plate, bowl and baking cover) and assigned one of the following values: positive (+), negative (-) or insignificant (o). The table summarizing the results is included in the final discussion (see Table 41).

Interior / Exterior

If the forms in this group were used as lids, their properties would suit the purpose perfectly with one exception. The handling of some forms has probably been difficult in the absence of a knob. Although some can be lifted from the ring-shaped protrusion, which is usually 3-6 mm high, the grip it offers is not particularly good. 422 Naturally, those who believe that the presence of protrusion excludes the vessel use as a lid have eagerly stressed this point,423 whereas arguments negating the automatic identification of such vessels as plates have also been expressed.424 The protruding rim alone, however, may have increased the surface area enough to improve the vessel controllability,425 while the probable association of ring-shaped protrusions with thickened angular rims belonging to the Hayes 196 form series (group A)426 gives additional support to this interpretation. In other major forms (groups B and C) the vessel rim is usually projecting and the protrusion has been replaced by convex upper exterior surface showing one to three shallow concentric grooves. Closer examination shows that these grooves were purposely incised to exterior surface after the vessel had been turned, if they were not produced in the turning process itself.427 While the projecting rim has been claimed to exclude the use of these vessels as plates,428 if used as a plate, a grooved convex bottom must have been unstable, if not unserviceable. On the same grounds, it can be doubted whether a ring-shaped protrusion, usually 7-9 cm in diameter, improved enough the vessel stability to permit its use as a plate.429 The idea that some of these forms would have served as baking covers is the most improbable one.430 A fair amount of locally produced baking covers, characterized by a high domed body and a protruding flange located on the vessel exterior only a few centimeters above the rim, have been identified from the late Republican and early Imperial contexts in West-Central Italy. None of the forms in this group, however, resemble these vessels even remotely, as they are too small and shallow and thus, do not offer sufficient volume needed for bread baking or cooking.431 It is equally dubious whether the Hayes 23B shallow casserole (form J), which has been pushed into the limelight as another potential African baking cover candidate (see section 3.9.3), was better suited for such purpose.

Untreated

Smoothed

Impermeable

TOTAL

Untreated

799

50

20

869

Smoothed

41

68

8

117

2

4

5

11

844

120

33

997

Impermeable TOTAL

76

Table 36. Surface finishes of African cookware lids

Forms

Rim area

–

X

X

X

–

–

X

–

Exterior surface

–

–

X

X

X

X

–

–

Interior surface

–

–

–

X

–

X

X

X

72,8 %

11.2 %

TOTAL

5,4 % 4,8 % 2,7 %

The examination of vessel surfacings, which is intended to provide an alternative data set concerning the use of these forms, has been simplified by grouping this material into three broad categories: 1) untreated: surfaces that received at maximum a haphazard smoothing after the forming of the vessel, 2) smoothed: the vessel surface was smoothed, but the procedure did not result in an impermeable layer blocking the absorption of fluids through the ceramic body, 3) impermeable: surfaces – like burnishing, smoothing with self-slip, slip or their combination – significantly reducing the permeability of vessel surface. Their cross-tabulation by vessel side (Table 36) leads to the following observations. Firstly, as only 1% of the examples in the study assemblage have an impermeable interior surface, these vessels were hardly intended to be used as plates or bowls.432 An impermeable interior surface stops or at least slows significantly the rate in which the vessel contents are absorbed to its walls. This is highly important, as residues in ceramic body may affect the vessel performance and alter the flavor of freshly cooked meals. Had this form been a baking cover, the distribution of surface treatments should roughly follow the pattern of ordinary cooking vessels, due to similarity of required vessel performance characteristics. This is not the case, however, as the study assemblage does not include any examples with a corrugated exterior surface, which has been shown to increase cooking utensils' crack propagation resistance to repeated thermal shocks.433 Besides, if improved heat retention capacity had been one of the desired properties, the interior surface would have been made impermeable also in this case.434 As the study assemblage contradicts also this possibility, it is safe to conclude that on the basis of surfacing, these vessels were meant to be used as lids. Although we lack archaeological experiments focusing on the ideal performance characteristics of lids, their characteristics very unlikely differed from

1,2 % 1,2 % 0,7 %

- soot absent; X soot present

Table 37. Combinations of soot distribution on African cookware lids (N=997)

the properties expected from their modern counterparts. Today, a good earthenware lid must a) fit properly, b) be proportionally related to a vessel that carries it, c) create an impression that it belongs together with the piece it is topping, and d) also fulfil its function. Therefore, in a well-performing twopieced cooking set the food is constantly basted by condensation dripping off the inside of the lid back into the casserole.435 Although the fit could also have been achieved through careful measurements, unpleasant surprises resulting from the differential drying shrinkage between casseroles and lids was avoided by applying the same clay paste to the production of both African cookware forms. In addition to this practical constraint, the choice of the same clay was very likely influenced by aesthetics. As the general commonness of soot deposition in this group (Table 37) is examined, in most cases(84 %) either the vessel is not sooted or soot deposition is detected only on the rim area. The latter pattern, where the distribution of soot is limited to the outer edge of the rim is characteristic of lids, of which this was the only vessel part exposed to flames and fumes when used on top of a cooking vessel. If these forms were baking covers, one would expect to find vessels with sooted exterior surface in abundance, because in use they would have been covered with hot coals.436 On the other hand, the prevalence of examples showing no evidence of soot deposition supports the identification of these vessels as plates or bowls, but it also requires an explanation for the mechanism resulting in a soot cover on some of the rims. The fact that in most cases soot is associated with vessel use is reflected in the study assemblage by the rarity of such examples, where the presence of this deposition on fracture surfaces points strongly at its post-depositional origin. The possibility that the intensity of soot deposition could provide additional information on the vessel use was also considered. The intensity of soot cover between vessel surfaces is not overly anomalous (Table 38), though the homogeneity may partially

Zone / Density

Absent

Sparse

Patchy

Moderate

Dense

Rim area

77.5 %

9.2 %

7.6 %

4.4 %

1.3 %

Exterior surface

85.7 %

6.2 %

5.2 %

2.2 %

0.7 %

Interior surface

91.7 %

3.5 %

3.1 %

1.3 %

0.3 %

77

Table 38. The intensity of soot on different surface areas

Chapter 3

Wear-type

No wear

1

2

3

4

5

TOTAL

72.3 %

8.7 %

16.9 %

0.7 %

0.9 %

0.4 %

result from the cleansing of the study material with water and brushes. The possibility that some combinations in the intensity of soot deposition would correlate with certain vessel forms or variants was also tested with negative results. In all, with the major concentrations on the rim area, the soot deposition evidence suggests that these objects were predominantly used as lids. The interpretation is further supported by a frequent correlation between the extent of carbon deposition and blackened orlo annerito surface. This correlation possibly reflects the endeavors of the producers to increase the attractiveness of their cooking vessels in the eyes of consumers with products, where blackened surface likely masked some of the soot deposited on vessel in use. The presence and the distribution of use-wear on vessel surfaces were also examined, but chipping was the only type of use-wear identified. In many cases, minute round chips had been removed from substantially large areas of vessel surface (Plate 2b) and on the basis of the extent and the intensity of these abraded areas, six use-wear patterns were distinguished (Figure 16, Table 39). The first observation to be made is that ca. 72% of the rims do not show use-wear of any kind, while in the rest of the cases the distribution of abraded areas is highly indicative of the vessel's function. In general, areas showing most often traces of use-wear are the rim edge and the interior surface implying that they likely belonged to the lower part of the vessel. Hence, the use of these vessels as plates can be excluded once more, especially when none of the ringshaped protrusions (cf. form F) in the study assemblage show a trace of abraded protrusion edges,

Table 39. The frequency of different types of usewear (N=997)

the examples of which should be plentiful in the opposite case. In a closer examination some interesting patterns emerge from the distribution of use-wear. Contrary to expectations, increased presence of use-wear does not relate to vessels within a certain size range, but with certain variants. Half of the vessels in Hayes 182/195 series (variants B.I and C.I) show traces of use-wear, whereas the percentage of abraded examples in other lid variants is notably lower (Table 40). One explanation for such an anomalous distribution – besides apparent differences in sample size – might be that use-wear is often associated with variants without a knob or ring-shaped protrusion crowning the upper exterior surface. Hence, the variants of the Hayes 182/195 series were possibly more difficult to place on the top of a casserole than variants equipped with a handle device of some kind. The difference in the share of worn examples – likely bearing witness to the toughness of Central Tunisian fabric (see sections 2.4-5) – between the two productions of the Hayes 196 standard lid (variants A.II and C.II) is also worth noticing. As the evidence presented in previous paragraphs is summarized (Table 41), it becomes evident that the vessels of this group functioned most likely as lids. In terms of various parameters, including the vessel performance, alternative uses presented for this group in archaeological literature are evidently of secondary importance. Of course, the possibility that some lid variants may have had a secondary use as plates, bowls or even as baking covers cannot be excluded. Even so, these vessels were designed to fulfil their function as lids; and it is the primary function that matters in the search of appropriate description for a

Wear type Variant

N=

1

2

3

4

5

Wear %

A.I

191

15

17

-

-

-

16.8 %

A.II

284

30

52

6

2

1

32.0 %

A.III

99

12

18

1

2

-

33.3 %

B.I

95

12

34

-

2

2

52.6 %

B.III

37

-

5

-

-

1

16.2 %

B.IV

33

3

6

-

1

-

30.3 %

C.I

19

3

6

-

-

-

47.4 %

C.II

106

4

9

-

-

-

12.3 %

C.III

39

4

4

-

-

-

20.5 %

C.IV

53

2

12

-

-

-

26.4 %

78

Table 40. The frequency of use-wear on the most common lid variants

Forms

1.

2.

3.

4.

additional use as serving dishes, especially when it came to dishes prepared with egg that tended to stuck in cooking vessel.438 In this respect, it is especially attempting to see the carefully executed Hayes 181 pan and the Hayes 182 lid as a cooking set439 that had a secondary use on the table. The use with a lid is equally suggested by the distribution of occasional traces of use-wear, which in these pans are most often limited to the top of the rim.440 However, there is no consensus on the mode of cooking or on the foodstuffs processed with the Hayes 181 pan. Like many other scholars, Bats and González Villaescusa have proposed that le plàt a four or cazuela o fuente de horno was intended to be used in an oven,441 and the interpretation is supported by the low number of sooted examples in the study assemblage. The distribution of sooted patches in the rest of the cases shows that the Hayes 181 pan was at least occasionally used over a fire.442 This is also the conclusion reached by Rivet,443 who sees the preparation of cakes, pies and omelets as well as the cooking of grain, lentils and beans as suitable uses for this pan form. In addition, the Hayes 181 pan is also said to have been utilized in bread baking.444 The use of the Hayes 26/181 pan-casseroles did not differ much from the use of Hayes 181 pans as far as the use-alteration evidence is concerned. The study material does not include any examples with a sooted exterior surface and traces of use-wear are also substantially rare. Because the distribution of attrition marks is strictly limited to the topmost part of the vessel rim, small pan-casseroles were evidently used with a lid, just like the other components of the Hayes 26/181 series. However, some examples of the North Tunisian pan-casserole have a sagging bottom, which seems to become even more frequent towards the end of the production.445 Therefore, they could have also been used on a cooking stand instead of being utilized exclusively in oven cooking. In addition to the strong presence of examples with a sagging base, the use alteration evidence reveals a customary use over a fire. About 20% of the material show patchy soot deposition on the exterior surface, which cannot result from post-depositional factors. Although the use-wear is often limited to the utmost part of the lip, in many examples the chipped surface extends either to the upper exterior or interior surface. The most likely explanation for a wear pattern

= slightly worn = medium wear = heavily worn

5.

Figure 16. Types of use-wear detected from African cookware lids. given vessel form. Otherwise, we are obliged to fill the pages of archaeological site reports with obscure and complex definitions, of which the definition pancasserole (see section 3.3.2) is only a mild example, taking into account every possible use a vessel might have had. Hence, it is hoped that the use of definitions like “lid-bowl” and “plate-lid” (piatto-coperchio, platotapadera) will cease in the future, just like it happened with the definition ceramica ad orlo annerito nearly two decades ago. 3.9.2 Pans From a technological point of view, the most characteristic feature of pans and pan-casseroles is the finish given to the interior surface. In many cases the vessel has first been either slipped or wet-smoothed and thereafter burnished.437 Both combinations were evidently designed to improve the vessel performance, as the resulting anti-stick surface prevented cooked substances from adhering to vessel interior. There are also indices that some cookware forms had an

Form

Rim

Body

Base

Fabric

Finish

Use

Occur.

Plate

-

+

o

-

-

o

-

Bowl

o

-

o

-

-

o

-

Lid

+

+

-

+

o

+

+

Baking cover

-

-

-

o

o

-

-

79

Evaluation of the effect - negative o indifferent or unknown + positive

Table 41. The use of groups A-D as indicated by various parameters

Chapter 3

of this kind is a lid, either the Hayes 195 or 196 (cf. A.III & B.I) used to cover these vessels. Because wear-marks do not form patterns on the interior surface, where they are present occasionally, their origin is hard to determine. The risk of abrasion, for example, is high both when the contents are stirred and when the vessel is washed in the end of the usecycle. Washing is also a potential explanation for the external wear, although similar marks may have been produced by a cooking stand set underneath the vessel. Although, there are fewer sooted examples than in the group of medium-sized pan-casseroles, this hardly signifies the different usage of large pancasseroles, because the distribution of use-attrition marks is very similar to the previous group. Instead, the increased occurrence of abraded lips indicates that also large pan-casseroles were used with a lid. The corresponding lid form is most likely the late Hayes 195 variant B.IV, which is not only contemporaneous, but also the sole North Tunisian cookware lid reaching frequently the same dimensions with large pan-casseroles (see Figure 10). The increase in the relative amount of pans and pan-casseroles at the expense of shallow casseroles (see section 3.8.1) may reflect the transition from puls to bread, as the Hayes 23B shallow casserole is the most likely candidate for pultarius, the casserole used in the cooking of puls.446 The grinding methods were significantly improved during the Imperial period due to introduction of large mills, which improved the quality of flour and decreased the importance of home baking.447 Instead of nourishing themselves with puls or home-baked bread (panis artopticus), people were increasingly consuming bread prepared in numerous bakeries of Rome. Puls started to lose ground to bread in the first half of the 2nd century AD, but grain was replaced by bread among the provisions distributed for nothing to the poor of Rome only under the reign of Aurelian.448

without a lid. Openness guaranteed immediate access to the cooked substance, which was most likely puls, the ancient equivalent of pap.450 Hence, the contents were easy to stir with a spoon or other table cutlery and clotting was avoided with the rapid addition of water. The study assemblage partially supports this idea. In the case of Hayes 23A, traces of use-wear are both too scarce and haphazardly distributed to witness a customary use with a lid, but the evidence concerning the Hayes 23B leaves enough room for a different conclusion. Nevertheless, in an article on the baking covers of Roman Italy, Roberts has pointed out another possible function for the Hayes 23B shallow casserole identifying it as a baking cover instead of a shallow casserole.451 The following arguments are said to sustain the interpretation: 1) the protruding flange at the junction of the upper wall and the vessel bottom is similar to flanges frequently found in baking covers, the function of which was to retain hot ashes on top of the vessel together with corrugated dome (i.e. bottom); 2) the rim and the upper exterior surface are frequently blackened in a way that does not seem to result from uneven firing conditions; 3) there is not enough evidence of a lid variant that would match with the Hayes 23B shallow casserole; 4) the form lacks a lid seating, which is characteristic of many other African cookware casseroles; 5) the fabric in the Hayes 23 series is somewhat coarser than the usual African Red Slip ware -fabric. To my mind, this argumentation is not acceptable. Firstly, flanged protrusions of the Hayes 23B series are very shallow, rarely measuring more than 5 mm in width, for example in comparison to 23 cm wide flanges of late Roman baking covers from Matrice and Molise.452 Other than giving strength to the weakest part of the vessel, the protrusion of the Hayes 23B shallow casserole was more likely designed to facilitate stacking both in the kiln and the ship's cargo hold, as well as to increase the stability of the vessel on cooking stand. The forthcoming discussion on vessel manufacturing and firing techniques (sections 4.1-3) will demonstrate both that the blackening of the vessel exterior chiefly results from firing conditions and that the corrugated exterior bottom was essential in order to improve the thermal shock resistance of cooking vessels. As previously indicated (cf. variant A.II), certain variants of the Hayes 196 lid were most likely used together with the Hayes 23B. The absence of a lid seating is not exceptional either, as shown by African cookware pans and pan-casseroles (cf. section 3.3), which were certainly used with a lid. Finally, the argument putting weight on the coarse fabric is somewhat pointless, as it results from the old convention to classify the Hayes 23 shallow casseroles as African Red Slip ware, thanks to their

3.9.3 Casseroles The considerable variability included in the group of casseroles may result in part from differences in the vessel use, which has been utilized as a criterion to divide the material into four groups.449 The first group includes wide-mouthed casseroles with rim showing a lid seating at the inner edge. When used, the casserole /lid combination produces a vaporous atmosphere that was essential for the cooking of certain foodstuffs, like fish. In the preceding discussion, the forms certainly belonging to Aguarod Otal's first group are Hayes 194 (K.I) and late Roman deep casserole (P.I), but as they form in practice the chronological extremes of the Hayes 197 series, this identification has also been revalued under appropriate entries. The Hayes 23A and the Hayes 23B shallow casseroles form together the second group, characterized as open casserole forms presumably used

80

Forms slip-covered interior surfaces.453 In all, there is no evidence to identify this form as a baking cover. The common denominator for the forms of the third group is high capacity, which is believed to indicate stewing of meat and/or vegetables in abundant liquid.454 Also these forms, only represented in the study assemblage by the Hayes 199 series (form Q), have regularly a pronounced lid seating at the inner edge of the rim. The majority of vessels in the fourth group, which comprises medium-sized deep casseroles, are equally equipped with a lid seating. The most dominant form is undoubtedly the Hayes 197 (form O), which is one of the three – the two others being the Hayes 196 lid and the Hayes 23B shallow casserole – most widely distributed forms of African cookware in the western Mediterranean. The vessels of the fourth group are thought to have been used to cook foodstuffs like sauces and mushes that required occasional stirring. Alternative uses suggested for the fourth group include light-frying and oven-cooking,455 of which the latter is most likely based on the apparently false impression that deep casseroles seldom show soot deposition on the vessel surfaces. However, especially the question about the use of the Hayes 197 minor (form N) is hard to answer unambiguously. Firstly, the fact that a vast majority of the examples are equipped with a corrugated exterior bottom implies the form was meant to be used directly over a fire, even though the study assemblage includes only two sooted examples. González Villaescusa has calculated the volumetric difference between the Hayes 197 minor, which he defines as a small casserole (cazuela pequeña), and the full-sized Hayes 197 deep casserole. While the capacity of the former hardly ever exceeds one liter, the latter can hold as much as 5-7 liters.456 A variation of such magnitude implies that these forms were certainly used for different purposes. The modest capacity of the Hayes 197 minor suggests that it was the ancient equivalent to the modern saucepan. In contrast to the Hayes 197 minor, the full-sized Hayes 197 may have been used to cook coarsely chopped meat or/and vegetables in an abundant amount of liquid.457 The aspect to be reconsidered in this context is the dominance of unfinished, rilled interior walls, which must have increased evaporative cooling by absorbing easily any liquid. For a cooking vessel this is a highly undesirable property and it must be wondered whether the rilling had a function of some kind, because otherwise it would have only promoted the sticking of the contents to the interior surface. For example, the material found in Ebusus was reported to have born carbonaceous encrustations on the interior surface, while a more complete example of Hayes 197 was found full of snail shells, though it is uncertain whether the latter remains are related to the vessel use or to its post-depositional history.458

3.9.4 Functional sets It is commonly thought that objects found in the same archaeological context were related to one another also in life assemblage. The assumption explains partially the popularity of quantitative analyses in archaeology, the purpose of which is often to identify functionally interrelated objects. This is usually done by comparing the vessel counts of different deposits, and assuming that the number of functionally interrelated forms should roughly correspond to one another. On the other hand, “significant” differences in the distribution of possibly related objects have occasionally led to alternative and fairly radical conclusions. Tomber, for example, based the identification of African cookware lids as baking covers on the late Roman deposit excavated near the Circus of Carthage, where their share of African cookware rims was ca. 61% by count and ca. 67% by weight. This evidence was enough to confirm that “the form had an additional function apart from its implied one.”459 In my opinion, the lid/casserole relationship is by no means straightforward, because refuse disposal is dictated by behavioral patterns. Thus, we can ask if the Romans threw away their unbroken lids with broken casseroles or vice versa? In other words, can we reasonably assume that casseroles and lids were broken and discarded at the same rate? Arguments both supporting and denying the relationship may be posed, but only one of them deserves to be mentioned here. If African cookware lids usually reached the consumer as a part of a set completed either by a pan or a casserole,460 at least the number of these forms in the life assemblage roughly corresponded to one another. Turning the attention to the study assemblage, the pattern emerging from diachronic comparisons of the quantities between this group and a group combining African cookware pans and casseroles is interesting (Table 42). As their number in each horizon is strikingly similar, the idea that sherd counts could actually be used to point out functionally interrelated vessel forms is hard to avoid.461 Another aspect deserving particular attention in this context, is the contemporaneous typological change of the Hayes 196 lid and the Hayes 197 deep casserole series. 462 As the size of lid rims increases considerably, the flanged deep casserole rim gets compressed against the exterior wall until it is reduced to hardly anything but a shallow thickening with a furrow that likely acted as a lid seating on the top. 463 This link may be an example of the methods by which the potters of Roman Africa sought to improve their products. As the width of the Hayes 197 rim was reduced, the usefulness of a “cooking set” was maintained by increasing the rim size of corresponding lid form. This was necessary to avoid problems that might have arisen from the combination of a lightweight lid topping practically a rimless casserole.

81

Chapter 3

Phase

Lids

Pans & casseroles

TOTAL

% of lids

Phase 6

197

273

470

41.9

Phase 7

182

228

410

44.4

Phase 8

294

314

610

48.2

Phase 9

118

131

249

47.4

Phase 10

42

51

93

45.2

Phase 11

164

142

306

53.4

TOTAL

997

1139

2136

46.7

Because these modifications likely facilitated both the manufacturing process and vessel transportation, the development may also be comprehended as an improvement in production, because in the long run it increased the output of workshops manufacturing African cookware. The question of co-occurrence can also be examined with evidence provided by Roman shipwrecks, of which the most significant example in this case is the early 3rd century AD merchantman of Trincere, located near Tarquinia on the coast of WestCentral Italy. The main body of its cargo consisted of African merchandise, pottery in particular. As altogether three fourths of slightly over a hundred sherds, were identified as African cookware, the high percentage is said to indicate that these vessels, including one lid and two casserole forms,464 were on board as a component of the cargo rather than as utensils belonging to the ship's crew.465 The number of lids and casseroles, 40:39, seems to suggest that the assemblage is formed two cooking sets, both including a lid and either Hayes 23B or Hayes 197 casserole. However, by appealing to difference of 4.93 cm in the average rim diameter of lids and shallow casseroles in Trincere wreck and by highlighting the absence of stripe-burnished surfaces in the former group, Pontacolone and Incitti managed to identify the three forms as a three-piece cooking set. Hence, the Hayes 23B shallow casserole would have served as the upper part of a chafing dish or a double boiler that was placed on the top of the Hayes 197 deep casserole and covered with a lid of the Hayes 196 series.466 To my mind this proposition, based on minor differences in the rim diameter of these three forms, is not valid for the following reasons. The use of the Hayes 23B shallow casserole in direct contact with fire is confirmed by numerous examples preserving a part of the vessel base, which

Table 42. The ratio of African cooking vessels and their lids

shows irregular sooting on corrugated exterior surface.467 The corrugated surface was designed to enhance the vessel performance in a same way that smoothing or slipping of vessel interior decreased the likelihood that food would adhere to the surface and prevented the absorption of fluids into the body of the vessel (see section 4.2). The rarity of Central Tunisian Hayes 197 deep casseroles in the study assemblage –only 12 examples in contrast to 254 examples in North Tunisian fabric, and 115 examples of the Hayes 23B in Central Tunisian fabric – is also hard to explain if these forms were produced, marketed and used together as a set with the Hayes 196 lid. It is also necessary to refer here to superfluous number of lids on board that would require explanation, while the evidence on Roman shipwrecks shows that the Trincere wreck is not by any means exceptional. The combination of these three African cookware forms has been found in many mid- or late Imperial shipwrecks (see section 5.2.4, Table 51). To conclude, the search of functionally related forms, which was essentially based on three parameters – namely frequency, fabric and vessel diameter – produced some interesting results. Some new links between African cooking vessels and lids, of which the combination of the Hayes 26/181 pancasserole (H.III) and the late version of the Hayes 195 lid (B.IV) is perhaps the most apparent one, were identified. The study has also shown that some forms that have long been grasped as functional sets, like the Hayes 196 lid and the Hayes 197 deep casserole, do not automatically belong together. There is also enough evidence to suggest that certain forms thought to have been used uncovered, such as the Hayes 23B shallow casserole, were more often than not topped with a lid. And, finally, the argument about the threepieced cooking set formed by the Hayes 196 lid, the Hayes 23B shallow casserole and the Hayes 197 deep casserole does not withstand critical evaluation.

82

Forms

CHAPTER 3 ENDNOTES 1

2

3 4 5 6

7

8

9

10 11 12 13

14

15 16 17 18 19

20

21

22 23

24 25 26 27 28

29 30

31

32 33 34 35 36 37 38 39 40

The discussion concerning the vessel function will be based on observations concerning the objects themselves, while studies approaching the topic through the interpretation of literary sources can be found elsewhere, e.g. Mazzeo Saracino et al. 1997, 205-209. Unsurprisingly, literary sources hint that a given cookpot could have been used not only to prepare various dishes but also for various purposes. The use of cumulative rim diameter distribution curves was necessitated by the need to equalize the bias resulting from the unequal number of finds. The starting hypothesis was that in case of functionally interrelated forms these distributions should correlate instead of reflecting form/diameter interaction (cf. Orton 1970, 347). Previously, a similar method has been explored in the study of African cookware by Gonzales Villaescusa (1990, 109-111, figs. 917). Peña 1999, 60; see also Gibbins 2001, 316. Orton et al. 1993, 73, 79. Hayes 1972. The decision to limit the number of examples was intentional as superfluous catalogues do not necessarily provide any additional value to the study, see e.g. Fernández Sotelo 1994. The term “turning” refers to a method, in which excess clay is removed from a rotating vessel, while “trimming” i s the same operation performed by hand with a stationary vessel, see Rye 1981, 74-88. van der Werff 1982, 121. In spite of later attempts (e.g. Leveau 1984, 459-460, fig. 246), the classification presented in the pages of the Late Roman Pottery is still the most elaborate one and will be referred to whenever possible throughout this study. The most recent classification (Aguarod Otal 1991, 241, 245-252) relies on the presence or absence of knob, for which the method has lately been criticized (Aquilué Abadías 1995, 66-67). Lamboglia 1950, 203. As mentioned earlier, the definition seeks to describe blackened rim zone, which resulted from the way vessels were stacked on top of one another in the kiln and the changes of atmosphere during the last stage of firing (Gandolfi 1994, 149). Dore 1989, 88. Fulford 1994, 64. Aguarod Otal 1991, 247-251. A similar approach has been used by Aquilué Abadías (1991, 972), who set the limit of rim width to 10 mm in order to distinguish the examples of Ostia III.332 from the lid variant Ostia I:261. Ostia I, 87; Ostia III, 419-420; Niemeyer (1968, 80) has opposed this view for its oversimplifying approach and stresses the variability included in this group, whereas Tomber (1986, 43) recognized only two variants instead of three. Tortorella 1981b, 210; Aquilué Abadías 1991, 973; Freed 1998, 21. A.I: 47/191= 24.6%; A.II: 41/390= 10.5%; A.III: 6/99= 6.1%. Freed 1998, 21, 31. Ostia III, 419-420; González Villaescusa 1990, 37, 52. Aquilué Abadías 1985, 210; Aguarod Otal 1991, 245-246; Aquilué Abadías 1995, 67; see also Aquilué Abadías 1987, 185. Italy: Zevi & Pohl 1970 (Ostia), 92; Dyson 1976, 126, no. 74 (Cosa); Olcese 1993, 136 (Ventimiglia). France: Rivet 1996, 336. Ostia II, 83; Ostia IV, 347-348; Hayes 1976, 49, no. 1; 59, no. 7; Carandini 1977, 24; Michelucci 1985, 99, nos. 948, 950; Fulford 1994, 64. Aquilué Abadías 1985, 210; 1995, 67; see also Aquilué Abadías 1987, 185; González Villaescusa 1990, 52. Luni II, 384; Ostia III, 419-420; Tomber 1986, 40-41, tabb. 1-3, Black-top 196 variants 1-2. The excavation of the 3rd century AD deposit at Ceuta yielded some 160 examples of the Hayes 196 (Fernández Sotelo 1994, figg. 22-30), only a few of which are classifiable as variant A.I. Olcese 1993, 249. Fulford 1984b, 197, no. 1. González Villaescusa 1990, 52. Dyson 1976, 126, no. 74. Ostia I, 86; Ostia III, 419-420; Hayes 1976, 62, no. 15; 1978, 38; Aquilué Abadías 1985, 211; 1989, 190; Fulford 1994, 52, 64. Luni II, 181. All the following dates have been suggested on the basis of archaeological evidence. Early 2nd century AD: Hayes 1976, 72, no.8; Marín Jordá 1995, 157, 162; Mid-2nd century AD: Fentress et al. 1983, 80-95; González Villaescusa 1990, 52; Hayes 1972, 208-209; Late 2nd century AD: Carandini 1970, 109. González Villaescusa 1990, 37-38; Pontacolone & Incitti 1991, 546; Bost et al. 1992, 182-183; Giannotta 1992, 57, no. 262; Olcese 1993, 136; Fernández Sotelo 1994, figg. 22-30. Hayes 1972, 208-209; Fentress et al. 1983, 80-95; Marín Jordá 1995, 157, 162. Hayes 1976, 68; Tomber 1989, 486, nos. 268-269. Fulford 1984b, 191, no. 11; Fulford 1994, 52, 64. Arthur 1987, 523. Aguarod Otal 1991, 247-248. Carandini 1970, 109. Hayes 1980, 518; Freed 1998, 30-31. Hayes 1972, 18, 208-209; Pontacolone & Incitti 1991, 547. Ostia I, 86.

83

Chapter 3

41

42

43 44 45

46 47

48 49

50 51 52 53 54 55 56

57 58 59

60 61

62 63 64

65 66 67 68 69 70 71 72

73 74 75

76

77 78 79

80 81

82 83

Therrien (1981, 103) has defined the variant A.III as Hayes 196b, but due to identification of three variants in this study, the Hayes 196c is an alternate definition used occasionally to signify this variant. At least the possibility that this custom would have been transferred from the production of similar tableware forms, like Hayes 99 and Hayes 104, can be excluded as they are dated from the early 6th century AD onwards, see Hayes 1972, 152-166. Aquilué Abadías 1989, 192; 1995, 67. Luni II, 183, 384; Tomber 1986, 44, no. 111; Nolla & Casas 1990, 197-199. Tortorella 1981b, 212; Villedieu 1984, 136; Aquilué Abadías 1989, 192; Dore 1989, 154; Giannotta 1992, 57-59, nos. 265-267; Olcese 1993, 249; Aquilué Abadías 1995, 67; Macias et al. 1997, 164 (Ostia I-261). Ostia IV, 104. González Villaescusa 1990, 52, 70. The identification of these examples is my own and in fact contradicts the statement given by González Villaescusa (p. 74) who stresses the absence of the Ostia IV forms, fig. 60 and 61 in his material. Ostia III, 419. Ostia IV, 104; Settefinestre II, 118; Hayes 1976, 50, no. 11; 68, no. 26; Riley 1981, 94, no. 27; Therrien 1981, 103; Carignani et al. 1986, 32; Aquilué Abadías 1989, 192; Tomber 1989, 486, nos. 271-273; Rakob 1991, 88; Aguarod Otal 1991, 248-249; Aquilué Abadías 1995, 67. Coletti 1998, 409. Fulford 1984b, 191, no. 11; 1994, 66, no. 15. Marín Jordá 1995, 162. Dore 1989, 157. Tomber 1986, 44, no. 103. Fulford 1984b, 191-193. Aquilué Abadías 1989, 194-200; Reynolds 1993, 136-138. The late date is unanimously suggested by the presence of this form among the surface debris of El Mahrine kiln site that was active in the Medjerda River valley from the early 4th to the mid-7th century AD, see Mackensen 1993, 439 (form 73.2). Tomber 1989, 490, no. 305; cf. also 481-484 and Sánchez Sánchez 1995, 275. Hayes 1976, 93. On africano grande amphorae, see Peacock & Williams 1986, 156. For this reason the introduction of this variant in Ostia I (87) is hard to understand, as it is said to have surfaces slipped with a brush (pennello). Therrien 1981, 102 (in this case the variant to be regarded as a “Hayes 195 variant” is no. 62). Tomber 1986, 43, no. 92; González Villaescusa 1990, 52, 69; Aguarod Otal 1991, 254-255; Fulford 1994, 64, no. 12; Sánchez Sánchez 1995, 268; Freed 1998, 31. Fulford 1984b, 191, no. 9; Aquilué Abadías 1995, 67; Sánchez Sánchez 1995, 268. Luni II, 182; Aguarod Otal 1991, 254-255; cf. also 383-384 and fig. 121. Therrien 1981, 102; Fulford 1984b, 191, no. 9; Tomber 1989, 486, nos. 265-267; González Villaescusa 1990, 52, 69; Fulford 1994, 64, no. 12; Aquilué Abadías 1995, 67; Sánchez Sánchez 1995, 268; Macias et al. 1997, 164 (Ostia I-262). Villedieu 1984, 136. González Villaescusa 1990, 52, 69. Hayes 1976, 68, no. 25; Tomber 1989, 486, no. 278. Aquilué Abadías 1991, 976. Tortorella 1981b, 213. Tomber 1986, 44, no. 104; Fulford 1994, 64, no. 12. Ostia IV, 104. Fulford 1984b, 191, no. 9; Aquilué Abadías 1989, 194-200; Tomber 1989, 486, no. 262; Aquilué Abadías 1991, 975. Hayes 1976, 64, no. 5; 68, nos. 23-25. Aquilué Abadías (1991, 975) has reported examples found in the mid-3rd century AD deposits at Tarragona. Luni II, 182; Ostia IV, 104; Hayes 1976, 64, no. 5; 68, nos. 23-25; Therrien 1981, 102; Tortorella 1981b, 213; Tomber 1989, 486, nos. 263-264; Although the study assemblage may include examples of South Tunisian and Tripolitanian cookware, the vast majority of forms produced elsewhere than in North Tunisia can be identified with substantial high degree of certainty as Central Tunisian products. Hence, the definition “Central Tunisian” will be used instead of “Central/South Tunisian or Tripolitanian” throughout this section. E.g. Hayes 1972, 18, 201-203; Pucci 1975, 61. Tortorella 1981b, 208. In a certain sense, the increased charm of burnished vessels is reflected by the establishment of stripe-burnished ware (ceramica da cucina polita a bande, see e.g. Gandolfi 1994, 151), an artificial subcategory of African cookware differing from the rest of the production group only by the peculiar surfacing. van der Werff 1982, 78; Dore 1989, 157. Luni II, 182; Hayes 1972, 18, 201-203; González Villaescusa 1990, 52, 69; Aguarod Otal 1991, 254-255; Aquilué Abadías 1995, 67; cf. also 383 fig. 121; Sánchez Sánchez 1995, 268. In some relatively remote places, like Valencia, it was only introduced in the mid-3rd century AD, Marín Jordá 1995, 161. Hayes 1972, 18, 201-203; Dore 1989, 157; Small 1994, 69, no. 61. González Villaescusa 1990, 52, 69; Aguarod Otal 1991, 254-255; Aquilué Abadías 1995, 67; Sánchez Sánchez 1995, 268.

84

Forms

84

Hayes 1972, 298; Hayes 1976, 88. Dore 1992, 120; Stirling & Ben Lazreg 2001, 229. In addition, the Hayes 182 lid was also well-attested together with the Hayes 181 pan at the Roman cemetery (site 10) excavated on the southeast edge of Leptiminus, see Dore 2001b, 206-208. 86 Peacock et al. 1989, 190. Tortorella 1995, 87 (referring to M. Bonifay, A. Oueslati, R. Paskoff, H. Slim, & P. Trousset 1992, Programme tunis-français d’étude du littoral de la Tunisie [bilan des travaux 1987-1990], Bulletin des travaux de l’Institut National du Patrimoine, comptes rendus [fascicule 5. Janvier-Juin 1990], 95-116). 87 Tortorella 1981b, 213. 88 Pucci 1975, 61, 105. 89 Hayes 1972, 18, 201-203; Pucci 1975, 61, 105; van der Werff 1982, 88; González Villaescusa 1990, 52, 69; Marín Jordá 1995, 161; Mattingly et al. 2001, 79. 90 Lund 1995, 558, form 28. 91 van der Werff 1982, fig. 4 no. 7 (burnished Byzacenan pottery); fig. 6 no. 11 (red Byzacenan commonware). 92 Even the identification of these examples can be questioned, because when overfired the North Tunisian fabric appears to be very similar to Central Tunisian one. The possibility is also suggested by the Hayes 197 deep casserole of presumed Central Tunisian origin (see Table 5), the chemical composition of which is anomalous and closer to the North Tunisian than the Central Tunisian fabric. 93 Peacock et al. 1990, 62. 94 Aquilué Abadías 1987, 187. 95 Aguarod Otal 1991, 237-238, 241-244, 267-269. 96 For examples with ring-shaped protrusion, see van der Werff 1982, fig. 4 no. 11, fig. 6 no. 13. 97 Pucci 1975, 73-74; Aquilué Abadías 1987, 28-30; Dore 1989, 154-156, see also fig. 42, 156. 98 Ostia III, 420; Ostia IV, 348; Lavoie 1989, 101, no. S88.47E2; Peacock et al. 1990, 62-64; Fulford 1994, 67, no. 16. 99 Lavoie 1989, 101, no. S88.47E2; Peacock et al. 1990, 62-64. 100 Pucci 1975, 73-74; van der Werff 1982, 89-90; Dore 1989, 154-156, see also fig. 42, 156. 101 Mackensen 1993, 439 (form 73.1). 102 Fulford 1984b, 197, no. 2; Fulford 1994, 66, no. 14. 103 Luni II, 182, CM 3735/1; Aguarod Otal 1991, 259. 104 Fulford 1984b, 189, no. 3. 105 Fulford 1984b, 189, no. 9; Tomber 1989, 488, no. 298. 106 Hayes 1976, 72, no. 17. 107 Luni II, 180; Gandolfi 1994, 151. 108 At Uzita solid lid knobs were found from layer II (ca. AD 80) and layer I (AD 190), while they were absent from the well that contained a fill dated between the late 2nd century and AD 250, see van der Werff 1982, 21-23. 109 Luni II, 180. 110 Aguarod Otal 1991, 245-251. 111 Aquilué Abadías 1995, 66-67. 112 Aguarod Otal 1991, 245-246. 113 Hayes 1976, 58, n. 60. 114 Aguarod Otal 1991, 248, see also van der Werff 1982, 117. 115 Aguarod Otal 1991, 250. 116 Pontacolone & Incitti 1991, 564; Bost et al. 1992, 189. 117 The turning of the ring handle has most likely been executed on a chuck, see Hopper 1986, 137 fig. 9.8, cf. Freed 1998. The same method is used to turn the footrings of bowls, see Rye 1981, 87 fig. 74, c.1-2. 118 Hayes 1978, 29; Tomber 1989, 486, no. 260; Pontacolone & Incitti 1991, 564. 119 The additional input in vessel finishing has also been observed by Fernández Sotelo (1994, 31 no. 306). 120 The two productions were distinguished already in the early 1980’s by Hayes (1980: xviii-xix) and van der Werff (1982, 88, 238), but probably because the form was not illustrated in the former publication and the latter is an unpublished manuscript written in Dutch, they are still commonly merged. Recently Hayes (1998, 10 note 4) himself has pointed out that some forms described in Late Roman Pottery were inadequately illustrated. 121 Luni II, 165; Lamboglia 1958, 275; Hayes 1972, 18; Peña 1999, 164. 122 Villedieu 1984, 132. 123 van der Werff 1982, 127. Burnished variants of the Hayes 181 pan, however, seem to have been introduced only i n the early 1st century AD, see van der Werff 1982, 80-81, 127. 124 Anselmino et al. 1986, 63; Tortorella 1995, 89. 125 Tortorella 1987, 299. 126 Luni II, 158-162, fig. 77; 165-166; 381; Ostia IV, 330-332; Lamboglia 1958 (form 9b), 274; Hayes 1972 (form 181), 18, 201; Carandini 1975, 52; Dyson 1976, 145-146; Neuru 1987, 177-179; Aquilué Abadías 1989, 192; 1991, 978; Carillo Díaz-Pinés & Murillo Redondo 1994, 1316. 127 Anselmino et al. 1986, 59. 128 Giannotta 1992, 57, no. 258. 129 Luni II, 158-162, fig. 77; 165-166; 381; Aquilué Abadías 1989, 192; Fernández Sotelo 1994, 27. 130 Macias et al. 1997, 164 (Lamboglia 9b). 131 Hayes 1972, 298; 1976, 88. 132 Dore 1992, 120; Stirling & Ben Lazreg 2001, 229; Mattingly et al. 2001, 79. The Hayes 181 pan was also wellattested at the Roman cemetery (site 10) excavated on the southeast edge of Leptiminus, see Dore 2001b, 206-208. 85

85

Chapter 3

133

Hayes 1972, 200. Neuru 1987, 177-179. During my short visit to Central Tunisia in February 1998 I was able to verify the presence of the Hayes 181 pan by screening the surface finds at the archaeological sites of Cillium, El Djem and Sufetula. By contrast, none of these sites produced examples of the North Tunisian Hayes 26/181 pan-casserole. 135 Pucci 1975, 61, 105. 136 Tortorella 1981a, note 21. 137 Tortorella 1983b, 130. 138 Hayes 1980, xix. 139 Neuru 1986, 71; Reynolds 1995, 87; Freed 1998, 31. 140 “Late version of Lamboglia 9A=Hayes 181 with straight wall”, Carignani et al. 1986, 32; cf. Aguarod Otal 1991, 260. 141 Although the definition "pan-casserole" is nearly as vague as "lid-plate" (see the discussion in section 3.9.1), the difference between the use of pans and casseroles is certainly not as distinct as the use of lids and plates. 142 Bats 1988, 69; González Villaescusa 1990, 61; Aguarod Otal 1991, 260. For IRSC, see e.g. Peña 1990; Hayes 1997, 78-79. 143 The typology of African cookware pans is fairly similar to large African Red Slip ware dishes, the diameter of which seems to increase as a function of time (Hayes 1972, 16). The sequence is begun in the late 1st/mid-2nd century AD with ARS A form Hayes 27 (Atlante I, 32; Hayes 1972, 51). It is followed by the mid-3rd century/early 4th century AD ARS C form Hayes 49 (Atlante I, 61-62; Hayes 1972, 67-68), the shape of which is strikingly similar to the Hayes 181 pan. The parallel for the Hayes 26/181 pan-casserole series is the Hayes 62A flat-based dish (Atlante I, 120, Hayes 107-109) produced in several fabrics (ARS C3, D1 and E). Interestingly also this form, dated between the mid-4th and the first quarter of the 5th century AD, was apparently produced in three sizes: 19-24 cm, 26-34 cm and 45-55 cm. Another late parallel is the flat-based Hayes 64 dish (Atlante I, 87; Hayes 109-111) in ARS D, a form dated to the early 5th century AD onwards (Hayes 1972, 111). Like some examples of the Hayes 26/181 pan-casserole, also this form has occasionally a convex bottom (e.g. Hayes 1972, 110 fig. 18 no 64.2). 144 González Villaescusa 1990, nos. 69-71. 145 Settefinestre II, 115; Fulford 1984b, 191, no. 6; Neuru 1986, 71; Tomber 1989, 444-446, tab. 1; Reynolds 1995, 87. 146 Neuru 1986, 71. 147 Sánchez Sánchez 1995, 269. 148 Aguarod Otal 1991, 262. 149 Taylor & Robinson 1996b, 248-253. 150 Lamboglia 1958, 275; Lamboglia 1963, 187. 151 Saguì 1979, 263; Hayes 1980, xviii-xix. 152 Aquilué Abadías 1987, 187-188. 153 Luni II, 165; Salomonson 1968, 124-125; Hayes 1976, 68; 1980, xviii-xix; Tomber 1986, 42, nos. 26-27; Aquilué Abadías 1987, 187-188; Tomber 1989, 446, nos. 8-11; Macias et al. 1997, 164 (Lamboglia 9a). 154 Saguì 1979, 263; Aquilué Abadías 1991, 979. 155 Cf. Therrien 1981, 98. 156 Neuru 1980, 203; Therrien 1981, 98; Tortorella 1981b, 215; Michelucci 1985, 21-22, no. 91; Tomber 1989, 446, nos. 12-13. 157 Hayes 1976, 59, no. 3; Neuru 1980, 201, nos. 19-20; Aquilué Abadías 1991, 979; Coletti 1998, 409; Lund (1995, 550) has even suggested that the production of this variant did not cease until AD 580/600. 158 On the definition of the term “carination”, cf. Gibson & Woods 1997, 118-119. 159 Young & Stone 1990, 198-202; Schiffer et al. 1994, 208, see also the discussion on surface finishes in section 4.2. 160 Cf. Lamboglia 1958, 277. 161 Luni II, 478, 485; Villedieu 1984, 131. 162 Carandini 1970, 108; Nin 1996, 271-272. 163 Aguarod Otal 1991, 269. Aquilué Abadías, who was the first scholar inclined to date the event to the reign of Vespasian, has lately accepted this hypothesis with caution, see Aquilué Abadías 1995, 69; cf. Aquilué Abadías 1987, 99, 192-193; 1991, 977. 164 Ostia IV, 330-332; Hayes 1972, 45-48; Aquilué Abadías 1985, 211-218; 1987, 36-38, 76-77, 99-101; Marín Jordá 1995, 155, 162; Freed 1998, 31. 165 Introductory: Luni II, 158-162, fig. 77. Terminal: Tomber 1986, 37 (citing Hayes' personal comment); Aquilué Abadías 1987, 36-38, 76-77, 99-101; Aguarod Otal 1991, 269; Carillo Díaz-Pinés & Murillo Redondo 1994, 1315. 166 Villedieu 1984, 24, 131. 167 Aquilué Abadías 1985, 211-218; 1991, 977; Marín Jordá 1995, 155, 162. At Uzita the form is present in layer I (AD 190), but absent from the early 3rd century AD fill found in a well, see van der Werff 1982, 227 table 7. 168 Luni II, 158-162, fig. 77. 169 Pucci 1975, 58; González Villaescusa 1990, 32-35. 170 Tortorella 1981b, 217; Tomber 1989, 446. 171 Aguarod Otal 1991, 269. 172 Tortorella 1981a, note 21. 173 Ostia II, 85; Ostia III, 415; Tomber 1986, 45, no. 163. 174 Sánchez Sánchez 1995, 270. 175 E.g. López Mullor et al. 1997, p. 71, fig. 1, no. 7. 176 Aquilué Abadías 1991, 977; Bourgeois & Mayet 1991, 249. 134

86

Forms

177

Aquilué Abadías 1995, 69; cf. Aquilué Abadías 1985, 212; 1987, 190; Aguarod Otal 1991, 267; Marín Jordá 1995, 159. This form is also known as Lamboglia 10A, see Lamboglia 1958, 277. 178 Rakob 1991; Freed 1998, 31. 179 Tomber 1986, 36. 180 Hayes 1972, 45-48; Fentress et al. 1983, 80-95; González Villaescusa 1990, 51-54, 68. 181 Luni II, 158-162, fig. 77. 182 Fentress et al. 1983, 80-95; Pontacolone & Incitti 1991, 547-548; Giannotta 1992, 57, no. 255. 183 Hayes 1972, 18, 45-48. 184 Luni II, 158-162, fig. 77; Villedieu 1984, 132; González Villaescusa 1990, 32-34, 68; Aguarod Otal 1991, 267. 185 Aquilué Abadías 1991, 978; 1995, 69; cf. Aquilué Abadías 1985, 212; 1987, 190. 186 Tortorella 1981b, 217; Aquilué Abadías 1987, 190, cf. Aquilué Abadías 1991, 977. 187 Aguarod Otal 1991, 267; Marín Jordá 1995, 161. 188 Villedieu 1984, 131-132. 189 Sánchez Sánchez 1995, 270. 190 Cf. Villedieu 1984, 132. 191 González Villaescusa 1990, 32, 68. 192 Luni II, 166. 193 Tortorella 1981a, note 21. 194 Taylor & Robinson 1996b, 252, tab. 3, sample 23. 195 Leveau 1984, 452-454. 196 Neuru 1987, 183; Neuru 1990, 256. 197 I base this conclusion on a brief examination of the surface finds at these archaeological sites. 198 Hayes 1972, 18, 208; González Villaescusa 1990, 69; Pontacolone & Incitti 1991, 548. 199 Rita 1990, 20. 200 Cf. Luni II, 485; Villedieu 1984, 131-132. 201 Michelucci 1985, 195, no. 912. 202 Aquilué Abadías 1987, 26, 76-77, 106-108. 203 Cf. Tomber 1989, 446, no. 7. 204 Pontacolone & Incitti 1991, 547; Bost et al. 1992, 180-181. 205 Giannotta 1992, 57, nos. 256-257. 206 Ostia III, 413, cf. form 60.1. 207 Ostia III, 414; Tortorella 1981a, 359; 1981b, 216; van der Werff 1982, 112; Aquilué Abadías 1991, 981; Fulford 1994, 58, no. 3; Freed 1998, 31. 208 Ostia III, 413-414. 209 Ostia II, 85. 210 Hayes 1972, 44, 207. 211 Freed 1998, 21, 31. 212 Zevi & Pohl 1970, 75-76; Dyson 1976, 123, no. 49; Aquilué Abadías 1985, 211; 1987, 190; Aguarod Otal 1991, 237, 241-242, 265; Aquilué Abadías 1991, 981; 1995, 68; Rivet 1996, 336. 213 Ostia II, 85; Ostia IV, 346-347; Carandini 1977, 24. 214 Salomonson 1968, 104, form A 11; Hayes 1972, 207; Dore 1988, 80; Dore 1989, 107. 215 Aguarod Otal 1991, 237, 241-242, 265; Aquilué Abadías 1991, 981; Coletti & Pavolini 1996, 409. 216 Ostia II, 85; Settefinestre II, 115; Tortorella 1981b, 211; Aquilué Abadías 1985, 211; 1987, 190; 1995, 68. 217 Lamboglia 1958, 276. 218 Hayes (1980, 518) specifies that this casserole was most likely produced in the region of Carthage. 219 Hopper 1986, 139. 220 González Villaescusa 1990, 40. 221 Hayes 1972, 18, 207. 222 Aguarod Otal 1991, 69; Aquilué Abadías (1995, 69) has stated that the fabric in this particular example is hardly African. 223 Ostia III, 416, cf. table 417; Carandini 1970, 109; Sagui 1980, 523, no. 144; Aguarod Otal 1991, 69; Bost et al. 1992, 182; Marín Jordá 1995, 164. 224 Carandini 1970, 109. 225 Atlante I, 35; Ostia IV, 101; Tortorella 1981b, 209. 226 Nolla & Aquilué Abadías 1984, 51; Aguarod Otal 1991, 69; Aquilué Abadías 1991, 979; 1995, 69. 227 Luni II, 181 fig. 79, 187. 228 Ostia IV, 101-102. 229 Villedieu 1984, 137. 230 None the less, all variants of the Hayes 193 in the study assemblage exceed in rim diameter the range (14.6-16.2 cm) that Nolla & Aquilué Abadías (1984, 52) have established for this form. 231 Bost et al. 1992, 182. 232 Tomber 1989, 490, no. 302. 233 Hayes 1976, 54, no. VI.2, fig. 2. 234 Luni II, 499; Dore 1989, 101, forms 37 & 53. Tomber (1986, 43) has defined this deep casserole form as the BlackTop cooking pot 1. 235 Tortorella 1981b, 211. 236 Fulford 1994, 53.

87

Chapter 3

237

Ostia III, 411; Zevi & Pohl 1970, 177; Tomber 1986, 37; Aquilué Abadías 1987, 55-56; Aguarod Otal 1991, 279; Marín Jordá 1995, 158. 238 Hayes 1976, 64; Hayes 1978, 38; Tomber 1986, 37. 239 Hayes 1976, 72, nos. 5-6; Aquilué Abadías 1987, 28. The short duration of production is equally supported by the fact that the late 2nd century finds of Luni and Porto Torres have been interpreted as residual, see Luni II, 185; Villedieu 1984, 137. 240 Aquilué Abadías 1987, 38; Dore 1988, 80; Fulford 1994, 60, no. 10. 241 Tomber 1986, 40-41 (tabb. 1-3), 43. 242 Sánchez Sánchez 1995, 272. 243 Ostia III, 412. 244 Dyson 1976, 118, no. 17. 245 Cf. Ostia I, 87. 246 González Villaescusa 1990, 35-36, 68. 247 Dore 1989, 101; Fulford 1994, 60. 248 Fulford 1994, 60. 249 González Villaescusa 1990, 52, 68. 250 Ostia III, 412. 251 Ostia III, 411-412. 252 Leveau 1984, 459-461, fig. 247. 253 Aguarod Otal 1991, 281-284. 254 The idea that these rims were formed by folding the clay against the exterior surface was first suggested by Niemeyer (1968, 80) in the late 1960's when the application of rim was the prevalent explanation, see e.g. Ostia I, 87; Zevi & Pohl 1970, 34; Vegas 1973, 22. For more recent references to the use of the latter technique, see e.g. Marín Jordá 1995, 162; Pasquinucci & Menchelli 1996, 512; Bernal Casasola 1998, 1348; Pons Machado 1998, 1340. 255 González Villaescusa 1990, 35-36. 256 The best-known examples of such tripods are possibly the ones displayed in the kitchen of Casa dei Vettii at Pompeii. 257 Hayes 1972, 209; González Villaescusa 1990, 69; Pontacolone & Incitti 1991, 547. 258 Moliner 1996, 239 fig. 4, 244-245. 259 Fulford 1994, 53. 260 Ostia I, 87; Carandini 1970, 108; Vegas 1973, 24-25; Aguarod Otal 1991, 281 (variant A); Marín Jordá 1995, 158159. 261 Marín Jordá 1995, 158-159. 262 Ostia I, 87; Carandini 1970, 108; Aquilué Abadías 1985, 212; Aquilué Abadías 1989, 193; González Villaescusa 1990, 35-36, 61; Aguarod Otal 1991, 281 (variant B); Aquilué Abadías 1995, 68. Cf. Terme del Nuotatore (Ostia I, 87), where no 1st century AD examples have been found. 263 Luni II, 185-186; Fentress et al. 1983, 80-95; Fulford 1984b, 181; Villedieu 1984, 136-137; Tomber 1986, 36; Dore 1988, 72, 80; Dore 1989, 101; Fulford 1994, 60; Marín Jordá 1995, 158-159. 264 Luni II, 185-186; Ostia IV, 346-347. The supply of this form may have been subjected to a temporary interruption i n the late 3rd/early 4th century AD, see Peña 1998, 12-13; Peña 1999, 135, 164. 265 Hayes (1972, 18, 209) dates the form from the 3rd century AD onwards, whereas Vegas (1973, 24-25) regards the final decades of the same century as the probable terminal date. At Fosso della Crescenza the latest examples of Hayes 197 have also been found in the late 3rd century AD contexts (Fentress et al. 1983, 80-95), whereas at Otranto this form is only present in contexts dating after AD 330-350 (Giannotta 1992, 57, no. 263). 266 Luni II, 185; Ostia II, 86; Ostia III, 410-411. 267 Fulford 1984b, 181; Villedieu 1984, 136-137; Aquilué Abadías 1985, 212; Aquilué Abadías 1989, 193; Tomber 1989, 482-483, table 3; González Villaescusa 1990, 35-36; Aguarod Otal 1991, 281; Aquilué Abadías 1995, 68; Reynolds 1995, 88; Pons Machado 1998, 1338. 268 Dore 1989, 100-102. 269 E.g. Dore 1989, 100-101; Hayes (1980, 518) thinks that the production took place in the surroundings of Carthage. 270 van der Werff 1982, 130-131 (see also pl. 43, no. 19); Dore 1988, 72, 80; Dore 1989, 101. For example, in the excavations of the Italian expedition at Sabratha (Pucci 1975, 69-73) the ratio of Hayes 197 finds was as low as 1:40 compared to local deep casserole forms. 271 Taylor & Robinson 1996b, 252, tab. 3, sample 30. 272 Cf. footnote 343. 273 Tomber 1989, 484, no. 252; Aguarod Otal 1991, 281; Pontacolone & Incitti 1991, 546. 274 Aguarod Otal 1991, 281. 275 E.g. Olcese 1993, 135. 276 Luni II, 185-186; Hayes 1976, 50, no. 10; Tortorella 1981b, 219. 277 Settefinestre II, 117; Therrien 1981, 98. 278 Hayes 1976, 58, no. 57. 279 Ostia III, 412; Hayes 1976, 68, no. 32; Therrien 1981, 98; Tomber 1989, 484, nos. 250-251. 280 Aquilué Abadías 1987, 197. 281 Luni II, 186. 282 Ostia III, 416; Marín Jordá 1995, 163. 283 Luni II, 186. 284 Aguarod Otal 1991, 286.

88

Forms

285

Macias et al. 1997, 153, 169, fig. 5, no. 8. Luni II, 499; Sagui 1980, 525, no. 142, cf. also fig. 8, 522-523; Tomber (1986, 43 no.88) introduced this variant as Black-top cooking pot 2. 287 Hayes 1972, 210. 288 Marín Jordá 1995, 158. 289 The similarity of the variants Q.I-II has already been stressed by Saqui (1980, 525, no. 142, cf. also fig. 8, 522523). 290 Aguarod Otal 1991, 238. 291 Hayes 1972, 210; Tortorella 1981b, 219. 292 Hayes 1976, 72, no. 11; Tomber 1986, 41 tabb. 2-3. 293 Hayes 1972, 210; Dore 1989, 101, 124; Rakob 1991, 119; Fulford 1994, 60, no. 23; Marín Jordá 1995, 158. 294 Luni II, 186; CM 2868/5. 295 Sagui 1980, 525, no. 143, cf. also fig. 8, 522-523. 296 Aguarod Otal 1991, 237-238, 277. 297 Aquilué Abadías 1989, 197. 298 Marín Jordá 1995, 162. 299 Macias et al. 1997, 157-158, 164 (Ostia I-270). 300 Dore 1989, 130. 301 Luni II, 181, fig. 79. 302 Hayes 1978, 25. 303 Sagui 1980, 523. 304 Fulford 1984b, 183, no. 11. 305 Fulford 1984b, 183, no. 5; Dore 1989, 102-103; Mackensen 1993, 438 (form 67.2), see also 626-627. 306 Aquilué Abadías 1989, 194. 307 Tortorella 1981b, 220. 308 Dore 1989, 132. 309 Neuru 1980, 201 no. 24. 310 Aquilué Abadías 1995, 70; Marín Jordá 1995, 161. 311 Fulford 1984b, 183, no. 7. 312 Mackensen 1993, 438 (form 67.2, see tab. 81 no 1), see also 626-627. 313 Tortorella 1981b, 220-221, forms CVIII, 1-3. 314 Lavoie 1989, 101, no. S88.46E1. 315 Hayes 1972, 203-204. 316 Cf. Dore 1989, 126-128. 317 Pucci 1975, 69-71. 318 Dore 1989, 126. 319 For complete examples, see e.g. Hayes 1972, p. 202, fig. 35, no. 183.4; Tortorella 1981b, tav. CIX, nos. 2-6. 320 The chemical reaction causing the deposition of the scum will be discussed in detail in section 5.3. 321 Hayes 1972, 298; van der Werff 1982, 90 322 Dore 1988, 82. 323 Central Tunisia: Henchir ech Choggaf, Henchir el Guellal, Zeglass and Dahar Slima (Peacock et al. 1989, 197); Tripolitania: Sidi-as-Sid (Arthur 1982, 67, fig. 7, nos. 20-21; Peacock et al. 1990, 62, 75). 324 Mahjoubi et al. 1973, 64-65. 325 Mahjoubi et al. 1973, 64-65. 326 Dore 1989, 102, 126. 327 van der Werff 1982, 90; Dore 1988, 82. 328 Dore 1989, 120 (forms 58 & 59); Pucci 1975, 69-71. 329 Luni II, 186; Mahjoubi et al. 1973, tab. III; Dore 1989, 100-102; Bourgeois & Mayet 1991, 250; Dore 1992, 138; Sánchez Sánchez 1995, 272. 330 Mahjoubi et al. 1973, 11-12, 64-66; van der Werff 1982, 84, 120; Dore 1989, 126. 331 Tortorella 1981b, 224; Dore 1992, 138; Sánchez Sánchez 1995, p. 272. 332 Dore 1989, 102. 333 Hayes 1972, 18. 334 Ostia III, 416, cf. also table in 417; Ostia IV, 346-347; Pucci 1975, 70. 335 van der Werff 1982, 89, 125. 336 Berenice, Caesaraugusta, Carthage, Luni, Ostia and Sabratha. 337 van der Werff 1982, pl. 51, no. 5; see also Aguarod Otal 1991, 273. 338 Tomber 1986, 46, no. 193; Aguarod Otal 1991, 273. 339 Riley 1979, 250-251, no. 454. 340 Pucci 1975, 70. 341 van der Werff 1982, 119-120, 208. 342 Dore 1992, 138, cf. his discussion on the Hayes form 182, p. 120. 343 Luni II, 185; Ostia III, 411-413. 344 Luni II, 186; Dore 1989, 101; Aguarod Otal 1991, 277. 345 Lavoie 1989, 91-93. 346 Aguarod Otal 1991, 294. 347 Dore 1989, 123. 286

89

Chapter 3

348

Hayes 1976, 68, no. 34; Aquilué Abadías 1989, 194-200; Aguarod Otal 1991, 294. Hayes 1976, 68, no. 35; Villedieu 1984, 138; Aquilué Abadías 1989, 194-200. 350 Pucci 1975, 70. 351 Fulford 1984b, 183, no. 5. 352 Aquilué Abadías 1991, 985. 353 Dore 1989, 133. 354 Cf. Aquilué Abadías 1991, 985-989. 355 Dore 1989, 112. 356 Tomber 1989, 476. 357 Leveau 1984, 462, fig. 248, no 10715. 358 Hayes 1972, 36-38; Carandini 1975, 48-49, 52-53. 359 Hayes 1976, 68; Tomber 1989, 446, nos. 2-3. 360 González Villaescusa 1993, 158. 361 Coletti & Pavolini 1996, 396-398; Pavolini 1998, 394. 362 González Villaescusa 1993, 158. It remains to be resolved, whether the presence of rare forms indicates the importance of the location they were found in (ibid., 158-159). 363 González Villaescusa 1993, 153-157. Tortorella (1995, 89) has doubted this identification, which would make it one of the earliest exports in Central Tunisian cookware besides the Hayes 181 pan. 364 González Villaescusa 1993, 155, fig. 3, no. 3. 365 González Villaescusa 1993, 157. 366 Aguarod Otal 1991, 300; Marín Jordá 1995, 158. 367 This is proposed by Aquilué Abadías (1995, 69-70) who has recently presented a detailed discussion on this form, cf. González Villaescusa 1993, 157. 368 Neuru 1980, 201 no. 30; Fulford 1984b, 209, no. 29. 369 Aguarod Otal 1991, 238, 300; González Villaescusa 1993, 41. 370 Whittaker 1989, 12. 371 González Villaescusa 1990, 41, 54, 74; González Villaescusa 1993, 157-158. 372 For Aegean cookware forms with handles, see e.g. Hayes 1983, 123-125; West-Central Italian e.g. Dyson 1976, passim. 373 Aguarod Otal 1991, 348 (cookpots), 349-350 (Hayes 131), 351-352 (Hayes 200) & 353-355 (pitcher). 374 Aguarod Otal 1995, 137, fig. 9, nos. 9-11. 375 Aquilué Abadías 1995, 70. 376 Aguarod Otal 1991, 238. 377 González Villaescusa 1990, 41, 53; Aquilué Abadías 1995, 70. 378 Aguarod Otal 1991, 297. 379 Hayes 1972, 178. 380 Coletti & Pavolini 1996, 410. 381 Freed (1998, 31), however, has identified both these forms as bread baking pans. 382 Ostia III, 416-417. 383 Anselmino et al. 1986, 59. 384 Dore 2001a, 75-76, 82-84. 385 Tortorella 1981b, 209. 386 This is indicated by the presence of late African cookware forms, e.g. B.IV (Hayes 195 lid) and H.III (Hayes 26/181 pan-casserole), in the early horizons. 387 Hayes 1976, 59, no. 3; Neuru 1980, 201, nos. 19-20; Lund 1995, 550; Coletti 1998, 409. 388 Ermeti 1998, 611. 389 Aquilué Abadías 1991, 968-970, 987-991. 390 Hayes 1972, 14; Aquilué Abadías 1991, 990. 391 Hopper 1986, 142. 392 Hayes 1972, 18. Another topic worth examining would be whether a difference in vessel size can be detected between African cookware exports and items directed for local consumption. 393 Orton et al. 1993, 190. 394 Carandini (1983, 150) has linked the great size of some vessel forms introduced in ARS C with a hypothesized change in eating and serving habits that were altered in favor of collective eating. But this development could also result from other factors, such an increase in the size of an average Roman household. 395 Fontana (1998) has recently presented an extensive summary on the Italian imitations of African tableware; Pavolini 1995, 117; Tortorella 1995, 96-97; Pasquinucci & Menchelli 1996, 507-510; Bernal Casasola 1998, 1342, 13491351. 396 Fulford 1994, 53. 397 Luni II, 183; Castañer Masoliver et al. 1990, 170. 398 Tortorella 1995, 95-96; Dodinet & Leblanc 1988, 138-141; Freed 1998, 62 no. 137 (see also p. 60, fig. 8, no. 137). 399 Aquilué Abadías 1987, 52, fig. 15, no. 13; Aguarod Otal 1991, 47, 413-414; Alonso de la Sierra Fernández 1995, 158-159; Sánchez Sánchez 1995, 267, 274, fig. 13, nos. 28-31; Serrano Ramos 1997, 220-221; Bernal Casasola 1998. 400 Tortorella 1995, 97 (referring to Vivien G. Swan, Un tradition de Gallia Narbonnensis à York au début du IIIe siècle, in SFECAG, Actes du congrès de Versailles, 1993, p. 371ff.); Alaimo et al. 1997, 55, 67. 349

90

Forms

401

E.g. Pasquinucci & Menchelli 1996, 509, fig. 5 no. 32; 511, fig. 7, nos. 49-52; Pasquinucci et al. 1998, 14091410. 402 González Villaescusa 1993, 155, see also, 156, fig. 4, no. 5. 403 Aquilué Abadías 1987, 80; Bernal Casasola 1998, 1351. 404 Picon & Olcese 1995, 112; Tortorella 1995, 95-96; Peña 1999, 164. 405 Panella 1986, 440; Bernal Casasola 1998, 1350. 406 Tortorella 1995, 95-96; Bernal Casasola 1998, 1351; Fontana 1998, 83, 96. 407 Serrano Ramos 1997, 220. The most obvious example is the patina cenerognola cover on the exterior surface, even though imitations with the intentional reproduction of this cover also exist, Fontana 1998, 83. See also section 5.3.1. 408 Anselmino et al. 1986, 62, see also Villedieu 1984, 135; Panella 1999, 189-190. 409 Ricci 1986, 85. 410 Fontana 1998, 96. 411 Cf. Dodinet & Leblanc 1988, 143. 412 Dodinet & Leblanc 1988, 136-139, 142 fig. 10; Passelac 1996, 371, 374. 413 Serrano Ramos 1997, 220; 229, figs. 9, 12-13. Bernal Casasola 1998, 1341, 1349-1350. 414 Moreno Almenara & Alarcón Castellano 1994, 1298, figs. 1-2. 415 The author is currently preparing a manuscript on the West-Central Italian cookware assemblage of the Palatine East excavations. See also Peña 1999, 123 form 45.2. 416 The latest imitations of African cookware, of which I am aware of, have been found in the early 7th century AD context on coastal North Etruria, see Pasquinucci et al. 1998, 1409-1410. 417 Lamboglia 1950, 203; Ostia I, 86-87. 418 Ostia III, 418; Ostia IV, 103; Vegas 1973, 49; Pucci 1975, 73; van der Werff 1982, 121; Anselmino et al. 1986, 59; Bats 1988, 70; Aquilué Abadías 1989, 201. 419 Zevi & Pohl 1970, 92; Hayes 1972, 18, 200-209. 420 Tomber 1989, 483-484; see also Hayes 1997, 78. Although the literary sources hint that baking covers, used for baking bread and maturing other provisions, were used commonly throughout the Roman world (see Cubberley et al. 1988, 100-102; Cubberley 1995), the archaeological evidence concerning them is fairly scarce. Due this imbalance, today the re-identification of known vessel forms as baking covers has been a kind of fashion, which has led t o confusing results. For example, instead of vessels identified here as lids, for Roberts the form assumed to be a baking cover (testa) is the Hayes 23B shallow casserole (Cubberley et al. 1988, 114-116), which in my opinion was actually used with some of these lids. 421 On the criticism concerning the use of this method, see Rice 1990, 4-7. 422 In July 1997 I was able to experiment the handling of this form in Rome with an exceptionally well-preserved ringshaped protrusion. Due to requirements in optimum performance in heat conduction and cooking, even the modern earthenware cooking vessels are customarily equipped with small-sized handles, lugs and knobs: “so often one finds handles and knobs which are next to impossible to pick up when cold, let alone when they are hot and the user i s wearing oven mitts”, (Hopper 1986, 176). 423 Luni II, 180; Niemeyer 1968, 80; Gandolfi 1994, 151. 424 Zevi & Pohl 1970, 92; Fernández Sotelo 1994, 30-31. 425 van der Werff 1982, 88. 426 E.g. Tomber 1986, 57, fig. 7, no. 197. 427 As I presented some results of my work at the University of Oulu in the fall of 1997, it was suggested that these grooves could also result from a device used to turn the exterior surface. Although the idea itself is alluring, I have not had yet any means to examine this possibility. 428 Zevi & Pohl 1970, 96. 429 Cf. Aguarod Otal 1991, 237, 248. Similar ring bases are frequent in African Red Slip ware, but their weight as an evidence is uncertain, because with the exception of some early ARS A bowls and the 6th century AD bowl form Hayes 99, they customarily exceed ring-shaped protrusions of African cookware in diameter. A small ring foot i s certainly not suitable for a soup bowl, as any movement of liquid content makes the vessel unstable and highly vulnerable for a breakage, Hopper 1986, 134. 430 Cf. Tomber 1989, 483-484. 431 In addition to a high domed body, the rim diameter range of baking covers is 20-50 cm and they usually have an opening on the top to facilitate the regulation of the heat, see Cubberley 1995, 58. 432 Although Vegas (1973, 51) suggested that the prevalence of unfinished surfaces confirms the use of this group as lids, the interpretation has also been negated by Niemeyer (1968, 80) and Aguarod Otal (1991, 237). 433 Schiffer et al. 1994, 207-208; Rice 1996a, 148. 434 Schiffer 1990, 377-379; Schiffer et al. 1994, 204-205. 435 Hopper 1986, 141-142, 176-177 (see also fig. 4.1, p. 177). 436 On the use of baking covers, see Frayn 1978; Cubberley et al. 1988; Cubberley 1995. In the light of the present knowledge, it is curious that Tomber (1989, 484) takes sooted interior surfaces as evidence supporting the identification of these forms as baking covers. 437 Lamboglia 1958 (form 9b), 274; Hayes 1972 (form 181), 18, 200; Peña 1999, 135 (form 49.1). The additional input in vessel finishing has resulted in publications, in which the Hayes 181 pan is classified as African Red Slip ware or it is said to belong a hypothetical intermediate class between ARS and African cookware, see e.g. Carandini 1975, 52; Stirling & Ben Lazreg 2001, 229. 438 Mazzeo Saracino et al. 1997, 206-211.

91

Chapter 3

439

The exportation of Hayes 182 lid (see form C.I) began approximately at the same time with its counterpart, the Hayes 181 pan (e.g. González Villaescusa 1990, 69). An example of their contemporaneous introduction can be pointed out in the mid-3rd century AD contexts of Valencia (Marín Jordá 1995, 161-163). 440 Cf. Peña 1999, 134. 441 Bats 1988, 69; González Villaescusa 1990, 61; Aguarod Otal 1991, 260. The popularity of the Hayes 181 pan around the western Mediterranean from the late 2nd to the 4th century AD may support this identification, as the properties of Central Tunisian cookware fabric may have been better suited for use in an oven than over a fire. 442 Cf. Fernández Sotelo 1994, 27. 443 Rivet 1996, 337. 444 Mayeske 1988, 154; Freed 1998, 31. During my visit to Central Tunisia in February 1998, I was told that the modern equivalent of the Hayes 181 pan is used for this purpose. Unfortunately, I had no means or opportunity to verify this information. 445 Ostia IV, 77, 347. 446 Dosi & Schnell 1992b, 87-88. 447 Dosi & Schnell 1992a, 53-57 448 Rickman 1980, 205-206; Garnsey 1999, 120-125. 449 Aguarod Otal 1991, 237-238, 241-244, 265-269. 450 Bats 1988, 62. Freed (1998, 31), however, has identified these forms as bread baking pans. Puls was the way i n which grain – the base of an average Roman diet (Rickman 1980, 4; Garnsey 1983, 118; Mayeske 1988, 149; Braun 1995, 25) – was consumed in the Roman world before the Late Empire. Emmer wheat (far) ground to flour, mixed with water and salt, was cooked to puls (Dosi & Schnell 1992a, 14; Garnsey 1999, 120; cf. PLIN. Nat. 18.83), whereas the use of barley instead of wheat resulted in polenta (Dosi & Schnell 1992a, 15; cf. CATO Agr. 156.5), a less appreciated variety of pap. Both varieties were sometimes enriched with a variety of cheap ingredients like beans, lentils and onions. The popularity of puls is reflected in the term pultiphagonides used by Plautus (Poen. 54) t o describe the Romans due their culinary preferences. 451 Cubberley et al. 1988, 114. Although Roberts refers regularly in his discussion to the Hayes 23B, his argumentation should also apply to the Hayes 23A, since the shape of the rim, the role of which is insignificant in this context, i s the only feature distinguishing these forms from one another. 452 Cubberley et al. 1988, 104-105, figs. 1-2; Cubberley 1995, 57 fig. 5.1, 59 fig. 5.2. 453 E.g. Carandini 1975, 52; Rita 1990, 20. 454 see also Hopper 1986, 176. 455 Rivet 1996, 337. 456 González Villaescusa 1990, 67-68. 457 Aguarod Otal 1991, 237-238; see also Hopper 1986, 176; Nin 1996, 285. 458 González Villaescusa 1990, 35-36, 52. 459 Tomber 1989, 483-484, see also Hayes 1997, 78. 460 Cf. Annis & Geertman 1987, 164-165, 182-185. 461 On the quantification of archaeological ceramics, see e.g. Orton 1993. 462 In this context the term “Hayes 197 series” refers to form groups M (proto-Hayes 197’s) and O (Hayes 197’s). 463 Although the changes in the typology of the Hayes 196 lids and the Hayes 197 deep casseroles have been observed by several scholars (e.g. Aquilué Abadías 1991, 982-983: Dore 1989, 100-102; Gandolfi 1994, 151), only Castañer Masoliver et al. (1990, 171) has paid any attention to the evident association of these changes. 464 Hayes 196 lid (40 pcs), Hayes 197 deep casserole (26 pcs) and Hayes 23B shallow casserole (13 pcs). 465 Pontacolone & Incitti 1991, 544, 555. 466 Pontacolone & Incitti 1991, 547-548, see also Peña 1999, 150 note 288. 467 Aquilué Abadías 1987, 192-193.

92

Production

4 PRODUCTION 4.1 FORMING As indicated in the preceding chapter, all the African cookware vessels were formed on the potter's wheel. The use of this technique is shown by various marks, mainly horizontal striations and ridges, visible both on the exterior and interior surface.1 It does not seem likely that the potter's wheel would have been used to shape only certain vessel parts,2 because similar striations and ridges were equally observed in the body sherd material. Thus, African cookware was normally shaped with the following method, which most centers producing pottery for exportation in Roman Africa used, irrespective of their geographical location or date. Not surprisingly, as forming methods are probably the most conservative segment of the potter's craft,3 it is still the most widely followed procedure in wheel-throwing. First, the vessel body was thrown right side up, whereupon the rim was formed by folding the topmost part of the clay over and pressing it against either of the surfaces.4 Thereafter, the vessel was removed from the wheel and allowed to dry enough to become leather-hard. Then, the vessel was inverted and reattached to the wheel-head so that the excess clay, which had supported the weight of the wall during the forming process could be turned off the bottom with a knife or other suitable tool (Plate 2d). The two-staged forming process, in which turning follows throwing, was destined to even the vessel thickness and ensure more homogeneous drying.5 If necessary, this stage also involved the turning of a ring handle or foot in the middle of the exterior surface. Alternatively, finishing touches were applied to the bottom of the vessel with a combed tool producing a corrugated surface. The protruding carination (Plate 2e), which is especially characteristic of shallow casseroles, was also formed by turning at this stage. Another interpretation of the same evidence suggests that the presence of a carination indicates the junction of two separately formed vessel parts.6 In fact, the presence of a carination in some African cookware vessel forms, like the Hayes 23 shallow casserole and the Hayes 197 deep casserole, led Schuring to propose that production was partly based on the use of moulds. Three possible production methods were suggested to her by a professional potter.7 Firstly, the base could have been formed in a mould, while the upper wall was thrown from a coil of clay attached to the edge of the bottom. Secondly, the vessel could have been placed inside a mould after preliminary free throwing and consecutively thrown again by pressing the clay against the wall on the mould with a ribbed instrument. Finally, the vessel

may have been formed by throwing it over a mould. Although the use of the second method provides a seemingly convincing explanation for the frequent presence of deep corrugations on the interior surface of the Hayes 197 deep casserole, moulds were unlikely used in the production. Firstly, any experienced potter could have produced these vessels, the shape of which is not complex, by throwing. The turning of the vessel bottom, on the other hand, would be superfluous stage of production if moulds were used. The combined use of moulding and wheel-throwing is risky, as the joint of a coil and moulded base would form a weak area, highly susceptible to cracking in high temperatures. Finally, free throwing with fast wheel has recently been shown to produce an irregular rilling with occasional “clay barbs” and droplets, which are in many cases seen on the interior surface of the Hayes 197 deep casserole (Plate 2f).8 Variation in the depth and shape of these corrugations indicates that sometimes the clay was lifted rapidly off the wheel-head only by hand (Plate 2f). At times the process seems to have been aided with a flat-edged tool producing less rounded ridges. The latter method allows the potter to pull up a greater amount of clay at once, as greater pressure could be used over a larger surface area. In general, this fast wheel technique is especially suitable for the production of restricted forms, while the shaping of large to very large plate-like forms, such as the Hayes 181 pan or the Hayes 26/181 pan-casserole, had to be carried out with slower rotational speed (ca. 50 rpm).9 The rounded shape of large temper particles in African cookware facilitated the forming of the vessel as well as the production of rounded shapes. The contrast is evident to the form repertory of WestCentral Italian cookwares, the production of which was chiefly based on the use of heavily tempered clay including various large angular minerals of volcanic origin, like sanidine. While rims of African cookware vessels have predominantly been finished by hand to shapes that preconceive the use of substantial compressional forces by the potter, the flanged rims of large West-Central Italian casseroles have nearly always been turned to angular shapes.10 The only exception to this rule is the group of West-Central Italian casseroles, rich in rounded quartz sand temper and presenting less angular rim profiles.11 In all, the sedimentary production environment of Roman Africa may have been advantageous also in this respect. The forming processes of African cookware may also be reviewed in relation to their usefulness, because the shape of a pot has a direct effect on its

93

Chapter 4

use. In the case of cookwares it has been stressed that the vessel profile should not include any abrupt changes, as they are the most susceptible parts to be broken either by thermal stresses or sudden impacts. Thus, it is somewhat stunning that some of the most common African cookware vessel forms, including the Hayes 197 deep casserole and the Hayes 23 shallow pans, always show a pronounced angular transition between the vessel bottom and the upper wall. Although the design was clearly influenced by metal cooking utensils,12 it may also reflect the search for equilibrium between the ease of manufacture, vessel durability and heat efficiency. A convex vessel bottom certainly improved the heat transfer efficiency of a vessel set on a tripod over a fire, but the angular joint with the vertical upper wall impaired its compactness. Similarly, the introduction of innovations, for example the use of moulds, has produced both advantages and disadvantages. Moulds may not only increase the output of a workshop by speeding up production,13 but also significantly decrease the use-life of the products, which are put together from several separately formed parts. The forming process can also be examined from the chronological point of view in order to determine the changes that have possibly influenced the vessel typology. The historical significance of this approach has occasionally been negated, because the changes do not seem to result from conscious experiments,14 as they are detectable only on the scale of several centuries. But experimentation was needed to develop new products,15 as the suitability of new forms was likely first explored locally, before they were turned to items for exportation. Additional investments to the production infrastructure were required to elaborate manufacturing techniques and to improve the quality of products. In all, as previously indicated on several occasions (see chapter 3), the forming processes of many African cookware vessels seem to have developed in a more cost-efficient direction. Only three examples are needed to underline the different aspects of this tendency: 1) the elaboration of the set formed by the Hayes 196 lid and the Hayes 197 deep casserole indicated by the simplification of both rim forms; 2) the development leading from the convex-walled Hayes 181 pan to the more open and straight-walled late Hayes 26/181 pan-casserole series; and 3) the substitution of a ring-shaped handle attachment in

cookware lids with a plain convex top, possibly indicating the use of a turning device of some kind. In spite of differences in vessel size and shape, even a heterogeneous collection of pottery may also result from contemporary output. In such cases the possible mechanisms behind the variability are manifold. Firstly, the coexistence of several workshops involved with the production of the same type of vessel may introduce variability. Secondly, it is unrealistic to presume that the potters of Africa Proconsularis would have been able to produce cooking utensils of invariable size without advanced technical equipment. The third explanation, which at first is also the most tempting one, is that cookware vessels were made in graded sizes on purpose, because a set of vessels nestled inside one another is easy to transport.16 The results obtained by González Villaescusa, who was able to demonstrate that the size of these vessels usually varies within certain limits, seem to support this interpretation.17 In terms of more conclusive evidence, one is obliged to refer to the late 1st century AD shipwreck Skerki Bank F located in the deep sea between Carthage and Rome, some 50 kms WNW of Sicily. The ship was carrying a large consignment of African cookware comprising several forms, but more importantly, they had been stacked by form: “lids and lid-plates nestled one into another, and casseroles nested one on top of another, according to size.”18 Another example of the latter custom can be pointed out from the Cabrera III shipwreck, where a set of Hayes 14B African Red Slip ware bowls was found stacked inside one another.19 Although these finds do not imply that the custom of nestling one vessel within another was a customary practice, its origins has been traced to Asia Minor, where finewares were produced in graded sizes in the 1st century AD.20 The longevity of the custom and its modern adaptation to cookware production were shown by an ethnoarchaeological study performed in the late 1980's at the village of Pabillonis (Sardinia).21 There, however, the potters were most likely adjusting their production according to demand, because most of the middlemen distributing their products traveled either on foot or by horse-drawn cart. Besides the difference in the scale of production, we can reasonably assume that too much variability in size probably produces problems both as the vessels are formed and stacked into a kiln. Further research is obviously needed to elucidate this question.

4.2 FINISHING During the last decade various archaeological experiments have shown that the performance characteristics of a cookware vessel depend strongly on

the way its surfaces have been finished. These studies have been evidently inspired by ceramic engineering, which has become a major branch of industry since 94

Production

the early 1980's. As a result of this renewed interest in ceramics, researchers have realized that even ancient pottery vessels could be analyzed as tools instead of utilizing them solely to trace down the history of a given area or to fix dates for layers, structures etc.22 Bearing this in mind, the rest of this section has been structured as follows. The interior and exterior finishes will be introduced separately, even though most techniques were applied to both surfaces. The effect of a certain finish, however, might be strongly dependent on the side it is applied to. Thus, the other, and perhaps more challenging, aim of the present section is to review how the expected performance characteristics of a cookpot established with various experiments are reflected in an actual archaeological assemblage.23 Because cookwares should obviously fulfil their purpose efficiently,24 it is fairly safe to assume that their performance characteristics received special attention. In the Roman world cooking vessels were used almost invariably by setting them on a metal tripod over a fire. Therefore, much of the variability in surface finishing techniques has to result from factors other than vessel use. For example, one could reasonably argue that the processing of certain foodstuffs preconceives a vessel with a surface finish designed to meet particular requirements. To my mind, this kind of reasoning is acceptable, because it is unimaginable that the potters would not have attempted to improve the performance characteristics of their products through the careful selection of raw materials (cf. chapter 2). Before the discussion on different surface finishes and their effect on the use of African cookware vessels, it is necessary to briefly introduce the three fundamental properties, which potters most likely tried to manipulate: permeability, heating efficiency and thermal shock resistance. Permeability is the measure indicating the rate at which a substance, either a gas or a liquid, is transferred through the ceramic body from one surface to another. In general, this assumes the existence of channel pores or pore links between vessel surfaces. Permeability is not equal to porosity, however, because a vessel may be fairly porous without being permeable, if most of the pores are closed and thus do not belong to apparent porosity.25 In a cookpot, high permeability is a negative feature, because it increases evaporative cooling – the rate at which liquids and gasses slow the transfer of heat from one surface to another – as additional energy is required to boil off the diffused water from the ceramic body.26 Elevated permeability negatively affects the heating efficiency – naturally a desired characteristic of a well-performing cookpot – of a vessel, by decreasing the rate of heat transfer. Thermal shock resistance, which is the third property, indicates the ability of a vessel to withstand thermal shocks

that result from temperature differences between the exterior and interior surface. It is also related to permeability, because as permeability decreases the magnitude of thermal shocks increases. Thus, the potters had to be familiar with various techniques of surface finishing as they sought to manufacture cookpots with good heating efficiency, low permeability and elevated thermal shock resistance. Next, the characteristic finishes of the interior surface in African cookware are introduced together with ones found usually on both surfaces. Thereafter, the discussion is concluded with the introduction of techniques found only on the exterior surface. 4.2.1 Interior surface Perhaps the most crucial factor affecting the performance characteristics of a cookpot, is the finish given to its interior surface. As a vessel is heated, the interior surface is subjected to extensive tensile stresses, which may shatter the object fairly easily unless necessary precautions are taken. Cracking, which appears usually first on the interior surface, can be prevented with a finish that reduces the permeability of the interior surface, and also improves the resistance of a vessel against thermal spalling.27 However, a cookpot with a permeable interior surface offers better resistance to thermal shocks, because a vessel body saturated with water reduces the heat efficiency. For the above-mentioned reasons, the reduction of the permeability of the interior surface is a necessity that is searched for by various methods. The quickest and perhaps also the easiest way to do this is to smooth a substantially wet vessel surface with a hand or a sponge28, though the method is obviously not the most effective one. Many vessels in the study assemblage, especially those belonging to the last phase of North Tunisian cookware tradition, have an interior surface that is something in between unfinished and slipped (Plate 2a).29 The peculiar nature of this finish, which is characteristic of many African cookware fabrics, has been previously observed by Hayes, who believes that such surface was obtained by wet-smoothing the vessel.30 This idea is supported by the study assemblage, which includes numerous vessels with the interior surface displaying a nonuniform slip-like luster that is lacking on such diagnostic points as undercuts and carinations.31 The previous statement does not imply, however, that slipping would be a completely nonexistent surface finish in African cookware. In fact, a significant portion of pans and pan-casseroles belonging both to the North and Central Tunisian tradition are slipped.32 The presence of a slip is often indicated by its color and composition, both of which usually differ from the ceramic body. This can be seen in a SEM backscatter image (Plate 1h), which shows

95

Chapter 4

a thin layer of slip – approximately 50-60 µm in thickness – applied to a ceramic body with frequent quartz inclusions. The slip was often derived from the same clay that was used to produce the vessel body,33 whereas vessels with a slip cover on the interior surface only, were probably produced by first pouring the slip inside the vessel and then dipping the rim into the solution. Slip, resin coating and other heavy-duty surface finishes have been shown to significantly improve the heating efficiency of cookware vessels.34 The other quality sought by the application of a slip was to decrease the chances that cooked substances will adhere to vessel surfaces by reducing surface irregularities.35 In this respect, slipped cookware vessels are comparable to modern kitchen utensils, which are sold with a wide selection of anti-stick surface coatings. Several vessels, both of Central and North Tunisian origin, were examined with the electron probe microanalyzer to determine the possible differences in the chemical composition between the slip and the vessel body (Table 43). Previously, slips have been observed to show constantly higher Al/Al+Si versus Fe/Fe+Ca+Mg values compared to the ceramic body,36 possibly due to the removal of quartz and calcite/dolomite (Ca/Mg)CO3 from the clay through gravitational settling. The removal of carbonates (calcite/dolomite) may also result from the levigation process, which redistributes calcium to the particles belonging to silt and sand range.37 This idea is supported by an additional observation, according to which poor slips (i.e. less glossy ones) show somewhat higher concentrations of calcium and magnesia. The lack of gloss is thought to result from the disrupting effect, which may partly result from lime spalling, that carbonates cause to otherwise strongly aligned clay minerals.38 In this sense, both samples are good examples of CaO/MgO rich slips that are also rich in potassium, which is usually found in higher concentrations in slip in contrast to clay body.39 The electron probe microanalyzer scans also coincide fairly well with the results reporting higher amounts of iron and potassium as well as lower calcium content both in the slip of Greek figured

pottery and Gallic Sigillata. The red color of the slip in African cookware is a further indication of its low calcium content, because in the opposite case the iron would form calcium iron silicates instead of remaining hematite. Calcareous clay body, however, is suggested to be more suitable for slipping, because its thermal expansion coefficient matches better the one of partially vitrified slip.40 Thus the risk of slip-crazing during cooling will be decreased. This is possibly reflected in African cookware, in which burnishing is a more common surface finish in non-calcareous North Tunisian products, while the slip is found more often on calcite-rich Central Tunisian vessels. Occasionally the interior surface was not only slipped but also burnished (Plate 2c),41 though examples showing the exclusive use of the latter technique are considerably more abundant in the study assemblage. If the combination of these techniques was used, the procedure is thought to have begun by covering the interior surface with a slip, which was subsequently burnished as the vessel was rotating on the wheel-head.42 However, as this sequence will result in a slip of inconsistent thickness and accentuate any irregularity of the underlying surface, the procedure may as well have taken place in reverse order. In African cookware, the combination is found in significant quantities only in pans and pan-casseroles belonging either to the Central Tunisian Hayes 181 or the North Tunisian Hayes 26/181 series. The resulting shiny and compact surface with reduced permeability probably had both a decorative and practical function.43 The conception of burnishing has slightly changed since the early 1980's, when it was commonly viewed as a feature without any effect upon the usefulness of a vessel.44 Today, it is known that burnishing may enhance the adherence of a slip to the vessel surface – otherwise hampered by differences in the shrinkage rate of the clay body and the slip45 – provided that it is executed carefully and the firing temperature remains low. As neither of these preconditions is met in African cookware, it is more likely that the optical effect produced by alternating bands of lustrous and matt surface was of some advantage in the selling of the product.

Table 43. Comparisons of slip/vessel body composition (%) Fabric I.1 I.4 II.1 II.2

Form

Type

Na2O MgO

Al2O 3

SiO2

K2O

G → G → H.II → I.I →

body slip body slip body slip body slip

0.573 2.854 0.658 2.808

22.004 54.580 2.733 4.012 0.741 0.011 23.741 55.101 3.208 0.868 0.550 -

0.666 2.431 2.126 1.571

19.697 48.297 2.437 1.601 0.648 0.048 0.020 0.052 5.023 0.007 0.080 19.649 42.596 6.969 1.420 0.669 0.020 0.025 0.023 7.651 0.005 0.090

0.178 1.571 0.223 1.413

23.967 47.096 1.954 0.628 0.545 0.031 0.018 23.832 42.242 2.566 0.595 0.783 0.091 0.020

0.170 1.652 0.211 1.586

23.023 48.105 1.963 1.072 0.689 0.055 0.042 0.050 3.822 0.020 0.012 21.150 45.529 1.835 0.525 0.892 0.121 0.016 4.667 0.061 0.024

96

CaO

TiO2 V 2O 3 CrO 3 MnO FeO -

NiO

ZnO

0.042 8.018 0.007 0.074 8.292 0.020 0.093

-

2.972 0.027 0.031 5.424 0.005 -

Production

Tortorella already had observed that the function of these vessels is clearly indicated by the distribution of the slip, which is usually strictly limited to the interior surface.46 It has also been deduced that slipped African cookware forms are either casseroles or pans.47 The idea is acceptable, although there are two important exceptions, the so-called lid-plates and the Hayes 197 deep casserole, which do not show any evidence of a finish reducing the permeability of the interior surface. In the former group, the permeability of the interior surface must be considered a negative feature inhibiting the use of these vessels as plates, because an unfinished interior surface absorbs flavors easily. Although they might have occasionally served as plates in the same way that cowboys exploited their handguns as hammers, it would be reasonable to define this class in the future as lids. Similarly, the proposition that certain surface finishes would correlate with particular African cookware lid forms has recently been refuted by Aguarod Otal, who was able to demonstrate rather convincingly that corrugated, smoothed and unfinished surfaces have a random distribution in this class.48 The interpretation of the Hayes 197 deep casserole is markedly more difficult, because besides being unfinished, it is nearly always – in 85% of the cases in the study assemblage – covered with deep rilling marks that most likely resulted from free throwing with fast wheel (Plate 2f, see also section 4.1). In cooking vessels such a combination is impracticable for two main reasons. Firstly, an unfinished interior surface is permeable and thus scarcely heat efficient as much of the heat is consumed to evaporate the moisture absorbed by the walls. Secondly, the rilling on the interior very likely promoted the sticking of the contents to the vessel surface. In all, just by judging the interior finish it is hard to imagine a task for which this form would have been suited, because it seems to exclude both the cooking of liquids and mushes.

Central Tunisian Hayes 182 lid with an outward flanged rim (C.I). From the aesthetic point of view this is understandable, because with the Hayes 181 pan the lid variant in question formed a set that may have had a secondary use on the table (see form G). Just as the exterior surface of Hayes 182 is often slipped and burnished, so is the interior surface of its counterpart. Burnishing without a slip is also found on the exterior surface of many other forms than the Hayes 182 lid. Not only is it frequent on the exterior of Central Tunisian variants of the Hayes 23B and the Hayes 181, but also the North Tunisian set of the pan and lid – the Hayes 26/181 pan casserole and the Hayes 195 lid – and even some Hayes 197 deep casserole variants seem to bear it. However, the study assemblage does not give any support to the idea suggesting that the use of this technique would correlate with a coarser clay paste.50 In light of the recent experiments, it is also conceivable that cookware vessels were burnished to increase their strength. A compression resulting from the differential shrinkage rate of burnished areas by contrast with the vessel body is believed to absorb some of the torsion and impact forces, and the increase in the strength is reported to be as much as 20%.51 However, burnishing will work most efficiently when applied to the opposite side of the wall receiving impact forces, which is usually the exterior surface.52 The only surface finish found exclusively on the exterior surface of African cookware is a set of shallow concentric corrugations covering the bottom of most shallow and deep casseroles forms (Plate 2e, 2g). In this context, Hayes has made an important chronological observation, according to which corrugated exterior and slipped interior surfaces were introduced simultaneously.53 The combination of these finishes is necessary, because if the permeability of a vessel is reduced by slipping the interior, this will result in thermal spalling on the exterior unless preventive measures are taken. The danger can be overcome by enhancing the exterior surface with corrugations that provide more resistance against thermal spalling than other surface finishing techniques or unfinished surfaces. At present it is unclear whether or not this results from the increase of the exterior surface area.54 The anti-spalling quality is not the only function of a corrugated exterior surface. It may also have strengthened the vessel against strains developed on drying stage or improved its thermal shock resistance.55 The mechanisms behind these qualities are not yet known, although textured surfaces are supposed to distribute stresses differently in comparison with smoothed ones.56 It has also been pointed out that corrugations increase the surface area of the vessel bottom in African casseroles by ca. 10%,57 and thus would have improved their heating efficiency.58

4.2.2 Exterior surface The number of finishes applied to the exterior surface of African cookware vessels exceeds by one those used to the vessel interior. Compared to the function the exterior finish of cooking vessels is somewhat less important. For example, it does not improve the heating efficiency of the vessels effectively, unless the interior is made impermeable. If the permeability is reduced from the exterior surface only, it may well improve heating efficiency but also cause thermal spalling, which is without doubt a negative side effect.49 Hence, it is understandable that in terms of finishing, practically all the African cookpots but the Hayes 197 were given an impermeable vessel interior. Compared to the vessel interior, the exterior surface is seldom slipped. In fact, the only form customarily showing a slipped exterior surface is the

97

Chapter 4

This idea has been refuted in archaeological experiments, which have succeeded in showing that a corrugated exterior surface rather decreases than increases the heating efficiency of a cookpot. The same experiments have also indicated that a corrugated vessel surface does not increase evaporative cooling significantly, which was long regarded as one of its disadvantages.59 The alternative theories concerning the function of these corrugations pay more attention to other practical aspects than the heating properties. For example, corrugations may reduce the warping of the vessel in firing by imparting its strength and rigidity. On the other hand, corrugated exterior bottoms of the Hayes 23B shallow casserole and the Hayes 197 deep casserole have been proposed not only to promote the vessel stability in use, but also to facilitate their handling.60 Another possibility is to assimilate these corrugations with decoration, because textured surfaces are not as easily obscured by the soot as the plain ones, even though texturing makes keeping the surfaces clean particularly difficult.61 The observation is supported, to a certain extent, by the study assemblage, where examples of African cookware are generally less sooted than the West-Central Italian ones. This convenience of surface finishes may have even increased the market value of African cookware.

Finally, attention must be turned to the absence of all three forms of visual/informative surface alteration: decorations, inscriptions and stamps. In African cookware this is hardly casual, because when this question is reviewed against the Roman cookware production as an integrated whole, the absence of these features from cookpots becomes more apparent. Besides some pie-crust decorated forms, the date of which is mostly Republican, decorated cookware vessels are practically absent, neither frequent nor extremely rare, especially when compared to table- and utilitarian wares.62 Stamped maker’s marks, on the other hand are rare in cookware and absent in African cookware. To my knowledge, not a single stamped African cookware vessel has been published to date. The general lack of stamps on ceramic products of Roman Africa probably reflects the mode of production, which in all likelihood differed markedly from the hierarchical system used in France and Italy. As the presence of maker’s marks is clearly related to the organization of the workshops, this theme will be touched on more profoundly in section 4.4. Inscriptions are also rare but not non-existent. In addition to one letter on the top of a lid at Palatine East (Plate 6:29), a longer inscription scratched postcocturam to indicate the owner of the vessel (LICIPO) was found in Murillo de Calahorra in Spain.63

4.3 FIRING The method of firing pottery in the Roman world was strongly dependent upon the scale of production, because what was good for a small household production did not necessarily meet the requirements of nucleated workshops or manufactories, which produced vessels for exportation. Mass-produced pottery was predominantly fired in large updraught kilns – of which many local variants existed throughout the Empire – where the fuel and the products to be fired are separated from one another by a partition perforated with vent holes.64 Roman Africa was not an exception. Recently conducted excavations at a pottery production site near Leptiminus have shown that a workshop may have operated with several kilns, some of which were even reserved for firing certain pottery types only.65 Furthermore, it is hardly chance that in present-day North Africa the area of the most intensive use of updraught kilns coincides with the former provinces of Zeugitana and Byzacena,66 the major pottery producing areas of Roman Africa. Since a detailed examination of the structure of updraught kilns is beyond the scope of this study, attention must be turned to the firing process, which can be divided into two parallel themes: the firing atmosphere and the firing temperature. These topics

will be introduced independently, because their examination requires a different set of variables. As the firing atmosphere is discussed, the main emphasis is put on the changes visible in the cross section of the ceramic body, whereas the appearance of various mineral inclusions is of secondary importance. Reliable information on the firing temperature, on the contrary, can only be obtained by examining the state of certain marker minerals and the condition of the ceramic body, whereas color determinations are useful only if the samples have been re-fired to fixed temperature intervals. 4.3.1 Atmosphere The study assemblage does not offer overwhelming possibilities for speculation on the nature of firing atmosphere, because with the exception of the last stage of firing, over 90 % of the vessels were fired in a completely oxidizing atmosphere. The duration of firing was usually long enough to result in a red ceramic body without a somewhat darker core, which is sometimes seen in Central Tunisian pottery (Plate 1g), but considerably more often in West-Central Italian cookware. Hence, the only evidence on the alteration of the firing atmosphere can bee seen on the uppermost exterior surface (see Plate 2h), which is 98

Production

frequently blackish or brownish in color due to the reducing atmosphere created in the kiln during the last stage of firing.67 Because the vessels were loaded into the kiln in stacks, only a small portion of their exterior surface was exposed to a reducing atmosphere, as the reconstruction of the small stack of Hayes 23B shallow casseroles clearly shows (Figure 17).68 The resulting reduced zone was once used as a criterion to separate the class of African cookware into two fractions: a) ceramica a patina cenerognola i.e. “sootpatined surface ware” including all forms but lids and b) ceramica ad orlo annerito i.e. “blackened-rim ware” lids.69 The study assemblage supports the hypothesis concerning the origin of this surface, as it shows that the tone of patina cenerognola is strongly dependent on the fabric of the vessel. Whereas the reddish-orange North Tunisian fabric is customarily covered with the patina cenerognola proper that ranges from brownish to black in color, the dark red Central Tunisian fabric often bears an additional whitish-gray surface deposition (Plate 2i), which can be defined as a scum. As will be subsequently demonstrated, this kind of surface results from the application of sodium chloride used to neutralize the disadvantages produced by calcareous inclusions in the Central Tunisian fabric. As the grayish-black patina is a regular feature on exterior surfaces of African cookware vessels, it was more likely produced by intentional actions rather than an accidental spot reduction taking place in those parts of a kiln, where the circulation of the air was restricted due to tightly packed ware.70 Hence, the layer may have also had a function of some kind,71 at least the quality of the patina has been observed to improve in the course of time.72 In addition to several explanations, all of which stress the functionality of this surface,73 the absence

of this firing effect from the imitations of African cookware is said to show that these vessels were designed to fulfil another function than their prototypes.74 If the function is determined by the presence of a patina cenerognola, the natural implication is that the surfacing itself is functional. Firstly, a blackened exterior surface may improve the heat retention capacity of a vessel.75 Although the observation is probably valid, whether or not the property itself was improved enough to be recognized by the users remains to be resolved. The idea regarding the external blackening as some kind of decoration is more bizarre, even though aesthetics may have played some role in the design of cookwares. The possibility that African table- and cookwares could have been distinguished from one another on the basis of this surface is also worth mentioning. Because the black surface efficiently masks the soot deposited on the vessel exterior, patina cenerognola may also have been a convenient way to improve the appearance of a cookpot. Burnished bands on the vessel exterior have also been related to the presence of patina cenerognola, as this surface treatment is supposed to strengthen the bond between the wall and the cover.76 Although the idea itself is very interesting, it cannot be accepted for the following reasons. Firstly, as methods of surface alteration patina and burnish were probably destined to modify different properties: the former heat efficiency and the latter strength. Secondly, the study assemblage does not only include dozens of burnished vessels lacking any trace of patina cenerognola, but also the vast majority of the examples bearing the latter surfacing (84.6 %) are not burnished at all. The divergent color of burnished stripes on the vessel exterior is more likely explained by their compact surface that created a sort of protective layer against the reducing atmosphere in the last stage of firing.

Figure 17. The mechanism of patina cenerognola –deposition shown with a stack of Hayes 23B’s.

99

Chapter 4 Table 44. Two comparisons of patina cenerognola/vessel body composition (%) Fabric II.1

II.1

Form

Type

Na2O MgO

Al2O 3

SiO2

K2O

A.I

body

0.216 1.378

20.764 50.191 1.873 0.531 0.575

→

patina

0.603 0.931

20.133 57.337 5.883 0.479 0.641 0.017

J.II

body

0.215 1.268

22.363 46.205 1.706 0.481 0.688

→

patina

0.200 1.326

20.540 48.885 1.759 4.741 0.507 0.057 0.006 0.039 3.166 0.016 0.055

The possibility that patina cenerognola could result from the application of slip-like solution to the surface can also be excluded. In the sections examined with a scanning electron microscope, no transition was observed between the body and the exterior surface. Moreover, the series of electron probe microanalyzer scans taken from the surface towards the interior of the vessel does not replicate changes in chemical composition that are characteristic of a slipped vessel (Table 44). In fact, the decrease in the percentage of aluminum and the increase in calcium towards the exterior surface would suggest exactly the opposite. Neither do the results support the idea about the reaction of lime and iron in high firing temperatures, which would make the latter substance to volatilize.77 Although patina cenerognola covered exterior surfaces are also found in Central Tunisian products, a more characteristic feature for them is a white surface coating or scum (Plate 2i), which is present both independently and with the patina cenerognola. 78 The regional character of the surfacing is suggested by casseroles found in the excavations of the Roman cemetery at Raqqada (Henchir el Ouiba),79 which all bear this feature. The surfacing has been experimentally replicated by Peacock, who demonstrated that it resulted from soluble salts present in clay, or from salt water or raw salt used to prepare calcareous clay paste.80 Experiments have also shown that the formation of salt scum is highly dependent on the drying process. Drying may also explain why scum is usually found only in the rim area, because besides the migration of salts from the body to the surface, the way in which the vessels have been stacked in the kiln has very likely contributed to the creation of this grayish or yellowish zone near the rim.81 The experiments carried out within the modern brick industry have confirmed the mechanism by showing that scum deposition is intimately related to the use of sodium chloride in the preparation of clay paste.82 As mentioned in section 2.5.2, adding salt to the ceramic paste was a preventive measure to avoid limespalling, but when deposited on the exterior surface and exposed to a high firing temperature, it also produced a whitish scum, which is highly typical of Central Tunisian amphorae. Even as an accidental byproduct, it may have also served practical purposes, as

100

CaO

TiO2 V 2O 3 CrO 3 MnO FeO -

-

0.026 0.003 2.854 -

0.010 1.935

NiO

ZnO

-

0.005

-

0.008

0.012 0.000 3.379 0.018 0.052

white-surfaced vessels are proposed to have attracted customers more than vessels without such coating.83 Alternatively, the use of salt water has been seen as a convenient way to spare rare and thus valuable stocks of fresh water. The presence of a white surface coating does not imply, however, that the vessel was manufactured on a coastal production site, because saline water was also available at kiln sites located near the shore of a saline lake, like lake Sebkhet Sidi El Hani in Central Tunisia.84 The most fascinating theory concerning the use of salt has been put forward by Rye, according to whom the preparation of food in a salt-tempered vessel may have improved its taste.85 Even so, the presence of whitish scum on the exterior surface probably did not affect food preparation, since the surface was more likely in contact with the fire than with the ingredients. 4.3.2 Temperature The firing temperature of archaeological ceramics has often been sought by refiring sherds in determined temperature intervals and monitoring the change in their color. The method suits well for the testing of iron-rich clays,86 which is what most North African clays are. The application of this method has shown that African Red Slip ware was usually fired to circa 1000 °C,87 while other experiments comparing the firing temperature of African cookware with that of West-Central Italian cookwares have indicated that African pottery was often fired at a higher temperature than the corresponding production of the Italian peninsula. Further experiments have also pointed out that in Roman Africa tablewares were usually fired at higher temperatures than cookwares.88 Still, neither this observation nor the absence of patina cenerognola in African Red Slip ware establish enough firm ground for a straightforward claim that table- and cookwares were rather fired in separate loads than mixed together in a kiln. If the finer Red Slip ware was fired in saggars or placed loosely over fairly tightly packed cookware vessels at the top of an updraft kiln, which gets a good deal hotter in firing than the bottom, the above-described differences may have been obtained in a single firing.89 The absence of patina cenerognola from most examples of Hayes 197 minor, a small and relatively fragile casserole that must have been one of the uppermost forms in a kiln

Production

load, possibly reflects this custom. Reduced surfaces similar to African cookware occasionally seen in African amphorae,90 on the other hand, could indicate that they were fired together at least from time to time. The firing temperature of African cookware, according to Schuring, usually ranges from 850 °C to 1050 °C, while van der Werff’s proposition for Central Tunisian coarsewares is 990-1070 °C.91 The considerable variation in firing temperature is explained by the use of large kilns, which are known to be susceptible to substantial temperature variations even in a single firing.92 The decrease in the firing temperature observed towards the early Medieval period, on the other hand, has been taken as evidence of small scale producers making an effort to save fuel by employing smaller kilns.93 High firing temperature produces a compact ceramic body,94 that not only affects the performance characteristics of the pottery, but also raises the production costs by increasing the consumption of fuel per firing.95 Therefore, the difference in the firing temperature between African table- and cookwares has been said to indicate the cost-efficient thinking of the potters, who attempted to reduce their expenses by firing cookwares at a lower temperature.96 Although this explanation is acceptable, it is equally possible and perhaps even more probable that the production of too tough and thus friable cookware fabrics was avoided with lower firing temperature. This must have been the case in Central Tunisia, where the combination of high firing temperature and calcite-rich clay would have resulted in vessels that were too rigid to be used for cooking. 4.3.3 Fuel Due to scale of the pottery production in Roman Africa, large quantities of fuel were obviously needed to operate the kilns. Although pottery has traditionally been seen as one of the main reasons for deforestation in antiquity,97 a more ecological view of this industry stressing its symbiotic relationship with the environment, has emerged lately.98 Today, pottery kilns in Roman Africa are assumed to have been fired with agricultural by-products – such as olive prunings and pressings – that were certainly plentiful both in North and Central Tunisia.99 The idea was first based on ethnoarchaeological data, of which the most interesting were the bits of information on pottery production in present-day Tunisia. In the modern villages of Nabeul and Moknine as well as in the island of Djerba, both the waste of olive pressing and the prunings of date palms and olive trees are extensively exploited in the ceramic industry.100 While the use of alternative heat sources in the ceramic pyrotechnology seems to be a universal phenomenon,101 the exploitation of by-products of oleoculture is common around the Mediterranean. The

101

use of olive prunings as a fuel of pottery kilns has been documented e.g. in Bailén, Southern Spain and in the western shore of the Messenian gulf, Greece.102 Today, the use of agricultural by-products as fuel in Roman Africa is also supported by archaeological evidence, as deposits of ash containing carbonized olive pits have been encountered at several excavation areas in Leptiminus.103 While the nature of these deposits was doubted at first, they were consecutively interpreted as dumps of used fuel resulting from the intensive and well-organized exploitation of agricultural by-products in pottery production.104 The most indicative find was made from a service chamber within a large kiln complex, where five transport amphorae filled with burned olive pits, twigs and olive wood charcoal were found.105 In all, the evidence from Leptiminus suggests that by-products of oleoculture were exploited in this area on regular basis.106 The most important question regarding the use of agricultural by-products in Roman Africa concerns their availability. When it comes to olive pressings, the answer is depended mainly on two factors: the average annual yield of an olive tree, and the quantity of pressings produced per every kilo of pressed olives. An estimate of 20 kilograms of olives per tree is a good starting point regarding the annual yield, as both the biannual variation of the yield and the effects of drought, immature and decrepit trees are taken into account.107 At the estimated density of 40 trees, a hectare will yield 800 kilograms of olives.108 When 80 metric tons of olives – the annual produce of one square-kilometer olive orchard – is processed to oil, it will result in 32 tons of olive pressings, 32 tons of water and 16 tons of olive oil. By extrapolating this figure to the province of Byzacena, even at the 10% density, the area of 4000 square-kilometers has hosted as many as 10 million trees.109 Thus, the annual amount of production can be estimated as 200 000 tons of olives, the processing of which yielded 80 000 metric tons of olive pressings.110 In addition to solid olive pressings, another potential source of fuel was amurca, a sticky, black liquid residue that had also other uses in the Roman world.111 Only some decades ago in Peloponnesos, Greece, this sludge (pirini) was greatly appreciated as fuel by potters due to good heat and little ash produced in addition to the reduced firing time of the kiln.112 Olive pressings share these properties, probably due their residual oil content.113 Clippings and prunings must have been another preferred source of fuel, because they “do not build up a bulky mass of glowing slow-burning charcoal in the combustion chamber as would heavier wood”114. Yet another potential source of fuel may have been esparto (alpha) grass, the growth of which has been observed to coincide with areas of olive cultivation.115 Finally, the possibility to cultivate other crops between olive trees

Chapter 4

may have contributed to the use of cereal straw and chaff as fuel.116 Like today,117 the fuel of pottery kilns in Roman Africa seems to have consisted of agricultural residues, shrubs and basically everything that was available but wood. The reason was not necessarily the scarcity of firewood, but the availability of inexpensive and superior quality alternatives for this purpose.118 An olive orchard is relatively easy to maintain, because the most labor consuming tasks are the harvest and the consequent pruning of the trees, both of which take place in winter.119 Prunings are quick to dry, whereas firewood may need a year to dry properly.120 Although more firewood was probably consumed

in the use than in the production of cooking vessels in Roman Africa, the abundance of fuel in general is most likely related to the favorable climate, which saw long production seasons in contrast to the northern shores of the Mediterranean.121 In fact, pottery was most likely produced in Africa Proconsularis all year long, just like it is done today in the nucleated workshops of Nabeul and Moknine.122 A question certainly requiring further attention is, whether the seasons dictated the type of pottery produced. It can be assumed that amphorae were most likely produced in the summer, while more damp months directed the production towards smaller forms i.e. cook- and finewares.123

4.4 PRODUCTION SITES AND PRODUCERS 4.4.1 The nature of evidence This section will address the location of pottery production sites and the status of the potters in the Roman North Africa. Neither of these themes has received much attention until lately. Particularly embarrassing is our poor understanding of the organization of pottery production,124 especially when we think of the abundance of Red Slip wares and amphorae on many post 1st century AD sites in the western Mediterranean. As the study of production sites has also been conditioned by these two groups of pottery,125 they represent the only realistic possibility to infer something about the location of cookware production sites in Roman Africa. Otherwise, we have no other option than to repeat the common statement according to which the main production centers of African cookware were located with a fair degree of certainty within the boundaries of modern Tunisia.126 The identification of the source of a given form on the basis of geographical frequency is also hard due to lack of publications presenting the pottery found in important urban centers other than Carthage.127 Carthage is also referred to as an important pottery production center, although no archaeological evidence referring to the existence of Roman kiln sites has been found to date.128 However, while the surroundings of Carthage – including the Gulf of Tunis and the valleys of River Medjerda (Bagrada) and River Miliane (see Figure 18) – have been said to possess the best resources to host pottery production in North Tunisia,129 the paucity of suitable clay deposits in the Gulf of Tunis may have favoured the production in the Medjerda River valley.130 This is likely reflected in calcium-poor North Tunisian fabric, which suggests rather continental than marine origin of the clay.131 These expectations have also been met in the River Medjerda valley, where pottery production centers of varying certainty have been identified near the town of Tebourba (Thuburbo Minus) and on 102

several rural sites. The idea that most production centers were located in northern Tunisia132 is equally supported by the study assemblage, where the North Tunisian production outnumbers the Central Tunisian one by a ratio of ca. 1:3. Cookware vessels with a fabric resembling ARS A, A/D or D have usually been identified as North Tunisian, while similarities to ARS A/C, C as well as to some variants of ARS D are said to indicate Central Tunisian origin.133 In the case of Central Tunisian production sites, coastal location in the proximity of the main ports of exportation has usually been emphasized.134 The area south of Sahel does not seem to have offered many possibilities for large-scale pottery production with the exception of the island of Djerba, where the exploitation of local high-quality clay deposits probably began by Roman times.135 Certain surface finishes, such as stripeburnishing or slip, and firing effects – patina cenerognola deposition in particular – have also been linked with the vessel origin.136 The finds made in Uzita, Leptiminus and Pheradi Maius – a town located on the coast midway between Leptiminus and Carthage – have given concrete evidence on the production of stripe-burnished pottery in coastal Central Tunisia.137 Differences in the surface finish are also claimed to identify the output of workshops utilizing an identical mixture of clay and temper.138 Finally, the location of production sites may also be reviewed through the geomorphology of Tunisia, because the access to sufficient supplies of clay, water and fuel generally regulate the distribution of the craft.139 Although the information acquired by studying the output and other external evidence may be of great help, definitive answers can only be obtained by examining the actual remains of pottery production sites. In spite of a promising beginning in the

Production

Utica

Carthage Thuburbo Minus

ZEUGITANA

TUNIS Henchir el Biar

El Mahrine

dje

rd

a

Oudna

Mi l ia ne

Me

Nabeul

Thuburbo Maius

Siliana

Thugga

Pheradi Maius

Hadrumetum Uzita Raqqada

Leptiminus Moknine

Sidi Marzouk Tounsi

BYZACENA

Sullecthum Thysdrus

Henchir es Srira Bararus Sufetula Acholla

fig. 19

Cillium

SFAX Thanae

TUNISIA Modern city Ancient town/site GABÈS

Modern pottery Roman kiln site(-s)

DJERBA Guellal

Chott el Djerid

Province border

Figure 18. Tunisia: modern cities and potteries, Roman towns and kiln-sites 1890's,140 production sites were long taken into account only in regional surveys, which did little more than enrich the catalogue of sites.141 They were only brought into focus in the mid-1980's, when the British-Tunisian survey in Central Tunisia managed to locate over thirty kiln sites (Figure 19), mainly on the

103

basis of surface finds.142 Before the next stage of kiln site studies – involving large-scale archaeological excavations on selected sites – was reached, Mackensen published an important study based on the surface material of Late Roman production site complex at El Mahrine, North Tunisia.143 The first

Chapter 4

SULLECTHUM

HADRUMETUM KAIROUAN UZITA

THYSDRUS SUFETULA

BARARUS

output includes cookwares producing other pottery

project focusing on the excavation of pottery production sites has been launched only recently in Central Tunisia, where a Tunisian-American-Canadian team has explored a concentration of production sites located on the margins of the Roman city of Leptiminus (Lamta).144 The accumulation of first hand data has also started to result in the publication of summaries covering the production site evidence from a certain geographical area or chronological period.145 Both the kiln site evidence and the compositional analyses of pottery have confirmed that cookwares were often produced in the same workshops with tablewares and amphorae.146 Therefore, there is an attempt to couple the rapid decline of African cookware exportation with the roughly synchronous change in the location of production, that is said to have taken place in Central Tunisia by the early 5th century AD. Although the change is not apparent in all areas, production centers located on the coast near the principal ports were slowly replaced by rural inland kiln sites that very probably belonged to large farms or villas.147 The shift has been proposed to reflect a new, less centralized and less controlled mode of production.148 It has also been associated with the invasion of Vandals, which is known to have changed the agricultural economy of Zeugitana and Byzacena.149 The evidence on pottery production in Roman Africa can therefore be divided into two types by location. The differences in the mode of production are probably reflected in this dual distribution of kiln sites, which is examined next more closely. 4.4.2 Urban nucleated workshops The first production site type is associated with the phase when African provinces were bound to supply the city of Rome yearly with an impressive quantity of fiscal goods, which appear to have been accompanied by commodities meant to be sold for profit (see section 5.3.1). As a considerable part of the 104

Figure 19. Roman kilnsites located by the British-Tunisian survey in Central Tunisia (after Peacock et al. 1990, p. 60 fig. 1).

income was derived from the production of transport amphorae utilized in the shipment of fiscal loads, a kiln site certainly profited from coastal location adjacent to a major port or a center collecting exported foodstuffs.150 Especially in Central Tunisia, pottery production sites dating earlier than the 5th century AD have been noticed to form clusters around the ports linking active routes of redistribution or trade.151 Besides geographical proximity, such clusters or nucleated industries152 usually benefited from an extended production season and developed ancillary services, including an ensured supply of fuel and reduced transportation costs. Hence, it is not surprising that most amphora-producing kiln sites on the Sahel coast were involved with the production of other coarsewares, although their distribution is neither consistent nor patterned.153 The work of the international expedition at Leptiminus has been particularly important for the better understanding of coastal pottery production sites, as the chronology of the sites excavated so far covers the period of pottery production in Roman Africa. In addition to the main product, amphorae, the output of these sites seems to have included coarsewares, finewares and even building materials like tiles and pipes.154 The largest site with eight kilns, of which six are aligned and several have been excavated, dates from the 1st to the 3rd century AD.155 Cookware wasters, especially casseroles, and unfired coarseware rim sherds found in connection with certain kilns, suggests that each pottery type may have been fired in a separate kiln (cf. section 5.3).156 The abundance of kilns at Leptiminus also shows that a large-scale production relied on the flexibility offered by the use of several kilns, because even in favorable conditions the firing cycle of a kiln may occur over several days.157 Production based on one high-capacity kiln, on the other hand, would have been hazardous due to accelerated consumption of fuel,

Production

decreased control over firing and thus increased firingloss rate. It is plausible that such dense clusters of kiln sites on the coast were formed of independent workshops.158 The rarity of stamps in Roman pottery made in African provinces is thought to reflect decentralized production, as a manufacture based on small units does not need a formal craft organization or a system to identify the output of an individual worker or workshop.159 Had the production of pottery been organized effectively, the full-time professional potter did not have anything else to sell than his labor and therefore he was completely dependent on the merchant.160 An indicative feature is the prevalence of toponomastic stamps in rarely stamped African amphorae between the 1st and 4th century AD, which is said to refer to a production controlled by the city or the state.161 The possibility that at least some operations related to pottery production were centralized if not communal is strongly suggested by the Leptiminus excavations. There, in addition to features related to the collection of water, a rectangular (14.5 x 14.5 m) clay preparation tank holding ca. 300 tons of clay was discovered in the base of a dry wadi (stream) adjacent to production sites (cf. section 5.3).162 4.4.3 Decentralized rural production The importance of North Africa, Central Tunisia in particular, as one of the pivotal areas in the Roman economy, started to decline sometime between the late 3rd and the 4th century AD.163 The ecological stress resulting from intensive oleoculture was probably reflected in declining harvests,164 causing the retreat of olive groves from the coast of Sahel towards inland areas. The resulting development, which was further complicated in the early 5th century AD by the Vandal conquest of Roman Africa, possibly led to the decentralization of pottery production. In Leptiminus, kiln sites that had previously ringed the city were relocated near the seashore, where their new location is said to have guaranteed access both to salt water and the sizeable harbor of the city.165 Elsewhere, nucleated industries were increasingly replaced by estate production where kiln sites were predominantly located on the margins of intensively cultivated areas, large enough to guarantee constant fuel supply.166 A countryside village adjacent to a large villa (vicus circa villam) as well as the villa itself probably had enough economic potential to host the production of pottery.167 A system with an independent craftsman contracting over a certain production with the landowner (possessor fundi) was particularly well suited for estate production in Africa Proconsularis.168 First of all, a landowner could better exploit the assets of the estate without great expenses or direct involvement through the brick or pottery

105

production.169 The landowner could also have had access to external markets by owning means of transportation or the pottery was purchased by itinerant merchantmen.170 Secondly, the estate manager (vilicus), regardless of whether the workshop manager (officinator) was his subordinate, collaborator or alter ego,171 could fill gaps in the agricultural calendar with the laborintensive tasks the landowner was responsible for: the supply of raw materials and fuel.172 If pottery production had to be suspended in the winter due to an unfavorable weather conditions, workers could participate in olive growing in its most laborintensive season.173 Estate production may also have led to specialization, because every estate did not necessarily produce amphorae, as some estates probably produced enough surplus to supply neighboring estates.174 Traditionally, pottery production has been looked upon as a craft based primarily on the exploitation of slave labor, which is said to have been ineffective, because of its inflexibility and lack of coordination.175 Still, it has been seen as the prevalent mode of production in the Roman economy, although the supporting arguments are often one-sidedly based on the situation in Italy during the late Republican or early Imperial periods. The problem with the provinces is that we possess neither supporting or denying evidence on the use of slave labor, although the operation of large estates probably depended on them.176 But while the high number of slaves in a given province does not automatically imply their involvement in pottery production, the exploitation of slave labor cannot be excluded,177 just by assuming that the mode of production in Roman Africa had to be more effective than in Italy (see also section 5.3.1). Whatever the exact mode of production was, the proximity of a kiln site to the center of agricultural production or refinery, especially to those practicing oleoculture, was a considerable advantage with regards to the availability of fuel, manpower and capital.178 The idea that pottery production is related to oleoculture, because it provided a means of livelihood in those 5-10 years before a planted olive tree can be harvested for the first time,179 is more difficult to accept. But the symbiosis of the agriculture and pottery production is particularly evident in Central Tunisia, where the distribution of the late Roman kiln sites follows fairly closely the pattern of estates involved in the production of olive oil.180 Another example is the El Mahrine kiln-site, which is located in the Medjerda River valley near Tebourba in North Tunisia. The site was probably incorporated into a large agricultural estate (fundus), the main product of which could have been olive oil, since the area is dominated today by olive groves and orchards.181 Production consisted mainly of African Red Slip ware, while the rare examples of African

Chapter 4

Distance (kms) Route

land

river

sea

Costs

Vessels

Carthage-Ostia

-

-

570

570

1000

Hadrumetum-Ostia

-

-

665

665

858

Hadrumetum-Carthage-Ostia

-

-

798

798

715

Thuburbo Minus-Carthage-Ostia

-

93

570

933

611

Thuburbo Minus-Carthage-Ostia

53

-

570

2054

278

Sufetula-Hadrumetum-Ostia

152

-

665

4921

116

Sufetula-Hadrumetum-Carthage-Ostia

152

-

798

5054

113

cookware are forms that have been seldom found on the northern shore of the Mediterranean. The products of the El Mahrine kiln site were predominantly transported by land on wagons and carts, while the Medjerda River was only navigable in the wintertime.182 As this production site is located some 45 kilometers southwest of Carthage, the scarcity of its products (North Tunisian casserole variants R.II & R.IV) in the study assemblage indicates that at least the distance affected the distribution of cookwares. This did certainly not result from the absence of transport infrastructure, because in this terrain with few rivers, a great deal of statesubsidized agricultural products (annona) had to be taken from one place to another by a wagon post (cursus clabularius).183 The combined transportation by land and sea was by no means exceptional either.184 4.4.4 Transportation costs The new economic reality, the main principle of which was independence, directed the pottery production more tightly towards amphorae and tablewares. The making of transport amphorae was self-evident, because self-sufficiency on them was reflected in reduced costs of transportation, while the fabrication of other kinds of ceramic goods produced extra income.185 As many production sites were located in fairly remote places, the production of tablewares was certainly more profitable per piece when compared to robust and heavy cookware vessels. The difference can be illustrated by applying the sea/river/land transport

Table 45. A comparison of transportation costs

cost index (1/4.9/28) developed by Duncan-Jones and approved by Greene, to the shipment of cookpots from various locations of Roman Africa to Ostia along different routes (Table 45).186 The comparison of transportation costs underlines both the advantageousness of a production site located on the coast and the disastrous effect the land transportation had on the cost of merchandise. Thus, the shipment costs of a thousand cookpots from Carthage, the most important outlet,187 to Ostia would have been equivalent to the costs of shipping slightly over a hundred cookpots from Sufetula to Ostia. In fact, the land transportation of goods from Sufetula to Hadrumetum alone was over six times more expensive than their maritime transportation from Hadrumetum to Ostia.188 In spite of high transportation costs, the manufacture of African Red Slip ware for exportation continued in inland production centers from the mid5th to the early 7th century AD.189 This is clearly reflected by the Sidi Marzouk Tounsi kiln site, which has produced evidence on the manufacture of exported African Red Slip ware forms.190 Its location in Central Tunisia over a hundred kilometers distant from the coast underlines the reason why the site, dating between the late 4th and the first quarter of the 6th century AD, did not yield any evidence on the production of cookwares. Thus, besides world politics, both economic and ecological factors seem to have contributed to the cessation of African cookware production.191

4.5 EVIDENCE OF SPECIALIZATION? Despite the transparent ethnocentrism included in the following definition,192 the features introduced above classify African cookware as a high-quality massproduced good with optimized performance characteristics. It remains to be shown, however, that its success was not only dependent on craft

106

specialization but also on standardized production achieved through the utilization of certain techniques of manufacture.193 Specialization is a topic that has recently been the focus of archaeological ceramology to such a extent that four types have been distinguished: site, resource, function and producer-

Production related specialization.194 The one the production of African cookware corresponds most closely with is resource specialization, but in contrast to a normal case the production of African cookware was based not only on the availability of suitable clay and temper, as also the abundance of suitable fuel and favorable climate were probably other factors boosting the production. The limited number of exported forms, reflected by the absence of unidentified examples of African cookware in the study assemblage, probably indicates functional specialization, which becomes even more apparent when production is contrasted to the contemporary cookwares of the West-Central Italy. Although much of the variability included in the latter class is explained by the geographical proximity of the production centers to Rome, it may be equally presumed that West-Central Italian potters were forced to concentrate on the manufacture of forms either fulfilling local needs or filling the gaps in the North African form repertoire. The conditions for product specialization were exceptionally good on the coast of the present-day Tunisia, as the production was centralized195 in nucleated workshops, the principal task of which was to provide containers for the large-scale exportation of bulk goods. Some workshops seem to have been so specialized that they produced only three or four kinds of vessels.196 Many other features characteristic of specialized production – including efficiency, riskaverse production strategy and skill – are not similarly observable from the output.197

ø 13

15

17

19

21

West-Central Italian

23

25

Standardization, which has often been used as an index for specialization, is usually thought to be reflected in the uniformity of vessel dimensions. Uniformity and the use of the potter's wheel are also said to indicate specialized mass-production.198 If the evolution-like process towards an increasingly homogeneous production is defined as “standardization” and the outcome of that process as “uniformity”,199 the study assemblage itself reflects rather the uniformity than standardization. The process of standardization can be examined either by comparing the uniformity of multiple productions or, alternatively, by introducing a diachronic dimension into the analysis.200 In the present case, the former analysis was carried out by comparing the rim diameter distribution of the Hayes 23B shallow casserole in both African cookware fabrics with corresponding production in West-Central Italian cookware. The chronological consistency of African cookware production, on the other hand, is reviewed through the rim diameter distribution of Hayes 196 lids. The rim diameter distribution of the Hayes 23B shallow casserole was chosen to be the subject of a further production analysis, because the study assemblage included a sufficient quantity of examples in all three cookware productions: Central Tunisian (115 pcs), North Tunisian (230 pcs) and West-Central Italian (131 pcs). The examination of the data (Figure 20) leads to the following observations. North Tunisian production comes closest to the normal

27

29

Central Tunisian

31

33

35

37

North Tunisian

Figure 20. The rim diameter distribution of the Hayes 23B shallow casserole in three productions: West-Central Italian, Central Tunisian and North Tunisian. 107

Chapter 4

ø 14

16

18

20

Phase 6

22

24

26

28

Phase 7

30

32

34

36

38

Phase 8

40

42

Phase 11

Figure 21. The rim diameter distribution of North Tunisian A.II (Hayes 196) lid by phases. distribution, though the twofold size of the sample has certainly contributed to the smoothing of the curve edges. Also the Hayes 23Bs of Central Tunisian origin form a fairly homogeneous group, while most variability is included in the group of the WestCentral Italian imitations. Thus, at least on the level of this fairly straightforward comparison, the study assemblage supports the idea about the higher level of standardization in the production of African cookware. The question about standardization as a chronological process is somewhat more difficult to examine, because while the study assemblage consists of some twenty-one hundred cookware vessels, only the Hayes 196 lid with its three variants (A.I-III) has a roughly homogeneous quantitative distribution in all chronological phases. While the number of North Tunisian products representing the two last phases is close of being quantitatively insufficient, the evidence of the remaining four phases is also difficult to interpret. Although the material belonging to the two early 4th century AD phases 7-8 is clearly more uniform than either in the preceding or subsequent group (Figure 21), it would be foolish to declare the early 4th century AD as the age of standardized cookware production in North Tunisia. The interpretation of both analyses is problematic, because the results are affected by “cumulative blurring”. In other words, despite the apparent homogeneity the study assemblage is actually a mixture of the output of multiple workshops. Chronological differences are usually the other factor increasing the degree of the variability in 108

the assemblage, even if the producers had been specialists.201 The variability included in the output of a single workshop may in fact be substantial. Firstly, production can be affected by the demand, although in the case of exported African cookware vessels the relationship between the producer and consumer was hardly intimate.202 Secondly, a certain amount of variability was probably introduced during production due to physiognomic differences, because most potters control the dimensions of the vessels with measurements based only upon the size of their hands.203 In the event that the process of standardization had been set in motion by an external agent, such as government, the above-mentioned restrictions do not necessarily apply. So far, the increased involvement of the government in pottery production has been observed by Riley in his study of late 2nd/early 3rd century AD cookware production in Cyrenaica.204 The possibility will be worth examining in the future by contrasting the present assemblage with the early Imperial finds of the Palatine East excavations, which include a good selection of North Tunisian cookware from the late 1st to the early 3rd century AD. Meantime, I can only refer to the heterogeneous distribution of rim diameters in phase 11 (Figure 21), which may be interpreted as a symptom of less controlled production. If so, the development is roughly synchronous with the retreat of the kiln sites from coast to the inland and the establishment of a new production structure based on self-sufficient estate economy.

Production

CHAPTER 4 ENDNOTES 1

2 3 4

5

6 7 8 9 10 11 12

13

14 15 16

17

18

19 20 21 22 23 24 25 26 27 28 29

30 31 32 33 34

35 36 37 38 39 40 41

42 43 44 45 46 47 48

Due to lack of contrary archaeological evidence, Roman potters used most likely a hand-operated wheel (Peacock 1982, 55-57; Mackensen 1993, 65-67, cf. Swan 1984, 50-51). The first pictorical representations of a kick wheel, in which the wheel-head is joined to a heavy fly wheel by a vertical iron shaft, date only to the 16th century AD (Cuomo di Caprio 1995, 148; McCarthy & Brooks 1998, 29-30, 54). Henrickson 1991, 523-524. Arnold 1979, 735; Rye 1981, 5. Even if the method was fraught with danger as the air trapped between the surfaces may expand in firing and shatter the vessel, its use is witnessed by the regular presence of small cavity visible in the middle of the rim section. Still, the possibility that some of these rims were formed by retaining a thickened rim as the pot was thrown cannot be excluded. When using this method, the potter faces a constant risk to run out of clay (Hopper 1986, 13). McCarthy & Brooks 1998, 23. Flat bases represent a risk as they may warp to convex shape on drying stage, see Hopper 1986, 134. Gibson & Woods 1997, 118. Schuring 1988, 34-35, 40. Courty & Roux 1995, 28-30, fig. 5a. Cuomo di Caprio 1995, 148-151. See also Tite 1999, 217; Tite & Kilikoglou 2002, 4. E.g. Peña 1999, 131 no. 59; fig. 35. We owe much of our knowledge on Roman metal cooking utensils to Campanian towns buried by the eruption of Vesuvius in AD 79, see e.g. Annecchino 1977; Tassinari 1996. The use of moulds does not require such a highly skilled labor as throwing on a wheel. Therefore, the forming of vessels can be undertaken both men and women as well as children, Tite 1999, 217. Ostia III, 411-412. Carandini 1970, 114. Hayes 1994, 114; Peña 1999, 30-31. Riley (1980, 76) adds that most forms seem to have been produced only in two sizes. González Villaescusa 1990, 67-69. Even some imitations of African cookware have been reported to fall into two or three size classes, see Dodinet & Leblanc 1988, 138-139. Ballard et al. 2000, 1612; McCann 2001, 259. It should be observed, however, that the shape of many African cookware forms, like the Hayes 23 shallow casserole or the Hayes 196 lid, allows to stack vessels with the same diameter one on top of another, see figure 16. Bost et al. 1992, 180; 187, fig. 45, nos. 1-4. Hayes 1997, 25-26. Annis & Geertman 1987, 164-165, 182-185. Braun 1983, 107-108. The importance of such comparisons has recently been stressed by Rice (1996a, 142). On the terminology related to ceramics and heat, see Bronitsky 1987. The difference between permeability, porosity and density is discussed by Rice (1987, 230-232). Schiffer 1990, 379-380. Schiffer et al. 1994, 200-209. van der Werff 1982, 130. A slip is a fluid suspension of fine clay and water applied to a vessel in the leather-hard stage by pouring, dipping or painting. Rice 1987, 482; Gibson & Woods 1997, 249-250. Hayes 1972, 14. Rye 1981, 57. Gandolfi 1994, 151. Schuring 1988, 30-32, fig 23b. Schiffer 1990, 377-380. The possibility that slip would also improve the strength of the vessel has been brought up by Wallace (1989, 38). Aquilué Abadías 1991, 976. Freestone 1982, 103-104. Blackman 1992, 118-120. Freestone 1982, 104. Tite et al. 1982, 116; Blackman 1992, 118; Tite 1992, 117. Tite et al. 1982, 111 tab. 1, 115 tab. 3, 118-119. In burnishing, the vessel is rubbed in the leather-hard stage with a hard round-edged tool, such as a beach pebble, t o align and compact the clay minerals on the surface so that reflected light becomes concentrated and may be perceived as luster. Rye 1981, 21; Rice 1987, 138-139; Wallace 1989, 33. Therefore, none of these lustrous stripes were painted onto the exterior surface, as Lamboglia (1958, 277) suggested some thirty years ago. See also Aquilué Abadías 1991, 976. Hayes 1972, 200. Gibson & Woods 1997, 113-114; Rice 1987, 472. Rye 1981, 3. Rice 1987, 150. Tortorella 1981b, 208. Gandolfi 1994, 151. Aguarod Otal 1991, 250.

109

Chapter 4

49 50 51 52 53 54 55 56 57 58 59 60 61 62

63 64

65 66 67

68 69 70 71 72 73 74

75 76

77 78

79 80 81

82 83 84 85 86 87 88 89 90 91 92

93 94 95 96 97

98

Schiffer 1990, 378-380. González Villaescusa 1990, 74. Wallace 1989, 34-36. Wallace 1989, 37. Hayes 1972, 18. Young & Stone 1990, 198-202; Schiffer et al. 1994, 208. Schuring 1988, 35, 44; Young & Stone 1990, 202; Schiffer et al. 1994, 208-211. Schiffer et al. 1994, 208. Schuring 1988, 35. Rice 1987, 232. Young & Stone 1990, 198-202. Pontacolone & Incitti 1991, 548; Schiffer et al. 1994, 210. Young & Stone 1990, 202; Schiffer et al. 1994, 210. For example, decorated body sherds make up only 1.5 ‰ of the diagnostic fragments of West-Central Italian cookwares found in the Late Roman deposits of the Palatine East exavations, and compared to the total amount of body sherds, they are non-existent. Aguarod Otal 1991, 342, fig. 82.3. As the structure and operation of the Roman updraught kiln has been described both in many introductory texts (e.g. Peacock 1982, 67-73) and special studies (e.g. Cuomo di Caprio 1972; Swan 1984; Tecnología 1992), unnecessary repetition is avoided here. Stirling & Ben Lazreg 1997, 372; Stirling & Ben Lazreg 2001, passim. Schütz 1992, 164 fig. 4; see also Stirling & Ben Lazreg 2001, 225. Schuring 1986, 188; 1988, 36-40; Aguarod Otal 1995, 136, Mason & ‘Amr (1995, 635) attribute this effect to the composition of clay. Aguarod Otal 1991, 236, fig. 13; Hayes 1997, 64. Luni II, 180; Hayes 1972, 205 and 209; Gandolfi 1994, 149. Cf. Mason & ‘Amr 1995, 635. González Villaescusa 1990, 51. Luni II, 499. Schuring 1988, 36-40. Aguarod Otal 1991, 245. The idea concerning the absence of this surfacing from the imitations of African cookware has proved false, as vessels of local production showing slipped surfaces and a patina cenerognola cover have recently been published in Spain, see Moreno Almenara & Alarcón Castellano 1994, 1289-1293; Revilla Calvo et al. 1997, 105, 116, nos. 90-91; Serrano Ramos 1997, 220. Arnold 1985, 23. “nell'applicare la patina cenerognola (andrebbe chiarito cosa essa significhi da un punto di vista strettamente tecnico) si produceva con uno strumento non meglio identificato, forse involontariamente una specie di politura a strisce, sulle strisce prodotte da liscatura, la patina cenerognola si fissava meglio, cosi che quando sulle parti non polite la patina cenerognola finiva per scomparire, restava effetto particolare, ma non originale di una patina cenerognola a strisce.” Ostia II, 86. Schuring 1984, 167. Hayes (1972, 46, 289) was probably trying to describe this coating/scum in his discussion on the Hayes 23B shallow casserole, which is chiefly a Central Tunisian product: “On this form alone the exterior of the rim and wall i s regularly fired grey, indicating probably that it did not merit the careful firing given to the tablewares.” Mahjoubi et al. 1973, 64, note 141. Peacock 1984c, 263-264; Schuring 1984, 153. See also van der Werff 1982, 413. Schuring 1988, 38. Sometimes, patina cenerognola cover has been seen to result from the darkening of salt scum, the extent of which is said to correlate with the zone that has not been turned (van der Werff 1982, 265 note 339; Peña 1999, 133, 150 note 285). Laird & Worcester 1956, 557. Rye 1976, 122; Peacock 1984b, 15. Peacock et al. 1990, 66; cf. Peacock & Williams 1986, 45. Rye 1976, 122. See also Arnold 1985, 138-139. Tite 1995, 38. Hayes 1972, 295. Schuring 1988, 38-40. Matson 1972, 218; Mackensen 1993, 437. On the use of saggars, see Mackensen 1993, 88-94; Tortorella 1995, 81. Schuring 1988, 38-40. van der Werff 1982, 420. Tite 1995, 38. Differences up to 150 °C has been observed in the firing temperature of a single vessel, see Tite 1999, 190. Schuring 1986, 202-203 Schuring 1986, 198; 1988, 38-39, fig. 37, cf. van der Werff 1982, 415. Schuring 1988, 6. Schuring 1988, 38-40. Wertime 1983; Raven 1993, 90; Smith W. 1998, 192. The extensive brush and wood cover of North Africa was destroyed by man only a century ago, Shaw 1981, 393. Smith W. 1998, 201; Ben Lazreg 2001, 436.

110

Production

99

Mackensen 1993, 54-56, 484; 1998a, 355; 1998b, 30. It goes without saying that agricultural by-products may have also had other uses, e.g. as fertilizer, animal feed or fuel of households and baths, see Mattingly 1996, 225; Ben Lazreg 2001, 436. 100 Peacock 1982, 25, 41; Stone et al. 1999, 313; Ben Lazreg 2001, 436. In the island of Djerba annual prunings are enough to support the local workshops throughout the year. 101 Rye 1976, 112; 1981, 104; Arnold 1985, 31-36; Rice 1987, 162, 176; Smith W. 1998, 192. 102 Curtis 1962, 496; Matson 1972, 213. 103 Smith W. 1998, 193; Rife 2001, 301; Stone et al. 2001, 217. 104 Smith W. 1998, 193-194, 199-200; Rife 2001, 301; Smith W. 2001, 434-435; cf. Smith W. 1998, 198. 105 Stone et al. 1999, 311-313; Stirling & Ben Lazreg 2001, 227-228. This find is possibly explained by local custom to use ash and carbonized olive pits as an ingredient of water-proof mortar, see Ben Lazreg 2001, 436. 106 Smith W. 2001, 434-435, see also Mattingly 1996, 223. 107 Mattingly 1988a, 41 note 37; Matingly 1993, 492. 108 Mattingly (1988a, 45) has used aerial photographs to estimate the density of olive trees in Roman olive orchards. 109 Mattingly 1988a, 45; Mattingly 1988b, 183-184; Mattingly 1993, 484. 110 Today, the same area holds 15 million trees with the average density of 22 trees per hectare. Even more humongous figures are reached, if the estimate is based on the consumption of olive oil in the Roman world. With the consumption of ca. 20 litres per capita, the annual production of 1,600,000 metric tons of olive oil was required t o meet a demand created by 25-50 million consumers around the Mediterranean (Mattingly 1988a, 34; Mattingly 1994, 140). From the 8,000,000 metric tons of olives needed to produce this amount of oil, ca. 3,200,000 tons were turned to olive pressings. For the sake of comparison, the amount of olive pressings produced in Greece in 1993 was 682,000 metric tons (Koukios 1999, 5). 111 Amouretti 1993, 467-473; Mattingly 1996, 225. 112 Matson 1972, 202, 219. 113 Mason and ‘Amr (1995) report of several experimental firings carried out with reconstructed Roman pottery kiln and olive pressings as the main fuel. 114 Matson 1972, 219. 115 Hitchner 1993, 501; Mackensen 1993, 55-56. 116 Mattingly 1994, 140; Smith W. 1998, 195; 117 Matson 1972, 202. 118 Mason & ‘Amr 1995, 633-635; Smith W. 1998, 191-192. Hedges and loppings of brushwood were still preferred i n the 16th century England, where proper firewood was certainly available, see McCarthy & Brooks 1998, 46-48. 119 Mattingly 1994, 140; Ben Lazreg 2001, 436. The following 2-3 months involved the production of olive oil. It was accomplished with presses that had a maximum production capacity of 5,000-10,000 kgs of olive oil per season (Mattingly 1993, 484, 492-496). While it would be extremely interesting to estimate the amount of pottery that could have been fired with 10-20 tons of olive pressings resulting from this process, I leave this topic untouched due to lack of reliable data, see Peacock 1982, 25; DeLaine 2001, 263. 120 Peña 1992, 102. 121 In Peloponnesos, Greece, the production season of modern potters lasts from April to October (Matson 1972, 220). At Vasanello, in Central Italy, traditional pottery workshops were active around the year. However, the rate of production slackened during the winter as increased humidity lengthened the time needed for the drying of the products (Peña 1993, 95-98). In general, the climate regulates effectively pottery production, see Arnold 1985, 6198. 122 Peacock 1982, 39, 41. This analogy is based on the observation about the similarity of present climate to that of Roman times, e.g. Shaw 1981, 380, 395; Mattingly 1994, 13. This contrasts the common belief that the climate i n North Africa would have been more favorable in Roman times than it is today. 123 The regulating effect of the seasons was observed at Vasanello, where the production of decorated forms took place i n the wintertime, see Peña 1992, 120. 124 Mattingly & Hitchner 1995, 201. It is hoped that the increasing frequency of intensive production site studies will result in deepening our knowledge also on this topic, see Manacorda & Panella 1993, 63; Panella 1996, 11. 125 E.g. Mackensen 1998b; Mackensen & Schneider 2002. 126 Hayes 1972, 296. 127 Peacock 1984b, 6. 128 Tortorella 1995, 82. Some ash deposits found in the excavations of Avenue Habib Bourgiba may be considered equal to dumps of used fuel found in Leptiminus, Smith W. 1998, 193; see also section 4.3.3. 129 Carandini 1970, 114-115; Tortorella 1987, 295. Interestingly, the same area is customarily defined as the granary of Africa Proconsularis, Warmington 1954, 55; Rickman 1980, 109, Garnsey 1988, 248. 130 Mackensen 1993, 52-54. 131 Peña 1992, 110. 132 Tortorella 1981b, 210. 133 Hayes 1980, 518; Tortorella 1981b, 209; van der Werff 1982, 123; Schuring 1988, 3, 9-25. 134 van der Werff 1982, 126. 135 Peacock 1982, 41. 136 Hayes 1972, 298; Tortorella 1981b, 209; Schuring 1988, 3. 137 Ostia I, 31, 45-46; van der Werff 1982, 78-79; Tortorella 1981b, 209; 1987, 299; 1995, 86; Taylor & Robinson 1996a, 233; cf. Gandolfi 1994, 151. 138 Schuring 1988, 3.

111

Chapter 4

139

Arnold 1985, passim; Mackensen 1993, 54-56. The evidence from the Ksour Essaf region (Sahel) suggests, however, that the distribution of kiln sites is not always conditioned by the proximity of clay resources, Peacock et al. 1989, 183. 140 On the Roman pottery production site studies in North Africa, see Tortorella 1995, 79-88; Mackensen 1998b, 2325. 141 Mackensen (1993, 21-23) claims that the lack of interest in the study of production sites has customarily been paired with the uncritical use of survey material. 142 Peacock et al. 1989; 1990; Peacock & Tomber 1991. 143 Mackensen 1993. Mackensen (1993, 463, 484, 495) dates the production from the late 3rd/early 4th century AD with AD 330/340 as an absolute terminus ante quem all the way to AD 680/690. 144 Ben Lazreg & Mattingly 1992; Stirling & Ben Lazreg 1997; Mattingly et al. 2001 145 Mackensen 1998a & 1998b. 146 Ostia II, 84; Peacock & Tomber 1991, 291; Olcese 1993, 138; Stirling 2001, 31, 65-67; Stirling & Ben Lazreg 2001, 228-229. 147 Peacock et al. 1989, 199-201; Peacock et al. 1990, 59, 82-84; Peacock & Tomber 1991, 291. 148 Peacock et al. 1989, 200; see also Gibbins 2001, 322-324. 149 Peacock & Tomber 1991, 291-292. While the Vandal conquest had a minimal effect on pottery trade, the Byzantine reoccupation with newly imposed annona and other taxes may have diminished or even interrupted the exportation of African pottery. Tortorella 1995, 89, 101. 150 Carandini 1970, 114-115; Carandini 1983, 152; Mattingly 1988a, 42; Peña 1999, 163. 151 Stone et al. 2001, 218. 152 Peacock 1982, 39-43. 153 Peacock & Tomber 1991, 291. 154 Stirling 2001, 31, 65-67; Stirling & Ben Lazreg 2001, 228-229. 155 Stirling & Ben Lazreg 1997, 372; Stone et al. 1999, 311-312; Stirling & Ben Lazreg 2001, 221-222, 228. 156 Stone et al. 1999, 311-312. 157 The firing cycle covers the process from the loading of unfired pots to the unloading of fired pots. Depending on the size of the kiln and the firing temperature desired this may take 4-7 days. While the actual firing does not usually last longer than 12 hours, the kiln often requires several days to cool down, see Matson 1972, 219; McCarthy & Brooks 1998, 54. 158 Peacock & Williams 1986, 41, cf. Mattingly 1988a, 42. 159 Greene 1992, 57; see also Peña 1992, 95; Manacorda & Panella 1993, 61. 160 Carandini 1986, 9. 161 Carandini 1970, 106, 115; Manacorda & Panella 1993, 57-60, fig. 2; Gibbins 2001, 322-324. While these stamps identify four Central Tunisian ports – Hadrumetum (Sousse), Leptiminus, Sullecthum and Thanae – and one North Tunisian port – Neapolis – as potential pottery production sites, only the surroundings of Leptiminus and Sullecthum have produced confirmatory archaeological evidence, see Peacock et al. 1989, 183-196; Dore 1992, 155. 162 Brown & Mattingly 2001, 387-389, 395; Smith M. 2001, passim. 163 Carandini 1970, 103. 164 Mattingly 1996, 237. 165 Stirling 2001, 68-69 Besides a 4th century AD site involved in the production of amphorae, the excavations at Leptiminus have produced evidence of a Roman bath complex that was turned to industrial use – including a pottery production unit – between the 6th and 7th century AD. The phenomenon of converting baths and other public buildings into potteries began in Roman North Africa during the 5th century AD and flourished in the following two centuries, Stirling 2001, 69-71. 166 Fulford 1987, 69. The shift that can be dated to the late 4th/early 5th century AD does not necessarily reflect a decline in production. While the economy of Roman North Africa remained strong, the town as an institution seems to have lost its importance, see Stone et al. 2001, 219. 167 Carandini 1970, 114-115. 168 Mackensen 1993, 473-475. 169 Harris 1993b, 22-23; 1993c, 188; Mackensen 1993, 486, cf. Whittaker 1989, 538-539. 170 Greene 1986, 165; 1992, 58; Mackensen 1993, 497. The production process was more likely a business of many independent participants than kept “in house”, although the use of middlemen could lead to complex legal situations, Paterson 1998, 159-163. 171 Aubert 1993, 173-175. 172 Greene 1992, 56. 173 Mattingly 1996, 221. 174 Peacock & Williams 1986, 41, cf. Mattingly 1988a, 42. 175 Pucci 1983, 115-117. 176 Harris 1993b, 25-26; 1993c, 188. 177 Tortorella 1987, 304. 178 Carandini 1970, 114-115; Carandini 1983, 152; Pucci 1983, 114; Peacock et al. 1989, 200; Mackensen 1993, 5456. 179 Cf. Peacock et al. 1990, 83. 180 Panella 1993, 629. 181 Mackensen 1998b, 30-33. 182 Mackensen 1993, 53-58.

112

Production

183

Wickham 1988, 191, see also Peña 1999, 34-35. Normally this meant the use of ox-drawn cart that could operate at the maximum speed of 3 kms per hour and transport 50-70 modii (340-480 kgs.) of wheat at a time, Rickman 1980, 13, 120-121. 184 Laurence 1998, 130-134, 143. 185 Peacock et al. 1990, 83. 186 Duncan-Jones 1974, 366-369; Greene 1986, 40; cf. Laurence 1998, 135. In the following calculations, the costs of river transportation have been revised according to DeLaine (2001, passim) as 3.9 for downstream and 7.7 for upstream. For example, a consignment of one thousand African cookpots was exported from a kiln site located in the vicinity of Thuburbo Minus via Carthage to Rome. If it was transported to Carthage by river, the journey involved 93 kms of downstream transportation by using the River Medjerda. This was followed by 570 kms of maritime transportation. Hence, the total transportation cost per unit can be calculated as (3.9x93)+(1x570)≈933. The effect of additional shipment costs on the number of the vessels to be exported, in comparison to a consignment exported directly from Carthage, can be derived by using the following formula (570/933)x1000 vessels ≈ 611 vessels. 187 Hayes 1980, 518. 188 Sufetula-Hadrumetum: 28x152=4256; Hadrumetum Ostia: 1x665=665 189 Mackensen 1998b, 23 (see also p. 24, fig. 1). 190 The site and products of Sidi Marzouk Tounsi have been described in Peacock et al. 1990, 66-74; Mackensen 1998a; Mackensen 1998b, 25-30; Mackensen & Schneider 2002, 130-134, 151-155. 191 Cf. Tite 1999, 225. 192 Cf. Rice 1990, 5-8; Arnold & Nieves 1992, 93; Rice 1996b, 180. 193 Ostia III, 410; Schuring 1988, 3. 194 Rice 1991, 262-263. 195 Rice 1991, 264. 196 Hayes 1972, 14-16. 197 Rice 1991, 268. 198 Arnold & Nieves 1992, 108-109; Blackman et al. 1993, 61; Courty & Roux 1995, 17. 199 Rice 1996b, 179-180. 200 Arnold & Nieves 1992, 94. 201 Blackman et al. 1993, 74; Rice 1996b, 181-182. 202 In this production scheme, the producers are informed about the needs of consumers only through middlemen, see Aubert 1993, 171. 203 Arnold & Nieves 1992, 108-112; Blackman et al. 1993, 75. 204 Riley 1980, 73-75.

113

Distribution and consumption

5 DISTRIBUTION AND CONSUMPTION

5.1 QUANTIFICATION OF THE STUDY ASSEMBLAGE 5.1.1 Methods and background Quantification is a branch of archaeology that has only recently been subjected to considerable development, although the method itself was first applied to pottery studies in the late 19th century.1 Much of the recent progress can be attributed to a group of British scholars, headed by professor Clive Orton, whose long-term work with statistics has shown the value of accurately quantified data and the bias included in the most common measures used in pottery quantification: sherd counts, weights and estimated number of vessels.2 Some years ago this research culminated in the introduction of a new quantification method – including several not readily understandable terms like “estimated vessel equivalent (eve)” and “pottery information equivalent (pie)” – together with the release of a relevant software, The Pie-slice computer package,3 for the manipulation of the data. The ultimate goal of the method is to establish the proportions of the life assemblage from a sample, the kind of which any archaeological assemblage ultimately is. Surprisingly, the improvement of these research tools has not directly resulted in their adoption in archaeological interpretation. For example, the abovementioned software package has seldom if ever been put into practice in connection with the study of a large pottery assemblage from the Mediterranean. Of the many possible explanations, only two should be brought out here. Firstly, the nucleus of the model proposed by Orton and Tyers is a complex mathematical formula, which is barely understandable without at least a moderate familiarity with archaeological statistics. In fact, the method has even been objected for being overly theoretical.4 Secondly, considering the present state of pottery quantification in the Roman Mediterranean, the estimation of vessel equivalents and the counting of pie-slices may be futile. The enormous amounts of pottery produced by urban excavations have usually directed scholars to tackle the material with traditional methods of quantification,5 like sherd counts and weights, to which the results obtained with more refined methods cannot be directly compared.6 Therefore, the quantitative evidence from other sites will be discussed separately in the following section. In the present case, the data for the quantification was obtained with a method, in which an estimated vessel equivalent was determined for each measurable rim sherd family.7 Their identification has been accomplished subjectively, but also carefully grouping 115

together all the sherds that seem to belong to the same vessel. Thus, the minimum number of vessels, which has been regarded as one of the most reliable and consistent means of analysis,8 can be determined both as a sum of estimated vessel equivalents and nucleated sherd assemblage counts. The third method included in the analysis for determining the variability arising from the use of different measures is sherd counts. A comparison of this kind was deemed necessary due to somewhat alarming results of previous case studies. The study of the coarseware assemblage of Berenice, for example, has explicitly shown that methods intended to estimate the absolute number of vessels regularly produce inconsistent results.9 Therefore, the ultimate aim of the following analysis is to establish the relative proportions of the Late Roman cookware assemblage of the Palatine East excavations as percentages.10 Although the pie-slice method is not exploited directly, its influence is obvious. In addition, possible changes in the exportation of the two African cookware fabrics has been examined both by comparing them to one another and introducing the group of miscellaneous African cookware fabrics as an additional element into the analysis. 5.1.2 Results and interpretations The first observation emerging from the crosstabulation of the three main cookware fabrics with three different methods of quantification (Table 46) is the close resemblance of sherd counts (sherd) and estimated minimum number of vessels (evrep ii). This is not a surprise considering the fragmentary nature of the study material. The calculated minimum number of vessels (evrep i), on the other hand, differs markedly from the other two groups, possibly due to a different state of preservation of various contexts.11 Another factor that may skew the picture, this time in the favor of West-Central Italian production, is the effect of the vessel shape and size on the probability of preservation. At least the large thick-walled flanged casseroles of West-Central Italian production have likely withstood mechanical wearing better than medium-sized African cookware casseroles with slightly thickened rims.12 The most important observation to be made, however, is that all three methods of quantification produce a similar overall pattern characterized by the steady decline in the amount of African imports. The growing importance of the West-Central Italian products is the most readily observable

Chapter 5

Table 46. The study assemblage quantified with three methods (%) Cookware fabric / Phase

P6

P7

P8

P9

P10

P11

sherd

53.8

61.3

58.3

71.4

81.3

89.3

evrep i

65.3

74.6

71.0

77.1

89.0

94.5

evrep ii

52.8

58.9

58.4

70.1

80.9

88.4

sherd

6.2

13.0

10.6

8.1

4.7

3.1

evrep i

5.9

7.6

6.2

6.7

2.6

1.5

evrep ii

6.2

9.4

10.4

8.4

4.8

3.1

sherd

39.9

25.7

31.1

20.5

14.0

7.6

evrep i

28.8

17.8

22.8

16.2

8.4

4.0

evrep ii

38.8

30.9

31.1

21.5

14.1

7.4

Campanian cookware

evrep ii

0.4

-

-

-

-

-

Internal Red Slip cookware

evrep ii

1.8

0.6

0.1

-

0.2

0.1

Aegean cookware

evrep ii

-

0.2

-

-

-

-

Hand-built cookware

evrep ii

-

-

-

-

-

1.0

West-Central Italian cookware

Central Tunisian cookware

North Tunisian cookware

development as the quantitative distribution of cookwares is examined production by production. The explanation for this phenomenon, which seems to be independent of the measure used, can scarcely be attributed to the Vandal conquest of Roman Africa in the early 5th century AD, because the first signs of decline are already detected in the mid-4th century AD. Thus, it is reasonable to refer to other external factors, like the exhaustive effect of oleoculture resulting in the transfer of kiln sites from coastal areas to inland. On the other hand, changes in consumption affected the distribution of imported pottery as likely as changes related to production. The city of Rome was not the capital of the Empire after AD 330 and its economy increasingly suffered both from barbarian attacks and not later than the early 5th century AD from the decrease in population. All this likely directed the demand in favor of the local production.13 Another remarkable aspect in the assemblage is the almost complete absence of Aegean cookware and other imported cookwares other than of African provenance. As both forms and fabrics of Aegean cookware are reasonably easy to identify,14 the possibility that the phenomenon could actually reflect the real situation in the past is hard to exclude. The following conclusion has been made on the basis of marginal status of Aegean cookware in the markets of Ostia and Rome.15 As the city of Rome was strongly dependent on the western Mediterranean products, the lack of Aegean pottery other than transport amphorae signifies that the importation of agricultural products was not customarily accompanied by the importation of table- and cookwares of the same origin.16 With regard to other minor fabrics, the gradual disappearance

116

of both Campanian and Internal Red Slip cookware was expected in the same way as the late introduction of Hand-built cookware. As the study assemblage is cross-tabulated by chronological horizon and the method of quantification (Table 47), it is noticed that the elimination of other cookwares from the analysis improves the resolution by which the relationship of North and Central Tunisian production can be examined. The picture is not affected by the inclusion of a small number of miscellaneous finds either, the nature of which is very heterogeneous. The resulting pattern is very simple: as the amount of African cookware diminishes in comparison with the West-Central Italian cookware, the Central Tunisian production practically doubles its share in comparison to the North Tunisian production. But when speaking of the absolute number of vessels, the latter fabric is clearly superior to the former, a feature that recalls the distribution of corresponding tablewares.17 Some problems require further attention. One of them is the low share of Central Tunisian production, as, considering the importance of Byzacena as a supplier of agricultural products, the situation should be exactly the opposite. A probable explanation is the relatively late appearance of exported Central Tunisian pottery, of which the introduction of African Red Slip C in the late 2nd century AD is probably the best example. It is hardly a coincidence that at approximately the same time African olive oil was first distributed to the poor of Rome by Antoninus Pius.18 The reign of Septimius Severus, during which olive oil was included in annona on regular basis, was also the heyday of Central Tunisian pottery

Distribution and consumption

Table 47. The relative distribution of African cookware productions (%) Central Tunisian

North Tunisian

Miscellaneous

Phase

Date (AD)

srd

ei

e ii

srd

ei

e ii

srd

ei

e ii

6

270/290-300/310

13.5

16.8

13.7

86.5

83.2

86.3

-

-

-

7

300/310-310/320

33.0

29.4

22.8

65.2

68.7

74.8

1.8

1.9

1.4

8

310/320-320/325

25.1

20.9

24.7

73.5

77.5

73.8

1.4

1.6

1.5

9

320/325-350/360

27.9

28.7

27.8

70.9

70.2

71.0

1.2

1.1

1.2

10

350/360-400/425

25.0

23.3

25.5

75.0

76.7

74.5

-

-

-

11

400/425-500/525

29.0

26.7

28.7

69.8

71.7

70.0

1.2

1.6

1.3

production. Thus the explanation that the exportation of olive oil would not have allowed parasitic pottery exports must be rejected. The connection between table- and cookwares is significantly more problematic, because as the markets for the African Red Slip ware C were slowly dying in the 4th

century, the share of corresponding cookware was growing steadily.19 By the end of the 5th century AD, the number of African cookware exports was subjected to an evident decline, but the axis between Carthage and Rome did not start to lose its significance until the beginning of the 6th century AD.20

5.2 EVIDENCE FROM OTHER SITES The evidence on the distribution of African cookware from other sites than the Palatine East excavations is presented in this section. The discussion proceeds from the city of Rome and its main port, Ostia, through the rest of the Italian peninsula to other regions of the western Mediterranean including France and Spain, provinces of Roman Africa other than Zeugitana and Byzacena, and finally cross-regional surveys. The important pieces of evidence provided by many Roman shipwrecks in which African cookware was on board either as a part of the cargo or the crew's personal belongings, is introduced separately. Instead Fosso della Crescenza

of attempting to cover all the occurrences of African cookware ever published in printed form, this section concentrates on sites and contexts considered to be illustrative of the various aspects related to the distribution and consumption of this pottery. Particular attention will be paid to the suggested closing dates for the importation of African cookware as well as to the relation of this production to other African pottery, other imported cookwares and cookwares of local manufacture. Finally, the section is also intended as a background for the following discussion on the importance of African cookware.

er Tib

Nomentum

Veii

Caere ad Sextum Fidenae

Tibur

Ficulea

via S. Cannizzaro

ad Turres

Anio

Collatia Gabii

Fregenae

Ti

be

r Vigna Barberini Magna Mater

Tusculum

Portus

Ficana

PALATINE EAST

Schola Praeconum

Caelius domus S. Stefano Rotondo

ad Decimum Ostia Lungotevere Testaccio Monte Testaccio Laurentum

Velletri

117

Figure 22. Rome, Ostia and the lower Tiber river valley.

Chapter 5

5.2.1 Rome and Ostia The city of Rome provides us with several points of reference to which the quantitative evidence from the Palatine East excavations can be compared (Figure 22). The possibility of tracing the pace of the disappearance of African cookware from the declining markets of urbs aeterna is of special interest (Table 48). Although the comparison lacks pure 4th century AD material from elsewhere in Rome than the Palatine Hill,21 the evidence does not show clear signs of the declining importance of African cookware imports until the late 5th century AD.22 However, the nature of this pottery has been taken for granted in the early 5th century deposit of Schola Praeconum (SP I),23 although the fact that the group represents less than 1% of the total assemblage does not automatically imply its residuality. If compared only with cookwares, phase 10 of the Palatine East excavations confirm that the share of 17.7% is not by any means exceptional for an early 5th century AD deposit. African cookware has also been found in Rome in the 5th/early 6th century AD deposits, but its scarce presence seems to support the idea, according to which the production did not persist longer than the second half of the 5th century AD.24 This tendency has also been said to reflect the decline in the importance of African products in the markets of Rome during the 5th and 6th century AD, even though the city was still an important center of consumption, in spite of its constantly decreasing population.25 Besides the excavations of a late Roman domus on the Caelian hill, the other published deposit of the Schola Praeconum excavations (SP II),26 provides further confirmation of the extinction of African cookware imports.

In contrast to its mother city Rome, the port of Ostia provides us with a series of contexts practically covering the extent of African cookware production from the late 1st century to the late 6th/early 7th century AD.27 The importance of African imports among cookware products is evident at Ostia from the late 1st century AD onwards.28 Although the production is already well attested in the late 1st/early 2nd century AD (Table 49), the importance of African cookware had certainly risen by the Hadrianic period.29 By the late 2nd century AD its dominant position was well established both in comparison with WestCentral Italian and other imported cookwares, which at that time came usually from the Aegean or Campania.30 By the late Severan period the proportions had risen dramatically, as African production comprised practically all of cookwares. Although the 4th century AD witnessed a slight decrease in quantity, it is uncertain whether this drop should be regarded as a prelude to the decline in the importation of African cookware, which actually started in the 5th century AD.31 The flow of African tablewares and amphorae to Ostia, on the other hand, seems to have continued considerably longer without major interruptions. The development is also reflected by the changing ratio of African cook- and tableware products in various deposits. While in late 2nd century AD cooking vessels represented nearly three fourths of all African pottery, the proportion of these two functional classes was more or less in balance by the 4th century AD and only from the 5th century AD onwards were the majority of vessels in this class tableware. The significant evidence regarding the end of the exportation of African cookware to Rome and Ostia

Table 48. Examples of quantified deposits with African cookware in Rome. Site

Deposit

Date (AD)

Vigna Barberini

5B

190-210/220

Vigna Barberini

5A

Vigna Barberini

Share

Nos.

References

14.2% (A)

10

Villedieu 1998, 29-30, 41.

210/220-240

14.9% (A)

9

Villedieu 1998, 29-30, 39.

2B1

250-350

10.0% (A)

52

Villedieu 1998, 25, 33.

Vigna Barberini

2B2

250-350

13.4% (A)

191

Villedieu 1998, 26-27, 36.

SW Palatine Hill

-

l. 4th c.

15.2% (A)

168

De Rossi & Mandarini 1998.

Lungotevere Testaccio

US 201

l. 4th c. (398?)

19.6% (C)

-

Meneghini 1987-1988.

S. Stefano Rotondo

-

l. 4th - e. 5th c.

36.7% (A)

91

Martin 1989; 1991-1992.

Lungotevere Testaccio

US 106

e. 5th c. (408?)

23.0% (C)

-

Meneghini 1987-1988.

Schola Praeconum

I

e. 5th c.

17.7% (A)

139

Whitehouse et al. 1982.

Temple of Magna Mater

-

400-425

7.0% (B)

20

Carignani et al. 1986.

Caelian Hill (domus)

-

480-550

5.2% (A)

8

Pacetti & Sfrecola 1989.

Schola Praeconum

II

600

2.0% (C)

12

Whitehouse et al. 1985.

Date: e. = early, m. = mid; l. = late; Share: (A) = of cookwares; (B) = of cookwares and African Red Slip ware; (C) = of the total assemblage; Nos. = number of sherds.

118

Distribution and consumption

Table 49. Examples of quantified deposits with African cookware at Ostia and Portus. Site

Area

Date (AD)

Share

Nos.

References

Ostia, Terme del Nuotatore

Area NE

90-120

29.8 % (A) 181

Coletti & Pavolini 1996, 413.

Ostia, Terme del Nuotatore

Area NE

120-140

45.5 % (A) 412

Coletti & Pavolini 1996, 414.

Ostia, Terme del Nuotatore

Area NE

160-190

56.0 % (A) 3324

Coletti & Pavolini 1996, 413.

Ostia, Terme del Nuotatore

?

230-250

97.9 % (A) 7741

Panella 1991, 296, fig. 14.

Ostia, Terme del Nuotatore

?

4th c.

95.7 % (A) 428

Panella 1991, 296, fig. 14.

Pianabella (Ostia antica)

Area 3000

400-450/70

37.0 % (B) -

Paroli et al. 1993, 281, fig. 4.

Pianabella (Ostia antica)

Area 3000

450-550

27.0 % (B) -

Paroli et al. 1993, 281, fig. 5.

Portus

Area I

500-550

54.7 % (A) 8 1

Coletti 1998, 403, fig. 6.

Pianabella (Ostia antica)

Area 3000

550-e. 7th c.

3.0 % (B) -

Paroli et al. 1993, 281, fig. 7.

Portus

Area I

7th c.

2.8 % (A) 6

Coletti 1998, 403, fig. 7.

Date: e. = early, m. = mid; l. = late; Share: (A) = of cookwares; (B) = of cookwares and tablewares; (C) = of the total assemblage; Nos. = number of sherds.

includes material from Pianabella (Ostia Antica) and Portus. At Pianabella, the share of African cookware in a sequence formed of several chronologically distinct deposits is clearly a declining one. Still, the extension of African cookware production to the early 7th century AD is less likely than the possibility that the few examples found in the latest deposits are residual. Generally speaking, the evidence from Ostia shows that the intimate relationship between the capital and its port is not reflected in an identical distribution of imported commodities,32 as the share of African cookware products at Ostia is constantly higher in comparison with Rome.33 The observation is of great importance,34 as it may signify that the state-controlled distribution of grain and olive oil was actually a crucial precondition for the exportation of African pottery. But the main port of the Empire, Portus, was also an important center of redistribution for merchandise.35

5.2.4), were just on the way from Carthage to Rome or if their actual destination had been the island itself. The dependence on maritime transportation is certainly reflected in Sardinia, which was probably one of the intermediate stops on the trade route between Carthage and Rome. The availability of African goods ensured by its favorable location has been confirmed by archaeological data. The excavations that took place at Porto Torres in the northwestern part of the island confirmed that these products were present in quantities comparable to those of contemporary deposits of Rome and Ostia. For example, African production represents 40% of cookwares in a context

Luni Albintimilium Pisa

Pesaro

S. Gaetano di Vada

5.2.2 Italy The evidence from South Italy and Sicily does not lend direct support to the notion that the geographical proximity to Africa Proconsularis would imply the strong presence of African cookware in post-1st century AD contexts. For example, only a moderate amount of African cookware has been found in a context dated after AD 330 at Otranto (Puglia),36 which is one of the few reported occurrences of this pottery reported from the southern peninsula (Figure 23). The evidence from Sicily is even more sporadic, even though excavations at Agrigento, Segesta and Piazza Armerina, have succeeded in producing a fair amount of African cookware.37 The picture would possibly be more focused if we could tell whether the many ships that sunk with African cookware on board within the territorial waters of Sicily (see section

119

Roselle Cosa

Suasa Lugnano in Teverina Viterbo

Portus

ROME Ostia Sperlonga

Porto Torres

Puteoli

S.Giacomo degli Schiavoni S.Vincenzo al Volturno S. Maria Capua Vetere Pompeii S.Giovanni di Ruoti Otranto

Segesta Piazza Armerina Agrigento CARTHAGE

Figure 23. The map of Italy with sites mentioned in the text.

Chapter 5

dated to the third quarter of the 2nd century AD and by the late 3rd/early 4th century AD this figure has risen to 67%.38 Therefore, it is hardly surprising that the site also yielded similar, though possibly residual material from late 5th century AD deposits, which are dominated by cookwares of local manufacture.39 Evidence concerning the distribution of African cookware in the Bay of Naples is scarce, mainly because the heyday of this pottery dates after the eruption of Vesuvius in AD 79. Moreover, many contexts in which African cookware has been found cannot be dated accurately.40 Not surprisingly, the northern shores of the coastal West-Central Italy have been well supplied with African cookware, as the capital of the Empire supported its surroundings through vast amounts of imported pottery. Cosa, where the earliest deposit (22II) containing significant amounts of African cookware has been dated from the early 1st century AD onwards, is a case in point of favorable geographical location near the coast.41 An indication of the strong presence even in ager Cosanus is offered by the villa of Settefinestre, where African products made up 40% of cookwares in deposit dated to the late 2nd/early 3rd century AD.42 But as usual, when found in the late Roman and early Medieval deposits of the late 5th and 6th century AD, African cookware has automatically been considered residual.43 African cookware is also well distributed in coastal North Tyrrhenia starting from the late 1st/early 2nd century AD.44 At the excavations of Horrea di Vada Volaterrana in località S. Gaetano di Vada, African cookware counts for 24% of the pottery finds in contexts dated between the 1st and the 3rd century AD and 30% in contexts dated from the 4th to the early 7th century AD.45 Similarly, the excavations carried out in the early 1960's in Giardino dell’ Arcivescovado at Pisa produced two deposits of great interest, neither of which can be dated precisely. Especially the layer E2, which has been defined as a mixture of post-1st century AD materials, shows with its 80% share the status of this production in this area.46 In contrast to the coast, the lack of cookware imports at inland sites of West-Central Italy is striking. The difference can be illustrated with several examples, of which the most indicative is perhaps the case of Località Rebibbia, via S. Cannizzaro. Although the site with its multiple deposits extending from the early 1st to the 5th century AD is located just ten kilometers east northeast of Rome, the find assemblage included only four sherds of African cookware, 47 all of which are most certainly residual. An even more strict pattern has been observed at the Roman villa of Lugnano in Teverina (Umbria) and at the sites of Asinello and “Le Masse di San Sisto” in Viterbo (Latium). The excavations of several deposits dated between the late 2nd and the mid-6th century AD at these sites yielded substantial quantities of African

120

tableware, while not a trace of African cookware was present.48 Hence, it is plausible to suggest that African cookware was mainly distributed through the coastal centers and, therefore, it rarely reached inland. The difference between the coast and the inland can also be observed on the eastern side of the Apennines, though the absolute amount of African pottery imports is there markedly inferior.49 The scarcity of African cookware in the Adriatic region during the 4th and 5th centuries AD has not only been expounded as an indication of increased transportation costs but also as a reflection of declining cookware production in Africa; even African Red Slip wares were then widely distributed.50 Thus, the proximity to the area of production may have been of considerable importance in terms of the distribution of cookwares in the late Roman period, while the Adriatic and Tyrrhenian can be distinguished as two economically and culturally different regions.51 The slightly delayed date of the introduction of African pottery in the Adriatic region in contrast to the Tyrrhenian coast not only supports this idea,52 but suggests that the distribution mechanism behind it was free commerce (see section 5.3.1). When moving towards northern Italy, the most important site regarding the subject of this study is Ventimiglia (Albintimilium), located on the coast of northern Liguria, where African pottery was diffused relatively early, even if the examples predating the late 1st century AD are regarded as intrusive.53 The peak in the quantity of African cookware, however, was reached only in the mid-4th century AD, when it stands for 16% of all commonware finds.54 As in other regions, this was followed by a sharp decline in the amount of material, as shown by the 5th century AD deposits in which only 5% of the finds are classifiable as African cookware. The situation is vaguely similar to both the Tyrrhenian and the Central Adriatic coast. Trace amounts of African cookware have also been found in Lombardia and Piemonte, which are regions reachable only through the use of land transportation.55 Therefore, it is even more reasonable to believe that the trade in this ware was strongly affected by transportation costs. 5.2.3 Western Mediterranean In France56 the earliest layers containing African cookware, which have been encountered on sites like Aix-en-Provence (Aquae Sextiae), Lattes, Ambrussum and Fréjus (Figure 24), have been dated to the second half of the 1st century AD. In spite of the quantitative insignificance of African cookware in the late 1st/early 2nd century AD deposits, it evidently supplemented the batterie cuisinie consisting mainly of the local products.57 A decisive change took place by the late 2nd/early 3rd century AD, whereupon African production seems to have occupied certain segments of cookware markets.58 However, in the late 2nd century

Distribution and consumption

Murillo de Calahorra

Lattes

Aque Sextiae Frejus Marseilles

Mas-Saintes-Puelles Caesaraugusta (Zaragoza)

century, although imported pottery generally occurs in lower quantities in inland sites than in centers located on the coast.63 The importance of this production on the Iberian peninsula during the late Imperial period is shown by several assemblages – Sant Marti d'Empúries, VilaRoma and Iluro – including relatively high percentage of African cookware. In particular, the 22.4% share of all the pottery finds in the mid-5th century AD context at Iluro is markedly higher than what could be reasonably expected from a production thought to have ceased during the period in question.64 For now, the cookware products from Africa Proconsularis seem to be substantially rare in the western provinces of Roman Africa, although the picture may be skewed by the small number of archaeologically explored sites. The deposits unearthed at the Forum of Cherchel confirm the scarce distribution of both African table- and cookware products in this area, although the evidence from the excavations of a villa located at Nador may force us to re-evaluate the interpretation in the future.65 Both table- and cookwares were certainly exported to the province of Tripolitania, but from the late 3rd century AD onwards they were gradually replaced by local production.66 The excavations of Berenice in the province of Cyrenaica have shown that this area had firm trade connections with the Roman east, which is reflected both in the sporadic presence of African cookware and the increasing number of Aegean cookware imports, which took place at the expense of local pottery production.67 In fact, it has been argued that Cyrenaica was isolated from Africa Proconsularis and Tripolitania by unfavorable winds and sea currents, as well as by cultural differences. The comparison of pottery found in Sabratha (Tripolitania) and Berenice (Cyrenaica) demonstrated that the distribution of Tripolitanian products as well as imported pottery found at Sabratha witnessed strong contacts with the Roman west, while the corresponding evidence from Berenice indicates intensive trade with Crete and other centers of economic importance in the northeastern

Puig Rodon (Gerona) Celsa

Baetulo Tarraco (Tarragona)

Pollentia

Valencia

Ebusus

Javea Colonia Patricia Corduba Malaga Ceuta

Caesarea (Cherchel)

Tipasa

Figure 24. The map of the Western Mediterranean with sites mentioned in the text. AD at sites located not further than ten kilometers away from the coast, the presence of this pottery is only marginal.59 Still, the late 2nd century AD material found in the excavations of Roman villa at Mas-Saintes-Puelles, Bernardis (Languedoc) has verified the presence of this pottery, though in very modest amounts – circa 8% of the identified vessels – on sites located over 100 kilometers away from the coast.60 The first cookware products of African manufacture seem to appear in the Iberian peninsula as early as in the Julio-Claudian period with a frequency corresponding up to 5% of all the commonwares. Much of the evidence regarding the early importation of this ware derives from a sequence of deposits excavated at Baetulo dating from the late 1st century AD onwards.61 Still, African cookware reached a dominant position over the Spanish markets only in the mid-3rd century AD, after the introduction of several new vessel forms.62 In the meantime, the volume of African cookware imports had grown fairly steadily (Table 50). The volume of trade seems to have remained somewhat stabile during the late 2nd

Table 50. Examples of quantified deposits with African cookware in Spain. Site

Area

Date (AD)

Baetulo

-

l. 1st c.

Baetulo

-

Orive (Colonia Patricia)

Share

Nos.

References

5-12 % (B)

-

Aquilué Abadías 1985, 219-220

125-150

14-22 % (B)

-

Aquilué Abadías 1987, 26, 46

Corte 3

175-200

14.5 % (A)

328

Sant Marti d'Empúries

5000

400-450

8.2 % (B )

67

Carrillo Díaz-Pinés & Murillo Redondo 1994, 1311 Aquilué Abadías 1997, 85

Vila-Roma

-

440-450

16.7 % (B )

-

Aquilué Abadías 1989, 426

Iluro

UE 3000

450

22.4 % (C)

-

Revilla Calvo et al. 1997, 104.

Date: e. = early, m. = mid; l. = late; Share: (A) = of cookwares and tablewares; (B) = of commonwares; (C) = of the total assemblage; Nos. = number of sherds.

121

Chapter 5

Mediterranean.68 This explains the virtual absence of Cyrenaican cookware products in the study assemblage. In cross-regional surveys, attention has mostly been paid to the distribution of African Red Slip ware and the results are customarily based on sherd counts,69 which is not necessarily the most applicable method in this kind of analysis. However, they seem to strengthen the prevailing theory concerning the distribution of African pottery. After a period of rapid increase in the number of African tableware exports, which lasted from the late 1st century AD to ca. AD 160-190, the first significant decrease post-dated the reign of the Severan dynasty.70 Following a momentary depression around AD 300, trade in African pottery starts to flourish again, possibly due to the introduction of African Red Slip ware D. Having reached a new peak in the AD 380's and 390's, the volume of trade began to decline slowly towards the mid-5th century AD, whereupon the production of African cookware is traditionally presumed to cease. On the other hand, the decline in the exportation of African cookware may have already started in the latter half of the 4th century AD, although the cookware production itself seems to have continued in Roman Africa on a local scale well after the 5th century AD.71 This is also the picture based upon the study assemblage, where the share of African cookware is subjected to a steady decline between the late 4th and early 6th century AD.

Monaco A

5.2.4 Shipwrecks Roman shipwrecks, especially those examined with underwater excavations, is one of the few sources that offers in favorable conditions undisturbed primary material – an archaeological snapshot – which was not formed through deliberate selection, as is the case with grave goods.72 However, this does not rule out the fact that shipwrecks are subjected to pre- and postdepositional processes just as conventional archaeological layers.73 Our picture of ancient seamanship is also likely skewed both by the uneven distribution of underwater activity and the unwillingness of certain countries to make public the wrecks located within their territorial waters.74 At present, only a few reports have been published on wrecked ships that were carrying African cookware on board either as a component of cargo or utensils used by the ship's crew.75 Although the rarity may reflect the insignificance of African cookware in the Roman world, the picture has possibly been influenced by the excavation reports, in which the full potential of the pottery finds has only recently been exploited.76 The pottery evidence of all the Roman ships from the western Mediterranean that had African cookware on board as they went down is summarized in the following table (Table 51).77 The first observation emerging from the data is that the total number of shipwrecks found outside the trade route between Rome and Carthage is considerable (Figure 25). This makes redistribution and commerce seem to have been more probable mechanisms of

Procchio Porto Azurro A

Cala Culip IV

Punta Ala Giglio Porto Marritza

Trincere ROME

Cales Coves Porto Cristo Cabrera III

Colonia de Sant Jordi C Fontanamare A Skerki Bank

Cefalu

CARTHAGE

Plemmirio B Camarina A

122

Ognina

Figure 25. Roman shipwrecks with African cookware.

Distribution and consumption

Table 51. Roman shipwrecks with African cookware in the Mediterranean Shipwreck

Region

Date (AD)

ACw

ARS

Amphorae

Cabrera III

Majorca Balearic Is.

3rd c. ca. 257

H 23 B H 193 H 196

H 14/17 H 40 H 50

Africana Almagro 50 & 51c Beltrán 68 & 72 Dressel 20 & 23

Cales Corves

Menorca Balearic Is.

150-200 350-e. 5th c.

H 23 B, 196 H 197

H 58 B, 63 H 67, 91

-

Camarina A

S Sicily

l. 2nd c.

H 195, 197

-

Africana I, II

Cefalu

N Sicily

5th-6th c.

ns

ns

Keay 53-55, 62

Colonia de Sant Jordi C

Majorca Balearic Is.

250-300 (?)

oa

Fontanamare A

SW Sardinia

305-310 (?)

WNW Italy

200-225

H 31 - 32 L 40, 52? -

Africana II

Giglio Porto

oa pc ns

Les Laurons B

Marseilles S France

175-200

H 23 B

H8A

Pélichet 47

Marritza Monaco

N Sardinia S France

Ognina

Syracusa SW Sicily

75-125 late 3rd c. (?) 200-250 AD 210-215

as oa pc H 23 B H 197

ns H3 C, 27 H 50 A -

Plemmirio B

Syracusa SW Sicily

200

H 23 B, 196 H 181 183-4

-

Porto Azurro A

Elba WNW Italy

H 181, 195 H 196, 197

Majorca Balearic Is.

H 23 B H 196 H 197

H 3 C, 8 A H 58 B, 59 A H 61 A H3C H 58 B H 67, 99 A

Africana II D

Porto Cristo

150-200 250-300 320-350 150-200 300-400 e. 5th c.

Procchio

Elba WNW Italy

150-200

H 23 B H 196, 197

H3B H9A

Africana I Pélichet 47

Punta Ala

WNW Italy

250-300

H 23B oa pc

H 50 A

Africana II B-D Beltrán 72 Dressel 20, 30 (?)

Skerki Bank F

Deep-sea

l. 1st c.

H 196 H 194

-

Dressel 3, 9 Empoli, Pompeii?

Trincere

Tarquinia WNW Italy

l. 2nd c. e. 3rd c.

H 23 B H 196 H 197

-

Africana II Keay I A Benghazi MR 1

Africana II B, II D

Africana II A, B

Africana II A Pélichet 47 Africana I Almagro 50 Kapitän I-II Africana I, II A Mauretanian

Tripolitana III

KEY: Date: e. = early, m. = mid; l. = late; ACw: African cookware forms; H = Hayes; ns = not specified; oa = orlo annerito; pc = patina cenerognola; as = sigillata ‘a strisce’; ARS: African Red Slip ware forms; H = Hayes; L = Lamboglia. Wrecks in which African cookware traveled as part of the cargo are indicated in boldfaced letters.

transportation in contrast to the collection of the state tax. In many cases the cargo also includes merchandise of diverse origins. For example, transport amphora finds derive from many areas on the Mediterranean and their surmised contents also vary greatly: wine, olive oil, figs and fish products. The abundance of mixed cargoes suggests that they result from long trade routes involving several intermediate

123

stops between the ports of departure and arrival. During these stops, some of the merchandise was obviously unloaded, sold and replaced with local items of trade.78 However, only some wrecks produced indisputable evidence of African cookware as a part of the cargo,79 whereas in most cases this material is associated with the belongings of the ship's crew.80 The clearest case favoring the former interpretation is the recently documented late 1st

Chapter 5

century AD wreck Skerki Bank F located in the deep sea between Carthage and Rome, some 50 kms west northwest of Sicily. The large consignment of African cookware this ship was carrying did not only compriseseveral forms but it has also been characterized as “very well defined and probably (was) packed separately as a unit”.81 Hence, the Skerki Bank F wreck is an important indication towards the possibility that African cookware was, at least from time to time, an integral component of the cargo rather than a space filler or an utensil used by the ship's crew. Another case of certain importance is the Camarina wreck, in which the examples of African cookware found on board were most likely unused. Also in the early 3rd century AD wreck, near Trincere that lies in the a shallow water off the Tarquinian coast, the cargo the ship had been carrying consisted very likely of African merchandise, but only the pottery was preserved in the seawater. The pottery of the cargo consisted mainly of African cookware in addition to transport amphorae carrying fish sauce and

wine, and few examples of North African mortaria.82 Although the sample is very probably biased, because only the material on the surface of the wreck was collected, the fact that the site is located well off the Carthage-Rome trade route makes it more difficult to associate this wreck with the state controlled distribution of goods than with redistribution or commerce.83 The most important piece of the negative evidence regarding the importance of African cookware is without a doubt the late 4th century AD deep-water shipwreck called ‘ISIS’, also located at Skerki Bank, some 50 kms west northwest of Sicily. The vessel had left Carthage in order to sail to Ostia with a compound cargo consisting mainly of olive oil and fish sauce amphorae from Tunisia, Calabria and Asia Minor, but neither table- nor cookware vessels that could have been interpreted as items of trade were found.84 Therefore, the lesson of the ‘ISIS’ is that the transportation of agricultural products was not automatically accompanied by accessory pottery of the same origin.85

5.3 AFRICAN COOKWARE AND THE NATURE OF ROMAN TRADE This section is aimed to examine the significance of African cookware as a component of the Roman trade by reviewing the results and ideas introduced in preceding chapters against different modes, by which goods were distributed inter-regionally in the Roman world.86 Theories concerning the distribution of agricultural and craft products from Roman Africa to other parts of the Mediterranean will be introduced first. Both the role of the state as a redistributional force and recent archaeological evidence stressing the importance of commerce will be emphasized. Thereupon, attention will be turned to the importance of pottery as an indicator of the trade of perishable goods, because even the pottery of the highest quality was seldom traded alone inter-regionally. This topic leads to the question regarding the value of pottery as an exported commodity. Ultimately, the quality and value of African cookware is discussed in light of other African pottery as well as competing cookwares traded in the western Mediterranean during the late Imperial period. 5.3.1 Exchange, redistribution or commerce? The Roman economy is hard to fit into a model that would cover all the aspects related to the movement of goods in the Mediterranean in satisfactory manner.87 Therefore, the field has been left open for competing theories that in most cases emphasize the importance of one particular aspect, which is thought to have been decisive enough to maintain the functionality of 124

the system as a whole. The three most common catchwords have been exchange, redistribution and commerce, all of which must have had some importance in the distribution of goods in the Roman world.88 Until recently the prevailing concept on the economy in classical antiquity was based on the presumption that agriculture, notwithstanding of other indices, was always the pivot of the Roman economy.89 As the produce was consumed by the producers themselves or redistributed elsewhere more often than traded,90 the poorly organized transportation of agricultural products was strictly regulated. If interregional trade existed at all, it was small in volume and based either on luxury or subsistence goods supplied by the government. The items belonging to the latter group were only traded in case of crises and shortages,91 whereas low-priced provisions intended to be mass-marketed overseas hardly existed.92 Hence, the presence of both African table- and cookwares in Italy is explained to result from ‘internal supply’ taking place between important senatorial estates of Italy and Africa Proconsularis,93 while Roman society is said to have persisted with its heavily ostensible prejudices against trade and traders.94 Taxation has also been regarded as a primary stimulus for inter-regional trade in the Roman world,95 because taxpayers had to cash in the surplus they managed to produce in order to pay taxes. Hence, the provinces of North Africa have been counted in as tax-producing, relatively rich areas capable of

Distribution and consumption

supplying both the central administration and the Roman legions stationed in the frontier provinces. The sequence, in which peasants sold some of their products on local markets to artisans – involved in the production of more valuable, inter-regionally distributed goods – resulted in the increase of agricultural production, division of labor, number of artisans, size of towns, development of local markets and long-distance commerce.96 But taxes were more likely paid in kind, as the annual shipments of grain and olive oil from the provinces of North Africa to Rome show.97 In fact, the movement of goods out of the African provinces resulted to a great extent from the need of the state to feed the population of its overcrowded capital, which in case of food crises would have posed a serious threat to the supremacy of the elite.98 Agricultural products of Roman Africa, first grain and later also olive oil, became indispensable components of these efforts from the Flavian period onwards.99 The sudden predominance of African products has often been explained with the decline of the Italian economy, of which, however, there is no clear evidence.100 Instead, the significant growth of agriculture in Roman North Africa resulted likely both from environmental advantages – including the climate and a light but also less yielding soil – and the expansion and intensification of the existing production, which relied on small proprietors, tenancy and labourers of varying status.101 The acquisition of African provisions was carried out through cura annonae, a set of operations needed for the transportation of grain and other provisions, annona, 102 gathered as a tax in kind and distributed free of charge to the poor of Rome.103 As the state did not possess a merchant fleet of its own, the transportation of annona was handed to ship owners (navicularii), first hiring out under a sixyear contract for this purpose.104 From the 2nd century AD onwards the state established these contracts with ship owner colleges, which were easier to deal with than a large group of individual ship owners. Although colleges could also secure their privileges easily, the task itself became hereditary, burdensome and very hard by its conditions by the early 4th century AD.105 The shipment of grain to Rome had been originally profitable to ship owners, as the participation in this service gave an immunity from public obligations (munus publicum), a privilege that was later awarded to the shippers of olive oil.106 As the vast majority of Romans did not profit from annona, they were potential consumers of the commeatus privatus, commercial supplies these merchant ships could carry alongside the fiscal load.107 At least in the late Roman period these products were exempt both from harbor tax (portorium) and city customs tax (octavae),108 which would have otherwise increased their price by 16.5-17.5 %.109

125

Literary sources suggest that instead of potters, the transportation, distribution and marketing of pottery was taken care of by private merchants (navicularii-negotiatores),110 who could avoid collatio lustralis, a trade tax levied every fifth year from the Constantine era onwards,111 by participating in the transportation of annona. The flux of North African commodities – grain, wine and olive oil – resulting from state-controlled operations created an opportunity to exploit the overseas markets with other African products, like table- and coarsewares, horses, fish sauce and timber.112 Thus, the reasons why pottery traveled alongside everyday commodities were both the reduction of transport costs and the exemption from taxes when shipped together with fiscal loads. As a result, African pottery products were competitive with local products even when shipped to remote places.113 In many cases the association of annona and the exportation of African pottery seems particularly evident. The exportation of African Red Slip ware, for example, was strongly affected by the decrease in the exportation of African olive oil,114 whereas the reorganization of property and production in Africa Proconsularis following the Vandal conquest in the first half of the 5th century AD is said to have caused the disappearance of African cookware from the Mediterranean markets.115 The late 5th century AD examples of African cookware in the study assemblage as well as on other sites around the western Mediterranean contrast the latter idea. In fact, only the absolute volume of African pottery seems to diminish in the 5th century AD, whereas its relative share among the pottery imports remains more or less the same.116 Also the opposite conclusion, stressing that the Vandal occupation of Roman Africa stimulated intense commerce rather than caused the decline of agricultural or pottery production, has been reached on the basis of both pottery and coin evidence.117 The possibility of estate owners to sell the produce instead of giving it away in form of taxes was short-lived,118 as already in AD 533 the Byzantine troops led by general Belisarius re-conquered North Africa and the system of taxation was temporarily re-established.119 Although the marginal costs of exportation were certainly crucial to merchants as they took advantage of the grain annona,120 African pottery would be found exclusively in the major centers of consumption, like the city of Rome or the military encampments receiving provisions from Africa, unless a certain fraction of it was redistributed from Ostia or traded on the basis of free commerce.121 The relatively short and safe trade route between Rome and Africa Proconsularis is the other factor contributing to the strong presence of African imports in Ostia.122 The wide dispersion of African cookware shows, in any case, that a considerable part of maritime transactions was based on commerce rather than on bilateral

Chapter 5

relationships of remote estates. Generally speaking, recent archaeological discoveries have shown that both inter-provincial and inter-regional trade in foodstuffs, minerals, metals and hand-made products were particularly intense in Roman North Africa, especially during the 2nd and 3rd century AD.123 These indices make the exclusion of profit-based commerce from the Roman Empire difficult, although it would be absurd to assume that pottery alone would have upheld the flourishing trade between Rome and Carthage.124 But the economy of Roman Africa was not unidirectional either as shown by abundant ruins of luxurious villas and grave monuments, which are an evident reflection of the increased power and wealth in North Africa,125 Economic growth also increased the political importance of these provinces and, together, these factors strongly contradict the traditional conception of the Roman economy either as a static and unchanged or primitive and underdeveloped entity.126 A part of trade was undoubtedly redistributional by nature and controlled by the state, but commerce and market economy also had a role in the Roman Empire.127 Interestingly, African cookware has been used as an example of ceramic ware, whose mode of exportation will not be fully understood until the degree of involvement of the state through the distribution of grain and olive oil is defined.128 5.3.2 Pottery as a proxy of Roman trade The evident association of pottery with perishable commodities has raised the question whether coarsewares could be used as an index of the intensity of their trade. When found outside of its presumed production area, pottery, which is seen in the most optimistic appraisals as one of the few reliable sources on the Roman economy,129 may indicate the direction and volume of commerce established through trade in perishable commodities.130 The regular presence of various table- and cookwares of “relatively humble status” at coastal sites along the Mediterranean has also been taken as an important sign of the volume and the sophistication of inter- and intra-provincial trade,131 while the absence of imported pottery has been said to indicate self-sufficiency.132 In more pessimistic views, it is pointed out that the volume of accessory items of trade does not necessarily tell anything about the volume of trade in the principal item.133 Also the usefulness of pottery as a proxy for trade in perishable commodities has been doubted, because the importance of compound cargoes is underlined by a considerable fraction of the evidence.134 It is therefore plausible that the exportation of pottery, coarsewares in particular, was partially independent from other items of trade sharing the same origin.135 Surprisingly, the ancillary character of ceramics to the perishable items of trade has been taken as a negative quality and the

126

information value of pottery has been belittled with arguments disclaiming its usefulness.136 The most repeated argument concerning the exportation of Roman pottery other than amphorae is, however, that it traveled in small batches as secondary cargo.137 Of the other constituents of a compound cargo, at least the principal item of trade is believed to have been more valuable than pottery.138 Compound cargoes resulted either from commercial supplies carried alongside the fiscal load in merchant ships or several stops the ship made before reaching its final destination.139 In some shipwrecks, pottery has been found piled between the stacks of amphorae to fill up the empty spaces left by the primary cargo, but alternatively pottery could also have been placed on top of it.140 Still, the proposition that pottery acted as ballast in ships carrying grain or olive oil as their primary cargo141 is not acceptable due to its relatively high stowage factor.142 Large cargoes composed of only one type of valuable traded item, on the other hand, would have hazarded the subsistence of tradesmen, because the risk of losing the cargo in a wreck was substantially high, and mixed cargoes were probably a convenient way to reduce the risk.143 An elucidating example of the mode pottery was traded in the Roman empire is offered by the Cala Culip IV shipwreck with 86 amphorae and over 4000 fineware vessels, both South Gaulish Sigillata and Baetican thin-walled ware, included in the cargo.144 This wreck makes the idea about coast-hopping vessels transporting goods from a port to another in somewhat random pattern easier to accept. In addition, the importance of pottery in the Roman trade is put in the right perspective by the observation that the some 4000 tableware vessels weighed only 10.5 % of the total cargo and occupied just 3.5 cubic meters in the hold. Generally speaking, the evidence on the maritime transport of traded goods is plentiful compared with the evidence on the mode pottery was distributed after it was unloaded from ships. The only well-known case comes from Pompeii (House IX, Insula V, Regio VIII), where on the 4th of October 1881 was witnessed the discovery of the charred remains of a wooden case including 90 vessels of South Gaulish Sigillata and 37 unused terracotta lamps of North Italian origin.145 This find implies that even small consignments of pottery might have contained vessels of different provenance and that trade in pottery was not necessarily based on direct contacts between the area of production and distribution. 5.3.3 The value of African pottery as exported commodity The production of pottery was highly unlikely an industry of major economic importance in antiquity, even though most archaeological assessments on the nature of Roman trade are based on the pottery evidence.146 Still, the impact exported pottery had on

Distribution and consumption

the economy of the producing region has been said to differ from one case to another. The importance of Italian and Gallic Sigillata industries has always been stressed, whereas widely distributed African pottery products are said to have had less economic weight, because their production took place in small workshops.147 As the interpretation is based on the absence of stamps from African pottery, it overemphasizes the significance of additional information offered by stamped Italic and Gallic Sigillata.148 The absence of stamps in a mass-product, like African Red Slip ware, distributed on a large scale, may signify that the importance of individuals in the production sequence had been reduced to the minimum. In addition, the production site excavated recently at Leptiminus with its eight kilns dated from the 1st to the 3rd century AD – the tip of the iceberg in our knowledge of the North African pottery production sites (see section 4.4.2) – certainly does not meet the definition of a small workshop. Hence, some producers have made their profit through the mass-production of low-cost commodities, instead of specializing themselves on expensive high-quality products.149 The abundance of African pottery exports in towns and rural sites located near the coast and principal shipping routes indicates the low-cost maritime transport as the main factor stimulating the movement of North African commodities to Rome and to other parts of the western Mediterranean.150 Conversely, exportation might have even raised the marketing value of pottery.151 The increase did not only result from the costs of transportation, as several other reasons may have also improved the competitiveness of African pottery as an exported item. These include larger holdings, minor investments and standardized medium-quality massproduction besides flexibility and effective coordination.152 On the other hand, the flourishing pottery industry of Roman Africa can be seen as an adaptation,153 by which the production reacted to the external pressures set by the climate, availability of raw materials including fuel. Because the literary sources do not contain any references to merchants of coarse pottery, it was not likely a central component of maritime trade.154 Still, the idea regularly repeated in the discussion concerning

the massive amounts of African cookware exports that wheel-made coarsewares would have had the status of tertiary goods, while finewares customarily belonged to the second rank,155 is somewhat vexing. Although the trade in coarse pottery certainly profited from the shipments of fiscal loads and trade in other perishable commodities, another item would have occupied the place of cookpots in the hold, had their quality and value been overwhelmingly low.156 Moreover, the idea about African cookware as a product of inferior quality and low marketing value compared to African tablewares should be rejected as a classic example of an evaluation based on the common presumption that plain hand-made artefacts are primitive and of low importance.157 The argument on the low quality of African cookware products is not acceptable.158 In terms of utilitarian ceramics, when the value of the product is determined by its usefulness,159 these vessels are welldesigned, well-executed and therefore probably something that an average Roman was inclined to use, if metal cooking utensils were beyond one’s economic realm or unsuitable for a given task.160 In terms of use-related properties, African cookware should be easy to classify as high-quality produce. Because the use of cooking vessels was an everyday necessity,161 improved performance characteristics probably raised the demand of African cookware outside Roman Africa. This partially contradicts the idea according to which pottery production in the Roman Empire was chiefly based on quantity, not on quality.162 Additional evidence contradicting the low value of coarsewares can be pointed out from Sardinia, where cooking vessels produced in the village of Pabillonis are distributed nearly over the entire island due to their extraordinary thermal properties.163 Improved use-related properties could also explain in part the success of African cookware,164 as fairly recent archaeological experiments have managed to point out that cooking vessels may actually be considered masterpieces of ancient engineering. Cooking vessels may not always be aesthetically pleasant, but nevertheless, they can still be functional objects of high quality.165 In all, the reasons why it was more convenient to purchase an African casserole instead of a locally produced corresponding form may have been plentiful.166

5.4 POTENTIAL CONSUMERS As exported pottery has always been the focal point for archaeologists studying the Roman economy and its distribution has nearly always been reviewed through the supply, hardly any attention has been paid to the demand. In general, coarse pottery is observed 127

to have gained wider distribution in areas that had well-established commercial contacts with the other parts of the Mediterranean,167 while other factors possibly affecting the distribution of pottery included: a) the direction of trade in principal exported items, b)

Chapter 5

the dependence of the recipient on imported provisions, particularly on state-controlled shipments of provisions and c) the nature and value of primary cargo.168 Still, even if pottery was marketed independently from other African products that had accompanied it in the hold, the actual problem is the total absence of information regarding the demand of the pottery in the Roman world.169 The situation originates partly in the lack of literary sources discussing the question, partly in the lack of scholarly interest to deal with this problem. A similar attitude has long characterized the majority of works written on the Roman kitchen or diet. For the charm of chronological exoticism, these books often content themselves to illustrate the culinary extravaganza offered by authors like Petronius Arbiter. Contrarily, the aim of this final section is to discuss the following two questions: a) who were the potential users of African cookware? and b) where was African cookware used? Although the answers to these questions seem at first obvious, it will be specifically pointed out that this is not the case, as the consumption of African cookware was largely depended on the commonness of kitchens and alternative cooking equipment in the Roman world. Private kitchens were by no means common in the Roman world, although the evidence from Pompeii shows that virtually every decent patrician atrium house had one. But atrium houses were not common either in Rome or Ostia during the Imperial period, as their number was only 1790 in contrast to 46,602 insulae, the Roman equivalent for a modern apartment building.170 The first or the first two floors of an insula were normally reserved for a single family apartment, paradoxically called domus, which could include a kitchen (culina), even though this was uncommon. Moreover, not only was the kitchen unfavorably located in a space underneath the stairs leading to the upper stories, but it could also include a toilet (latrina) without a partition.171 As culina-latrina, the combination of a kitchen and a toilet – a hygienic disaster in modern standards – is customarily found even in Pompeian style atrium houses, their co-existence was probably necessitated by several practical reasons. Firstly, the kitchen was connected with the cesspit of latrina with a sewer that carried away the water used in dish washing as well as fluids produced during to preparation of food, while the cook could also get easily rid of the solid waste of food processing.172 Secondly, with this arrangement the kitchen could deftly provide the hot water needed in the toilet.173 Roman kitchen had also religious functions, of which the worship of Vesta, Lares and Penates – the guardian spirits of the food in storage – are probably the best examples.174 Cooking took usually place on a 110-130 cm high platform, upon which the vessel was set on a metal tripod or a C-shaped construction

128

made of brick.175 The platforms are moderately common in Roman Campania, where their distribution has been observed to correlate with the social status of inhabitants, but rare to absent in Rome and Ostia, especially after the 1st century AD.176 The existence of a kitchen downstairs did not help the poor living on hire in the upper stories,177 which were customarily very crowded in spite of being divided into small rooms. As the estimated tenant density may have reached figures as high as three persons per ten square-meters,178 a family could live in a single room. While the preserved archaeological evidence on the upper stories of insulae is extremely scarce,179 these small rooms could not provide space for permanent structures – either an oven or a cooking platform – used in cooking. It is also improbable that each floor of an insula would have been equipped with a common kitchen. There is, however, some evidence on the use of portable metal hearths,180 although it is unclear whether they were meant to be used for cooking or just to keep different courses and plates warm. The possibility that braziers, the principal function of which was to provide the heating, were used in cooking has also been suggested.181 Hence, food was likely prepared with mobile equipment in any possible place, preferably next to window to expel fumes.182 Outlets both for firewood and charcoal (carbonariae tabernae) did exist throughout the town.183 Alternatively, the poor were forced to purchase their hot meals and even the hot water they needed from nearby food stands, cauponae and tabernae.184 The vessel needed to transport the meal from the nearest food outlet to home was probably owned by the customer,185 as the opposite case presupposes the existence of vessels, in which the name of the lessee would be indicated in one way or another. There are several conclusions to be made from the general absence of kitchens in the Roman world. Firstly, if the poor could not access the kitchen on regular basis or were not able to use alternative cooking facilities (i.e. braziers etc.), they needed very unlikely cooking utensils like African cookware. Therefore, the possibility that pottery shipped with the African grain would have been distributed for nothing through the channels that were also in charge of the distribution of free grain can been excluded.186 If the rich were accustomed to purchasing metal cooking vessels and the poor did not need any, the potential users of African cookware probably belonged to that part of the middle class, which was wealthy enough to purchase wheat and barley in a price fixed by the state.187 But after all, African cookware may have offered competent performance characteristics with a price that appealed to all levels of the Roman society.188

Distribution and consumption

CHAPTER 5 ENDNOTES 1 2 3 4 5 6 7 8

9 10

11 12 13 14 15 16

17 18 19 20 21 22 23

24 25

26

27

28 29

30 31 32 33 34 35 36 37 38 39 40

41 42 43

44

45

On the history of the discipline, see Orton 1993, 170-174. See e.g. Orton & Tyers 1990, 82-89. Orton & Tyers 1993. Pollard 1990, 77. Tomber 1993, 149. Orton 1975, 31; Orton & Tyers 1990, 83. Orton & Tyers 1990, 83, 105; Orton 1993, 173. Millet 1979b, 77-78 (methods e.i & e.ii), 79 table 2. The “estimated minimum number of vessels” method has been criticized for being overly based on “guesstimates” (Orton 1993, 175, 181), which is pretty much the truth. Riley 1979a, 101-102. The assemblage is a fairly integrated whole. Besides the two African cookware fabrics, the only component of certain quantitative importance is the West-Central Italian cookware. As the presence of other cookwares – Campanian cookware, Internal Red Slip cookware, Aegean cookware and Hand-built cookware – is clearly accessory by nature, their quantification is based only on the minimum number of vessels estimated from nucleated sherd families. Baxter & Cool 1997, 89-94. Clark & Schofield 1991, 100. Panella 1999, 183, 191. See e.g. Hayes 1983. Pavolini 1996, 229-230. The absence of Aegean cookware does not either support Moliner’s (1996, 250) idea of an ideal Roman earthenware cooking vessel panoply including African casserole with a lid, Aegean casserole and the African cooking plate, a term that has been used to define both the Hayes 23B shallow casserole and the Hayes 26/181 pan-casserole. Reynolds 1995, 16. Carandini 1970, 101; Reynolds 1995, 107. Reynolds 1995, 111. Carandini 1986, 10; Panella 1999, 191. See also Villedieu 1998, 25-33. Cf. Martin 1989, 481; Martin 1991-1992, 171; see also Meneghini 1987-1988, 360. Whitehouse et al. 1982, 58-59. The precise nature of the Schola Praeconum material is difficult to evaluate, because none of the sherds were illustrated. Pacetti & Sfrecola 1989, 490. Carandini 1970, 103; 1986, 10; Pacetti & Sfrecola 1989, 498-500. The estimates on the populace of Rome before the decline range from 630,000 to 1,000,000 individuals: Garnsey & Saller 1987, 83; Panella 1999, 183; Peña 1999, 3. The accuracy of these estimates, which have been derived from the number of persons receiving free grain (see section 5.3.1), has been doubted recently, see Purcell 1999, 137-139. Whitehouse et al. 1985, 177. This deposit has been reviewed by Hayes (1998, 12), who dates it to the early 6th century AD or to an indeterminate period in between the two suggested dates. Although this would not be possible without the excavations of the Terme del Nuotatore (Ostia I-IV; Panella 1991; Coletti & Pavolini 1996), in two fairly recent review articles attention has also been paid to the late ceramic sequences of both Ostia and Portus, which differ considerably in composition from the contemporary evidence i n Rome, see Paroli et al. 1993; Coletti 1998. Anselmino et al. 1986, 58-64; Pavolini 1998, 227. It is revealing that from the early 2nd century AD onwards, the African corn fleet sailed no longer to Puteoli but t o Ostia (see Rickman 1980, 18-19), where the construction of Trajan’s harbor had been completed. Panella 1986, 433-439; Coletti & Pavolini 1996, 413-415, tabs. 1-3. Panella 1991, 296, fig. 14. Pavolini 1996, 223 note 1, cf. Pavolini 1985, 200. Pavolini 1996, 227-236. Martin 1989, 482. Fentress & Perkins 1988, 213; Fontana 1991, 138-142. Giannotta 1992. Ampolo et al. 1971; Alaimo et al. 1997, 51. Panella 1986, 435-440. Panella 1986, 449. The famous imperial villa at Sperlonga, where African production constituted 15.5% of cooking vessels, is a good example of such site, see Sagui 1980, 474, 524 fig. 9. Dyson 1976, 161-167. Ricci 1986, 84-85, fig. 1. Fentress et al. 1991, 214. This convention has been questioned by Arthur (1987, 530-533), who rejects the possibility that the eight sherds of African cookware found in the late 5th century AD deposit at S. Maria Capua Vetere would be residual, although they represent only 1.4% of the assemblage. Although the second of a three-volume Luni excavation report (Luni II) introduces pottery finds of this area from the early Julio-Claudian era to the Middle Ages, the structure of the publication prevents efficiently its use here for quantitative comparisons. Pasquinucci & Menchelli 1996, 499, 504-505; Pasquinucci et al. 1998, 1408

129

Chapter 5

46 47 48 49

50 51 52 53 54 55 56

57

58

59 60 61 62 63

64

65 66

67 68 69

70 71 72 73 74

75 76

77

78 79 80 81 82 83 84 85

86

87

Pasquinucci & Storti 1989, 7-8, 79. Staffa 1986, 656-661; 662, fig. 397, nos. 183-184. Barbieri 1989, 117; Monacchi 1990; Barbieri 1996, 22-66. Of the individual sites in the Adriatic region, the Late Roman domus at Suasa is the only site that has produced African cookware even in moderate quantities, i.e. 20% of the diagnostic fragments of African pottery (Biondani 1992, 149, 167-168). The other evidence from the Adriatic coast is considerably more discouraging. For example, at S. Giovanni di Ruoti the excavations of four Late Roman middens dated between AD 375-525 (Freed 1983) have yielded plentiful evidence on other cookware imports, but only one African lid rim (Small 1994, 69, no. 61; see also Panella 1999, 208). The early 5th century AD deposits unearthed in the course of the excavations of a villa rustica at S. Giacomo degli Schiavoni (Albarella et al. 1993, 201-203) and the monastery of San Vincenzo al Volturno (Patterson 1987, 88, 101) – containing only a few examples of African Red Slip ware, while most cookware imports were of Aegean origin – are two parallel cases from Molise. Biondani 1992, 149. Albarella et al. 1993, 202. Biondani 1992, 173, see also Hayes 1994, 112. Olcese 1993, 154. Olcese 1993, 156-157, fig. 23, 180 fig. 27. Lavazza & Vitali 1994, 26 note 7, 31. The state of African cookware studies in France has been recently summarized in a collection of congress papers (Bats 1996) dealing with the commonwares from Campania and Gallia Narbonensis. Fiches 1996, 355, fig. 4; Nin 1996, 263, 270-272; Rivet 1996, 328, 336-337. For example, in a late 1st century AD deposit excavated at Marseilles, African cookware consisted of 61% of all imported domestic pottery, but only 2% in comparison to local cookwares, see Moliner 1996, 239 fig. 4, 244-245. Berato et al. 1986, 141-164. In a contemporary deposit excavated at Marseilles, African cookware makes up as much as 69% of imported cookwares and 31% of all pottery, see Moliner 1996, 239 fig. 4, 244-245. Fiches 1996, 355, fig. 4. Passelac 1996, 379, tab. 4. Aquilué Abadías 1985, 219. Marín Jordá 1995, 165. Reynolds 1993, 18; Carrillo Díaz-Pinés & Murillo Redondo 1994, 1301, 1316; Macias et al. 1997, 161, cf. Pons Machado 1998, 1340. This pottery is also present in deposits dated between the 4th and the early the 5th century AD at Málaga – where most examples belong to the Hayes 197 and Hayes 23 series (Navarro Luengo et al. 1997, 84) – and at Antiga Audiéncia (Tarraco) in a context dated to last quarter of the 5th century AD (Cau Ontiveros et al. 1997, 182). Fentress 1986, 25. Dore 1989, 61-63. The excavations of the Italian expedition at Sabratha have underlined both the momentary abundance of the Hayes 181/182 series and the rarity of North Tunisian products in Tripolitania, see Pucci 1975, 58. Riley 1979, 248; 1980, 73-74; Fulford 1989. Fulford 1989, 180, see also Clay 1992, 350-351. E.g. Fentress & Perkins 1988. Another serious attempt to summarize much of the pottery data has been published b y Reynolds (1995). His monograph on trade in the western Mediterranean brings together all the evidence on the distribution of pottery between AD 400-700, but its contribution to the study of African cookware is not as important as it is to the study of Late Roman hand-made cookwares. Fentress & Perkins 1988, 208-209. Panella 1986, 442. Gibbins 1990, 377; Parker 1992, 3; Adams 2001, 296-299. Parker 1984, 99-100; 1992, 2-4; Adams 2001, 296-298. Hence, the waters of Southern Gaul and Sicily are archaeologically much better known than, for example, the shores of Libya or Greece, see Parker 1984, 100-105; 1992, 9, see also Whittaker 1989, 537-538. Tortorella 1981a, 356. Cf. Gibbins 1990, 376. In this respect, Gibbins 1991; Bost et al. 1992 and McCann & Freed 1994 may serve as exemplary publications. The data is mainly based on Parker's catalogue of ancient shipwrecks in the Mediterranean (Parker 1992), but also o n Tortorella's article on shipwrecks containing African pottery (Tortorella 1981a). Additional data has been included from the following publications: Bost et al. 1992 (Cabrera III); Pontacolone & Incitti 1991 (Trincere); Gibbins & Parker 1986; Gibbins 1991 (Plemmirio B); Ballard et al. 2000; McCann 2001 (Skerki Bank F). Tortorella 1981a, 361; 1987, 283; see also Panella 1999, 209; Ballard et al. 2000, 1613-1614. Different constituents of a shipwreck are occasionally hard to distinguish, see Gibbins 1990, 384. Tortorella 1981a, 371; Gibbins & Parker 1986, 299; Parker 1990, 160; Gibbins 1991; Parker 1992, 29. Ballard et al. 2000, 1612-1613; McCann 2001, 259. Pontacolone & Incitti 1991, 544, 555. Pontacolone & Incitti 1991, 560. McCann 1994, 57; see also Ballard et al. 2000, passim. Conversely, shipwrecks containing African table- and cookwares without respective amphorae are said to indicate secondary trade, the direction of which was determined by the demand, Panella 1999, 209. The term ‘trade’ refers here to “organized, large-scale material transactions that took place regularly over relatively long periods of time and involved the flow of significant quantities of goods”, Tite 1999, 202. Morel 1983.

130

Distribution and consumption

88

Mattingly 1996, 238-239. Duncan-Jones 1974, 1-2; This concept was promulgated by a group of British historians, among which Moses I. Finley and Richard Duncan-Jones stood out as the most prominent figures. On the weak spots of “The Finley/Jones model”, see Greene 1986, 14-16, cf. Panella 1986, 445-446; Garnsey & Saller 1987, 43. 90 Garnsey 1999, 23. 91 Tomber 1993, 143. 92 Cf. Wickham 1988, 190. 93 Whittaker 1983, 176-177; 1988, 59, 67. Whittaker (1988, 60) estimates that 10-40% of the populace in Rome was either partly or totally depended on this ‘internal supply’. On the other hand, the distribution pattern of African exports, especially when found outside of Rome or the Italian peninsula, is far too irregular to have resulted from the exchange between senatorial estates, Carandini 1986, 14-15. 94 Duncan-Jones 1990, 46. 95 Mattingly & Salmon 2001b, 5. 96 Hopkins 1980, 101-105. To function, this model also calls for increased monetization, commercialization of exchange, elongated links between producers and consumers, rise of the specialist intermediaries and urbanization. See also Garnsey 1983, 128. 97 Garnsey & Saller 1987, 87. 98 Garnsey 1999, 2; Purcell 1999, 159. 99 Carandini 1970, 116; Fulford 1980, 73; Tortorella 1981a, 357; Carandini 1983, 149; Panella 1985, 182-185; Tortorella 1987, 304. The estimates on the amount of grain shipped annually to Rome vary greatly from 150,000200,000 (Garnsey 1983, 118, 128; Whittaker 1988, 61), through 270,000 (Rickman 1980, 10, 232) and 400,000 (Garnsey 1988, 231) to 500,000 (Raven 1993, 81-88) metric tons. 100 Rickman 1980, 103; Garnsey & Saller 1987, 59-63; Whittaker 1988, 52; Wickham 1988, 190. 101 Rickman 1980, 96; Shaw 1981, 390-391; Garnsey 1988, 44-46; Wickham 1988, 188-189; Hitchner 1993, 505506; Mattingly 1994, 138. On the landowning conditions in Africa Proconsularis, see Mackensen 1993, 479-486. 102 Fulford 1984c, 256-258. The wider meaning of the term annona, standing for the provisions needed to support a given population for a year, has recently been reduced to signify the maneuvers guaranteeing the annual food supply of the city of Rome, Pavolini 1986, 76. 103 This state controlled distribution system proves the existence of complex administrative and social structures in the Roman empire, Young 1981, 211. The office of the chief administrator, praefectus annonae, was created in the late Augustean period, while two subordinates, procurator annonae and procurator portus, were appointed only in the Claudian era. By the 4th century AD, when Roman Africa had become Rome’s greatest granary, the organization of annona had became the responsibility of praefectus annonae Africae, who worked as a subordinate of praefectus praetorio Italiae. After AD 331 praefectus annonae was also responsible for organizing the distribution of free grain in Rome, Rickman 1980, passim. In any time, Romans were obliged to supplement their grain-based diet with other sources of nutrition; or else they would have constantly suffered from the lack of the vitamins A, C and D. (Rickman 1980, 7). The exportation of African Red Slip wares A, C and D with annona did not relieve this problem, even though the explanation itself is extremely attempting in all of its pseudo-scientific beauty. 104 Rickman 1980, 72, 89; Panella 1985, 183-184; Pavolini 1986, 77; Garnsey & Saller 1987, 49-50; Raven 1993, 126; Reynolds 1995, 107; Garnsey 1999, 50; Peña 1999, 8. The eligible tonnellage of a ship to be used for this purpose was determined as 10,000 modii (ca. 70 tons), Garnsey 1988, 233. 105 Warmington 1954, 61-62; Rickman 1980: 73, 87-90, 228-230; Garnsey 1983, 127-128; 1988, 74; Peña 1999, 27. In the Theodosian code (AD 380) ship owners are bound to accept state cargoes from early April to early October, which was a severe restriction, as seafaring was suspended in the wintertime due to unfavorable sailing conditions, Rickman 1980, 15, 99. 106 Rickman 1980, 123; Garnsey & Saller 1987, 88; Whittaker 1988, 54; Duncan-Jones 1990, 162. 107 Harris 1993b, 16-17; Mattingly 1996, 239; Peña 1999, 8. 108 At least two kinds of tax, ansarium and foricularium, based either on the number of imports or their bulk volume, were levied, see Palmer 1980. 109 Peña 1999, 7; see also Warmington 1954, 61-62 Peacock and Williams 1986, 58; Tomber 1993. 110 A negotiator (large-scale wholesaler) could also act as a navicularius (shipper or financier of shipping) as well as the captain of the ship, but this was not always the case, Rickman 1980, 72, 124-126; Garnsey 1983, 125-126; Garnsey 1988, 70; Paterson 1998, 160. While some merchants specialized in the trade of certain commodities, others plied between certain centers of commerce, Rickman 1980, 142; Peacock 1982, 158-159. 111 Young 1981, 212-213; Greene 1992, 60; Peña 1999, 7. 112 Carandini 1970, 104, 116; Fulford 1980, 73; Peacock 1982, 158-159; Mattingly 1988a, 53; Greene 1992, 59; Panella 1993, 629; Mattingly & Hitchner 1995, 199-200; Reynolds 1995, 108. The infrastructure created for annona possibly lowered the market prices of foodstuffs and this was subsequently sought to offset in the producing provinces with pottery production, Fulford 1983, 11; Peña 1999, 6. 113 Panella 1986, 445; Olcese 1993, 160. 114 Carandini 1970, 104-105; Carandini 1983, 151-152. 115 Panella 1993, 643, cf. Tortorella 1986, 222. Also the contemporary peak in the amount of imported pottery at Carthage has been thought to result from the damage the conquerors caused to the agriculture of Africa Proconsularis, Fulford 1983, 10-12; Tortorella 1986, 214; Fulford 1987, 60; Clay 1992, 352. 116 Mattingly & Hitchner 1995, 200, 211. While the markets of Rome were momentarily lost due to the Vandal invasion and the demolition of annona, African pottery started to find its way to the eastern Mediterranean through interregional trade, Hayes 1980, 517; Fentress and Perkins 1988, 212-214. 117 Tortorella 1983, 19-21; Panella 1993, 641. 89

131

Chapter 5

118

Panella 1993, 651; Mattingly 1994, 215. Fulford 1980, 75. 120 Although annona as a system was beyond any economic reality, it provided stability to the Roman Empire, and at least enabled, if not even sustained economic growth that was an important precondition for the rise of commerce, Wickham 1988, 191-193; Mattingly & Salmon 2001b, 8-11. 121 Fentress & Perkins 1988, 213; Fontana 1991, 138-142, cf. Panella 1999, 209-210. 122 Fulford 1987, 63-64, 71. Rickman (1980, 71, 128) estimates that the journey of some 270 miles (432 km) in the direction of Carthage could have taken only two to three days (see PLIN. Nat. 19.3-4), while the trip in the opposite direction lasted from three days to a week, by using a safe route from Carthage - SE Sardinia - E Sardinia - Ostia; see also Pavolini 1986, 88. 123 Panella 1993, 614; Hopkins (1980, 120-124) has objected this idea by proposing that the general rise in insecurity of the mid-3rd century AD would have resulted in a decline in trade and successive decrease in the activity of artisans and merchants. This idea is contrasted, however, by African amphora stamps dated between the 1st and the 4th century AD, the number of which peaks in the 3rd century, Manacorda & Panella 1993, 57, see also Mattingly et al. 2001, 82. 124 Peacock 1982, 158. Interestingly, for Harris (1993c, 186) marble and olive oil are the two things that were rather redistributed than marketed. 125 Mattingly 1988a, 48, cf. Panella 1986, 443. 126 Mattingly 1988a, 48; Mattingly & Hitchner 1995, 199; Mattingly 1996, 224-225. Although, “Roman maritime commerce was related to economic trends rather than political events.” (Parker 1990, 154, 164-165; cf. Whittaker 1988, 68), the claim that commercial capitalism characterized by specialization, diversity of production modes, state-supported economics and demographic growth, was put into test for the first time in the Roman Empire (Carandini 1986, 15), is still somewhat exaggerated. 127 Mattingly 1988a, 34-35, 52, 56; Paterson 1998, 153-156; cf. Whittaker 1988, 52-53. 128 Fulford 1981, 202. 129 Pucci 1983, 105-107; Tomber 1993, 145. 130 Peacock 1982, 154; Greene 1992, 58, see also Fulford 1984c, 256-258; Pavolini 1985, 200. 131 Fulford 1987, 60-61, 66. Even the relatively tight correlation of African pottery with expenditure on public buildings has been proposed to reflect the general availability of capital, Fentress & Perkins 1988, 210-211. 132 Fulford 1983, 10-12. Therefore, the growing number of imported pottery in Roman Africa during the Late Roman period has been proposed to result from the declining local economy, although surplus would probably have been invested in other commodities than pottery during a boom. 133 Fulford 1987, 60-61. 134 Parker 1990, 163-165. 135 Fulford 1984c, 256; 1987, 70-71. The idea is supported by plentiful evidence. Firstly, while Aegean tablewares are virtually absent from western markets, the strong presence of Aegean amphorae is characteristic of the late Imperial period, Pavolini 1996, 230. There should also be plentiful archaeological evidence on table- and coarseware pottery produced in Roman Spain, which accompanied the Dressel 20 oil amphorae in their long journey from the Guadalquivir valley to the city of Rome, cf. Fulford 1987, 66. Secondly, the results of the Vinalopó valley ceramic survey indicate that African cookware was not customarily distributed with amphora-borne foodstuffs, Reynolds 1993, 13. Finally, if cookware exports merely indicate the trade in perishable commodities, we should possess evidence on the exportation of cookwares from Sardinia and Sicily: the supplementary granaries of Rome in the late Imperial period. 136 Duncan-Jones (1990, 31, 34-35), for example, claims that we lack information on the quantitative aspect of the trade, since pottery “is not a secure indicator of trading volume..., because it is abnormally resistant to destruction”, while to Hopkins (1980, 102 note 5) “archaeological evidence is uncorrectably biased by the survival of pots, which cannot have been so important in Roman economy.”, cf. Greene 1992, 58. It has also been pointed out that in the case of ceramic containers, transport amphorae in particular, the content and its package were not necessarily produced in the same area. But though the use of imported containers in bottling of local products together with other forms of recycling have been documented in the Roman Empire, the vast majority of amphorae were produced near the outlet of their contents, Duncan-Jones 1990, 34-35. 137 Peña 1999, 36. 138 Fulford 1980, 69; Peacock 1982, 158; Fulford 1986, 156-158; Greene 1986, 162; Fulford 1987, 60-61; Mattingly 1988a, 53; McGrail 1989, 357-358; Fontana 1991, 135; Gibbins 1991, 240; Tomber 1993, 143; Mattingly & Hitchner 1995, 200; Reynolds 1995, 116; Keay 1998: 141. 139 Parker 1984, 102-10 fig. 7; Parker 1992, 20-22; Reynolds 1995, 126. Items carried alongside fiscal loads were most likely redistributed to other ships only at Ostia (see Peña 1999, 8), because navicularii were bound to take the African grain there directly on pain of severe punishments (Rickman 1980, 202-203). Illustrative examples of compound cargoes are, e.g. the mid-3rd century AD wreck Cabrera III located near the island of Majorque that carried Spanish, Portuguese and Central Tunisian amphorae (Bost et al. 1992, 197-199) and the late 4th century AD wreck located off northwest Sicily that sunk with a compound cargo consisting of Tunisian, Aegean (or other eastern), Mauretanian and Calabrian amphorae (McCann 1994, 52-55). 140 Pucci 1983, 111-112; Pavolini 1985, 201; 1986, 92; Parker 1990, 28. 141 Fentress & Perkins 1988, 209; Mattingly 1994, cf. Clay 1992, 358. 142 In other words, pottery needs to occupy a considerable portion of the hold before the weight of cargo is significantly increased, McGrail 1989, 356, tab. 1. 143 Mattingly 1988a, 53; Tomber 1993, 143. 119

132

Distribution and consumption

144

This mixture of items of different provenance suggests that “we would probably do best to forget any idea that the goods were supplied in clean, unified cargoes direct from production centers.”, Millet 1993, 417-418. 145 Atkinson 1914, 28-29. 146 Peacock 1982, 154; Pucci 1983, 109. 147 Tomber 1993, 142. 148 Pucci 1983, 108. 149 Pucci 1983, 113; cf. Carandini 1983, 153 and Morel 1983, 67. 150 Peacock & Williams 1986, 64; Reynolds 1995, 106. The amount of African grain shipped to Rome each year, has been estimated as 27 million modii (1 modius = 6,82 kg), while the minimum capacity of a standard ship used t o transport annona in the late 2nd century AD was approximately 50.000 modii (Rickman 1980, 17). Although the tonnage of ships probably diminished in late antiquity (Panella 1999, 197), the transportation of the annual grain from Africa to Rome required at least 540 shiploads. The average ratio of three ships per day in the the six-month sailing season gives some idea of the economic pull created through this system. 151 Pucci 1983, 110. 152 Ostia III, 672; Carandini 1983, 146-147; Pucci 1983, 166; Carandini 1986, 7-9. Also the high quality of the early production is said to have contributed to the rapid success of African pottery to some extent, Carandini 1983, 147. 153 The application of Darwinian evolutionary approach to pottery studies has been promulgated by O'Brien et al. 1994. 154 E.g. Gill 1991, 42. The only literary evidence from the Roman period discussing the use and value of locally made coarse pottery is a group of Talmudic writings, see Adan-Bayewitz 1993. 155 Tortorella 1981b, 210; Reynolds 1995, 125-128; cf. Peacock 1982, 158-159; Fulford 1983, 6; 1984c, 256-258; 1989, 175-176. 156 Pucci 1983, 111; Fulford 1984c, 256-258. 157 Greene 1992, 55-56 158 Sagui 1980, 519; Greene 1986, 162. 159 Carandini 1975, 52-53. 160 Hence, demand may have already affected the production in antiquity, Morel 1983, 73. 161 Fulford 1983, 6; Fulford 1989, 175-176. 162 Panella 1993, 617. 163 Annis & Geertman 1987, 163, see also Peña 1992, 119. 164 Cf. Reynolds 1995, 130. 165 Panella 1996, 13. 166 Panella 1986, 446. 167 Tomber 1993, 143. 168 Reynolds 1995, 106, 129. 169 Fontana 1991, 135. 170 In the Late Roman period the word insula did not necessarily refer to a free-standing block of many dwelling units, but to an individual dwelling unit, see Purcell 1999, 151-152. 171 Salza Prina Ricotti 1980, 244, 281; Pavolini 1986, 181-182. 172 Scobie 1986, 410-411. 173 Dosi & Schnell 1992b, 85; Jansen 1997, 128-130. 174 Dosi & Schnell 1992a, 17; Foss 1997, 197-199. In addition, the social and the practical dimension, which kitchen might have had in the life of Roman family (see Varro 1.13.2), should not be forgotten. 175 Salza Prina Ricotti 1980, 242; Dosi & Schnell 1992b, 86-87. 176 Salza Prina Ricotti 1980, 239, 273, 280. 177 Scobie 1986, 401. Instead of being a homogeneous mass, the poor of Rome divided in to several classes, e.g. the good poor (the people) and the bad poor (the mob), Whittaker 1989, 8. On descriptions of the daily life in Rome, see MART. 7.61; PETR. 95; JUV. 3.190. 178 Scobie 1986, 427-428; Whittaker 1989, 10. 179 Pavolini 1986, 173. 180 Pavolini 1986, 181-182; Dosi & Schnell 1992b, 86-88. The recycling of metals has been offered as an explanation for the absence of these objects in the archaeological record, see Salza Prina Ricotti 1980, 281. 181 An argument possibly sustaining the use of braziers in the preparation of food is that many braziers in Pompeii have been discovered full of charcoal, while heating was certainly not needed in the August of AD 79 when Vesuvius erupted, see Salza Prina Ricotti 1980, 240-241, cf. Scobie 1986, 433. 182 Pavolini 1986, 181-182. 183 Dosi & Schnell 1992a, 43. Cooking may have been executed on hot coals, although some authors (e.g. MART. 2.90; 7.27) refer to the use of open flame. This may reflect a geographical difference, as the predominance of hot coals is associated with Campania, while open flame was used regularly in Latium. Salza Prina Ricotti (1980, 255, 274-275, 280-281). In large cooking-platforms, both modes may have been used simultaneously, Salza Prina Ricotti, 1980, 242. 184 Salza Prina Ricotti 1980, 270-273; Pavolini 1986, 181-182; Dosi & Schnell 1992a, 31-32; Dosi & Schnell 1992b, 40, 88. 185 Salza Prina Ricotti 1980, 270-273; Dosi & Schnell 1992b, 88 186 Cf. Panella 1986, 445-446. 187 Garnsey 1999, 32. Unfortunately, the distribution of African cookware in urban contexts of Rome and Ostia does not tell much of its status, as most examples of this pottery have been found in secondary fills of palaces, churches and baths. 188 Cf. Hayes 1972, 425.

133

6 SUMMARY

After the examination of some twenty-one hundred vessels from the Late Roman horizons of the Palatine East excavations (Rome), the following conclusions can be made about the cookware that was produced in the province of Africa Proconsularis from the early 1st century AD at least to the mid-/late 5th century AD. Firstly, the study has underlined the importance of versatility in the search of mechanisms behind the process that covers the life-span of a given ware from the selection of raw materials to its distribution and use. Generally speaking, attention should increasingly be paid in classical archaeology to development of archaeological ceramology outside of its traditional limits, where the research methods have been evolving recently in a fast pace. Despite of some difficulties at the outset, most of which were related to the encoding of variables – vessel fabric, soot deposition, firing etc. – the use of ceramic databases including detailed information on each individual vessel is highly recommended. By dividing the material into manageable units, in the present case chronological phases, the data can then be utilized to point out long-term developments from seemingly homogeneous material. The creation of substantial large units of reference is recommendable, as they tend to soft down the effect of probable contaminants – both residual and intrusive – included in the study assemblage. In spite of homogeneous raw materials used in production, African cookware is dividable at least into Central and North Tunisian fabrics, both of which comprise several fabric variants. The identification of variants was based on the compactness of clay matrix and the nature of inclusions. While both fabrics are rich in quartz sand, which is present both as a natural inclusion and temper, the compositional differences of the clay matrix as well as the overall non-plastic content have contributed to their dissimilar appearance. The abundance of calcite in dark red Central Tunisian fabric would have produced strong but overly rigid vessels, unless the effect was partially compensated by a firing in somewhat lower temperature compared to local tablewares. In yellowish-red North Tunisian fabric, the abundance of argillaceous rock fragments in some fabric variants did not produce problems of the same kind. Both fabrics were also shown to contain non-plastics of minor importance regarding the vessel performance, but which may act as important indicators in the characterization of the output of individual kiln sites. The typology of African cookware was studied by dividing the study assemblage into functional groups, forms and variants. This topic did not offer as

135

many possibilities for elaboration as was initially expected. The examination underlined the dual nature of the production. While the four principal forms – lids, pans, shallow and deep casseroles – were produced only by slightly retouching the vessel shape in the course of several centuries, the changing taste of the consumers was tested with the introduction of new forms and variants, most of which were only temporary. Hence, the combination of short- and longlived forms and variants may offer us some means to use African cookware to establish rough dates for archaeological deposits, at least for those lacking more readily datable material. In addition, the typology of some lid forms was observed to evolve simultaneously with certain pans and casseroles, which together with cumulative rim diameter charts were utilized to define functional sets. Although the study managed to pinpoint new sets as well as to confirm many defined previously, it also brought others under suspicion. The most obvious example of the last case is the widely propagated three-piece set consisting of North Tunisian lid, shallow and deep casserole. The matching number of African cookware lids with other vessels in study horizons hints that African cookware was most likely sold as sets. Flourishing agriculture, which has traditionally been interpreted as the most important factor contributing to the success of African pottery in the Late Roman period, was in fact significant to African cookware in several ways. The one-way traffic in grain and olive oil from Carthage to Rome did certainly boost the exportation of ceramics, but hardly generated pottery production on its own. The sedimentary environment of African Proconsularis ensured the availability of clay and tempering materials, while other than fresh water sources were most likely used for the production. Another substantial advantage, the importance of which has not been stressed sufficiently, is the favorable climate of North Africa that rendered possible pottery production throughout the year. In respect of the northern shores of the Mediterranean, the reduced fuel consumption of the households also left more room for crafts making use of advanced pyrotechnology. By specializing themselves on the exploitation of agricultural byproducts, of which the most important were the various residues of oleoculture, the potters of Roman Africa were able to ensure a continuous supply of combustible materials. The forming process of African cookware was based on the effective use of potter's wheel followed by complementary techniques of shaping, the most important of which was turning. The typology of

Chapter 6

principal vessel forms suggests that the forming process gradually developed to more cost-efficient direction, which is also reflected by the limited number of forms produced. As the production is compared to the ideal cookware vessel, which has been established recently through manifold archaeological experiments, both forms and surface finishes of African cookware are observed to correspond fairly often with the ideal. Only the angular transition between corrugated bottom and vertical upper wall is far from optimal regarding the thermal shock resistance, but as attention had to be paid also to the ease of manufacture and transportation, the production was obviously a continuous search for the balance between diverse requirements of production, distribution and use. In spite of fairly scarce archaeological evidence, it is proposed that the production of African cookware took chiefly place on kiln sites located either on the coast or riverside, as the ease of transportation guaranteed the access of the output to foreign markets. The manufacture of African cookware started to decline during the 4th century AD, possibly as a result of changes in agricultural production structure, which would explain better the development of this craft in Africa Proconsularis than a sudden stop caused by the Vandal conquest of Roman Africa in the second quarter of the 5th century AD. Thus, large production centers

136

making use of several kilns on the coast were increasingly replaced by smaller inland kiln sites, the primary function of which was to enrich the infrastructure of large agricultural production units. As the remote inland location could increase the costs of transportation up to ten times, supplementary pottery production was increasingly directed towards types of easy transportation, especially tablewares. The fact that the distribution of pottery was not solely dictated by the re-distribution of agricultural products is the lesson to be learned from the presence of African cookware on various coastal sites and shipwrecks scattered around the western Mediterranean. Its constant absence from inland sites, on the other hand, implies that the distribution abroad – just as the location of production sites in Roman Africa – was strongly affected by the mode of transportation. Still, quantities in which this production has been found in deposits dated between the 3rd and 5th century AD on several key sites is rather impressive. The appearance of locally made vessels with forms, and more occasionally also surface finishes, seeking to replicate African cookware is a further sign of the significant impact it had on foreign markets. As the production could certainly not compete with metal vessels, its success in the western Mediterranean was likely based on enhanced use-related properties in respect of local earthenware cooking utensils.

Bibliography

BIBLIOGRAPHY

Adams 2001 Adan-Bayewitz 1993 Adroer 1963 Aguarod Otal 1991 Aguarod Otal 1995 Aguarod Otal & Pérez Arantegui 1995 Alaimo et al. 1997

Albarella et al. 1993

Alonso de la Sierra Fernández 1995 Amouretti 1993 Amouretti & Brun 1993 Ampolo et al. 1971

Annecchino 1977

Annis & Geertman 1987 Anselmino et al. 1986 Aquilué Abadías 1985

Aquilué Abadías 1987

Aquilué Abadías 1989

Aquilué Abadías 1991

Jonathan Adams, Ships and boats as archaeological source material, in Gibbins & Adams 2001, 292-310. David Adan-Bayewitz, Common pottery in Roman Galilee. A study of local trade, Bar-Ilan studies in Near Eastern languages and culture, Jerusalem 1993. A.M. Adroer, La cerámica romana procedente de una necrópolis de Barcelona (Plaza Villa de Madrid), Rivista di Studi Liguri 29 (1963), 99-124. Carmen Aguarod Otal, Ceramica romana importada de cocina en la Tarraconense, Publicacion de la Institución Fernando el Católico 1.281, Zaragoza 1991. Carmen Aguarod Otal, La cerámica común de producción local/regional e importada. Estado de la cuestión en el Valle del Ebro, in Ibèrica 1995, 129-153. Carmen Aguarod Otal & J. Pérez Arantegui, Caracterización química de cerámica romana de cocina importada en la Tarraconense, in AA.VV., Estudis sobre cerámica antiga. Actes del simposi sobre cerámica antiga, Barcelona 1995, 145-146. R. Alaimo, G. Montana, R. Giarrusso, L. Di Franco, R.M. Bonacasa Carra, M. Denaro, O. Belvedere, A. Burgio & M.S. Rizzo, Le ceramiche comuni di Agrigento, Segesta e Termini Imerese: risultati archeometrici e problemi archeologici, in Santoro Bianchi & Fabbri 1997, 46-69. Umberto Albarella, Valeria Ceglia & Paul Roberts, S. Giacomo degli Schiavoni (Molise): an early 5th century AD deposit of pottery and animal bones from Central Adriatic Italy, Papers of the British School at Rome 61 (1993), 157-225. Juan Alonso de la Sierra Fernández, Cerámicas africanas de Córdoba, Anales de Arquelogia Cordobesa 6 (1995), 145-173. Marie-Claire Amouretti, Les sous-produits de la fabrication de l’huile et du vin, in Amouretti & Brun 1993, 463-476. Marie-Claire Amouretti & Jean-Pierre Brun (eds.), La production du vin et de l’huile en Méditerranée, Bulletin de correspondance hellénique. Supplément 26, Paris 1993. Carmine Ampolo, Andrea Carandini, Giuseppe Pucci & Patrizio Pensabene, La villa del casale a Piazza Armerina. Problemi, saggi stratigrafici e altre ricerche, Mèlanges de l'École Francaise a Rome 83 (1971), 141-281. Maria Annecchino, Suppellettile fittile da cucina di Pompei, Instrumentum domesticum di Ercolano e Pompei nella prima età Imperiale. Quaderni di cultura materiale 1 (1977), 105-114. Beatrice M. Annis & Herman Geertman, Production and distribution of cooking ware in Sardinia, in van As 1987, 154-196. Lucilla Anselmino, Caterina M. Coletti, Maria Luisa Ferrantini & Clementina Panella, Ostia. Terme del Nuotatore, in Giardina 1986, 45-82. Xavier Aquilué Abadías, Algunas concideraciones sobre el comercio africano. Tres facies características de la cerámica común africana de época alto-imperial, Empúries 47 (1985), 210-222. Xavier Aquilué Abadías, Las cerámicas africanas de la ciudad romana de Baetulo (Hispania tarraconensis), British Archaeological Reports: International Series 337, Oxford 1987. Xavier Aquilué Abadías, Les ceràmiques comunes de producció africana, in Un abocador del segle V d.C. en el Fòrum Provincial de Tàrraco, Taller escola d'arquelogia: Memòries d'excavació 2, Tarragona 1989, 190-204. Xavier Aquilué Abadías, Relaciones económicas, sociales e ideológicas entre el norte de África y la Tarraconense en época romana : las cerámicas de producción africana procedentes de la Colonia Iulia Urbs Triumphalis Tarraco, Ph.D. diss., University of Barcelona 1991.

137

Bibliography

Aquilué Abadías 1995 Aquilué Abadías 1997 Arnold 1979 Arnold 1985 Arnold & Nieves 1992

Arthur 1982 Arthur 1987 van As 1987

Atkinson 1914 Atlante I

Aubert 1993 Ballard et al. 2000

Barbieri 1989 Barbieri 1996

Bats 1988 Bats 1996

Baxter & Cool 1997 Beltran Lloris 1990 Ben Lazreg 2001 Ben Lazreg & Mattingly 1992 Berato et al. 1986

Bernal Casasola 1998

Biondani 1992

Xavier Aquilué Abadías, La cerámica común africana, in Ibèrica 1995, 61-72. Xavier Aquilué Abadías, Anàlisi comparativa de contextos ceràmics d'època tardoromana (segles V-VI), in Contextos 1997, 83-100. Dean E. Arnold, Comments, Current Anthropology 20 (1979), 735. Dean E. Arnold, Ceramic theory and cultural process, Cambridge 1985. Dean E. Arnold & Alvaro L. Nieves, Factors affecting ceramic standardization, in George J. Bey, III & Christopher A. Pool, (eds.), Ceramic production and distribution. An integrated approach, Boulder 1992, 93-113. Paul Arthur, Amphora production in the Tripolitanian Gebel, Libyan Studies 13 (1982), 61-72. Paul Arthur, Scavo in proprietà Carillo S.M.C.V.: Contributo per una conoscenza di Capua tardo-antica, Archaelogia Medievale 14 (1987), 517-535. Abraham van As (ed.), Knapsack full of pottery. Archaeo-ceramological miscellanea dedicated to H.J. Franken, Department of pottery technology newsletter 5, University of Leiden 1987. Donald Atkinson, A hoard of Samian ware from Pompeii, Journal of Roman Studies 7 (1914), 27-64. Andrea Carandini (ed.), Atlante delle forme ceramiche I: ceramica fine romana nel bacino Mediterraneo (medio e tardo impero), Enciclopedia dell'arte antica : classica e orientale, Roma 1981. Jean-Jaques Aubert, Workshop managers, in Harris 1993a, 171-181. R.D. Ballard, A.M. McCann, D. Yoerger, L. Whitcomb, D. Mindell, J. Oleson, H. Singh, B. Foley, J. Adams, D. Piechota & C. Giangrande, The discovery of ancient history in the deep sea using advanced deep submerge technology, Deep-Sea Research I 47 (2000), 1591-1620. Gabriela Barbieri, Ceramica romana da un insediamento rustico nei pressi di Viterbo, Rivista di Studi Liguri 55 (1989), 79-121. Gabriela Barbieri, Viterbo. Località Paliano. – Saggi di scavo presso le terme romane dette “Le Masse di San Sisto”, Notizie degli scavi di Antichità serie IX volume III-IV 1992-1993 (1996), 5-66. Michel Bats, Vaisselle et alimentation à Olbia de Provence, Revue archéologique de Narbonnaise. Supplément 18, Paris 1988. Michel Bats (ed.), Les céramiques communes de campanie et de narbonnaise (Ier s. av. J.-C. - IIe s. ap. J.-C.). La vaisselle de cuisine et de table, Collection du Centre Jean Bérard, Naples 1996. M.J. Baxter & E.M. Cool, Notes on some statistical aspects of pottery quantification, Medieval Ceramics 19 (1995), 89-98. Miguel Beltran Lloris, Guía de la cerámica romana, Zaragoza 1990. Nejib Ben Lazreg, A further note on the fuels used in this region, in Smith W . 2001, 436. Nejib Ben Lazreg & David J. Mattingly (eds.), Leptiminus (Lamta): a Roman port city in Tunisia, Report no. 1, Journal of Roman Archaeology Suppl. 4, Ann Arbor 1992. Jacques Berato, Marc Borreani, Pascal Lecacheur, Michel Pasqualini, Jean-Michel Theveny & Jacqueline and Yves Rigoir, Fouilles récentes à Toulon (Var). Quartier Besagne, 1985-1986, Documents d'Archéologie Méridionale 9 (1986), 135-166. Darío Bernal Casasola, La producción de cerámicas africanas de cocina en la Bética: aportaciones del taller de los Matagallares y de la alfarería de los Barreros (Salobreña, Granada), L’Africa Romana XII (1998), 1341-1353. Federico Biondani, Importazioni di ceramica africana e orientale a Suasa e nell’area medio e altoadriatica, Rivista di Studi Liguri 58 (1992), 147-178.

138

Bibliography

Blackman 1992

Blackman 1993

Bon et al. 1997

Bon & Jones 1997 Bost et al. 1992

Bourgeois & Mayet 1991 Bradley & Fulford 1984 Braun 1983 Braun 1995 Bronitsky 1987 Bronitsky & Hamer 1986

Brown & Mattingly 2001 Bullard 1985 Burollet 1967

Burragato et al. 1994a Burragato et al. 1994b

Butterworth 1956 Carandini 1970 Carandini 1975 Carandini 1977

Carandini 1983

James M. Blackman, The effect of human size sorting on the mineralogy and chemistry of ceramic clays, in Hector Neff (ed.), Chemical characterization of ceramic pastes in archaeology, Monographs in World Archaeology 7, Madison 1992, 113-124. James M. Blackman, Gil J. Stein & Pamela B. Vandiver, The standardization hypothesis and ceramic mass production: technological, compositional, and metric indexes of craft specialization at Tell Leilan (Syria), American Antiquity 58 (1993), 60-80. Sara E. Bon, Rick Jones, Bernice Kurchin & Damian J. Robinson, The context of the House of the Surgeon: Investigations in Insula VI,1 at Pompeii, in Bon & Jones 1997, 32-49. Sara E. Bon & Rick Jones (eds.), Sequence and space in Pompeii, Oxbow monograph 77, Oxford 1997. Jean Pierre Bost, Marta Campo, Dali Colls, Victor Guerrero & Françoise Mayet, L'épave Cabrera III (Majorque), Échanges commerciaux et circuits monétaires au milieu du IIIe siécle après Jésus-Christ, Publications du centre Pierre Paris 23, Paris 1992. Alain Bourgeois & F. Mayet, Belo VI. Les sigillées, Madrid 1991. Richard Bradley & Michael G. Fulford, Sherd size and dating of later part of sequence A, in Fulford & Peacock 1984, 43-47. David P. Braun, Pots as tools, in Arthur S. Keene & James A. Moore (eds.), Archaeological hammers and theories, New York 1983, 107-134. Thomas Braun, Barley cakes and emmer bread, in Wilkins et al. 1995, 25-37. Gordon Bronitsky, A computer-based system for measurement of thermal properties of archaeological ceramics, in van As 1987, 87-92. Gordon Bronitsky & Robert Hamer, Experiments in ceramic technology: the effects of various tempering materials on impact and thermal-shock resistance, American Antiquity 51 (1986), 89-101. Antony G. Brown & David J. Mattingly, A buried clay-filled pit linked to pottery production at Leptiminus (site 176), in Striling et al. 2001, 387-396. Reuben G. Bullard, Sedimentary environments and lithologic materials at two archaeological sites, in Rapp & Gifford 1985, 103-133. P.F. Burollet, General geology of Tunisia, in Lewis Martin (ed.), Guidebook to the geology and history of Tunisia, Petroleum exploration society of Libya. 9th annual field conference, Amsterdam 1967, 51-58. F. Burragato, O. Grubessi & Lorenzo Lazzarini (eds.), 1st European workshop on archaeological ceramics, Rome 1994. F. Burragato, Lorenzo Lazzarini & P. Di Russo, Caratterizzazione chimico-fisica delle anfore nord-africane del Monte Testaccio: nota preliminare, in Burragato et al. 1994a, 143-154. B. Butterworth, Lime blowing: some notes on the literature, Transactions of the British ceramic society 55 (1956), 532-544. Andrea Carandini, Produzione agricola e produzione ceramica nell'Africa di età imperiale, Studi Miscellanei 15 (1970), 97-119. Andrea Carandini, Storia di una forma ceramica africana. Un esempio di “selezione artigianale”, Studi Miscellanei 22 (1975), 47-56. Andrea Carandini, La sigillata africana, la ceramica a patina cenerognola e a orlo annerito, in Instrumentum domesticum di Ercolano e Pompei nella prima età Imperiale. Quaderni di cultura materiale 1 (1977), 23-24. Andrea Carandini, Pottery and the African economy, in Garnsey et al. 1983, 145162.

139

Bibliography

Carandini 1986 Carignani et al. 1986

Carillo Díaz-Pinés & Murillo Redondo 1994 Casas & Nolla 1993

Castañer Masoliver et al. 1990 Cau Ontiveros et al. 1997

Cau Ontiveros et al. 2002

Ciarrocchi et al. 1998

Clark & Schofield 1991

Clay 1992

Cogswell et al. 1996

Coletti 1998 Coletti & Pavolini 1996 Contextos 1997

Courty & Roux 1995

Crummy & Terry 1979 Cubberley 1995 Cubberley et al. 1988

Andrea Carandini, Il mondo della tarda antichitá visto attraverso le merci, in Giardina 1986, 1-19. Andrea Carignani, Alberto Ciotola, Francesco Pacetti & Clementina Panella, Roma. Il contesto del tempio della Magna Mater sul Palatino, in Giardina 1986, 27-43. José Ramon Carillo Díaz-Pinés & Juan Francisco Murillo Redondo, Un vertedero con céramica africana de cocina en Colonia Patricia, L'Africa Romana XI (1994), 1301-1319. Josep Casas & Josep Maria Nolla, Un conjunt tancat amb ceràmica africana a la villa romana dels Tolegassos (Viladamat; Alt Empordà), Empuries 48-50 (19861989, 1993), 202-213. Pere Castañer Masoliver, Joaquim Tremoleda Trilla & Assumpta Roure Bonaventura, Un conjunt ceràmic de finals del segle III d.C. a Vilauba (Camós, Pla de l'Estany), Cypsela VIII, Girona 1990, 157-191. Miguel Ángel Cau Ontiveros, Josep Giralt, Josep Maria Macias Solé, José Ignacio Padilla Lapuente & Francesc Tuset I Bertran, La céramica del nordeste peninsular y las Baleares entre los siglos V-X, in Gabrielle Démians d'Archimbaud (ed.), La céramique médiévale en Méditerranée. Actes du 6 e congrès de l'AIECM2, Aix-enProvence 13-18 novembre 1995, Aix-en-Provence 1997, 173-192. Miguel Ángel Cau Ontiveros, Peter M. Day & G. Montana, Secondary calcite in archaeological ceramics: evaluation of alteration and contamination processes by thin section study, in Kilikoglou et al. 2002, 9-18. Barbara Ciarrocchi, Catarina M. Coletti, Archer Martin, Lidia Paroli & Carlo Pavolini, Ceramica comune tardoantica da Ostia e Porto (V-VII secolo), in Sagui 1998, 383-420. R.H. Clark & A.J. Schofield, By experiment and calibration: an integrated approach to archaeology of the ploughsoil, in A.J. Schofield (ed.), Interpreting artefact scatters. Contributions to ploughzone archaeology, Oxbow Monograph 4 (1991), 93-105. Teresa Clay, Carthage et son commerce maritime dans l'antiquité tardive, Actes du Ve Colloque internationale d'historie et archéologie de l'Afrique du Nord, 1992, 349360. James W. Cogswell, Hector Neff & Michael D. Glascock, The effect of firing temperature on the elemental characterization of pottery, Journal of Archaeological Science 23 (1996), 283-287. Caterina M. Coletti, Ceramiche comuni dai contesti di Porto (VI e VII secolo), in Ciarrochi et al. 1998, 410-417. Caterina M. Coletti & Carlo Pavolini, Ceramica comune da Ostia, in Bats 1996, 391-419. Monserrat Comes i Solá, Josep M. Gurt i Esparraguera, Alberto López Mullor, Pepita Padrós i Martí & Mercè Roca i Roumens (eds.), Contextos ceràmics d'època romana tardana i de l'alta edat mitjana (segles IV-X), Arqueo Mediterrània 2/1997, Barcelona 1997. M.A. Courty & V. Roux, Identification of wheel throwing on the basis of ceramic surface features and microfabrics, Journal of Archaeological Science 22 (1995), 1750. Philip Crummy & Roger Terry, Seriation problems in urban archaeology, in Millet 1979, 49-59. Anthony Cubberley, Bread-baking in ancient Italy. Clibanus and sub testu in the Roman world: further thoughts, in Wilkins et al. 1995, 55-68. Anthony L. Cubberley, J. Lloyd & Paul Roberts, Testa and clibani: The baking covers of classical Italy, Papers of the British School at Rome 43 (1988), 98-119.

140

Bibliography

Cumberpatch 2001 Cuomo di Caprio 1972 Cuomo di Caprio 1995 Cuomo di Caprio & Vaughan 1993 Curtis 1962 DeLaine 2001 De Rossi & Mandarini 1998

Dodinet & Leblanc 1988

Dore 1988

Dore 1989

Dore 1992 Dore 2001a Dore 2001b Dore & Greene 1977 Dosi & Schnell 1992a Dosi & Schnell 1992b Dosi & Schnell 1992c Duncan-Jones 1974 Duncan-Jones 1990 Dyson 1976 Ermeti 1998 Fabbri et al. 1997 Falbe 1833 Feathers 1989

C. G. Cumberpatch, Comment on ‘Technological choices in ceramic production’, Archaeometry 43 (2001), 269-271. Ninina Cuomo di Caprio, Proposta di classificazione delle fornaci per ceramica e laterizi in area italiana, Sibrium 11 (1972), 371-375. Ninina Cuomo di Caprio, Osservazioni tecniche sul tornio da vasaio nell'antichità, Rivista di Archeologia 19 (1995), 145–152. Ninina Cuomo di Caprio & Sarah J. Vaughan, An experimental study in distinguishing grog (chamotte) from argillaceous inclusions in ceramic thin sections, Archaeomaterials 7 (1993), 21-40. F. Curtis, The utility pottery industry of Bailén, Southern Spain, American Anthropologist 64 (1962), 486-503. Janet DeLaine, Bricks and mortar. Exploring the economics of building techniques at Rome and Ostia, in Mattingly & Salmon 2001a, 230-268. Gianfranco De Rossi & Lucio Valerio Mandarini, Residui ceramici d’età romana da un contesto farnesiano nell’area sud ovest del Palatino, in Guidobaldi et al. 1998, 45-62. Marianne Dodinet & Jacques Leblanc, La production de céramiques gallo-romaines à “bords noircis” et à “patine cendrée” dans le Biterrois, Documents d'Archéologie Méridionale 11 (1988), 135-143. John N. Dore, Pottery and the history of Roman Tripolitania: evidence from Sabratha and the UNESCO Libyan valleys survey, Libyan Studies 19 (1988), 6185. John N. Dore, The coarse pottery, in John N. Dore & Nina Keay (eds.), Excavations at Sabratha 1948-1951, vol. II. The finds. Part. 1: The amphorae, coarse pottery and building materials, London 1989, 87-165. John N. Dore, First report on the pottery, in Ben Lazreg & Mattingly 1992, 115156. John N. Dore, The major pottery deposits following the disuse of the East Baths, in Striling et al. 2001, 74-98. John N. Dore, Selected finds, Site 10, 1991, in Striling et al. 2001, 202-214. John N. Dore & Kevin Greene (eds.), Roman Pottery Studies in Britain and beyond, British Archaeological Reports: British Series 30, Oxford 1977. A. Dosi & F. Schnell, Le abitudini alimentari dei Romani, Museo della civiltà Romana: Vita e costumi dei romani antichi 1, 2nd ed., Roma 1992. A. Dosi & F. Schnell, Pasti e vasellame da tavola, Museo della civiltà Romana: Vita e costumi dei romani antichi 2, 2nd ed., Roma 1992. A. Dosi & F. Schnell, I Romani in cucina, Museo della civiltà Romana: Vita e costumi dei romani antichi 3, 2nd ed., Roma 1992. Richard Duncan-Jones, The economy of the Roman Empire: quantitative studies, Cambridge 1974. Structure and scale in the Roman economy, Cambridge 1990. Stephen L. Dyson, Cosa: The utilitarian pottery, Memoirs of the American Academy at Rome 33 (1976). Anna L. Ermeti, Pesaro: ceramica tra VI e VII secolo da sterri urbani, in Sagui 1998, 611-616. B. Fabbri, S. Gualtieri & S. Santoro, L’alternativa chamotte/calcite nella ceramica grezza: prove tecniche, in Santoro Bianchi & Fabbri 1997, 183-190. Christian T. Falbe, Recherches sur l'emplacement de Carthage, Paris 1833. James K. Feathers, Effects of temper on strength of ceramics: response to Bronitsky and Hamer, American Antiquity 54 (1989), 579-588.

141

Bibliography

Fentress 1986

Fentress et al. 1983

Fentress & Perkins 1988 Fentress et al. 1991

Fernández Sotelo 1994 Fiches 1996 Fontana 1991

Fontana 1998 Foss 1997

Frayn 1978 Freed 1983 Freed 1994 Freed 1995 Freed 1998

Freestone 1982

Freestone 1995 Freestone & Middleton 1987 Freestone et al. 1985

Fulford 1980 Fulford 1981 Fulford 1983 Fulford 1984a

Elizabeth Fentress, The pottery, in Nacéra Benseddik & Timothy Potter, Fouilles du Forum de Cherchel. Rapport preliminaire, Bulletin d'Archeologie algerienne suppl. 4°, Alger 1986, 25-37. Elizabeth Fentress, Sheldon Judson, Thomas Blagg, Mariette de Vos & Paul Arthur, Excavations at Fosso della Crescenza 1962, Papers of the British School at Rome 51 (1983), 58-101. Elizabeth Fentress & Philip Perkins, Counting African Red Slip Ware, L'Africa Romana V (1988), 205-214. Elizabeth Fentress, Teresa Clay, Michelle Hobart & Matilda Webb, Late Roman and Medieval Cosa: The arx and the structure near the eastern height, Papers of the British School at Rome 46 (1991), 197-230. Emilio A. Fernández Sotelo, Un vertedero del siglo III en Ceuta, Transfertana. Monografias 1 (1994), Ceuta. Jean-Luc Fiches, Céramiques culinaires et vaisselle commune de table dans la région de Nîmes in, Bats 1996, 351-359. Sergio Fontana, Note sulla distribuzione della sigillata africana in Italia, in Edward Herring, Ruth Whitehouse & John Wilkins (eds.), Papers of the fourth confrence of Italian archaeology 2: The archaeology of power II 1991, 135-145. Sergio Fontana, Le “imitazioni” della sigillata africana e le ceramiche da mensa italiche tardo-antiche, in Sagui 1998, 83-100. Pedar W. Foss, Watchful Lares: Roman household organization and the rituals of cooking and eating, in Ray Laurence & Andrew Wallace-Hadrill (eds.), Domestic space in the Roman world: Pompeii and beyond, Journal of Roman Archaeology Suppl. 22 (1997), 197-218. Joan M. Frayn, Home-baking in Roman Italy, Antiquity 52 (1978), 28-33. Joann Freed, Pottery from the late middens at S. Giovanni, in AA.VV., Lo scavo di S. Giovanni di Ruoti ed il periodo tardo-antico in Basilicata, Roma 1983, 91-106. Joann Freed, The pottery from the late-Roman shipwreck, in McCann & Freed 1994, 21-48. Joann Freed, The late series of Tunisian cylindrical amphoras at Carthage, Journal of Roman Archaeology 8 (1995), 155-191. Joann Freed, Pottery report, in Colin Wells, Maureen Carroll, Joann Freed & David Godden, The construction of decumanus VI N and the economy of the early colony of Carthage, in Humphrey & Stone 1998, 18-63. Ian C. Freestone, Applications and potential of electron probe micro-analysis in technological and provenance investigations of ancient ceramics, Archaeometry 24 (1982), 99-116. Ian C. Freestone, Ceramic petrography, American Journal of Archaeology 99 (1995), 111-115. Ian C. Freestone & Andrew P. Middleton, Mineralogical applications of the analytical SEM in archaeology, Mineralogical Magazine 51 (1987), 21-31. Ian C. Freestone, N.D. Meeks & Andrew P. Middleton, Retention of phosphate in buried ceramics: an electron microbeam approach, Archaeometry 27 (1985), 161177. Michael G. Fulford, Carthage: overseas trade and the political economy c. AD 400700, Reading Medieval Studies 6 (1980), 68-80. Michael G. Fulford, Roman pottery: Towards the investigation of economic and social change, in Howard & Morris 1981, 195-208. Michael G. Fulford, Pottery and the economy of Carthage and its hinterland, Opus 2 (1983), 5-14. Michael G. Fulford, Chronology: A discussion of the pottery groups and their dating, in Fulford and Peacock 1984, 29-42.

142

Bibliography

Fulford 1984b Fulford 1984c Fulford 1986

Fulford 1987 Fulford 1989 Fulford 1994 Fulford & Peacock 1984

Fulford & Peacock 1994 Gabriel 1985 Gandolfi 1994 Garnsey 1983 Garnsey 1988 Garnsey 1999 Garnsey & Saller 1987 Garnsey et al. 1983 Gauckler 1897 Gazaille & Beaulieu 1981 Gianfrotta & Pomey 1980 Giannotta 1992

Giardina 1986 Gibbins 1990 Gibbins 1991

Gibbins 2001

Michael G. Fulford, The coarse (kitchen and domestic) and painted wares, in Fulford and Peacock 1984, 155-231. Michael G. Fulford, The long distance trade and communications of Carthage c. A.D. 400 to c. A.D. 650, in Fulford and Peacock 1984, 255-262. Michael G. Fulford, Pottery and precious metals in Roman world, in Michael Vickers (ed.), Pots & pans. A colloquium on precious metals and ceramics in the Muslim, Chinese and Graeco-Roman worlds. Oxford 1985, Oxford studies in Islamic art 3, Oxford 1986, 153-160. Michael G. Fulford, Economic interdependence among urban communities of the Roman Mediterranean, World Archaeology 19 (1987), 58-75. Michael G. Fulford, To East and West: The Mediterranean trade of Cyrenaica and Tripolitania, Libyan Studies 20 (1989), 169-191. Michael G. Fulford, The cooking and domestic wares, in Fulford & Peacock 1994, 53-77. Michael G. Fulford & David P.S. Peacock (eds.), The Avenue du President Habib Bourguiba, Salammbo: The pottery and other ceramic objects from the site, Excavations at Carthage: The British Mission I, 2, Sheffield 1984. Michael G. Fulford & David P.S. Peacock (eds.), The Circular harbour, north side: The pottery, Excavations at Carthage: The British Mission II, 2, Oxford 1994. Barbara L. Gabriel, SEM: A user's manual for materials science, Ohio 1985. Daniela Gandolfi, La produzione ceramica africana di età medio e tardo imperiale: terra sigillata chiara e ceramica da cucina, in Lusuardi Siena 1994, 127-156. Peter Garnsey, Grain for Rome, in Garnsey et al. 1983, 118-130 Peter Garnsey, Famine and food supply in the Graeco-Roman world. Responses to risks and crises, Cambridge University Press, Worcester 1988. Peter Garnsey, Food and society in classical antiquity, Cambridge University Press 1999. Peter Garnsey & Richard Saller, The Roman empire. Economy, society and culture, Great Britain 1987. Peter Garnsey, Keith Hopkins & Charles R. Whittaker (eds.), Trade in the ancient economy, London 1983. Philippe Gauckler, Rapport épigraphique sur les découvertes faites en Tunisie, Bulletin Archéologique du Comité des Travaux Historiques, 1897. Madeleine Gazaille & Ferdinand Beaulieu, Description partielle de la céramique 1979, in Senay 1981, 73-88. Piero A. Gianfrotta & P. Pomey, Archaeologia subacquea: storia, technice, scoperte e relitti, Milano 1980. Maria Teresa Giannotta, La ceramica africana e microasiatica, in Francesco D'Andria & David Whitehouse, Excavations at Otranto vol. 2. The finds, Lecce: Congedo 1992, 47-59. Andrea Giardina (ed.), Societa romana e impero tardoantico III: Le merci, gli insediamenti, Roma 1986. David J.L. Gibbins, Analytical approaches in maritime archaeology: A Mediterranean perspective, Antiquity 64 (1990), 376-389. David J.L. Gibbins, The Roman wreck of. c. 200 AD at Plemmirio, near Siracusa (Sicily): Third interim report. The domestic assemblage 2: Kitchen and table pottery, glass and fishing weights, International Journal of Nautical Archaeology 20.3 (1991), 227-246. David J.L. Gibbins, A Roman shipwreck of. c. AD 200 at Plemmirio, Sicily: evidence for north African amphora production during the Severan period, in Gibbins & Adams 2001, 311-334.

143

Bibliography

Gibbins & Adams 2001 Gibbins & Parker 1986

Gibson & Woods 1997 Gill 1991 González Villaescusa 1990 González Villaescusa 1993

Greene 1986 Greene 1992 Guidobaldi et al. 1998

Harris 1993a

Harris 1993b Harris 1993c Harris 1999 Hayes 1972 Hayes 1976 Hayes 1978 Hayes 1980 Hayes 1983 Hayes 1994

Hayes 1997 Hayes 1998 Heimann 1989 Heinrich 1965 Henrickson 1991

Hitchner 1993

David J.L. Gibbins & Jonathan Adams (eds.), Shipwrecks, World Archaeology 32/3 (2001). David J.L. Gibbins & Anthony J. Parker, The Roman wreck of c. AD 200 at Plemmirio, near Siracusa (Sicily): Interim report, International Journal of Nautical Archaeology 15.4 (1986), 267-304. Alex Gibson & Ann Woods, Prehistoric pottery for the archaeologist (2nd ed.), Guildford 1997. David W.J. Gill, Pots and trade: spacefillers or objects d'art, Journal of Hellenic Studies 111 (1991), 29-47. Ricardo González Villaescusa, El vertedero de la Avenida de España, 3 y el siglo III d. de C. en Ebusus, Trabajos del Museo Arqueologico de Ibiza 22, Eivissa 1990. Ricardo González Villaescusa, Origine et diffusion d’une forme peu courante de céramique africaine. Nouvelle contribution à la connaissance du commerce de Byzacène au IIIes. après J.-C., Antiquités Africaines 29 (1993), 151-161. Kevin Greene, The archaeology of the Roman economy, Berkeley 1986. Kevin Greene, Roman pottery, Interpreting the past 6, British museum press: London 1992. Federico Guidobaldi, Carlo Pavolini & Philippe Pergola (eds.), I materiali residui nello scavo archeologico : testi preliminari e atti della Tavola rotonda organizzata dall'École française de Rome ... : Roma, 16 marzo 1996, Collection de l'École Française de Rome 249 (1998). William V. Harris (ed.), The inscribed economy. Production and distribution in the Roman empire in the light of instrumentum domesticum, Journal of Roman Archaeology Suppl. 6, Ann Arbor 1993. William V. Harris, Between archaic and modern: some current problems in the history of the Roman economy, in Harris 1993a, 11-29. William V. Harris, Production, distribution and instrumentum domesticum, in Harris 1993a, 186-189. William V. Harris (ed.), The transformations of Urbs Roma in Late Antiquity, Journal of Roman Archaeology Suppl. 33, Ann Arbor 1999. John W. Hayes, Late Roman pottery, London 1972. John W. Hayes, Pottery: Stratified groups and typology, in Humphrey 1976, 47123. John W. Hayes, Pottery Report - 1976, in Humphrey 1978, 23-98. John W. Hayes, A supplement to Late Roman Pottery, London 1980. John W. Hayes, The Villa Dionysos excavations, Knossos: The pottery, Papers of the British School at Athens 78 (1983), 96-169. John W. Hayes, Eastern terra sigillata wares – Late Roman Red-slip wares and their competitors: filling in the Mediterranean picture, in Tony Hackens (ed.), Ancient and traditional ceramics, PACT 40 (1994), 107-116. John W. Hayes, Handbook of Mediterranean Roman pottery, Bath 1997. John W. Hayes, The study of Roman pottery in the Mediterranean: 23 years after Late Roman Pottery, in Sagui 1998, 9-21. Robert B. Heimann, Assessing the technology of ancient pottery: the use of ceramic phase diagrams, Archaeomaterials 3 (1989), 123-148. E. Heinrich, Microscopic identification of minerals, New York 1965. Robert C. Henrickson, Wheelmade or wheel-finished? Interpretation of “wheelmarks” on pottery, in Pamela B. Vandiver, James Druzik & George S. Wheeler (eds.), Materials issues in art and archaeology II, Materials research society symposium proceedings 185, Pittsburgh 1991, 523-541. Robert B. Hitchner, Olive production and the Roman economy: the case for intensive growth in the Roman Empire, in Amouretti & Brun 1993, 499-508.

144

Bibliography

Hoard et al. 1995

Hopkins 1980 Hopper 1986 Hostetter et al. 1990 Hostetter et al. 1991

Hostetter et al. 1994

Hostetter & Brandt 1999

Howard & Morris 1981 Humphrey 1976 Humphrey 1978 Humphrey 1981 Humphrey & Stone 1998 Ibèrica 1995 Jansen 1997 Jones 1994 Keay 1998 Keller et al. 1978

Kilikoglou et al. 1995

Kilikoglou et al. 1998

Kilikoglou et al. 2002

Kingery 2001

Robert J. Hoard, Michael J. O'Brien, Mohammad G. Khorasgany & Vellore S. Gopalaratnam, A materials-science approach to understanding limestone-tempered pottery from the midwestern United States, Journal of Archaeological Science 22 (1995), 823-832. Keith Hopkins, Taxes and trade in the Roman empire, Journal of Roman Studies 70 (1980), 101-125. Robin Hopper, Functional pottery. Form and aesthetic in pots of purpose, Radnor: Chilton 1986. Eric Hostetter, Thomas N. Howe & J. Rasmus Brandt, Excavation of the Late Roman buildings, Bolletino di Archeologia 3 (1990), 89-91. Eric Hostetter, Thomas N. Howe, J. Rasmus Brandt & Archer St.Clair, Palatino. Versante nordorientale; complesso tardoromano, Bolletino di Archeologia 9 (1991), 47-55. Eric Hostetter, Thomas N. Howe, J. Rasmus Brandt, Archer St.Clair & Maryline Parca, A Late Roman domus on the north-eastern slope of the Palatine-hill, in Rome Papers: The baths of Trajan Decius, Iside e Serapide nel Palazzo, a late domus on the Palatine, and Nero's Golden House, Journal of Roman Archaeology Suppl. 11 (1994), 131-182. Eric Hostetter & J. Rasmus Brandt, Excavations on the NE slope of the Palatine Hill (1989-1994): a summary, Societas archaeologiae classicae Fennica. Newsletter 2/99 (1999), 8-11. Hilary Howard & Elaine L. Morris (eds.), Production and distribution: A ceramic viewpoint, British Archaeological Reports: International Series 120, Oxford 1981. John H. Humphrey (ed.), Excavations at Carthage conducted by the University of Michigan vol. 1, Ann Arbor 1976. John H. Humphrey (ed.), Excavations at Carthage conducted by the University of Michigan vol. 4, Ann Arbor 1978. John H. Humphrey (ed.), Excavations at Carthage conducted by the University of Michigan vol. 6, Ann Arbor 1981. John H. Humphrey & David Stone (eds.), Carthage papers. Journal of Roman Archaeology Suppl. 28 (1998). AA.VV., Ceramica comuna romana d'època Alto-Imperial a la Península Ibèrica. Estat de la qüestió, Monografies Emporitanes VIII, Empuries 1995. Gemma Jansen, Private toilets at Pompeii: appearance and operation, in Bon & Jones 1997, 121-134. Richard E. Jones, Current trends and issues in Mediterranean ceramic studies, in Burragato et al. 1994a, 13-21. Simon Keay, African amphorae, in Saqui 1998, 141-155. J. Keller, W.B.F. Ryan, D. Ninkovich & R. Altherr, Explosive volcanic activity in the Mediterranean over the past 200.000 yr as recorded in deep-sea sediments, Geological Society of America Bulletin 89 (1978), 591-604. Vassilis Kilikoglou, G. Vekinis & Y. Maniatis, Toughening of ceramic earthenwares by quartz inclusions: an ancient art revisited, Acta Metallurgica et Materialia 43 (1995), 2959-2965. Vassilis Kilikoglou, G. Vekinis, Y. Maniatis & P.M. Day, Mechanical performance of quartz-tempered ceramics: Part I, strength and toughness, Archaeometry 40 (1998), 261-279. Vassilis Kilikoglou, A. Hein & Y. Maniatis (eds.), Modern trends in scientific studies on ancient ceramics. Papers presented at the 5th European meeting on ancient ceramics, Athens 1999, British Archaeological Reports: International Series 1011, Oxford 2002. David W. Kingery, Comment on ‘Technological choices in ceramic production’, Archaeometry 43 (2001), 273-277.

145

Bibliography

Klein & Hurlbut 1993 Koukios 1999

Laird & Worcester 1956 Lamboglia 1941 Lamboglia 1950 Lamboglia 1958 Lamboglia 1963 Laurence 1998

Lavazza & Vitali 1994

Lavoie 1989 Leveau 1984 López Mullor et al. 1997 Lund 1995

Luni II Lusuardi Siena 1994 Macias et al. 1997

Mackensen 1993

Mackensen 1998a Mackensen 1998b Mackensen & Schneider 2002 Maggetti 1994

Cornelis Klein & Cornelius S. Hurlbut Jr., Manual of mineralogy (21st ed.), New York 1993. Emmanuel G. Koukios, Standardization of solid biofuels in Greece. Summary and recommendations of the country report, 31.1.2002 http://www.biomasseinfo.net/downloads/GR.pdf. R.T Laird & M. Worcester, The inhibiting of lime blowing, Transactions of the British ceramic society 55 (1956), 545-563. Nino Lamboglia, Terra sigillata chiara, Rivista di Studi Liguri 7 (1941), 7-22. Nino Lamboglia, Gli scavi di Albintimilium e la cronologia della ceramica romana. Campagne di scavo 1938-1940, Bordighera 1950. Nino Lamboglia, Nuove osservazioni sulla “terra sigillata chiara” (Tipi A e B), Rivista di Studi Liguri 24 (1958), 257-330. Nino Lamboglia, Nuove osservazioni sulla terra sigillata chiara. II. La sigillata chiara C, Rivista di Studi Liguri 29 (1963), 145-212. Ray Laurence, Land transport in Roman Italy: costs, practice and the economy, in Helen Parkins & Christopher Smith (eds.), Trade, traders and the ancient city, Routledge: Ebbw Vale, 129-148. Antonella Lavazza & Maria Grazia Vitali, La ceramica d'uso comune: problemi generali e note su alcune produzioni tardoantiche e medievali, in Lusuardi Siena 1994, 17-54. Carl Lavoie, Les traces d'ateliers de poterie a Sullecthum, in Sullecthum I, Cahiers des études anciennes 22 (1989), 91-109. Philippe Leveau, Caesarea de Maurétanie: une ville romaine et ses campagnes, Collection de l'École Française de Rome 70, Rome 1984. Albert López Mullor, Xavier Fierro Macia & Àlvar Caixal Mata, Ceramica dels segles IV al X procedent de les comarques de Barcelona, in Contextos 1997, 59-82. John Lund, Hellenistic, Roman and Late Roman finewares from the Segermes valley – forms and chronology, in Søren Dietz, Laila Ladjimi Sebaï & Habib Ben Hassen (eds.), Africa Proconsularis: Regional studies in the Segermes valley of Northern Tunisia, vol. 2: Pottery, numismatics and the antiquarian data, Publications of the Collection of Near Eastern and Classical Antiquities – The National Museum of Denmark, 447-629. Antonio Frova (ed.), Scavi di Luni II. Relazione delle campagne di scavo 19721973-1974, Roma 1977. Silvia Lusuardi Siena (ed.), Ad mensam. Manufatti d’uso da contesti archeologici fra tarda anthicità e medioevo, Udine 1994. Josep M. Macias, Joan J. Menchon, Josep M. Puche & Josep A. Remolà, Nous contextos ceràmics del segle IV i inicis del V en la província de Tarragona, in Contextos 1997, 153-177. Michael Mackensen, Die Spätantiken Sigillata und Lampentöpfereien von el Mahrine (Nordtunesien): Studien zur Nordafrikanischen Feinkeramik des 4. bis 7. Jahrhunderts, Münchener Beiträge zur Vor und Frühgeschichte, Band 50, München 1993. Michael Mackensen, New evidence for Central Tunisian red slip ware with stamped decoration (ARS style D), Journal of Roman Archaeology 11 (1998), 355-370. Michael Mackensen, Centres of African red slip ware production in Tunisia from the late 5th to the 7th century, in Sagui 1998, 23-39. Production centres of African red slip ware (3rd-7th c.) in northern and central Tunisia, Journal of Roman Archaeology 15 (2002), 121-158. Marino Maggetti, Mineralogical and petrographical methods for the study of ancient pottery, in Burragato et al. 1994a, 23-35.

146

Bibliography

Mahjoubi et al. 1973

Malacitanae 1997 Manacorda & Panella 1993 Mannoni & Sfrecola 1990

Marín Jordá 1995 Martin 1989 Martin 1991-1992 Mason & ‘Amr 1995

Matson 1972

Mattingly 1988a

Mattingly 1988b Mattingly 1993

Mattingly 1994 Mattingly 1996

Mattingly et al. 2001 Mattingly & Hitchner 1995 Mattingly & Salmon 2001a Mattingly & Salmon 2001b Mayeske 1988

Mazzeo Saracino et al. 1997

McCann 1994

Ammar Mahjoubi, Jan Willem Salomonson & Abdelmajid Ennabli, La necropole romaine de Raqqada, Collection notes et documents 8.2. Tunis, Institut National d'Archeologie et d'Arts 1973. AA.VV., Figlinae Malacitanae. La producción de cerámica romana en los territorios malacitanos, Área de arqueología. Universidad de Málaga 1997. Daniele Manacorda & Clementina Panella, Anfore, in Harris 1993a, 55-64. Tiziano Mannoni & Sergio Sfrecola, Circolazione dei prodotti ceramici sulla base delle analisi mineralogico-petrografiche, in Tony Hackens & Marta Miró (eds.), Le commerce maritime romain en Méditerranée occidentale, PACT 27, 313-318. Carmen Marín Jordá, La cerámica de cocina africana: concideraciones en torno a la evidencia Valenciana, in Ibèrica 1995, 155-165. Archer Martin, L’importazione di ceramica africana a Roma tra il IV e il V secolo (S. Stefano Rotondo), L’Africa romana VI (1989), 475-483. Archer Martin, Sondages under S. Stefano Rotondo (Rome): the pottery and other finds, Boreas 14-15 (1991-1992), 157-178. James R.B. Mason & Khairieh ‘Amr, An investigation into the firing of Nabataean pottery, in Khairieh ‘Amr, Fawzi Zagadine & Muna Zaghoul (eds.), Studies in the history and archaeology of Jordan V: art and technology throughout the ages, Amman 1995, 629-636. Frederick R. Matson, Ceramic studies, in William A. McDonald & George R. Rapp Jr. (eds.), The Minnesota Messenia expedition: reconstructing a Bronze Age environment, University of Minnesota Press 1972. David J. Mattingly, Oil for export? A comparison of Libyan, Spanish and Tunisian olive oil production in the Roman empire, Journal of Roman Archaeology 1 (1988), 33-56. David J. Mattingly, Megalithic madness and measurement. Or how many olives could an olive press press?, Oxford Journal of Archaeology 7 (1988) 177-95. David J. Mattingly, Maximum figures and maximizing strategies of oil production? Further thoughts on the processing capacity of Roman olive presses, in Amouretti & Brun 1993, 483-498. David J. Mattingly, Tripolitania, Ann Arbor 1994. David J. Mattingly, First fruit? The olive in the Roman world, in Graham Shipley & John Salmon (eds.), Human landscapes in classical antiquity. Environment and culture, London 1996, 213-253. David. J. Mattingly, David Stone, Lea Striling & Nejib Ben Lazreg, Leptiminus (Tunisia). A ‘producer’ city?, in Mattingly & Salmon 2001a, 66-89. David J. Mattingly & R. Bruce Hitchner, Roman Africa: an archaeological review, Journal of Roman Studies 85 (1995), 165-213. David J. Mattingly & John Salmon (eds.), Economies beyond agriculture in the classical world, Routledge: London 2001. David J. Mattingly & John Salmon, The productive past. Economies beyond agriculture, in Mattingly & Salmon 2001a, 3-14. Betty Jo Mayeske, A Pompeian bakery on the Via dell’Abbondanza, in Robert I. Curtis (ed.), Studia Pompeiana & classica in honor of Wilhelmina F. Jashemski. Volume I: Pompeiana, New York 1988, 149-165. Luisa Mazzeo Saracino, Noris Morandi, Maria Cristina Nannetti & Mariangela Vergari, Una produzione di ceramica da cucina da Suasa (AN): esame tipologico e studio archeometrico per la definizione della funzione, in Santoro Bianchi & Fabbri 1997, 200-213. Anna M. McCann, The Late Roman shipwreck: Summary and conclusions, in McCann & Freed 1994, 49-58.

147

Bibliography

McCann 2001 McCann & Freed 1994

McCarthy & Brooks 1998 McGrail 1989 Meneghini 1987-1988

Michelucci 1985 Middleton & Freestone 1991 Millet 1979a Millet 1979b Millet 1987 Millet 1993 Mitchell & Hart 1989

Moliner 1996 Monacchi 1990

Morel 1983

Moreno Almenara & Alarcón Castellano 1994 Navarro Luengo et al. 1997

Neuru 1980 Neuru 1986 Neuru 1987 Neuru 1990 Niemayer 1968 Nin 1996

Anna M. McCann, An early imperial shipwreck in the deep sea off Skerki Bank, Rei Cretariae Romanae Fautorum Acta 37 (2001), 257-264. Anna M. McCann & Joann Freed, Deep water archaeology: a Late–Roman ship from Carthage and an ancient trade route near Skerki Bank of Northwest Sicily, Journal of Roman Archaeology Suppl. 13, Ann Arbor 1994, Michael R. McCarthy & Catherine M. Brooks, Medieval pottery in Britain AD 900-1600, Leicester University Press: Avon 1998. Sean McGrail, The shipment of traded goods and of ballast in antiquity, Oxford Journal of Archaeology 8 (1989), 353-358. R. Meneghini 1987-88, Lungotevere Testaccio. Produzioni e importazioni a Roma tra la fine del IV e il V secolo d.C., Bullettino della commissione archeologica di Roma 92 (1987-1988), 360-363. Maurizio Michelucci, Roselle. La domus dei mosaici, Montepulciano 1985. Andrew P. Middleton & Ian C. Freestone (eds.), Recent developments in ceramic petrology, British Museum Occasional Paper 81, London 1991. Martin Millet (ed.), Pottery and the archaeologist, University of London, Institute of Archaeology, Occasional Paper 4, London 1979. 1979b, How much pottery, in Millet 1979a, 77-80. Martin Millet, A question of time. Aspects of the future pottery studies, University of London. Institute of Archaeology Bulletin 24 (1987), 99-108. Martin Millet, Samian from the sea: Cala Culip shipwreck IV, Journal of Roman Archaeology 6 (1993), 415-419. Richard S. Mitchell & Steven C. Hart, Heated mineral mixtures related to ancient ceramic pastes. X-ray diffraction study, in Ralph O. Allen (ed.), Archaeological chemistry IV, Advances in Chemistry 220 (1989), 145-155. Martin Moliner, Les céramiques communes à Marseille d'après les fouilles récentes, in Bats 1996, 237-255. Daniela Monacchi, Lugnano in Teverina (Terni). Locale Poggio Gramignano Saggi di scavo di una villa rustica romana, Notizie degli Scavi di Antichità serie VIII volume XL-XLI 1986-1987 (1990), 5-35. Jean-Paul Morel, La céramique comme indice du commerce antique (réalités et interprétations), in Peter Garnsey & Charles R. Whittaker (eds.), Trade and famine in classical antiquity, Cambridge 1983. Maudilio Moreno Almenara & Francisco J. Alarcón Castellano, Producciones cerámicas locales o regionales de época tardía en Colonia Patricia Corduba. El yacimento de Cercadilla, L'Africa Romana XI (1994), 1285-1300. Ildefonso Navarro Luengo, Luís-Efrén Fernández Rodríquez & José Suárez Padilla, Cerámicas comunes de época tardorromana y bizantina en Málaga, in Malacitanae 1997, 79-93. Lucinda Neuru, Late Roman pottery: a fifth century deposit from Carthage, Antiquités Africaines 16 (1980), 195-211. Lucinda Neuru, Cooking wares of the Byzantine epoch at Carthage, in Pierre Senay (ed.), Carthage VIII, Cahiers des études anciennes 18 (1986), 69-79. Lucinda Neuru, Red-slipped wares of south-western central Tunisia: new evidence, Rei Cretariae Romanae Fautorum Acta 25-26 (1987), 175-188. Lucinda Neuru, The pottery of the Kasserine survey, in Robert B. Hitchner, The Kasserine archaeological survey 1987, in Antiquités Africaines 26 (1990), 255-259. Hans Georg Niemayer, Schwarzrandware aus Köln, Kölner Jahrbuch für Vor- und Frühgeschichte 9 (1968), 79-82. Núria Nin, Le vaisselier du site de l'établissement thermal à Aix-en-Provence (50 av. n.è.-fin du Ier s. de n.è.): ébauche d'un faciès culturel, in Bats 1996, 257-287.

148

Bibliography

Nolla & Aquilué Abadías 1984 Nolla & Casas 1990

Nolla & Puertas 1988

O'Brien et al. 1994

Olcese 1993 Olcese 1995

Oleson & Oleson 1987

Orton 1970 Orton 1975 Orton 1993 Orton & Tyers 1990 Orton & Tyers 1993 Orton et al. 1993 Ostia I Ostia II Ostia III Ostia IV Pacetti & Sfrecola 1989

Palmer 1980

Panella 1985

Panella 1986

Josep M. Nolla & Xavier Aquilué Abadías, Notes sobre una cassola de cuina de ceràmica africana procedent d'Empuries, Informació Arqueòlogica 42 (1984), 51-57. Josep Maria Nolla & Josep Casas, El material ceràmic d'importació de la villa Roma de Puig Rodon (Corça, Baix Empordà), d'epoca severiana a la Baixa Antiguitat, Cypsela VIII (1990), 193-218. Josep Maria Nolla & M. Concepció Puertas, Ceràmiques africanes i material d'importacio baix-imperial del jaciment del Camp de la Gruta (Torroella de Montgrí, Baix Emporda) Estudis sobre Temes del Baix Empordà 7 (1988), 29-77. Michael J. O'Brien, Thomas D. Holland, Robert J. Hoard & Gregory L. Fox, Evolutionary implications of design and performance characteristics of prehistoric pottery, Journal of Archaeological Method and Theory 1 (1994), 259-304. Gloria Olcese, Le ceramiche comuni di Albintimilium. Indagine archeologica e archeometrica sui materiali dell'area del Cardine, Firenze 1993 Gloria Olcese (ed.), Ceramica romana e archeometria: lo stato degli studi. Atti delle Giornate Internazionale di Studio Castello di Montegufoni (Firenze), 26-27 aprile 1993, Firenze 1995. Martha I. Oleson & John P. Oleson, Coarse wares, in Anna M. McCann, Joanne Bourgeois, Elaine K. Gazda, John P. Oleson & Elizabeth L. Will (eds.), The Roman port and fishery of Cosa. A center of ancient trade, Princeton 1987, 243248. Clive R. Orton, The production of pottery from a Romano-British kiln site: a statistical investigation, World Archaeology 1 (1970), 343-358. Clive R. Orton, Quantitative pottery studies. Some progress, problems and prospects, Science and Archaeology 16 (1975), 30-35. Clive R. Orton, How many pots make five?–An historical review of pottery quantification, Archaeometry 35 (1993), 169-184. Clive R. Orton & Paul A. Tyers, Statistical analysis of ceramic assemblages, Archeologia e calcolatori 1 (1990), 81-110. Clive R. Orton & Paul A. Tyers, A user's quide to Pie-slice, Institute of Archaeology, University of London 1993. Clive R. Orton, Paul A. Tyers & Alan Vince, Pottery in archaeology, Cambridge 1993. Andrea Carandini et al., Ostia I. Le Terme del Nuotatore, scavo dell'ambiente IV, Studi Miscellanei 13 (1968). Fede Berti et al., Ostia II. Le Terme del Nuotatore, scavo dell'ambiente I, Studi Miscellanei 16 (1970). Andrea Carandini et al., Ostia III. Le Terme del Nuotatore, scavo dell'ambiente V e di un saggio nell’area SO, Studi Miscellanei 21 (1973). Lucilla Anselmino et al., Ostia IV. Le Terme del Nuotatore, scavo dell'ambiente XVI e dell’area XXV, Studi Miscellanei 23 (1977). Francesco Pacetti & Sergio Sfrecola, Ceramiche africane di VI secolo provenienti da una domus tardoantica del Celio. Sintesi storica e indagine mineralogica, L’Africa romana VI (1989), 485-503. Robert E.A Palmer, Customs on market goods imported into the city of Rome, in John D'Arms & E. Christian Kopff (eds.), The seaborne commerce of ancient Rome: Studies in archaeology and history, Memoirs of the American Academy in Rome 36 (1980), 217-233. Clementina Panella, I commerci di Roma e di Ostia in età imperiale (secoli I-III): le derrate alimentari, in Misurare la terra: centuriazione e coloni nel mondo romano V . Città, agricoltura, commercio: materiali da Roma e dal suburbio, Modena 1985, 180-189. Clementina Panella, Le merci: produzioni, itinerari e destini, in Giardina 1986, 431-459.

149

Bibliography

Panella 1991 Panella 1993 Panella 1996 Panella 1999 Parker 1984 Parker 1990

Parker 1992 Paroli et al. 1993

Pasquinucci et al. 1998

Pasquinucci & Menchelli 1996

Pasquinucci & Storti 1989 Passelac 1996

Paterson 1998

Patterson 1987

Pavolini 1985

Pavolini 1986 Pavolini 1995 Pavolini 1996

Pavolini 1998 Peacock 1977a

Clementina Panella, Un contesto di età antonina dalle Terme del Nuotatore di Ostia, Studi Miscellanei 28 (1991), 283-297. Clementina Panella, Merci e scambi nel Mediterraneo tardoantico, in Andrea Carandini et al. (eds.), Storia di Roma III.2, 613-697. Clementina Panella, Lo studio delle ceramiche comuni di età romana: qualche riflessione, in Bats 1996, 9-15. Clementina Panella, Rifornimenti urbani e cultura materiale tra Aureliano e Alarico, in Harris 1999, 183-215. Anthony J. Parker, Shipwrecks and ancient trade in the Mediterranean, Archaeological Review from Cambridge 3 (1984), 99-113. Anthony J. Parker, The pattern of commerce as evidenced by shipwrecks, in Tony Hackens & Marta Miró (eds.), Le commerce maritime romain en Méditerranée occidentale, PACT 27 (1990), 147-168. Anthony J. Parker, Ancient shipwrecks of the Mediterranean and the Roman provinces, British Archaeological Reports: International Series 580, Oxford 1992. Lidia Paroli, Archer Martin, Barbara Ciarrocchi & Helen Patterson, Produzione e circolazione di ceramiche tardoantiche ed altomedievali ad Ostia e Porto, in Lidia Paroli & Paolo Delogu (eds.), La storia economica di Roma nell’alto medioevo alla luce dei recenti scavi archeologici, Biblioteca di archeologia medievale 10 (1993), 203-246 Marinella Pasquinucci, D. Alessi, S. Bianchini, A. Del Rio & Simonetta Menchelli, Circolazione di merci africane nel Tirreno settentrionale (I-VII sec. d.C.), L’Africa Romana XII (1998), 1401-1421. Marinella Pasquinucci & Simonetta Menchelli, Il vasellame di uso comune (I-VII sec. d.C.) rinvenuto negli horrea di S. Gaetano di Vada (Rosignano M.mo, Li): ricerche archeometriche, morfologiche ed analisi quantitative, Archeologia e Calcolatori 7 (1996), 497-514. Marinella Pasquinucci & Simonetta Storti, Pisa antica. Scavi nel giardino dell’Arcivescovado, Pondedra 1989. Michel Passelac, Céramiques communes gallo-romaines en Languedoc occidental: exemples de production et de consommation (fin Ier s. av. notre ère-IIe s. de notre ère), in Bats 1996, 361-387. Jeremy Paterson, Trade and trades in the Roman world: scale, structure, and organisation, in Helen Parkins & Christopher Smith (eds.), Trade, traders and the ancient city, Routledge: Ebbw Vale 1998, 149-167. Helen Patterson, The late Roman and early Medieval pottery from Molise, San Vincenzo al Volturno. The archaeology, art and territory of an early Medieval monastery, British Archaeological Reports: International Series 252, Oxford 1987, 83-110. Carlo Pavolini, I commerci di Roma e di Ostia nella prima età imperiale: merci d'accompagno e carichi di ritorno, in Misurare la terra: centuriazione e coloni nel mondo romano V. Città, agricoltura, commercio: materiali da Roma e dal suburbio, Modena 1985, 200-205. Carlo Pavolini, La vita quotidiana a Ostia, Roma 1986. Carlo Pavolini, Il commercio della ceramica comune: anticipazioni da una ricerca in corso sul materiale Ostiense, in Olcese 1995, 115-126. Carlo Pavolini, Mercato ostiense e mercato romano: alcuni contesti ceramici a confronto, in Anna Gallina Zevi & Amanda Claridge (eds.), Roman Ostia revisited, Rome 1996, 223-242 Carlo Pavolini, Forme chiuse in ceramica comune del VI-VII secolo nei magazzini di Ostia, in Ciarrocchi et al. 1998, 391-394. David P.S. Peacock (ed.), Characterization and trade in Roman and later ceramics, London 1977.

150

Bibliography

Peacock 1977b Peacock 1982 Peacock 1984a Peacock 1984b Peacock 1984c Peacock et al. 1989

Peacock et al. 1990 Peacock & Tomber 1991

Peacock & Williams 1986 Peña 1990

Peña 1992 Peña 1998 Peña 1999

Picon & Olcese 1995 Pollard 1990 Pons Machado 1998 Pontacolone & Incitti 1991 Press & Siever 1986 Prothero & Schwab 1996 Pucci 1975

Pucci 1983 Purcell 1999 Rakob 1991 Rapp & Gifford 1985 Raven 1993

David P.S. Peacock, Ceramics in Roman and Medieval archaeology, in Peacock 1977a, 21-34. David P.S. Peacock, Pottery in the Roman world: an ethnoarchaeological approach, New York 1982. David P.S. Peacock, Introduction, in Fulford & Peacock 1984, 1-5. David P.S. Peacock, Petrology and origins, in Fulford & Peacock 1984, 6-28. David P.S. Peacock, Seawater, salt and ceramics, in Fulford & Peacock 1984, 263264. David P.S. Peacock, Fathi Bejaoui & Nejib Ben Lazreg, Roman amphora production in the Sahel region of Tunisia, in Maurice Lenoir, Daniele Manacorda & Clementina Panella (eds.), Amphores Romaines et histoire economique: Dix ans de recherche, Collection de l'École Française de Rome 114 (1989), 179-222. David P.S. Peacock, Fathi Bejaoui & Nejib Ben Lazreg, Roman pottery production in Central Tunisia, Journal of Roman Archaeology 3 (1990), 59-84. David P.S. Peacock & Roberta Tomber, Roman amphora kilns in the Sahel of Tunisia: petrographic investigation of kiln material from a sedimentary environment, in Middleton and Freestone 1991, 289-304. David P.S. Peacock & David F. Williams, Amphorae and the Roman economy, New York 1986. J. Theodore Peña, Internal Red-slip cookware (Pompeian red ware) from Cetamura del Chianti, Italy: Mineralogical composition and provenience, American Journal of Archaeology 94 (1990), 647-661. J. Theodore Peña, Raw material use among nucleated industry potters: the case of Vasanello, Italy, Archeomaterials 6 (1992) 93-122. J. Theodore Peña, Aspects of residuality in the Palatine East pottery assemblage, in Guidobaldi et al. 1998, 5-19. J. Theodore Peña, The urban economy during the Early dominate. Pottery evidence from the Palatine Hill, British Archaeological Reports: International Series 784, Oxford 1999. Maurice Picon & Gloria Olcese, Per una classificazione in laboratorio delle ceramiche comuni, in Olcese 1995, 105-114. Richard Pollard, Quantification: towards a standard practice, Journal of Roman Pottery Studies 3 (1990), 75-79. Octavio Pons Machado, La cerámica norte africana del poblado talayotico de Son Catlar, L’Africa Romana XII (1998), 1335-1340. Laura Pontacolone & Mauro Incitti, Un relitto con carico di merci africane di età imperiale alle Trincere (Tarquinia), L'Africa Romana VIII (1991), 543-570. Frank Press & Raymond Siever, Earth (4th ed.), New York 1986. Donald R. Prothero & Fred Schwab, Sedimentary geology. An introduction to sedimentary rocks and stratigraphy, New York 1996. Giuseppe Pucci, La ceramica, in Antonio Di Vita, Paola Procaccini & Giuseppe Pucci, Lo scavo a Nord del mausoleo punico-ellenistico A di Sabratha, Libya Antiqua 11-12 (1975), 7-111. Giuseppe Pucci, Pottery and trade in the Roman period, in Garnsey et al. 1983, 105-117. Nicholas Purcell, The populace of Rome in late antiquity: problems of classification and historical description, in Harris 1999, 135-161. Friedrich Rakob (ed.), Karthago I, Die deutschen Ausgrabungen in Karthago, Mainz am Rhein 1991. George Rapp Jr. & John A. Gifford (eds.), Archaeological geology, New Haven 1985. Susan Raven, Rome in Africa, London 1993.

151

Bibliography

Revilla Calvo et al. 1997

Reynolds 1984

Reynolds 1993 Reynolds 1995 Ricci 1986 Rice 1987 Rice 1990

Rice 1991

Rice 1996a Rice 1996b Rickman 1980 Rife 2001 Riley 1979a Riley 1979b Riley 1980

Riley 1981 Rita 1990

Rivet 1996

Rüger & Niemeyer 1962 Rye 1976

Rye 1981 Saguì 1979 Saguì 1980

V. Revilla Calvo, C. Marti Garcia, J. García Rosello, J. Pera Isern, J.A. Cerdà Mellado & J. Pujol Del Horno, El nivell amortització del Cardo Maximus d’Iluro, in Contextos 1997, 101-119. Paul Reynolds, African Red Slip and late Roman imports in Valencia, in Thomas F.C. Blagg, Rick F.J. Jones & Simon J. Keay, Papers in Iberian Archaeology, British Archaeological Reports: International Series 194, Oxford 1984, 474-539. Paul Reynolds, Settlement and pottery in the Vinalopó valley (Alicante, Spain) AD 400-700, British Archaeological Reports: International Series 588, Oxford 1993. Paul Reynolds, Trade in the western Mediterranean AD 400-700: The pottery evidence, British Archaeological Reports: International Series 604, Oxford 1995. Andreina Ricci, Cosa e il suo territorio (Etruria). Il contesto di Settefinestre, in Giardina 1986, 83-87. Prudence M. Rice, Pottery analysis: a sourcebook, Chigaco 1987. Prudence M. Rice, Functions and uses of archaeological ceramics, in W.D. Kingery (ed.), The changing roles of ceramics in society: 26.000 B.P to the present. Ceramics and Civilization V, American ceramic society: Westerville 1990, 1-10. Prudence M. Rice, Specialization, standardization, and diversity: A retrospective, in Ronald L. Bishop & Frederick W. Lange (eds.), The ceramic legacy of Anna O. Shepard, Niwot 1991, 257-279. Prudence M. Rice, Recent ceramic analysis: 1. Function, style and origins, Journal of Archaeological Research 4 (1996), 133-163. Prudence M. Rice, Recent ceramic analysis: 2. Composition, production and theory, Journal of Archaeological Research 4 (1996), 165-202. Geoffrey Rickman, The corn supply of ancient Rome, Oxford 1980. Joseph L. Rife, Excavations adjacent to the House of the Venus Mosaic (site 250), in Striling et al. 2001, 293-324. John A. Riley, The coarse pottery from Berenice, in Lloyd, John (ed.), Excavations at Sidi Khrebish, Benghazi (Berenice) II, Libya Antiqua Suppl. 5.2 (1979), 91-476 John A. Riley, The petrological investigation of Roman and Islamic ceramics from Cyrenaica, Libyan Studies 10 (1979), 35-46. John A. Riley, Industrial standardization in Cyrenaica during the second and third centuries AD: The evidence from locally manufactured pottery, Libyan Studies 11 (1980), 73-78. John A. Riley, The pottery from the cisterns 1977,1, 1977,2 and 1977,3, in Humphrey 1981, 85-124. María Cristina Rita, Un basurero romano hallado en Mahón: excavaciones de urgencia en la Plaza de la Conquista 1989, Treballs del Museu de Menorca 11 (1990). Lucien Rivet, Fonctions et faciès: étude comparée de quelques lots de céramiques provenant de Fréjus (Var), Mandelieu (Alpes-Maritimes), Aix-en-Provence et SaintJulien-les-Martigues (Bouches-du-Rhône), in Bats 1996, 327-350. Chr. Rüger & Hans Georg Niemeyer, Vorbericht über die Arbeiten in Centcelles 2: Die Keramik der Grabungen 1959-1961, Madrider Mitteilungen 3 (1962), 101-118. Owen S. Rye, Keeping your temper under control: Materials and the manufacture of Papuan pottery, Archaeology and physical anthropology in Oceania 11 (1976), 10637. Owen S. Rye, Pottery technology: Principles and reconstruction, Manuals on Archaeology 4, Washington D.C. 1981. Lucia Saguì, Ceramica africana al Museo Archeologico Nazionale di Sperlonga, Archaeologia Classica 31 (1979), 255-272. Lucia Saguì, Ceramica africana dalla “Villa di Tiberio” a Sperlonga, Melanges de l'École Française de Rome 92 (1980), 471-544.

152

Bibliography

Saguì 1998 Salomonson 1962

Salomonson 1968

Salomonson 1969 Sánchez Sánchez 1995 Santoro Bianchi & Fabbri 1997 Salza Prina Ricotti 1980 Schiffer 1990 Schiffer et al. 1994

Schuring 1984 Schuring 1986

Schuring 1987

Schuring 1988 Schütz 1992 Scobie 1986 Senay 1981 Serrano Ramos 1997 Settefinestre II Shaw 1981

Sillar & Tite 2000 Skibo 1992 Skibo & Schiffer 1995

Lucia Saguì (ed.), Ceramica in Italia: VI-VII secolo. Atti del Convegno in onore di John W. Hayes. Roma 11-13 maggio 1995, Firenze 1998. Jan Willem Salomonson, Late-Roman earthenware with relief decoration found in Northern Africa and Egypt, Oudheidkundige mededelingen uit het Rijksmuseum van Oudheden te Leiden 43 (1962), 53-95. Jan Willem Salomonson, Etudes sur le céramique romaine d'Afrique. Sigilée claire et céramique commune de Henchir el Ouiba (Raqqada) en Tunisie centrale, Bulletin Antieke Beschaving 43 (1968), 80-145. Jan Willem Salomonson, Spätrömische rote Tonware mit Reliefverzierung aus Nordafrikanischen Werkstätten, Bulletin Antieke Beschaving 44 (1969), 4-109. M. Angeles Sánchez Sánchez, Producciones importadas en la vajilla culinaria romana del Bajo Guadalquivir, in Ibèrica 1995, 251-279. S. Santoro Bianchi & B. Fabbri (eds.), Il contributo delle analisi archeometriche allo studio delle ceramiche grezze e comuni, Imola 1997. E. Salza Prina Ricotti, Cucine e quartieri servili in epoca romana, Rendiconti della pontificia accademia romana di archeologia 51-52 (1980), 237-294. Michael B. Schiffer, The influence of surface treatment on heating effectiveness of ceramic vessels, Journal of Archaeological Science 17 (1990), 373-381. Michael B. Schiffer, James M. Skibo, Tamara C. Boelke, Mark A. Neupert & Meredith Aronson, New perspectives on experimental archaeology: surface treatments and thermal response of the clay cooking pot, American Antiquity 59 (1994), 197-217. Josine M. Schuring, Studies on Roman amphorae I-II, Bulletin Antieke Beschaving 59 (1984), 137-195. Josine M. Schuring, The Roman, early Medieval and Medieval coarse kitchen wares from the San Sisto Vecchio in Rome, Bulletin Antieke Beschaving 61 (1986), 158207. Josine M. Schuring, Terra Sigillata Africana from the San Sisto Vecchio in Rome: aspects of fabric characterization. A preliminary assessment, in van As 1987, 5473. Josine M. Schuring, Terra Sigillata Africana from the San Sisto Vecchio in Rome, Bulletin Antieke Beschaving 63 (1988), 1-68. Ilse Schütz, Sistemas tradicionales de cocción cerámica en el norte de África, in Tecnologia 1992, 153-169. Alex Scobie, Slums, sanitation and mortality in the Roman world, Klio 68 (1986), 399-433. Pierre Senay (ed.), Carthage V, Cahiers des études anciennes 13 (1981). Encarnación Serrano Ramos, La producción cerámica de los talleres romanos de la Depresión de Antequera, in Malacitanae 1997, 217-232. Andreina Ricci (ed.), Settefinestre. Una villa schiavistica nell’Etruria Romana II. La villa e suoi reperti, Modena 1985. Brent T. Shaw, Climate, environment and history: the case of Roman Africa, in T.M.L. Wigley, M. Ingram & G. Farmer (eds.), Climate and history: studies in past climates and their impact on man, Cambridge 1981. B. Sillar & Mike S. Tite, The challenge of ‘technological choices’ for materials science approaches in archaeology, Archaeometry 42 (2000), 2-20. James M. Skibo, Pottery function: A use-alteration perspective, New York 1992, 147-171. James M. Skibo & Michael B. Schiffer, The clay cooking pot. An exploration of women's technology, in James M. Skibo, William H. Walker & Axel E. Nielsen, Expanding archaeology, Salt Lake City 1995.

153

Bibliography

Small 1994 Smith M. 2001 Smith W. 1998

Smith W. 2001 Staffa 1986

Steen-McIntyre 1985 Stirling 2001 Striling et al. 2001

Stirling & Ben Lazreg 1997

Stirling & Ben Lazreg 2001 Stone et al. 1999

Stone et al. 2001 Swan 1984 Tassinari 1996 Taylor & Robinson 1996a Taylor & Robinson 1996b Taylor et al. 1997

Tecnología 1992

Therrien 1981 Tite 1992 Tite 1995

Alastair M. Small, The excavations of S. Giovanni di Ruoti, vol. 1. The villas and their environment, Phoenix. Supplementary volume 33, Toronto 1994. Monica L. Smith, A water-collecting area on Dhahret Slama ridge, in Striling et al. 2001, 253-258. Wendy Smith, Fuel for thought: archaeobotanical evidence for the use of alternatives to wood fuel in late antique North Africa, Journal of Mediterranean Archaeology 11 (1998), 191-205. Wendy Smith, Environmental sampling (1990-94), in Striling et al. 2001, 420441. A.R. Staffa, Località Rebibbia, via S. Cannizzaro. Un punto di sosta lungo la via Tiburtina fra l’età di Augusto e la tarda antichita (circ. V), Bullettino della commissione archeologica di Roma 91 (1986), 642-678. Virginia Steen-McIntyre, Tephrochronology and its application to archaeology, in Rapp & Gifford 1985, 265-302. Lea Stirling, The East Baths and their industrial re-use in late antiquity: 1992 excavations, in Striling et al. 2001, 29-73. Lea M. Stirling, David J. Mattingly & Nejib Ben Lazreg (eds.), Leptiminus (Lamta) report no. 2. The East baths, cemeteries, kilns, Venus mosaic, site museum and other studies, Journal of Roman Archaeology Suppl. 41, Ann Arbor 2001. Lea M. Stirling & Nejib Ben Lazreg, Roman kilns and pottery production at Leptiminus, Tunisia: results of the 1995 and 1996 excavations, American Journal of Archaeology 101 (1997), 372. Lea M. Stirling & Nejib Ben Lazreg, A Roman kiln complex (Site 290): preliminary results of excavation 1995-98, in Striling et al. 2001, 220-235. David Stone, Lea M. Stirling & Nejib Ben Lazreg, Suburban land-use and ceramic production around Leptiminus (Tunisia): interim report, Journal of Roman Archaeology 11 (1999), 305-317. David Stone, Lea M. Stirling & Nejib Ben Lazreg, Introduction to Dhahret Slama and the southeastern suburban zone, in Striling et al. 2001, 215-218. Vivien G. Swan, The pottery kilns of Roman Britain, Royal commission on historical monuments supplementary series 5, London 1984. Suzanne Tassinari, Bouillir, mijoter et cuire, porter et mesurer: le rôle de la vaisselle métallique dans le cuisines de Pompei, in Bats 1996, 113-119. R.J. Taylor & V.J Robinson, Neutron activation analysis of Roman African Red Slip ware kilns, Archaeometry 38 (1996), 231-243. R.J. Taylor & V.J Robinson, Provenance Studies of Roman African Red Slip ware using neutron activation analysis. Archaeometry 38 (1996), 245-255. R.J. Taylor, V.J. Robinson & David J.L. Gibbins, An investigation of the provenance of the Roman amphora cargo from the Plemmirio B shipwreck, Archaeometry 39 (1997), 9-21. AA.VV., Tecnología de la cocción cerámica desde la antigüedad a nuestros dias, Ponencias del Seminario celebrado en el Museo de Alfarería en Agost (Alicante) del 4 al 6 octubre de 1990, Asociación de ceramología. Gisèle Therrien, Étude typologique de la céramique commune 1979, in Senay 1981, 89-154. Mike S. Tite, The impact of electron microscopy on ceramic studies, Proceedings of the British Academy 77 (1992), 111-131. Mike S. Tite, Firing temperature determinations – why and how?, in Anders Lindahl & Ole Stilborg (eds.), The aim of laboratory analyses of ceramics in archaeology, Kungl. Vitterhets Historie och Antikvitets Akademien. Konfrenser 34, Stockholm 1995, 37-42

154

Bibliography

Tite 1999

Tite et al. 1982

Tite & Kilikoglou 2002 Tite et al. 2001

Tomber 1986

Tomber 1989 Tomber 1993 Tortorella 1981a

Tortorella 1981b Tortorella 1983a Tortorella 1983b

Tortorella 1986 Tortorella 1987

Tortorella 1995

Vázquez de la Cueva 1985 Vegas 1973 Vekinis & Kilikoglou 1998

Veniale 1994 Villa 1994 Villedieu 1984

Mike S. Tite, Pottery production, distribution, and consumption – the contribution of the physical sciences, Journal of archaeological method and theory 6 (1999), 181233. Mike S. Tite, Ian C. Freestone, N.D Meeks & M. Bimson, The use of scanning electron microscopy in the technological examination of ancient ceramics, in Jaqueline S. Olin & Alan D. Franklin (eds.), Archaeological ceramics, Washington D.C. 1982, 109-120. Mike S. Tite & Vassilis Kilikoglou, Do we understand cooking pots and is there an ideal cooking pot?, in Kilikoglou et al. 2002, 1-8.. Mike S. Tite, Vassilis Kilikoglou & G. Vekinis, Strength, toughness and thermal shock resistance of ancient ceramics, and their influence on technological choice, Archaeometry 43 (2001), 301-324. Roberta S. Tomber, Pottery from the south side of the circular harbour, Bulletin du Centre d'études et de documentation archeologique de la conservation de Carthage 7 (1986), 34-58. Roberta S. Tomber, Pottery from the 1982-83 excavations, in John H. Humphrey (ed.), The circus and byzantine cemetery at Carthage I, Ann Arbor 1989, 437-519. Roberta S. Tomber, Quantitative approaches to the investigation of long-distance exchange, Journal of Roman Archaeology 6 (1993), 142-166. Stefano Tortorella, Ceramica di produzione africana e rinvenimenti archeologici sottomarini della media e tarda età imperiale: analisi dei dati e dei contributi reciproci, Melanges de l'École Française de Rome 93 (1981), 355-380 Stefano Tortorella, Ceramica da cucina, in Atlante I (1981), 208-228. Stefano Tortorella, Produzione e circolazione della ceramica africana di Cartagine (V-VII sec.), Opus 2 (1983), 15-30. Stefano Tortorella, La sigillata africana a Cartagine fra il 400 d.C. e la conquista vandala: i dati dello scavo della missione archeologica italiana, Bulletin du Centre d'études et de documentation archeologique de la conservation de Carthage. Dossier 1: Actes du Colloque sur la ceramique antique de Carthage (1983), 125-139. Stefano Tortorella, La ceramica fine da mensa africa dal IV al VII secolo d.C., in Giardina 1986, 211-225. Stefano Tortorella, La ceramica africana: un riesame della problematica, in Pierre Leveque & Jean-Paul Morel (eds.), Ceràmiques hellénistiques et romaines II, Paris 1987, 297-327. Stefano Tortorella, La ceramica africana. Un bilancio dell’ultimo decennio di ricerche, in Pol Trousset (ed.), Productions et exportations africaines: actualités archéologiques en Afrique du Nord antique et médiévale, Actes du VI e colloque international sur l’histoire et l’archéologie de l’Afrique du Nord (Pau, octobre 1993 – 118e congrès), Paris 1995, 79-102. Ana Vázquez de la Cueva, Sigillata africana en Augusta Emerita, Monografías Emeritenses 3, Mérida 1985. Mercedes Vegas, Ceramica comun romana del Mediterraneo occidental, Barcelona 1973. G. Vekinis & Vassilis Kilikoglou, Mechanical performance of quartz-tempered ceramics: Part II, Hertzian strength, wear resistance and applications to ancient ceramics, Archaeometry 40 (1998), 281-292. F. Veniale, Raw materials and manufacturing processes in ancient ceramic artefacts, in Burragato et al. 1994a, 55-72. Luca Villa, Le anfore tra tardoantico e medioevo, in Lusuardi Siena, 335-431. Françoise Villedieu, Turris Libisonis. Fouille d'un site romain tardif à Porto Torres, Sardaigne, British Archaeological Reports: International Series 224, Oxford 1984.

155

Bibliography

Villedieu 1998

Volpe 1998 Waagé 1933 Waagé 1948 Wallace 1989 Warmington 1954 van der Werff 1982 Wertime 1983

Whitbread 1986 Whitehouse et al. 1982

Whitehouse et al. 1985

Whittaker 1983 Whittaker 1988 Whittaker 1989 Wickham 1988 Wilkins et al. 1995 Williams 1979 Williams 1984

Williams 1990

Young 1981 Young & Stone 1990 Zeggio & Rizzo 1998

Zevi & Pohl 1970

Françoise Villedieu, Les fouilles de la Vigna Barberini (secteur A). Apport de l’analyse du mobilier résiduel à l’interprétation des données de la fouille, in Guidobaldi et al. 1998, 21-43. Rita Volpe, Aqua Marcia. Il contesto generale e due casi particolari, in Guidobaldi et al. 1998, 237-249. Frederick O. Waagé, Excavations in the Athenian Agora. The Roman and Byzantine Pottery, Hesperia 2 (1933), 279-328. Frederick O. Waagé, Hellenistic and Roman Tableware of North-Syria, in Antiochon-the-Orontes IV:I (1948), 1-60. D. Wallace, Functional factors of mica and ceramic burnishing, in Gordon Bronitsky (ed.), Pottery technology: ideas and approaches, Boulder 1989, 33-39. Brian H. Warmington, The North African provinces from Diocletian to the Vandal conquest, Cambridge 1954. Jacobus Hermanus van der Werff, Uzita. Vondstenmateriaal uit een antieke nederzetting in Midden-Tunesië, Ph.D. diss., University of Utrecht 1982. T.A. Wertime, The furnace versus the goat: the pyrotechnologic industries and Mediterranean deforestation in antiquity, Journal of Field Archaeology 10 (1983), 445-452. Ian K. Whitbread, The characterisation of argillaceous inclusions in ceramic thin sections, Archaeometry 28 (1986), 79-88. David Whitehouse, Graeme Barker, Richard Reece & David Reese, The Schola Praeconum I: the coins, pottery, lamps and fauna, Papers of the British School at Rome 37 (1982), 53-101. David Whitehouse, Lorenzo Constantini, Federico Guidobaldi, Siro Passi, Patrizio Pensabene, Simon Pratt, Richard Reece & David Reese, The Schola Praeconum II, Papers of the British School at Rome 40 (1985), 163-210. Charles R. Whittaker, Late Roman trade and traders, in Garnsey, Hopkins & Whittaker 1983, 163-180. Charles R. Whittaker, Trade and aristocracy in the Roman Empire, Opus 4 (1988), 49-75. Charles R. Whittaker The poor in the city of Rome, Land, city and trade in the Roman Empire, Variorum collected studies series CS408, Norfolk 1989, VII.1-25. Chris Wickham, Marx, Sherlock Holmes and Late Roman commerce, Journal of Roman Studies 78 (1988), 183-193 John Wilkins, David Harvey & Mike Dobson (eds.), Food in antiquity, Exeter 1995. David F. Williams, Ceramic petrology and the archaeologist, in Millet 1979a, 7376. David F. Williams, Petrological analysis of mortar and mud brick, in Henry R. Hurst, & Stephen P. Roskams (eds.), Excavations at Carthage: The British Mission I.1. The Avenue du President Habib Bourguiba, Salammbo; The Site and Finds Other than Pottery, Sheffield 1984, 219-221. David F. Williams, The study of ancient ceramics: the contribution of petrographic method, in Tiziano Mannoni & Alessandra Molinari (eds.), Scienze in Archeologia, Firenze 1990, 43-63. C.J. Young, The value of the study of pottery in the Late Roman period, in Howard & Morris 1981, 209-220. Lisa C. Young & Tammy Stone, The thermal properties of textured ceramics: an experimental study, Journal of Field Archaeology 17 (1990), 195-203. Sabina Zeggio & Giorgio Rizzo, I materiali residui come indicatori della storia di un sito: Il caso della fossa di fondazione dell’Arco di Costantino, in Guidobaldi et al. 1998, 125-148. Fausto Zevi & Ingrid Pohl, Ostia. Saggi di Scavo, Notizie degli Scavi di Antichità serie VIII volume XXIV supplemento I (1970).

156

Plate 1. African cookware fabrics

a) Central Tunisian fabric I.2. Rough surface. 10x.

b) Central Tunisian fabric I.1. Polished surface. 20x.

c) North Tunisian fabric II.1. Rough surface. 10x.

d) North Tunisian fabric II.1. Polished surface. 20x.

e) North Tunisian fabric II.2. Rough surface. 10x.

f) North Tunisian fabric II.3. Polished surface. 20x.

g) Central Tunisian fabric I.3 with reduced core. Rough surface. 10x.

h) SEM-section of Central Tunisian fabric with slipped vessel surface. ca 15x.

a

b

c

d

e

f

g g

h Plate 2. Surface finishes and use-wear in African cookware:

i

a) Shallow casserole showing smoothed interior with light pitting and scratches. b) Pan with heavily worn interior surface. c) Stripe-burnished interior surface. d) Turned exterior surface. e) Shallow casserole with corrugated bottom and protruding carination. f) Rilled interior surface of the Hayes 197 deep casserole. g) Deep casserole with corrugated exterior bottom. h) Pan-casserole with patina cenerognola cover on the exterior surface below the rim. i) Lid with whitish scum cover on rim; soot deposition on the underside.

APPENDIX 1. FIND CATALOGUE

The alternative numbering given after sector code and context number in the find catalogue (Acc. #) follows the system established for “accessioned” finds of the Palatine East excavations. In other words, finds considered to be representative enough for publishing were separated from the rest of material and assigned with an unique catalogue number, the function of which was to ease their identification in the following stages of finds processing. The following abbreviations will be used in this catalogue: Dimensions sd= sherd(s) height x width rw= width of the rim rh= height of the rim vh= height of the vessel wh= height of the upper wall ph= height of patina cenerognola ø= vessel diameter

Color ic= interior surface color ec= exterior surface color es= exterior slip color sc= section color pc= patina cenerognola color oa= orlo annerito color br= the color of burnished surface

1 . Form variant A.I: Medium-sized, open form with strongly inward inclined, slightly convex wall joining smoothly with even rim and round lip. Excluding turned upper exterior wall, the vessel surfaces have been left unfinished. PEA 36; Acc. # 5742; Pl. 3, no. 1. Dimensions: sd 8,6 x 6,1 cm; ø 22 cm; rw 0,7 cm. Color: ec 2.5YR 5.5/8, ic 2.5YR 5.5/8, sc 2.5YR 5/8. Condition: good, substantially clean surfaces, sharp edges Comparanda: cf. no. 2. 2 . Form variant A.I: Medium-sized, open form with slightly convex upper wall joining smoothly with round, thickened rim. Lower exterior surface and interior surface unfinished, upper section of the exterior surface turned. PEA 36; Acc. # 1051; Pl. 3, no. 2. Dimensions: sd 13,0 x 7,8 cm; ø 30 cm; rw 1,2 cm. Color: ec 2.5YR 5/6; ic 2.5YR 5/6; sc 2.5YR 4/8. Condition: good; clean fracture surfaces and slightly worn fracture surface edges. Comparanda: Aguarod Otal 1991, pp. 305-310, figs. 4550 (several examples); Aquilué Abadías 1991, fig. 238, no. 7; Dyson 1976, fig. 49 no. 22II-77; Freed 1998, p . 58, fig. 7, no. 122; Fulford 1984b, p. 196, fig. 75, nos. 1.1-2; González Villaescusa 1990, fig. 19, no. 50; Hayes 1976, p. 51, fig. 1, no. I.3; p. 61, fig. 5, no. IX.7; Tomber 1986, p. 54, fig. 4, nos. 93-95; Tortorella 1981b, tav CIV, nos. 1-3. 3 . Form variant A.II: Medium-sized, open form with slightly angular, thickened rim and inward inclined, slightly convex body joining with a disk knob at a sharp angle. Unfinished exterior surface showing vague marks of smoothing on the upper section; wide wheel-ridging on the vessel interior. As indicated by the disk knob, which is the only one in the study assemblage belonging to a complete vessel profile, the vase was thrown rim side up and subsequently cut from a wheel with wire. PEA 22; Acc. # 607; Pl. 3, no. 3. Dimensions: ø 21 cm; vh 5,9 cm; rw 1,0 cm, knob: ø 5 cm, height 1,1 cm. Color: ec 10R 5.5/6; ic 10R 5.5/6; sc 10R 5.5/6. Condition: good, ca. 70% of the vessel preserved; sharp fracture surface edges. Patchy soot cover on rim and lower interior surface. Comparanda: cf. no. 4. 4 . Form variant A.II: Medium-sized, open form with a completely preserved profile. The topmost section of the exterior surface is crowned by a shallow ring-shaped protrusion, rectangular in profile. Outward inclined, slightly convex wall joining smoothly with a slightly

thickened rim. Unfinished interior and lower exterior surface. Ring base and the adjacent upper exterior surface turned. Blackish orlo annerito deposition on rim zone. PEA 105; Acc. # 3919; Pl. 3, no. 4. Dimensions: ø 19 cm; rw 1,0 cm; ring base height 0,5 cm. Color: ec 5Y 4/1; ic 5YR 5/4; sc 7.5YR 4/0; oa 7.5YR 4/0. Condition: very good, most of the vessel preserved, sharp fracture surfaces, light post-depositional scum. Comparanda: Aguarod Otal 1991, pp. 305-311, figs. 4551, passim.; Fulford 1984b, p. 190, fig. 72, nos. 11.1-4; Hayes 1972, p. 206, fig. 36, nos. 196.1 and 5; Ostia I, tav. XII, no. 260; Ostia III, tav. XLII, no. 332; tav. XLIII, nos. 333-334; Tortorella 1981b, tav CIV, nos. 5 & 7. 5 . Form variant A.III: Medium-sized, open form with slightly convex wall joining smoothly with thickened rim. Turned exterior surface, unfinished interior surface. Due to heavy soot deposition on the exterior surface and the rim, their color cannot be determined. Nevertheless, the rim seems to be covered by a white/brownish orlo annerito deposition. PEA 20; Acc. # 1047; Pl. 3, no. 5. Dimensions: sd 10,1 x 5,2 cm; ø 41 cm; rw 1,8 cm. Color: ec not available due to completely sooted exterior surface; ic 10R 5.5/6; sc 10R 5/6. Condition: Fairly good, a thin dirt deposition on fracture surfaces, sharp fracture surface edges. Comparanda: cf. no. 6. 6 . Form variant A.III: Medium-sized, open form with strongly inward inclined, slightly convex wall ending i n a heavily thickened rim with a convex lip. A smoothed rim, turned upper exterior, and unfinished lower exterior and interior surfaces. The rim is covered with a blackishbrown orlo annerito deposition with small sooted patches on the underside. PEA 127; Acc. # 5755; Pl. 3, no. 6. Dimensions: sd 4,7 x 7,8 cm; ø 29 cm; rw 1,8 cm. Color: ec 2.5YR 5.5/7; ic 2.5YR 5.5/7; sc 2.5YR 5/8; oa 2.5YR 4/2. Condition: medium; clean surfaces, but heavily worn surface edges. Comparanda: Aguarod Otal 1991, p. 311, fig. 51, nos. 26; Fulford 1984b, p. 190, fig. 72, no. 11.5; González Villaescusa 1990, fig. 21, no. 61; fig. 22, no. 55; Hayes 1976, p. 51, fig. 1, no. II.11; p. 53, fig. 2, no. V.14; p . 67, fig. 8, no. 26; Ostia I, tav. XII, no. 261; Ostia III, tav. XXI, no. 108a; Ostia IV, tav. XI, no. 60; Tomber 1986, p. 54, fig. 4, no. 111; Tomber 1989, p. 487, fig. 13, nos. 271-273; Tortorella 1981b, tav CIV, nos. 8-9.

1

2

3

4

5

6

Plate 3. 1-6 North Tunisian lids: plain or thickened rim (1-2 A.I; 3-4 A.II; 5-6 A.III). Scale 1:2.

7 . Form variant A.IV: Medium-sized, open form with strongly inward inclined straight wall ending in a triangular rim thickened on the exterior. Unfinished exterior surface with corrugations; pronounced corrugations on interior surface. Vague traces of white orlo annerito surface on rim. PEB 186; Acc. # 5750; Pl. 4, no. 7. Dimensions: sd 6,1 x 3,7; ø 20 cm; rw 1,0 cm. Color: ec 2.5YR 5.5/6; ic 2.5YR 6/6; sc 2.5YR 4/5; oa 7.5YR 7/2. Condition: good; substantially clean and sharp fracture surfaces, dense sooting underneath the rim. Comparanda: Aquilué Abadías 1989, p. 191, fig. 88, no. 5.2; Dore 1989, p. 153, fig. 41, no. 109.1379; Fulford 1984b, p. 190, fig. 72, no. 12.2; Mackensen 1993, p . 628 nos 73.2-4; fig. 82 nos 1-3; Reynolds 1993, pl. 58, no. 672; Riley 1981, p. 95, no. 29; Sánchez Sánchez 1995, p. 266, fig. 11, no. 6; Tomber 1986, p. 54, fig. 4 , no. 103; Tomber 1989, p. 491, fig. 15, no. 491. 8 . Form variant B.I: Large, open form with strongly inward inclined, slightly convex lower wall ending at a rim forming a hook with a rounded lip on the exterior surface. Unfinished interior surface with some haphazard traces of smoothing. Smoothed lower exterior surface with pronounced self-slip. Turned upper exterior surface. Two shallow, but sharp-edged (cut) furrows in the middle of the exterior surface. A small hole (ø .25-.4 cm) of indeterminate function, was pierced at the leather-hard stage from interior to exterior through the vessel wall with a pointed tool. Grayish-black orlo annerito cover on the vessel rim. PEB 270; Acc. # 5745; Pl. 4, no. 8. Dimensions: sd 9,7 x 9,2 cm; ø 35 cm, rh 0,9 cm; rw 1,8 cm. Color: ec 2.5YR 5/7; ic 2.5YR 5.5/7; sc 2.5YR 5.5/8; oa 2.5YR 4/0. Condition: good; clean surfaces and fracture surfaces but slightly abraded fracture surface edges. Comparanda: cf. no. 9. 9 . Form variant B.I: Medium-sized, open form with the rim folded over and thickened on the exterior to form a hook-shaped protrusion with a rounded lip. Slightly convex vessel wall and domed vessel body. Haphazard smoothing marks on otherwise unfinished interior surface. Lower part of the exterior surface unfinished, upper section turned. Two shallow, but sharp-edged (cut) furrows in the middle of exterior surface. Distinct black orlo annerito cover on the vessel rim. PEA 105; Acc. # 3920; Pl. 4, no. 9. Dimensions: ø 31 cm; rh 1,0 cm; rw 1,6 cm; 7 joining sherds. Color: ec 2.5YR 6/7; ic 5YR 6/6; sc 5YR 6/6; pc 5YR 3/1. Condition: good; partial soot cover on the rim and lower exterior surface, clean fracture surface and sharp fracture surface edges. Comparanda: Aguarod Otal 1991, pp. 312-313, figs. 5253 (several examples); Dyson 1976, fig. 62 no. LS96; Fernández Sotelo 1994, p. 73, fig. 20, nos. 37-41; Fulford 1984b, p. 190, fig. 72, nos. 9.1-2; Hayes 1972, p. 202, fig. 35, nos. 182.2-3; Ostia I, tav. XII, no. 262; Ostia III, tav. XXVIII, no. 170; Peacock et al. 1989, p . 211, fig. 12, nos. 11-12; Tortorella 1981b, tav CV, nos. 3-4. 1 0 . Form variant B.I: Large, open form with strongly inward curving, slightly convex wall ending in a thickened rim. The top of the hooked rim has been cut

off. Completely turned exterior surface with smoothed lower section. Haphazard smoothing marks on the otherwise unfinished interior surface. Shallow furrow o n the interior surface adjacent to the junction with the rim. Rim covered both by thick layer of whitish scum and patches of soot, which also extend to the lower exterior surface. PEB 222; Acc. # 5748; Pl. 4, no. 10. Dimensions: sd 5,2 x 6,0 cm; ø 32 cm; rh 0,8 cm; rw 1,2 cm. Color: ec 2.5YR 5.5/6; ic 2.5YR 5.5/8; sc 2.5YR 5.5/8; oa 10YR 8/3. Condition: good; clean surfaces with sharp fracture surface edges. 1 1 . Form variant B.I: Large, open form with curving, slightly convex wall forming a dome-shaped body. Thickened rim turning to a hook with pointed tip on the exterior surface. Unfinished, but fairly smooth surfaces. Thick, yellowish-white scum cover on the vessel rim. Hardly visible patches of soot on the exterior surface. PEB 341; Acc. # 5752; Pl. 4, no. 11. Dimensions: 2 non-joining sherds, sds 5,1 x 4,3 cm, 6,2 x 3,9 cm; ø 34 cm; rh 1,3 cm; rw 1,4 cm. Color: ec 2.5YR 5/7; ic 2.5YR 5/7; sc 2.5YR 4.5/8; oa 10YR 7/3. Condition: good; clean surfaces with sharp fracture surface edges. Comparanda: Dore 1989, p. 141, fig. 39, no. 96.1119; Hayes 1976 p. 67, fig. 8, no. 25; Tomber 1989, p. 487, fig. 13, no. 278. 1 2 . Form variant B.II: Large, open form with steep straight wall joining with an everted rim and hanging lip. Shallow, curved step at the junction of the rim and the vessel interior. Raised tapering rim with pointed lip. Unfinished interior and lower exterior surface; smoothed upper exterior surface. The information concerning surface finishes is not clear due to complete soot cover, most of which is probably of post-depositional origin. Yellowish-white scum deposition on the vessel rim. PEA 186; Acc. # 5754; Pl. 4, no. 12. Dimensions: sd 6,8 x 4,9 cm; ø 39 cm, rh 0,9 cm; rw 1,4 cm. Color: sc 2.5YR 5.5/8. Condition: good in spite of dense soot cover; sharp fracture surface edges. Comparanda: Aquilué Abadías 1991, fig. 239, no. 2 ; Giannotta 1992, p. 58, fig. 3:4, no. 261 (?); Schuring 1988, p. 68, fig. 43, no. 175; Tortorella 1981b, tav CV, no. 1. 1 3 . Form variant B.III: Large, open form with slightly convex, domed body joining with a raised, hook-shaped rim with a hanging lip. Sharp protruding ridge at the junction of the rim and the interior surface. Turned and partly smoothed exterior surface with two shallow furrows in the middle of the body. Smoothed upper interior surface, but unfinished lower section. Grayishblack orlo annerito cover on the vessel rim. PEA 12; Acc. # 5756; Pl. 4, no. 13. Dimensions: sd 8,8 x 9,2 cm; ø 32 cm, rh 1,2 cm; rw 1,7 cm. Color: ec 2.5YR 5/7; ic 2.5YR 5.5/6; sc 2.5YR 5/7; oa 10YR 3/1. Condition: good; substantially clean surfaces with sharp fracture surface edges, faint traces of soot on the rim. Comparanda: Aquilué Abadías 1989, p. 191, fig. 88, no. 5.11; Aquilué Abadías 1991, fig. 239, no. 1; Fulford 1984b, p. 191, fig. 72, no. 9.3; Fulford 1994, p. 65, fig.

7

8

9

10

11

12

13

Plate 4. 7 North Tunisian lids: plain or thickened rim (A.IV); 8-13 North Tunisian lids: everted rim (8-11 B.I; 12 B.II; 13 B.III). Scale 1:2.

4.8, no. 12.4; Giannotta 1992, p. 58, fig. 3:4, no. 260; Nolla & Puertas 1988, p. 61, fig. 9, no. 2; Ostia IV, tav. XI, no. 61; Reynolds 1984, p. 503, fig. 18.9, no. 189; Sánchez Sánchez 1995, p. 266, fig. 11, no. 5; Tomber 1986, p. 54, fig. 4. 1 4 . Form variant B.IV: Very large, open form with convex lower wall ending in a large hemispherical rim rolled against the exterior surface. The interior wall joins with the rim in a sharp angle. Unfinished lower exterior surface with occasional smoothed stripes, the upper exterior surface first turned, then smoothed; unfinished interior surface. Grayish-brown orlo annerito cover o n the vessel rim. PEB 206; Acc. # 5751; Pl. 5., no. 14. Dimensions: sd 11,9 x 8,7 cm; ø 47 cm; rh 1,5; rw 1,8 cm. Color: oa from 2.5YR 4/0 to 2.5YR 4/2. Condition: very good; clean surfaces with sharp fracture surfaces. Comparanda: Aguarod Otal 1991, p. 312, fig. 52, no. 3 ; Aquilué Abadías 1991, fig. 238, no. 8; Gandolfi 1994, p . 150, fig. 12, no. 4; Hayes 1976, p. 64, fig. 7, no. XIII.5; p. 67, fig. 8, nos. 23 and 24; Ostia IV, tav. XI, no. 59; Sánchez Sánchez 1995, p. 271, fig. 12, no. 10; Tomber 1989, p. 487, fig. 13, nos. 263-264; Tortorella 1981b, tav CV, no. 5. 1 5 . Form variant C.I: Medium-sized, open form with fairly straight lower wall ending in a nearly rectangular, thickened rim – of which the most has consecutively been turned away – through diagonal transition on the exterior. Smoothed exterior surface with turned upper section, unfinished interior surface. A thin layer of whitish-gray scum topping a layer of blackish orlo annerito deposition. PEB 281; Acc. # 5801; Pl. 5, no. 15. Dimensions: sd 6,8 x 5,6 cm; ø 37 cm; rh 1,8 cm; rw 1,1 cm. Color: ec 10R 4.5/7; ic 10R 4.5/7; sc 10R 4/8; oa 7.5YR 7/2. Condition: good, clean surfaces with substantially sharp fracture surface edges and clean fracture surfaces Comparanda: Dore 1989, p. 153, fig. 41, no. 104.1291; Lund 1995, p. 617, fig. 22, ARS 182 variants 2-5; p . 619, fig. 24, Segermes form 28. 1 6 . Form variant C.III: Large, open form with fairly straight lower wall thickening to slightly angular rim. Turning with a shallow furrow on the exterior surface adjacent to the rim produced a protruding pointed lip. Traces of turning visible on the lower exterior surface, smoothed upper exterior surface and unfinished interior surface. A layer of whitish scum underneath a dense layer of soot on the rim. PEB 186; Acc. # 5743; Pl. 5, no. 16. Dimensions: sherd 8,7 x 6,5 cm; ø 34 cm; rh 1,0 cm; rw 1,3 cm. Color: ec 10R 5/6; ic 2.5YR 4/4; pc 10YR 7/4. Condition: good; clean surfaces and sharp fracture surface edges. Comparanda: Aquilué Abadías 1987, p. 186, fig. 50, no. 5; Peacock et al. 1990, p. 63, fig. 2, no. 25. 1 7 . Form variant C.IV: Large, open form with slightly convex lower wall joining with an almond-shaped rim i n a sharp angle. The uppermost part of the exterior surface turned, smoothed lower section and partly smoothed interior surface. Whitish-gray layer of scum on the vessel rim covers in part the blackish orlo annerito surface.

PEA 127; Acc. # 5744; Pl. 5, no. 17. Dimensions: sd 6,3 x 7,4 cm; ø 33 cm; rh 1,7; rw 1,1 cm. Color: ec 2.5YR 4.5/5; ic 2.5YR 4.5/5; sc 2.5YR 4/4; oa 10YR 4/1 and 10YR 8/2. Condition: good, clean surfaces excluding slightly dirty exterior surface, sharp fracture surface edges. Comparanda: Alonso de la Sierra Fernández 1995, p. 173, fig. 10, no. 72; Aquilué Abadías 1987, p. 25, fig. 6, no. 12; Dore 1989, p. 153, fig. 41, nos. 104.3709 and 104.1063 (?); Fulford 1994, p. 66, fig. 4.9, no. 16.2; Ostia I, tav. XII, no. 263; Peacock et al. 1990, p. 63, fig. 2, no. 26; p. 65, fig. 3, no. 15; Tortorella 1981b, tav CV, no. 9. 1 8 . Form variant D.I: Medium-sized, open form with straight, diagonal lower wall ending in a small rim with a protruding thickening on the interior surface forming a rounded lip. Unfinished surfaces with the vessel interior showing shallow wheel-ridging. Brownish orlo annerito cover on the vessel rim. PEA 124; Acc. # 5746; Pl. 5, no. 18. Dimensions: sd 6,2 x 4,5 cm; ø 21 cm; rh 0,6 cm; rw 0,9 cm. Color: ec 2.5YR 5.5/8; ic 2.5YR 5.5/8; sc 2.5YR 5/6; oa 10R 4.5/3.5. Condition: medium; all the surfaces covered by patches of post-depositional scum, dirty fracture surfaces with slightly worn fracture surface edges. Comparanda: Aguarod Otal 1991, p. 315, fig. 55, no. 5 ; Casas & Nolla 1993, p. 209, fig. 7, no. 11; Fulford 1984b, p. 196, fig. 75, no. 2.1; Fulford 1994, p. 66, fig. 4.9, no. 14.1; Mackensen 1993, p. 628, no 73.1; tab. 8 1 no 8; Oleson & Oleson 1987, fig. XI-87. K 87; Reynolds 1984, p. 503, fig. 18.9, no. 187. 1 9 . Form variant D.II: Medium-sized, open form with high dome-shaped upper wall turning smoothly to an everted horizontal rim with slightly angular lip. A shallow concavity on the underside the rim. Carefully smoothed vessel surfaces with large patches of self-slip. PEB 186; Acc. # 5753; Pl. 5, no. 19. Dimensions: sd 14,8 x 5,6 cm; ø 20 cm; rw 0,7 cm. Color: ec 2.5YR 5/7; ic 2.5YR 5/7; sc 2.5YR 5/8. Condition: good, clean surfaces with sharp fracture surface edges. Comparanda: Fulford 1984b, p. 188, fig. 71, no. 3.1; Luni II, fig. 259, nos. 3 and 4. 2 0 . Form variant D.II: Medium-sized, open form with slightly convex lower wall turning to a short everted rim with angular lip. Shallow concavity underside the rim. Turned upper exterior surface; the lower exterior and interior show vague traces of smoothed stripes o n otherwise unfinished surfaces. Blackish-gray orlo annerito deposition covers the rim. PEB 342; Acc. # 5800; Pl. 6, no. 20. Dimensions: sd 6,1 x 5,7 cm; ø 30 cm; rh 1,3 cm; rw 0,9 cm. Color: ec 2.5YR 4.5/2; ic 2.5YR 4.5/2; sc 5YR 4/3; oa 7.5R 3.5/0. Condition: medium; dirty surfaces with occasional patches of soot on exterior with slightly worn edges, dirty fracture surfaces. Comparanda: Fulford 1984b, p. 188, fig. 71, nos. 1.1-2; Hayes 1976, p. 71, fig. 10, no. 17; Tomber 1989, p . 489, fig. 14, no. 298. 2 1 . Form variant D.III: Small, open form with domed, convex wall ending in a thinned rim with rounded lip. Smoothed vessel surfaces with patches of soot, probably

14

15

16

17

18

19

Plate 5. 14 North Tunisian lids: everted rim (B.IV); 15-17 Central Tunisian lids (15 C.I; 16 C.III; 17 C.IV); 18-19 Miscellaneous African cookware lids (18 D.I; 19 D.II). Scale 1:2.

of post-depositional origin, visible on both sides. PEB 309; Acc. # 5773; Pl. 6, no. 21. Dimensions: sd 5,3 x 4,2 cm; ø 17 cm; rw 0,6 cm. Color: ec 2.5YR 5.5/6; ic 2.5YR 5.5/6; sc 2.5YR 5/6. Condition: medium, fairly clean surfaces with slightly worn fracture surface edges. 2 2 . Form variant D.IV: Medium-sized, open form with straight lower wall ending in a slightly downward everted, diagonal rim with pointed lip. Unfinished vessel surfaces, with blackish orlo annerito deposition topped by thin whitish scum covering the rim and the whole interior surface. Patches of soot on the exterior surface. PEA 93; Acc. # 5749; Pl. 6, no. 22. Dimensions: sd 6,4 x 3,0 cm; ø 30 cm; rw 0,7 cm Color: ec 2.5YR 5.5/6; ic 7.5YR 4/0; sc 10R 6/8; oa 7.5YR 4/0. Condition: fairly good; clean surfaces and substantially sharp fracture surface edges. 2 3 . Form variant E.I: Small knob with a hole pierced pre-cocturam (before firing) through the top. Nearly complete component of a lid, cylindrical in shape that widens out a little toward the top of the vessel. Smoothed and slipped exterior surface, unfinished interior surface. PEA 62; Acc. # 5810; Pl. 6, no. 23. Dimensions: sd 2,4 x 2,1 m; ø 2,4 cm, hole ø 0,7 cm. Color: es 10R 5/8, ic not determined, sc 2.5YR 5.5/8. Condition: medium, the topmost part of the knob has flaked off in several places, but the vessel surfaces are fairly clean and fracture surface edges sharp. 2 4 . Form variant E.II: Small, complete disk knob with slightly concave top and sharp edges. Knob narrows down slightly towards the junction with the vessel wall. Smoothed, turned and slipped exterior surface, smoothed and slipped interior surface. PEA 33; Acc. # 5739; Pl. 6, no. 24. Dimensions: sd 2,2 x 2,2 cm; ø 2,2 cm, knob height 0,9 cm. Color: ec 2.5YR 5.5/8, ic 2.5YR 6/8, sc 2.5YR 6/6. Condition: good, clean surfaces with sharp fracture surface edges. Comparanda: cf. no. 25. 2 5 . Form variant E.II: Small, tall disk knob hollowed from the interior. Flat top with fairly sharp edges and wall tilting slightly inwards toward the junction with the vessel wall. Marks on the top indicate that the vessel was first thrown upside down and cut from a wheel with a wire, then the knob was turned and both surfaces smoothed. PEA 105; Acc. # 5811; Pl. 6, no. 25. Dimensions: sd 3,3 x 1,7 cm; ø 3,0 cm; knob height 1,1 cm. Color: ec 10R 5.5/8; ic 10R 5.5/8; sc 2.5YR 5.5/7. Condition: fairly good, clean surfaces, but slightly worn fracture surface edges. Comparanda: Aguarod Otal 1991, p. 305, fig. 45, nos. 12; Hayes 1976, p. 57, fig. 4, no. VII.60; Tortorella 1981b, tav XIV, nos. 15-16. 2 6 . Form variant F.I: Medium-sized, open form with a shallow protrusion on a convex vessel top tilted towards the center. Exterior surface turned on both sides of the ring handle, unfinished interior surface. PEA 16; Acc. # 5736; Pl. 6, no. 26. Dimensions: sd 6,4 x 6,3 cm; ø 8 cm, height 0,4 cm, width 0,7 cm. Color: ec 2.5YR 6/7; ic 2.5YR 6/6; sc 2.5YR 6/6.

Condition: fairly good, patches of whitish scum deposited on both surfaces, clean but worn fracture surface edges. Comparanda: Bost et al. 1992, p. 189, fig. 47, no. 16; Casas & Nolla 1993, p. 209, fig. 7, nos. 10-11; Fernández Sotelo 1994, p. 86, fig. 33, nos. 1-2 & 6 ; González Villaescusa 1990, fig. 19, no. 51; Hayes 1976, p. 67, fig. 8, no. 29; Luni II, fig. 120, no. 5; Oleson & Oleson 1987, fig. XI-89. K 89; Schuring 1988, p. 67, fig. 42, no. 160. 2 7 . Form variant F.II: Medium-sized, open form preserving part of a ring handle and the upper wall. Convex wall bearing a shallow protrusion with rectangular profile and flat lip. Upper exterior surface turned; unfinished lower exterior and interior surfaces. PEA 104; Acc. # 5735; Pl. 6, no. 27. Dimensions: sd 10,3 x 10,5 cm; ø 6 cm; height 0,4 cm; width 0,6 cm. Color: ec 2.5YR 5.5/8; ic 2.5YR 5.5/8; sc 2.5YR 5/6. Condition: good; clean surfaces with sharp fracture surface edges. Comparanda: cf. no. 28. 2 8 . Form variant F.II: Medium-sized, open form preserving part of a ring handle and the upper wall. Convex wall bearing a shallow protrusion with rectangular profile and flat lip. Upper exterior surface turned; unfinished lower exterior and interior surfaces. A possible graffito on the exterior surface inside the ring handle, executed post-cocturam and composed of two scratches that seem to form the Greek letter lambda. PEA 135; Acc. # 5741; Pl. 6, no. 28. Dimensions: sd 4,5 x 7,4 cm; ø 7 cm; height 0,5 cm; width 0,6 cm. Color: ec 2.5YR 5.5/8; ic 2.5YR 5.5/8; sc 2.5YR 5/8. Condition: fairly good, patches of whitish scum deposited on surfaces, but sharp fracture surfaces. Comparanda: Bost et al. 1992, p. 189, fig. 47, nos. 1415; Casas & Nolla 1993, p. 209, fig. 7, nos. 6-9; Dyson 1976, fig. 62 no. LS97; Fernández Sotelo 1994, p. 86, fig. 33 nos. 4-5 & 12; p. 87, fig. 34, nos. 16-17, 22-23; González Villaescusa 1990, fig. 19, nos. 52-53; Hayes 1978, p. 31, fig. 3, no. 28; Pontacolone & Incitti, p . 564, fig. 2, nos. 2-3; Schuring 1988, p. 67, fig. 42, nos. 157-159; Tomber 1989, p. 489, fig. 13, no. 260. 2 9 . Form variant F.III: Medium-sized, open form with convex upper wall bearing a small protruding ring-handle with rounded lip. Turned exterior surface; unfinished interior surface. The topmost section of the protrusion i s covered with a blackish patina cenerognola deposition. PEA 124; Acc. # 5733; Pl. 6, no. 29. Dimensions: sd 4,6 x 5,5 cm; ø 6 cm; height 0,5 cm, width 0,5 cm. Color: ec 2.5YR 6/7; ic 2.5YR 6/7; sc 5.5/6; oa 5YR 4/1. Condition: good; clean surfaces and substantially sharp fracture surface edges. Comparanda: Casas & Nolla 1993, p. 209, fig. 7, no. 5 ; Fernández Sotelo 1994, p. 87, fig. 34, no. 19; Oleson & Oleson 1987, fig. XI-88. K 88; Schuring 1988, p. 67, fig. 42, no. 161. 3 0 . Form variant F.IV: Small, open form with flat, slightly hollowed lid top turning smoothly to a convex upper wall. Turned exterior surface paired with unfinished interior surface. PEA 124; Acc. # 5733; Pl. 6, no. 30. Dimensions: sd 4,2 x 3,6 cm; ø 8 cm; height 0,2 cm.

20

21

22

23

24

26

25

27

29

28

30

Plate 6. 20-22 Miscellaneous African cookware lids (20 D.II; 21 D.III; 22 D.IV); 23-25 Lid knobs (23 E.I; 24-25 E.II); 26-30 Ring-handles (26 F.I; 27-28 F.II; 29 F.III; 30 F.IV). Scale 1:2.

Color: ec 2.5YR 5.5/7; ic 2.5YR 5.5/7; sc 2.5YR 5.5/7. Condition: medium; patches of whitish scum deposited on both surfaces, worn fracture surface edges. 3 1 . Form G: Large, open form with a wide flat base turning to a shallow wall with convex profile. Slightly thinned rim with pointed tip. Turned exterior surface with overlapping bands of burnishing and patina cenerognola deposition underneath the rim, smoothed or possibly slipped interior surface. A notch on the interior surface at the junction of the base and wall with two other concentric furrows closer to the center of the vessel. PEA 124; Acc. # 5726; Pl. 7, no. 31. Dimensions: sd 8,5 x 7,1 cm; ø 31 cm; rw 0,6 cm; vh 4,7 cm; ph 3,1 cm. Color: ec 2.5YR 5.5/8; ic 2.5YR 5/6; sc 2.5YR 5/7; pc 10YR 4/1. Condition: good; in spite of patchy sooting on the exterior base and lower wall, and traces of use-wear on the interior surface, the vessel has clean surfaces and sharp fracture surface edges. Comparanda: cf. no. 32. 3 2 . Form G: Medium-sized, open form with a wide flat base turning to a shallow convex wall with thinned rim and pointed lip. Turned exterior surface with overlapping bands of smoothing and patina cenerognola deposition underneath the rim, slipped interior surface. A notch o n the interior surface at the junction of base and wall. PEA 124; Acc. # 5727; Pl. 7, no. 32. Dimensions: sd 8,1 x 7,7 cm; ø 28 cm; rw 0,6 cm; vht 4,2 cm, ph 2,1 cm. Color: ec 2.5YR 4.5/4; ic 2.5YR 3.5/4, sc 2.5YR 4/6; pc 2.5YR 4/0. Condition: good, clean and sharp fracture surfaces; small patches of soot on the exterior base and lower wall. Comparanda: Aguarod Otal 1991, pp. 316-317, figs. 5657 (several examples); Alonso de la Sierra Fernández 1995, p. 173, fig. 10, nos. 73-75 (North Tunisian); Biondani 1992, p. 177, fig. 10, no. 2; Castañer Masoliver et al. 1990, p. 165, fig. 10, no. 5; Fernández Sotelo 1994, p. 68, fig. 15, no. 8; Gandolfi 1994, p . 150, fig. 12, no. 5; Giannotta 1992, p. 54, fig. 3:3, no. 258; Hayes 1972, p. 202, fig. 35, nos. 2, 12 and 13; Sánchez Sánchez 1995, p. 271, fig. 12, no. 11; Tortorella 1981b, tav CVI, nos. 3-4. 3 3 . Form variant H.I: Medium-sized, open form with straight, slightly outward inclined upper wall ending in a thinned rim with pointed lip. Smoothed exterior surface with wide band of patina cenerognola cover on the underside of the rim, smoothed interior surface with occasional patches of self-slip. PEA 33; Acc. # 5730; Pl. 7, no. 33. Dimensions: 2 sherds; ø 22 cm; rw 0,4 cm; wh 2,6 cm; ph 2,5 cm. Color: ec 5YR 5.5/6; ic 10R 5/6; sc 2.5YR 6/8; pc 5YR 4/1. Condition: medium; patches of calcareous scum deposited on otherwise clean surfaces, slightly worn fracture surface edges. Comparanda: Aguarod Otal 1991, p. 316, fig. 56, no. 4 ; Aquilué Abadías 1989, p. 193, fig. 89, no. 5.21; González Villaescusa 1990, fig. 24, no. 71; Hayes 1976, p. 53, fig. 2, no. 12; Hayes 1978, p. 57, fig. 14, no. 34; Macias et al. 1997, p. 169, fig. 5, no. 5; Ostia IV, fig. 1 ; Rakob 1991, p. 86, fig. 18, no 14; Schuring 1988, p . 67, fig. 42, no. 138.

3 4 . Form variant H.II: Large, open form with slightly sagging base joining with the upper wall through a curved transition. Nearly vertical upper wall with thinned rim and rounded lip. Smoothed interior surface with selfslipped areas and a shallow furrow at the junction of the base and wall. Two deeper, concentric furrows on the interior surface of the base. Turned exterior base, smoothed exterior wall and rim. Black patina cenerognola cover on the exterior surface below the rim. PEA 105; Acc. # 3921; Pl. 7, no. 34. Dimensions: sd 12,1 x 9,3 cm; ø 30 cm; vh 4,7 cm; rw 0,5 cm; ph 3,5 cm. Color: ec 2.5YR 5/7; ic 10R 5.5/8; sc 2.5YR 5.5/8; pc 2.5Y 3.5/0. Condition: good, patches of whitish scum deposited o n the interior surface, traces of soot on the exterior bottom, clean and sharp fracture surface edges. Comparanda: cf. no. 35. 3 5 . Form variant H.II: Large, open form with flat bottom turning to a slightly outward inclined upper wall. Slightly thinned rim with pointed lip. A shallow groove on the interior surface of the base. Turned exterior surface; smoothed and self-slipped (or slipped?) interior surface. A band of patina cenerognola covering the upper section of the exterior surface. PEA 20; Acc. # 1044; Pl. 7, no. 35. Dimensions: ø 37 cm; vh 5,9 cm; rw 0,6 cm; ph 3,4 cm. Color: ec lower zone 2.5YR 5.5/8; transition to pc 5YR 6/8; ic 2.5YR 7/8; sc 5YR 6.5/8; self-slip 2.5YR 5/6; pc 5YR 4/3. Condition: medium; although the sherd preserves a complete vessel profile, it also has worn fracture surface edges, chipped exterior surface and only partly preserved surfacing on the interior. Comparanda: Aguarod Otal 1991, pp. 316-317, figs. 5657 (several examples); González Villaescusa 1990, fig. 24, nos. 69-70; Hayes 1976, p. 67, fig. 8, no. 20; Ostia I, tav. I; Rakob 1991, p. 86, fig. 18, no. 13; p. 200, fig. 41, no. 2; Saguì 1979, p. 271, fig. 13; Tomber 1986, p . 51, fig. 1, nos. 26-27; Tomber 1989, p. 447, fig. 1, nos. 8-11. 3 6 . Form variant H.III: Large, open form with slightly outward inclined, tall upper wall ending in an even rim with rounded lip. Curved transition from the wall to a flat, slightly sagging base. The bottom of the base as well as part of the lower exterior wall were first turned, then smoothed. Smoothed upper exterior surface with burnished bands and an uneven band of patina cenerognola cover. Smoothed interior surface with patches of self-slip. PEB 209; Acc. # 5729; Pl. 7, no. 36. Dimensions: sd 4,9 x 7,5 cm; ø 31 cm; rw 0,7 cm; wh 5,3 cm; pc 4,4 cm. Color: ec 2.5YR 5/7; ic 2.5YR 5/6; sc 2.5YR 5.5/8; pc 5YR 3/1 - 2.5YR 5/5. Condition: good; clean surfaces and sharp fracture surface edges, patches of soot deposited on the vessel bottom and lower exterior surface. Comparanda: Fulford 1984b, p. 188, fig. 71, no. 6.3; Hayes 1976, p. 57, fig. 4, no. 3; Lund 1995, p. 616, fig. 21, ARS 26 late version; Neuru 1980, fig. 4, nos. 19 and 20; Sánchez Sánchez 1995, p. 271, fig. 12, no. 12; Therrien 1981, p. 146, nos. 43-44; Tomber 1989, p . 447, fig. 1, nos. 12-13; Tortorella 1981b, tav CVI, no. 5.

31

32

33

34

35

36 Plate 7. 31-32 Hayes 181 pan (G); 33-36 Hayes 26/181 pan-casserole (33 H.I; 34-35 H.II; 36 H.III). Scale 1:2.

3 7 . Form variant I.I: Medium-sized, open form with outward inclined, slightly convex upper wall ending in a plain, thickened rim with rounded lip. Sharp-edged protrusion on the exterior surface at the junction of the wall and base. Sagging base with corrugated exterior surface, otherwise smoothed surfaces with occasional patches of self-slip on the interior of the vessel. Both the exterior base and protrusion were turned before smoothing. Sharp furrow at the junction of the wall and base on the interior surface. PEB 222; Acc. # 5724; Pl. 8, no. 37. Dimensions: sd 6,9 x 5,0 cm; ø 21 cm; rw 0,7 cm; wh 3,8 cm. Color: ec 2.5R 5/7; ic 2.5R 5/7; sc 2.5R 5/6. Condition: good; clean surfaces, sharp fracture surface edges. Comparanda: Aguarod Otal 1991, pp. 326-327, figs. 6667 (several examples); Aquilué Abadías 1985, p. 215, fig. 3, nos. 9-13; p. 217, fig. 5, no. 1; Coletti & Pavolini 1996, p. 408, fig. 9, no. 8; Gandolfi 1994, p . 150, fig. 12, no. 7; Hayes 1972, p. 46, fig. 7, no. 11; Sagui 1980, no. 139; Tortorella 1981b, tav CVI, nos. 12-13. 3 8 . Form variant I.II: Medium-sized, open form with outward inclined, fairly straight upper wall ending in a plain thinned rim with pointed lip. Triangular protrusion with sharp edges at the junction of wall and base. Sagging base with turned and corrugated exterior surface. Otherwise smoothed surfaces with patches of self-slip o n the vessel interior. Shallow furrow at the junction of wall and base on the interior surface. PEA 20; Acc. # 1150; Pl. 8, no. 38. Dimensions: sd 10,1 x 7,8 cm; ø 22 cm; rw 0,5 cm. Color: ec 10R 4.5/8; ic 10R 5/8; sc 10R 5/8. Condition: medium; all the surfaces covered with patches of calcareous scum and fairly thick layer of soot, the origin of which is in all likelihood post-depositional, as it extends over fracture surfaces. Comparanda: Giannotta 1992, p. 54, fig. 3:3, no. 254; Hayes 1972, p. 46, fig. 7, no. 1; Luni II, fig. 114, no. 6 ; fig. 250, no. 4; Ostia II, tav. XVIII, no. 309; Reynolds 1984, p. 495, fig. 18.5, no. 121; Tomber 1986, p. 56, fig. 6, nos. 163. 3 9 . Form variant I.III: Medium-sized, open form with strongly outward inclined upper wall ending in a thinned rim with rounded lip. Corrugated exterior surface with a shallow concavity on the lower side, possibly anticipating a protrusion. Smoothed surfaces with burnished matt bands on the exterior. Faint traces of soot (possibly of post-depositional origin) on the interior surface. PEB 222; Acc. # 5771; Pl. 8, no. 39. Dimensions: sd 4,9 x 5,2 cm; ø 29 cm; rw 0,6 cm. Color: ec 2.5YR 4.5/6; ic 2.5YR 5/6; sc 2.5YR 5/6. Condition: good; substantially clean surfaces with sharp fracture surface edges. Comparanda: López Mullor et al. 1997, p. 71, fig. 1, no. 7. 40. Form variant J.I: Medium-sized, open form. Slightly sagging bottom with corrugated exterior surface. Shallow, angular protrusion on the exterior at the junction of the wall and the bottom. Slightly convex, outward inclined upper wall ending in an undercut rim with rounded lip thickened on the interior. Turned lower exterior wall and exterior bottom, smoothed upper exterior wall with occasional burnished bands and a band

of brown patina cenerognola. Slipped or self-slipped interior surface. The latter alternative seems more probable, as the interior surface at the junction of the wall and bottom is matt and bears a shallow groove. PEA 93; Acc. # 5720; Pl. 8, no. 40. Dimensions: sd 9,9 x 6,2 cm; ø 25 cm; rh 0,7 cm; rw 0,7; wh 5,1 cm. Color: ec 2.5YR 5/6; ic 10R 5/8; sc 5YR 5/5; pc 2.5YR 4/2. Condition: good; clean surfaces and sharp fracture surface edges; most of the exterior surface covered with calcareous scum. Comparanda: Aguarod Otal 1991, pp. 322-325, figs. 6265 (numerous examples); Aquilué Abadías 1987, passim.; Berato et al. 1986, p. 154, fig. 23, no. 15; Dyson 1976, fig. 57, no. LS34; Gandolfi 1994, p. 150, fig. 12, no. 6 ; Hayes 1972, p. 46, fig. 7, nos. 21 & 24; Michelucci 1985, p. 136, fig. 12, nos. 310 & 756; p. 137, fig. 13, nos. 167 & 912; Nolla & Puertas 1988, p. 57, fig. 8 no. 14; Tortorella 1981b, tav CVI, no. 10. 4 1 . Form variant J.II: Medium-sized, open form with outward inclined, slightly convex upper wall separated from slightly sagging bottom by shallow, angular protrusion on the exterior surface. Undercut rim with pointed lip thickened on the interior of the vessel. Turned lower exterior surface and bottom, smoothed upper exterior surface. A band of brownish patina cenerognola cover on the upper exterior surface and concentric shallow grooves on the exterior bottom. The slip on the interior surface has partly flaked off. PEA 105; Acc. # 3918; Pl. 8, no. 41. North Tunisian fabric. Dimensions: sd 9,7 x 11,2 cm; ø 28 cm; rh 1,3 cm; rw 0,8 cm. Color: ec 2.5YR 5/6; ic 10R 5/8; sc 2.5YR 5/8; pc 5YR 4/2. Condition: medium; sooted vessel bottom and lower exterior surface; worn fracture surface edges; thick whitish calcareous (?) scum cover, possibly of postdepositional origin, on the exterior surface Comparanda: cf. no. 42. 4 2 . Form variant J.II: Medium-sized, open form with outward inclined, slightly convex upper wall ending i n an undercut rim with pointed lip thickened on the vessel interior. Slightly sagging bottom separated from the upper wall by a conical protrusion on the exterior surface. The protrusion and the preserved part of the exterior bottom turned, smoothed upper exterior wall and interior surface showing occasional patches of self-slip. The upper exterior surface preserves traces of both patina cenerognola and salt scum, which have been worn off for the most part. PEA 20; Acc. # 1046; Pl. 8, no. 42. Central Tunisian fabric. Dimensions: sd 10,1 x 7,5 cm; ø 29 cm; rh 2,0 cm; rw 0,8 cm; wh 7,1 cm. Color: ec 10R 4/4; ic 10R 4.5/6; sc 10R 4.5/6; pc 5Y 8/2. Condition: fairly good; clean fracture surfaces and sharp fracture surface edges, patches of soot on the exterior surface near the protrusion. Comparanda: Aguarod Otal 1991, p. 322, fig. 62, nos. 1 & 6; p. 323, fig. 63, no. 2; Casas & Nolla 1993, p. 208, fig. 5, nos. 1-6; Coletti & Pavolini, p. 408, fig. 9, no. 7 ; Dyson 1976, fig. 57 no. LS33; González Villaescusa 1990, figs. 11-14, nos. 20-37; Hayes 1972, p. 46, fig. 7, no. 23; Villedieu 1984, p. 298, fig. 28.

37

38

39

40

41

42 Plate 8. 37-39 Hayes 23A shallow casserole (37 I.I; 38 I.II; 39 I.III); 40-42 Hayes 23B shallow casserole (40 J.I; 41-42 J.II). Scale 1:2.

4 3 . Form variant J.III: Medium-sized, open form with outward inclined upper wall separated from the slightly sagging bottom by an angular protrusion on the exterior surface. Thickened rim with pointed lip. Turned lower exterior surface, the latter of which also preserves faint concentric corrugations on the exterior. Occasional patches of self-slip on smoothed upper parts of the vessel, especially on the interior. A band of yellowishbrown patina cenerognola deposition underneath the rim on the exterior surface. PEB 309; Acc. # 5722; Pl. 9, no. 43. Dimensions: sd 9,6 x 7,6 cm; ø 29 cm; rh 1,7 cm; rw 0,8 cm; wh 7,0 cm; ph 2.8 cm. Color: ec 10R 4.5/6; ic 10R 5/6; sc 2.5YR 4/8; pc 7.5YR 5/4. Condition: poor, all the surfaces covered by a thick layer of calcareous scum, worn fracture surface edges, patches of soot both on the lower exterior surface and the exterior bottom. 4 4 . Form variant J.IV: Medium-sized, open form with outward inclined upper wall ending in an undercut rim thickened on the interior surface. Rim has a pointed lip and a shallow groove at the center of the interior surface. Angular protrusion on the exterior surface separates the upper wall from slightly sagging bottom. Turned protrusion and vessel bottom, smoothed exterior wall with two burnished bands in the middle, blackish-gray patina cenerognola cover on the upper exterior surface. Smoothed interior surface with random patches of selfslip. Shallow groove on the interior surface at the junction of the wall and the bottom. PEA 105; Acc. # 5721; Pl. 9, no. 44. Dimensions: 3 sherds; ø 27 cm; rh 1,8 cm; rw 0,8 cm; wh 8,0 cm. Color: ec 2.5YR 5/6; ic 10R 5/6; sc 2.5YR 4/4; pc 10YR 4/1. Condition: very good; sharp surfaces with clean fracture surface edges. Comparanda: Hayes 1972, p. 46, fig. 7, no. 25; Schuring 1988, p. 66, fig. 41, nos. 126-127; Tortorella 1981b, tav CVI, no. 11. 4 5 . Form variant J.V: Medium-sized, open form with outward inclined, slightly convex upper wall ending i n an undercut rim thickened on the interior. Rim with a pointed lip and three shallow grooves on the interior surface. Smoothed exterior surface with occasional matt

burnished bands and blackish-gray patina cenerognola cover under the rim. Smoothed interior surface. PEA 33; Acc. # 5723; Pl. 9, no. 45. Dimensions: sd 7,5 x 5,3 cm; ø 29 cm; rh 2,2 cm; rw 0,8 cm; ph 3,1 cm. Color: ec 10YR 5/7; ic 10R 5/6, sc 2.5YR 4/8; pc 2.5YR 4/0. Condition: good; clean surfaces but slightly worn fracture surface edges. 4 6 . Form K: Medium-sized, open form with vertical upper wall ending in a tall thickened rim with diagonal lip and a rounded ledge forming a profiled lid seating o n the interior surface. Smoothed surfaces with sparse burnished stripes and a band of brownish-gray patina cenerognola cover on the exterior surface. PEA 34; Acc. # 5768; Pl. 9, no. 46. Dimensions: sd 4,8 x 3,8 cm; ø 23 cm; rh 2,5 cm; rw 0,9 cm. Color: ic 2.5YR 5.5/6; sc 2.5YR 5.5/6; pc 5YR 4/1.5. Condition: medium; clean surfaces, but worn fracture surface edges. Comparanda: Aguarod Otal 1991, p. 321, fig. 61, nos. 34; Aquilué Abadías 1985, p. 214, fig. 2, nos. 3-4; Dyson 1976, fig. 47, no. 22II-49; Freed 1998, p. 58, fig. 7, no. 123; Fulford 1994, p. 59, fig. 4.4, no. 3.5; Hayes 1972, p. 206, fig. 36, no. 194.1; Ostia II, tav. XVIII, nos. 303304; Rivet 1996, p. 336, fig. 9, no. 7; Tomber 1986, p . 53, fig. 3, no. 81; Tortorella 1981b, tav CVI, no. 8. 4 7 . Form variant L.I: Medium-sized, open form with high convex wall ending in a slightly thickened rim with pointed lip. Carination at the junction of the upper wall and slightly sagging bottom. Smoothed exterior surface with occasional bands of matt burnishing, turned base. Smoothed interior surface with patches of self-slip. Most of the exterior surface covered by a deep black patina cenerognola deposition and a barely visible layer of whitish scum. PEA 96; Acc. # 5732; Pl. 9, no. 47. Dimensions: sd 10,9 x 7,7 cm; ø 17 cm; rw 0,7 cm; wh 7,5 cm, ph 7,0 cm. Color: ec 5YR 5/6; ic 2.5YR 5/7; sc 2.5YR 5.5/8; pc 10YR 4.5/1. Condition: very good; clean surfaces with sharp fracture surface edges. Comparanda: cf. no. 49.

43

44

45

46

47

Plate 9. 43-45 Hayes 23B shallow casserole (43 J.III; 44 J.IV; 45 J.V); 46 Hayes 194 deep casserole (K); 47 Hayes 193 deep casserole (L.I). Scale 1:2.

4 8 . Form variant L.I: Medium-sized, open form with high convex wall ending in a slightly thickened rim with pointed lip. Carinated junction separates the upper wall from slightly sagging bottom. Smoothed exterior surface covered by a layer of blackish-gray patina cenerognola deposition. Unfinished interior surface. PEA 105; Acc. # 5812; Pl. 10, no. 48. Dimensions: sds 5,3 x 6,5 cm & 3,9 x 6,2 cm; ø 19 cm; rw 0,6 cm; wh 6,3 cm; ph 6,1 cm. Color: ic 2.5YR 6/7; sc 2.5YR 5.5/7; pc 2.5 YR 4/0 5YR 4/1. Condition: good; substantially clean surfaces with sharp fracture surface edges. Comparanda: Aguarod Otal 1991, p. 345, fig. 85, nos. 12; Bost et al. 1992, p. 188, fig. 46, nos. 9-11; Gandolfi 1994, p. 150, fig. 12, no. 9; Hayes 1972, p. 205, fig. 36, no. 193.1; Luni II, fig. 124, no. 10; fig. 261, nos. 45; Ostia I, tav. XII, no. 273; Sagui 1980, no. 144; Tomber 1986, p. 53, fig. 3, no. 80; Tortorella 1981b, tav CVIII, no. 11. 4 9 . Form variant L.II: Medium-sized, open form with slightly convex, high wall ending in a rim with rounded lip. Shallow but wide furrow on the exterior surface right above the junction of the wall and base. Sagging bottom with turned exterior surface. Smoothed exterior wall covered by dense grayish-brown band of patina cenerognola, unfinished interior surface. Patches of soot both on the exterior wall and bottom. PEA 33; Acc. # 5731; Pl. 10, no. 49. Dimensions: sd 7,4 x 5,6 cm; ø 16 cm; wh 5,2 cm; rw 0,5 cm. Color: ic 2.5YR 6/6; sc 5YR 6/6; pc 5YR 3.5/2. Condition: medium; fairly clean surfaces with sharp fracture surface edges, a band of small subrounded chips worn off from the middle of the exterior surface, probably as a result of the vessel use. Comparanda: Bost et al. 1992, p. 188, fig. 46, no. 12; Tomber 1989, p. 491, fig. 15, no. 302. 5 0 . Form variant L.III: Medium-sized, open form with inward inclined straight wall turning to a slightly thinned rim with rounded lip. Smoothed bands on the exterior surface, completely smoothed interior surface. Dense soot cover on the exterior surface underneath the rim and patches of soot cover also the unsmoothed parts of the exterior surface.

PEB 271; Acc. # 5761; Pl. 10, no. 50. Dimensions: sd 3,6 x 4,2 cm; ø 23 cm, rw 0,8 cm. Color: ec 2.5YR 4.5/4; ic 5YR 5/5; sc 5YR 6/4. Condition: fairly good; clean surfaces with slightly worn fracture surface edges. 5 1 . Form variant M.I: Medium-sized, open form with slightly inward inclined upper wall ending in an undercut rim thickened on the exterior and a notched lid-seating on the rim top. Smoothed surfaces with a layer of blackish-brown patina cenerognola deposition on the exterior. PEA 33; Acc. # 5706; Pl. 10, no. 51. Dimensions: sd 6,3 x 2,3 cm; ø 22 cm; rh 1,1 cm; rw 1,2 cm. Color: ic 2.5YR 6/6; sc 2.5YR 5/6; pc 5YR 4/1. Condition: good; fairly clean surfaces with sharp fracture surface edges. Comparanda: cf. no. 52. 5 2 . Form variant M.I: Medium-sized, open form with deep vertical wall separated from sagging bottom by an angular joint. Protruding rim with a grooved lid seating on the top and a notch on the underside separating it from the exterior wall. Smoothed exterior surface with sparse burnished patches, unfinished interior surface with clearly visible wheel-turning marks. The dense blackishgray patina cenerognola cover on the exterior surface extends to the upper section of the vessel bottom. PEA 150; Acc. # 5183; Pl. 10, no. 52. Dimensions: sd 12,1 x 9,2 cm; ø 32 cm; rh 1,1 cm; rw 1,3 cm; wh 7,1 cm; ph 7,5 cm. Color: ec 2.5YR 5.5/8; ic 2.5YR 6/6; sc 2.5YR 6/7; pc 7.5R 4/0 to 2.5YR 4/0. Condition: good, clean surfaces and sharp fracture surface edges. Comparanda: Aguarod Otal 1991, p. 334, fig. 74, nos. 23, 5; Aquilué Abadías 1987, p. 25, fig. 6, nos. 10-11, 14; Aquilué Abadías 1991, fig. 241, no. 1; Hayes 1976, p . 65, fig. 7, no. XIII.3; p. 73, fig. 11, nos. 5-6; Ostia III, tav. XLII, nos. 324 & 327; Ostia IV, tav. X, no. 55; Tomber 1986, p. 53, fig. 3, no. 86; Tortorella 1981b, tav. CVII no. 5.

48

49

50

51

52

Plate 10. 48-50 Hayes 193 deep casserole (48 L.I; 49 L.II; 50 L.III); 51-52 Proto Hayes 197 deep casserole (51-52 M.I). Scale 1:2.

5 3 . Form variant M.II: Medium-sized, open form with slightly inward inclined upper wall ending in a short flanged rim with a shallow notch on the top forming the lid seating. Smoothed vessel surfaces with blackishbrown patina cenerognola cover on the exterior surface. PEA 12; Acc. # 5712; Pl. 11, no. 53. Dimensions: sd 2,5 x 3,1 cm; ø 19 cm; rh 0,7 cm; rw 1,3 cm. Color: ic 7.5YR 6/6; sc 2.5YR 6/6; pc 5YR 3/1. Condition: medium, scum deposition on fracture surfaces, slightly worn fracture surface edges. Comparanda: cf. no. 54. 5 4 . Form variant M.II: Medium-sized, open form with vertical, slightly inward inclined upper wall separated from a sagging bottom by an angular joint. Short flanged rim with a shallow depression on the rim top forming a lid seating. Lip with angular upper and rounded lower edge. Smoothed exterior surface, unfinished interior surface. Thick patina cenerognola cover on the upper wall, patches of soot on the exterior bottom. PEA 160; Acc. # 5804; Pl. 11, no. 54. Dimensions: sd 10,8 x 5,2 cm; ø 23 cm; rh 0,6 cm; rw 1,4 cm; wh 4,6 cm. Color: ec 2.5YR 6/0 & 2.5YR 3/0; ic 2.5YR 5.5/7; sc 2.5YR 5.5/8. Condition: medium, patches of calcareous incrustation on the upper exterior surface, substantially thick incrustation of uncertain origin on the exterior surface. Sharp fracture surface edges, but incrustated fracture surfaces. Comparanda: Aguarod Otal 1991, p. 334, fig. 74, nos. 4 & 6; Aquilué Abadías 1987, p. 37, fig. 10, no. 10; p. 62, fig. 19, no. 19; p. 196, fig. 54, no. 3; Dore 1988, p. 81, fig. 17, no. N5; Fulford 1994, p. 61, fig. 4.5, nos. 10.45; Luni II, fig. 259, no. 9; Ostia III, tav. XLII, nos. 326, 328-330; Tomber 1986, p. 53, fig. 3, no. 87. 5 5 . Form N: Small, open form with a slightly sagging bottom separated from low, vertical upper wall by an angular joint. Undercut rim thickened on the exterior. Shallow, grooved lid seating at the inner edge of pointed lip. Turned and smoothed exterior bottom, unfinished upper wall. Smoothed floor and rilled upper interior wall. PEA 96; Acc. # 5700; Pl. 11, no. 55. Dimensions: sd 5,3 x 4,2 cm; ø 15 cm; rh 1,0 cm; rw 0,7 cm; wh 3,0 cm. Color: ec 2.5YR 5/6; ic 2.5YR 5/6; sc 10YR 4/1. Condition: good, clean fracture surfaces with sharp fracture surface edges. Comparanda: cf. no. 56. 5 6 . Form N: Small, open form with slightly sagging bottom separated from low vertical wall by an angular joint. Protruding rim with grooved lid-seating at the inner edge of rounded lip. Turned and corrugated exterior bottom, smoothed exterior wall with brownish patina

cenerognola cover on the upper section and turned vessel lip. Smoothed interior bottom, unfinished interior wall. PEA 22; Acc. # 5701; Pl. 11, no. 56. Dimensions: sd 3,1 x 4,6 cm; ø 15 cm; rh 0,7 cm; rw 0,8 cm; wh 3,2 cm. Color: ec 2.5YR 6/6; ic 2.5YR 6/6; sc 2.5YR 5.5/8; pc 5YR 4/2. Condition: good, clean fracture surfaces with sharp fracture surface edges. Comparanda: Aguarod Otal 1991, pp. 336-342, figs. 7682 (several examples); Aquilué Abadías 1987, p. 196, fig. 54, no. 4; Dyson 1976, fig. 44, no. 22II-17; fig. 54, nos. LS8-10; Fulford 1994, p. 62, fig. 4.6, no. 26.1; González Villaescusa 1990, fig. 15, nos. 38-40; Hayes 1972, p. 206, fig. 36, no. 197.1; Ostia I, tav. XII, no. 266. 5 7 . Form variant O.I: Medium-sized, open form with vertical, slightly convex wall ending in an undercut, rounded rim thickened on the exterior and a grooved lid seating on the rim top near the inner edge. The upper wall separated from sagging bottom by sharp carination. Smoothed exterior wall with patchy, blackish-gray patina cenerognola cover and corrugated exterior surface. Wheel-turning or turning ridges on the interior surface. PEA 96; Acc. # 5703; Pl. 11, no. 57. Dimensions: sd 10,3 x 7,3 cm; ø 21 cm; wh 5,9 cm; rh 1,1 cm; rw 0,9 cm. Color: ec 2.5YR 5.5/8; ic 2.5YR 5.5/8; sc 2.5YR 5.5/8; pc 5YR 3.5/1. Condition: very good; clean surfaces and sharp fracture surface edges. Patchy soot cover on the exterior bottom and lower exterior wall. Comparanda: cf. no. 58. 5 8 . Form variant O.I: Medium-sized, open form with vertical, slightly convex upper wall ending in a rim with undercut thickening on the exterior and a pointed lip with grooved lid seating on the top near the inner edge. Smoothed exterior wall covered up by blackish-gray patina cenerognola deposition, wheel-turning or turning ridges on the interior surface. PEA 124; Acc. # 3086; Pl. 11, no. 58. Dimensions: sd 15,9 x 5,4 cm; ø 29 cm; rh 1,7; rw 1,3 cm. Color: ic 2.5YR 5/6; sc 2.5YR 5/6; pc 5YR 4/1. Condition: good; clean surfaces with sharp fracture surface edges. Comparanda: Aguarod Otal 1991, pp. 339-341, figs. 7981 (several examples); Berato et al. 1986, p. 154, fig. 23, no. 19; Dyson 1976, fig. 54, nos. LS6-7; Fulford 1994, p. 62, fig. 4.6, no. 26.3; González Villaescusa 1990, fig. 18, nos. 47-49; Ostia I, tav. XII, no. 265; Tomber 1989, p. 485, fig. 12, no. 252; Tortorella 1981b, tav. CVII no. 7.

53

54

55

56

57

58 Plate 11. 53-54 Proto Hayes 197 deep casserole (M.II); 55-56 Hayes 197 minor (N); 57-58 Hayes 197 deep casserole (O.I). Scale 1:2.

5 9 . Form variant O.II: Medium-sized, open form with vertical, slightly convex upper wall ending in a protruding rim with undercut thickening on the exterior. A grooved lid seating on the top of the vessel near the inner edge of a rounded lip. Sagging bottom separated from the upper wall by a sharp carination. Turned vessel bottom with corrugated exterior surface. Smoothed exterior wall covered with patches of blackish-gray patina cenerognola deposition, rilled interior surface. PEA 104; Acc. # 5704; Pl. 12, no. 59. Dimensions: sd 7,4 x 7,1 cm; ø 25 cm; wh 6,3 cm; rh 1,4 cm; rw 1,2 cm. Color: ec 2.5YR 5/6; ic 2.5YR 5.5/6; sc 2.5YR 5/6; pc 5YR 3/1. Condition: good; clean surfaces and sharp fracture surface edges; patchy soot cover on the exterior bottom. Comparanda; Aguarod Otal 1991, pp. 335-338, figs. 7578 (several examples); Aquilué Abadías 1987, p. 47, fig. 13, nos. 4-5; Casas & Nolla 1993, p. 210, fig. 9, nos. 24; Coletti & Pavolini 1996, p. 408, fig. 9, no. 1; Fulford 1994, p. 62, fig. 4.6, nos. 26.4 & 26.7; Luni II, fig. 192, no. 5; fig. 260, no. 4; Tortorella 1981b, tav. CVII no. 6. 6 0 . Form variant O.III: Medium-sized, open form with vertical wall ending in a protruding rim with undercut thickening on the vessel exterior. A grooved lid seating on the vessel top, near the inner edge of a pointed lip. Smoothed exterior wall covered with partially off-worn, brownish patina cenerognola deposition, rilled interior surface. PEA 135; Acc. # 5708; Pl. 12, no. 60. Dimensions: sd 9,4 x 4,6 cm; ø 23 cm; rh 1,3 cm; rw 1,1 cm. Color: ec 5YR 6.5/6; ic 2.5YR 6/8; sc 2.5YR 6/7; pc 5YR 4/3. Condition: poor; worn exterior surface and fracture surface edges, calcareous scum deposition on both sides. Comparanda: Luni II, fig. 260, no. 5. 6 1 . Form variant O.IV: Medium-sized, open form with vertical, slightly convex upper wall ending in a protruding, almond-shaped rim thickened on the vessel exterior. Rounded lip with a cut lid seating at the inner edge of the vessel top. Smoothed exterior wall covered up by blackish-gray patina cenerognola deposition, rilled interior surface. PEA 104; Acc. # 5709; Pl. 12, no. 61. Dimensions: sd 8,2 x 4,9 cm; ø 26 cm; rh 1,6 cm; rw 1,3 cm. Color: ic 2.5YR 5.5/8; sc 2.5YR 5/8; pc 2.5YR 4/0. Condition: good; substantially clean surfaces with sharp fracture surface edges.

Comparanda: Adroer 1963, p. 108, no. 7589; Dore 1988, p. 81, fig. 17, no. R1; Dore 1989, p. 127, fig. 34, no. 56.552; Gandolfi 1994, p. 150, fig. 12, no. 8; González Villaescusa 1990, fig. 16, no. 43; fig. 18, no. 45; Luni II, fig. 260, no. 6; Navarro Luengo et al. 1997, p.88, n o 5. Ostia IV, tav. X, no. 52; Tomber 1986, p. 53, fig. 3 , nos. 82 & 85. 6 2 . Form variant O.V: Medium-sized, open form with vertical upper wall ending in a protruding rim thickened on the vessel exterior. Shallow, wide furrow forms a lid seating on the rim top near the inner edge. Smoothed exterior surface showing occasional bands of burnishing and brownish patina cenerognola cover, rilled interior surface. PEA 20; Acc. # 5707; Pl. 12, no. 62. Dimensions: sd 6,3 x 5,2 cm; ø 34 cm; rh 1,6 cm; rw 1,3 cm. Color: ic 2.5YR 5/8; sc 2.5YR 5/8; pc 2.5YR 4.5/3. Condition: medium; clean surfaces, but worn fracture surface edges. Comparanda: Hayes 1976, p. 51, fig. 1, nos. I.2 & II.10; p. 53, fig. 2, no. V.13; p. 57, fig. 4, no. VII.57; Luni II, fig. 123, no. 6; fig. 260, no. 7; Michelucci 1985, p . 137, fig. 13, no. 760; Therrien 1981, p. 147, no. 46; Tortorella 1981b, tav. CVII no. 8. 6 3 . Form variant O.VI: Medium-sized, open form with vertical, slightly convex upper wall ending in a protruding rim with undercut thickening on the exterior surface. Grooved lid seating on the rim top near the inner edge. Sharp carination separates the sagging bottom from the upper wall. Turned bottom with corrugated exterior surface, smoothed exterior wall covered b y patches of blackish-gray patina cenerognola deposition, shallow rilling on the interior wall. PEA 33; Acc. # 5705; Pl. 12, no. 63. Dimensions: sd 10,6 x 10,1 cm; ø 25 cm; wh 8,1 cm; rh 1,2 cm; rw 1,0 cm. Color: ec 2.5YR 5/6; ic 2.5YR 5/6; sc 2.5YR 5/6; pc 10YR 3/1. Condition: good; in spite of sooted bottom and lower exterior wall the vessel has clean surfaces and sharp fracture surface edges. Comparanda: Aquilué Abadías 1987, p. 200, fig. 56, no. 2; Hayes 1976, p. 69, fig. 9, no. 32; Ostia III, tav. XXI, no. 108; Ostia IV, tav. X, no. 57; Rakob 1991, p. 86, fig. 18, no. 20; Schuring 1988, p. 67, fig. 42, no. 166; Therrien 1981, p. 147, no. 45; Tomber 1989, p. 485, fig. 12, nos. 250-251; Vázquez de la Cueva 1985, fig. 36, no. 163.

59

60

61

62

63

Plate 12. 59-63 Hayes 197 deep casserole (59 O.II; 60 O.III; 61 O.IV; 62 O.V; 63 O.VI). Scale 1:2.

64. Form variant O.VII: Medium-sized, open form with vertical, slightly convex upper wall ending in a thickened, protruding rim with triangular lip and a notched lid seating formed by two diagonal cuts on the rim top. Smoothed exterior wall with occasional stripes of burnishing and brownish-gray patina cenerognola cover, rilled interior surface. PEA 33; Acc. # 5710; Pl. 13, no. 64. Dimensions: sd 4,2 x 5,2 cm; ø 24 cm; rh 1,7 cm; rw 1,0 cm. Color: ic 2.5YR 6/6; sc 2.5YR 5.5/8; pc 5YR 4/2. Condition: good, clean surfaces and sharp fracture surface edges, patchy soot cover on the exterior surface, especially on the rim zone. Comparanda: Schuring 1988, p. 67, fig. 42, no. 164. 65. Form variant O.VIII: Medium-sized, open form with everted, protruding rim with pointed lip thickened on the exterior. A grooved lid-seating on the rim top near the inner edge. Strongly inward inclined upper wall identifies the form as a cookpot. Turned and smoothed exterior surface covered up by a deep black patina cenerognola deposition, unfinished interior surface with shallow rilling. PEA 16; Acc. # 5702; Pl. 13, no. 65. Dimensions: sd 15,8 x 3,0; ø 27; rh 1,6 cm; rw 1,1 cm. Color: ic 5YR 5/4; sc 10YR 4/8; pc 10YR 3/1. Condition: good, clean surfaces and sharp fracture surface edges. Comparanda: Luni II, fig. 123, no. 10; Marín Jordá 1995, p. 162, fig. 5, no. 13; Ostia I, tav. XII, no. 268; Tortorella 1981b, tav CVIII, no. 13. 6 6 . Form P: Medium-sized, open form with vertical upper wall ending in a thickened rim with pointed lip and a projecting lid-seating on the interior. Unfinished vessel surfaces. Brownish patina cenerognola cover o n the exterior surface. PEA 22; Acc. # 5718; Pl. 13, no. 66. Dimensions: sd 4,6 x 4,9 cm; ø 18 cm; rh 1,3 cm; rw 0,9 cm. Color: ec - not determined due to heavy soot cover; ic 2.5YR 5/6, sc 2.5YR 4/4.

Condition: medium; heavily sooted vessel surfaces and fracture surfaces, worn fracture surface edges. Comparanda: cf. no. 67. 6 7 . Form P: Medium-sized, open form with vertical, slightly convex upper wall ending in a thickened rim with rounded lip and projecting lid seating on the interior. Burnished stripes on otherwise unfinished exterior surface, rilled interior surface. PEA 67; Acc. # 5719; Pl. 13, no. 67. Dimensions: sd 6,5 x 8,1 cm; ø 29 cm; rh 2,3 cm; rw 1,7 cm. Color: ec 7.5YR 4/4; ic 5YR 5.5/4; sc 7.5YR 4/2. Condition: good; soot deposition on the exterior wall and rim, clean fracture surfaces and sharp fracture surface edges. Comparanda: Aguarod Otal 1991, p. 343, fig. 83, nos. 34; Dore 1992, p. 143, no. 25; Luni II, fig. 123, nos. 8 and 9; fig. 261, no. 2; Macias et al. 1997, p. 169, fig. 5 , no. 8. 6 8 . Form variant Q.I: Medium-sized, open form with inward inclined, slightly convex upper wall ending in a thickened, angular rim with shallow notched lid-seating on the vessel top near the inner edge. Smoothed exterior surface showing light ribbing ca. 2 cm underneath the rim. Unfinished interior surface preserving shallow wheel-ridging. Exterior surface and the lower section of the exterior rim covered by brownish patina cenerognola deposition. PEA 150; Acc. # 5805; Pl. 13, no. 68. Dimensions: sd 10,9 x 4,6 cm; ø 24 cm; rh 1,3 cm; rw 1,2 cm. Color: ic 2.5YR 6/6; sc 2.5YR 5.5/7; pc 2.5YR 4/2. Condition: medium; patches of dirt cover all the surfaces, flakes worn off from the exterior rim, worn fracture surface edges. Comparanda: Aguarod Otal 1991, p. 331, fig. 71, nos. 12; Dore 1989, p. 127, fig. 34, no. 53.550 (?); Fulford 1994, p. 61, fig. 4.5, no. 23; Hayes 1976, p. 73, fig. 11, no. 11; Rakob 1991, p. 119, fig. 23, no. 36; Reynolds 1984, p. 501, fig. 18.8, nos. 170-171; Sagui 1980, p . 544, no. 142; Tomber 1986, p. 53, fig. 3, no. 88.

64

65

66

67

68

Plate 13. 64-65 Hayes 197 deep casserole (64 O.VII; 65 O.VIII); 66-67 Late Roman deep casserole (P); 68 Hayes 199 deep casserole (Q.I). Scale 1:2.

6 9 . Form variant Q.II: Medium-sized, open form with inward inclined, slightly convex upper wall ending in a protruding rim thickened on the exterior. Flat, slightly inward sloping rim top with rounded lip. Angular transition between the rim and the interior surface, the exterior wall separated from the rim with a notch. Smoothed rim and exterior surfaces, unfinished interior surface. Part of the exterior surface covered by blackishbrown patina cenerognola deposition. PEA 104; Acc. # 5715; Pl. 14, no. 69. Dimensions: sd 6,3 x 3,4 cm; ø 23 cm; rh 1,1 cm. Color: ec 2.5YR 5/6; ic 2.5YR 6/7; sc 2.5YR 5/7; pc 5YR 4/1. Condition: good; substantially clean surfaces and sharp fracture surface edges. Comparanda: Aguarod Otal 1991, p. 332, fig. 72, nos. 1 , 3-4; Aquilué Abadías 1987, p. 198, fig. 55, no. 5; p . 200, fig. 56, no. 1; Macias et al. 1997, p. 169, fig. 5 , no. 4; Marín Jordá 1995, p. 162, fig. 5, no. 12; Nolla & Puertas 1988, p. 63, fig. 10, no 5; Reynolds 1984, fig. 18.8, no. 172; Sagui 1980, p. 544; no. 143. 7 0 . Form variant Q.III: Medium-sized, open form with inward inclined, convex upper wall ending in a thickened, angular rim with pointed lip and two shallow furrows on the flat rim top forming a lid seating. A notch underneath the rim separates it from the exterior wall. Unfinished exterior surface with matt smoothed bands and patches of darkened areas, which result more likely from the firing than vessel use. Unfinished interior surface. PEA 144; Acc. # 5762; Pl. 14, no. 70. Dimensions: sd 7,0 x 4,2 cm; ø 19 cm; rh 1,3 cm; rw 0,8 cm. Color: ec 2.5YR 4/2; ic 7.5YR 7/5; sc 2.5YR 5.5/6. Condition: good; clean surfaces with sharp fracture surface edges. Comparanda: Dore 1989, p. 127, fig. 34, no. 64.680; Hayes 1978, p. 25, fig. 1, no. 22; Luni II, fig. 124, no. 7. 7 1 . Form variant R.I: Medium-sized open form with inward inclined, slightly convex upper wall ending in a slightly everted rim with undercut thickening on the exterior. Rounded lip with notched lid seating near the inner edge. Unfinished vessel surfaces with shallow wheel-ridging on the interior. Patchy sooting on the exterior rim and wall. PEB 223; Acc. # 5763; Pl. 14, no. 71. Dimensions: sd 6,2 x 2,2 cm; ø 17 cm; rh 1,1 cm; rw 0,9 cm. Color: ec 2.5YR 6/6; ic 2.5YR 6/6; sc 2.5YR 5/7. Condition: fairly good; clean surfaces, but slightly worn fracture surface edges. Comparanda: Fulford 1984b, p. 182, fig. 68, no. 11; Sagui 1980, p. 544, no. 140. 7 2 . Form variant R.II: Medium-sized, open form with inward inclined, slightly convex upper wall ending in an everted rim with thickened, rounded lip. Smoothed bands and patchy soot deposition on the exterior wall and rim, unfinished interior surface. PEA 22; Acc. # 5764; Pl. 14, no. 72.

Dimensions: sd 4,5 x 4,6 cm; ø 21 cm; rh 0,8 cm; rw 0,6 cm. Color: ec 2.5YR 5/6; ic 2.5YR 6/6; sc 2.5YR 5/7. Condition: good; substantially clean surfaces with sharp fracture surface edges. Comparanda: cf. no. 73. 7 3 . Form variant R.II: Medium-sized, open form with inward inclined upper wall turning to an everted rim with rounded lip. The exterior rim turned with a tool that has resulted three shallow furrows. In spite of smoothed surfaces, the vessel interior has preserved some wheelthrowing marks. PEA 20; Acc. # 1049; Pl. 14, no. 73. Dimensions: sd 4,2 x 3,1 cm; ø 21 cm; rh 0,7 cm; rw 0,7 cm. Color: ec 2.5YR 5/7; ic 2.5YR 5/8; sc 2.5YR 5/7. Condition: good; clean surfaces and sharp fracture surface edges. Comparanda: Aguarod Otal 1991, p. 344, fig. 84, nos. 13; Aquilué Abadías 1989, p. 193, fig. 89, nos. 5.27-29; Dore 1989, p. 131, fig. 36, no. 70.682; Fulford 1984b, p. 180, fig. 67, no. 5.4 (?); Luni II, fig. 124, no. 6 (?); Macias et al. 1997, p. 169, fig. 5, no. 11; Sánchez Sánchez 1995, p. 274, fig. 13, no. 23; Tortorella 1981b, tav CVII, no. 11. 7 4 . Form variant R.III: Medium-sized, open form with inward inclined, slightly convex upper wall ending in an even, everted rim with somewhat angular lip. Turned lidseating on the rim interior. Smoothed exterior surface with horizontal zigzag decoration – incised at leatherhard stage – underneath the rim, unfinished interior surface. A layer of soot deposited on rim. PEA 33; Acc. # 5806; Pl. 14, no. 74. Dimensions: sd 5,2 x 3,1 cm; ø 24 cm; rh 1,4 cm; rw 0,8 cm. Color: ec 5YR 5.5/4; ic 10R 5/6; sc 10R 5/6. Condition: good, substantially clean surfaces and fairly sharp fracture surface edges. Comparanda: Dore 1992, p. 143, no. 18; Neuru 1980, fig. 5, no. 24; Mackensen 1993, tab. 80 no 6; Marín Jordá 1995, p. 161, fig. 4, no. 6. 7 5 . Form variant R.IV: Medium-sized, open form with inward inclined, slightly convex upper wall ending in a thickened, everted rim with rounded lip. Rim separated from the exterior wall by a notch and from the interior b y a shallow protrusion with angular edges. Smoothed exterior surface with traces of deep black patina cenerognola cover, unfinished interior surface. Wearmarks on the rim top and the upper exterior surface. PEA 135; Acc. # 5769; Pl. 14, no. 75. Dimensions: sd 4,9 x 5,4 cm; ø 28 cm, rh 1,1 cm; rw 1,7 cm. Color: ec 2.5YR 6/7; ic 2.5YR 6/7; sc 2.5YR 5.5/8; pc 2.5YR 5/0. Condition: bad; worn exterior surface with heavily worn fracture surface edges. Comparanda: Fulford 1984b, p. 182, fig. 68, no. 7.3; Lavoie 1989, p. 98, pl. IV, no. E1; Luni II, fig. 256, no. 10; Mackensen 1993, tab. 81 no 1; Tortorella 1981b, tav CVIII, no. 3.

69

70

71

72

73

74

75 Plate 14. 69-70 Hayes 199 deep casserole (69 Q.II; 70 Q.III); 71-75 Miscellaneous North Tunisian casseroles (71 R.I; 72-73 R.II; 74 R.III; 75 R.IV). Scale 1:2.

7 6 . Form variant S.I: Medium-sized open form with inward inclined, slightly convex upper wall ending in a short everted rim with undercut thickening on the exterior and pointed lip. Smoothed surfaces with sparse, matt burnished stripes and whitish-brown scum deposition resembling patina cenerognola cover on the exterior. PEB 357; Acc. # 5711; Pl. 15, no. 76. Dimensions: sd 6,1 x 4,5 cm; ø 26 cm; rh 1,1 cm; rw 1,2 cm. Color: ic 2.5YR 5/6; sc 2.5YR 4/8; pc 5YR 6/4. Condition: medium; patches of post-depositional calcareous scum covering all the surfaces, slightly worn fracture surface edges. Comparanda: Aguarod Otal, p. 345, fig. 85, nos. 1-2; Dore 1989, p. 127, fig. 34, nos. 58.574 & 58.3650; Dore 1992, p. 139, no. 13; Mahjoubi et al. 1973, fig. 19; Peacock et al. 1989, p. 221, fig. 2, no. 25; Peacock et al. 1990, p. 63, fig. 2, no. 23; p. 75, fig. 10, no. 12; Tortorella 1981b, tav. CIX, no. 10. 77. Form variant S.II: Medium-sized, open form with inward inclined, slightly convex upper wall ending i n slightly everted, elongated rim with undercut thickening on the exterior, broad groove forming a lid-seating o n the interior and rounded lip. Smoothed surfaces with the exterior covered by burnished, matt stripes and bands and a whitish-brown scum. PEA 96; Acc. # 5713; Pl. 15, no. 77. Dimensions: sd 5,0 x 4,3 cm; ø 29 cm; rh 1,2 cm; rw 1,1 cm. Color: ic 2.5YR 5/6; sc 2.5YR 4/8; pc 7.5YR 6.5/4. Condition: good; clean surfaces and substantially sharp fracture surface edges. Comparanda: Aguarod Otal 1991, p. 346, fig. 86, no. 1 ; Dore 1988, p. 81, fig. 17, no. S1; Dore 1989, p. 127, fig. 34, nos. 59.2474 & 59.2480; Dore 1992, p. 139, no. 14; Dyson 1976, fig. 55 no. LS11; Hayes 1972, p . 202, fig. 35, no. 184.1; Ostia I, tav. XII, no. 271; Pucci 1975, p. 68, fig. 64; Tortorella 1981b, tav. CIX, no. 9. 7 8 . Form variant S.III: Medium-sized, open form with strongly inward inclined, slightly convex upper wall turning to an everted, almost flanged rim with undercut thickening on the exterior, a broad furrow forming a lid seating on the interior and a rounded lip. Smoothed surfaces with sparse matt burnished bands and blackishbrown patina cenerognola cover on the exterior. Fully oxidized fabric without any traces of use-alteration. PEA 93; Acc. # 5714; Pl. 15, no. 78. Dimensions: sd 6,0 x 2,2 cm; ø 27 cm; rh 1,1 cm; rw 2,0 cm. Color: ic 2.5YR 5/6; sc 2.5YR 4/8; pc 5YR 4,1. Condition: medium; calcareous scum deposited on the interior and fracture surfaces, sharp fracture surface edges. Comparanda: Aguarod Otal 1991, p. 330, fig. 70, no. 2 ; Dore 1988, p. 81, fig. 17, no. P2A; Dore 1989, p. 121, fig. 31, no. 44.516; Dore 1992, p. 139, no. 8; Luni II, fig. 122, no. 8; Pucci 1975, p. 68, fig. 57; Riley 1979, fig. 100, no. 454; Tomber 1986, p. 57, fig. 7, no. 193; van der Werff 1982, pl. 25, no. 12; pl. 51, no. 5. 7 9 . Form variant S.IV: Medium-sized, open form with inward inclined, slightly convex upper wall ending in an angular rim with undercut thickening on the exterior and rounded lip. The rim top has been cut to form a lid seating. Smoothed surfaces with stripes of burnishing and a faint, whitish scum covers the exterior. PEB 340; Acc. # 5717; Pl. 15, no. 79. Dimensions: sd 2,9 x 4,9 cm; ø 19 cm; rh 1,0 cm; rw 1,1 cm.

Color: ec 2.5YR 5/6; ic 2.5YR 5/6; sc 2.5YR 4/8. Condition: good; despite the patchy soot cover on the rim and the lower section of the exterior wall, the vessel surfaces are substantially clean and fracture surface edges sharp. Comparanda: Aguarod Otal 1991, p. 332, fig. 72, no. 2 ; Dore 1989, p. 127, fig. 34, no. 54.2710 (North Tunisian fabric); Lavoie 1989, p. 97, no. S88.73C5; Luni II, fig. 124, no. 1; Tomber 1986, p. 54, fig. 4, nos. 109. 8 0 . Form variant S.V: Medium-sized, open form with inward inclined, slightly convex upper wall with an angular transition to everted, lozenge-shaped rim with undercut thickening on the exterior and angular lip. Shallow concavity on the rim top probably acted as a lidseating. Smoothed exterior surface with sparse burnished bands and a grayish-brown scum cover, unfinished interior surface. Reduced exterior surface in an otherwise oxidized fabric. PEA 135; Acc. # 5759; Pl. 15, no. 80. Dimensions: sd 6,6 x 4,0 cm; ø 19 cm; rh 0,8 cm; rw 1,4 cm. Color: ic 2.5YR 5/4; sc 2.5YR 4/6; pc 2.5YR 4/0. Condition: fairly good; somewhat dirty vessel surfaces with light wear on fracture surface edges. Comparanda: Aguarod Otal 1991, p. 347, fig. 87, nos. 12; Aquilué Abadías 1989, p. 195, fig. 90, no. 5.32; Dore 1988, p. 81, fig. 17, no. P2B; Dore 1989, p. 121, fig. 31, no. 47.602; Hayes 1976, p. 69, fig. 9, no. 34; Hayes 1978, p. 51, fig. 12, no. 30 (?). 8 1 . Form variant S.VI: Medium-sized, open form with inward inclined, slightly convex wall turning through an angular transition to an everted rim with sharply undercut thickening on the exterior and somewhat angular lip. Flattened rim top forms a lid seating. Smoothed exterior surface with darker, sparse matt burnished stripes, unfinished interior surface preserving shallow wheelridging. Fully oxidized fabric. PEB 180; Acc. # 5760; Pl. 15, no. 81. Dimensions: sd 6,4 x 4,8 cm; ø 26 cm, rh 0,8 cm; rw 1,4 cm. Color: ec 10R 5.5/6; bands 2.5YR 4.5/4; ic 10R 5/6; sc 2.5YR 4/4. Condition: good; substantially clean surfaces with sharp fracture surface edges. Comparanda: Aquilué Abadías 1989, p. 193, fig. 89, no. 5.30; Dore 1989, p. 110, fig. 25, nos. 51.3915 & 51.600; Dore 1992, p. 139, no. 12; Hayes 1976, p. 69, fig. 9, no. 35; Pucci 1975, p. 67, fig. 45; Villedieu 1984, p. 299, no. 39. 8 2 . Form variant S.VII: Medium-sized, open form with slightly inward inclined, convex upper wall turning to an even, slightly everted rim with somewhat angular lip. Lid seating formed by a broad furrow on the interior rim. Stepped thickening on the lower interior surface. Smoothed exterior surface with a faint whitish scum, unfinished interior surface. Fully oxidized fabric. Hardly visible darker traces of a dipinto (painted inscription) or resin cover on the exterior surface. PEA 12; Acc. # 5765; Pl. 16, no. 82. Dimensions: sd 8,0 x 4,5 cm; ø 25 cm; rh 1,0 cm; rw 0,7 cm. Color: ec 5YR 5/3; dipinto/resin 5YR 3/3; ic 2.5YR 5/6; sc 2.5YR 4.5/8. Condition: good, substantially clean surfaces with fairly sharp fracture surface edges. Comparanda: Fulford 1984b, p. 180, fig. 67, no. 5.1 (?); Tomber 1989, p. 491, fig. 15, no. 300.

76

77

78

79

80

81

Plate 15. 76-81 Central Tunisian deep casseroles (76 S.I; 77 S.II; 78 S.III; 79 S.IV; 80 S.V; 81 S.VI). Scale 1:2.

8 3 . Form variant T.I: Medium-sized, open form with inward inclined, slightly convex upper wall ending in a small everted rim with undercut thickening on the exterior and a spherical lip. Smoothed vessel surfaces without a trace of use-alteration, fully oxidized fabric. PEA 93; Acc. # 5808; Pl. 16, no. 83. Dimensions: sd 5,2 x 2,3 cm; ø 20 cm; rh 1,0 cm; rw 0,6 cm. Color: ec 2.5YR 5/6; ic 2.5YR 5/6; sc 2.5YR 5/6. Condition: good; fairly clean surfaces and sharp fracture surface edges. Comparanda: Aquilué Abadías 1991, fig. 241, no. 5; Dore 1989, p. 131, fig. 36, no. 77.663.

fully oxidized fabric. PEA 124; Acc. # 5809; Pl. 16, no. 87. Dimensions: sd 2,6 x 5,3 cm; ø 5 cm; rh 1,2 cm; rw 0,7 cm. Color: ec 5YR 6/6; ic 5YR 6/6; sc 2.5YR 6/6; pc 2.5YR 4/0. Condition: fairly good, substantially clean surfaces and sharp fracture surface edges Comparanda: Aguarod Otal 1991, pp. 353-355, figs. 9395; Casas & Nolla 1993, p. 211, fig. 10, nos. 1-3; Fulford 1984b, p. 208, fig. 81, no. 29; González Villaescusa 1990, fig. 27, nos. 77-78; Gonzáles Villaescusa 1993, pp. 154-155, figs. 2-3, nos. 1-4.

8 4 . Form variant T.II: Medium-sized, open form with vertical, slightly convex upper wall ending in a thickened rim with diagonal lid seating cut on the rim top. Unfinished surfaces with hardly visible wheelturning marks on the vessel interior. PEB 342; Acc. # 5807; Pl. 16, no. 84. Dimensions: sd 4,2 x 2,8 cm; ø 16 cm; rh 1,3 cm; rw 0,9 cm. Color: ec 2.5YR 5/6; ic 2.5YR 5/6; sc 2.5YR 4.5/5. Condition: good; clean surfaces and sharp fracture surface edges.

88. Form W: Fairly thick horizontal strap handle formed by cutting a strip of clay on both sides and profiling its exterior. Convex inner face, three slightly concave facets with sharp edges on the exterior. Light patches of brownish patina cenerognola -deposition on the exterior surface. PEA 20; Acc. # 5802; Pl. 16, no. 88. Dimensions: sherd 7,0 x 3,3 cm, hh 1,5 cm, hw 2,3 cm. Color: ic 10R 5/6; sc 5YR 5.5/6; pc 10R 4/3. Condition: fairly good, light dirt cover on surfaces, but sharp fracture surface edges and clean fracture surfaces.

8 5 . Form variant U.I: Medium-sized, open form with outward inclined, almost diagonal wall ending in a flanged rim with a pointed lip. Broad concavity on the top of the flange forms a lid-seating. Angular transition between the rim and the interior wall. Unfinished vessel surfaces with a band of blackish patina cenerognola deposition and patches of soot on the exterior underneath the rim. PEB 180; Acc. # 2644; Pl. 16, no. 85. Dimensions: sd 11,3 x 4,8 cm; ø 21 cm; rh 0,8 mm; rw 2,2 cm. Color: ec 2.5YR 5/6; ic 2.5YR 5.5/5; sc 2.5YR 5/7; pc 7.5R 4/0. Condition: good; clean surfaces and sharp fracture surface edges. Comparanda: Dore 1989, p. 110, fig. 25, no. 22.471; Leveau 1984, p. 462, fig. 248, no. 10715 ; Tomber 1989, p. 477, fig. 9, no. 214.

89. Form variant X.I: Small, possibly closed form with small, hollowed flat base joining with strongly outward expanding, slightly convex lower wall through short concave profiling. Smoothed exterior surface with a dense patina cenerognola cover and unfinished interior surface with whitish scum. Fully oxidized fabric. PEB 360; Acc. # 5737; Pl. 16, no. 89. Dimensions: sd 3,5 x 3,1 cm; ø 6 cm; base width 0,3 cm. Color: ec 10YR 3/1; ic 2.5YR 5/8; sc 2.5YR 6/6; scum on the interior 5YR 8/1. Condition: fairly good; thick patches of soot on the exterior surface, substantially clean fracture surfaces with sharp fracture surface edges.

8 6 . Form variant U.II: Medium-sized, open form with steep, slightly convex lower wall. Everted, protruding rim with undercut thickening on the exterior and rounded lip. Two furrows on the rim exterior. Traces of a protrusion on the upper exterior wall suggesting the presence of a ring knob or protruding carination. Turned exterior surface, smoothed interior surface and selfslipped rim. Possible traces of use-wear visible on the interior, where some subround chips have been worn off from the vessel surface. PEA 12; Acc. # 5757; Pl. 16, no. 86. Dimensions: sd 10,5 x 6,4 cm; ø 29 cm; rh 1,3 cm, rw 1,5 cm. Color: ec 2.5YR 5.5/8; ic 2.5YR 5/8; sc 2.5YR 5.5/8. Condition: good; substantially clean surfaces with sharp fracture surface edges. Comparanda: Carandini 1975, p. 51, fig. 6; Hayes 1976, p. 67, fig. 8, nos. 16-17; Tomber 1989, p. 447, fig. 1 , nos. 1-2; Tortorella 1981b, tav CV, no. 16. 8 7 . Form V: Medium-sized, closed form with narrow vertical neck turning to a slightly everted rim with a rounded lip. The irregular shape of the rim indicates that this sherd belongs to a pitcher with trefoil-shaped mouth. Vague traces of blackish patina cenerognola -like deposition on vessel lip. Unfinished vessel surfaces,

9 0 . Form variant X.II: Medium-sized, open form with a thick, flat base joining with slightly convex, outward inclined wall in a sharp angle. Smooth transition between the vessel base and wall on the interior. Turned and smoothed exterior surface, smoothed interior surface. Fully oxidized fabric without a trace of use-alteration. PEB 180; Acc. # 5803; Pl. 16, no. 90. Dimensions: sd 3,8 x 3,4 cm, ø 5 cm, base width 6 mm. Color: ec 10R 5/6; ic 2.5YR 5/6; sc 2.5 YR 5/6. Condition: medium; substantially clean surfaces, worn fracture surface edges and dirty fracture surfaces. 91. Form variant X.III: Medium-sized, closed form with steep, hollowed disk-base turning to strongly outward expanding, convex wall. The exterior base and a considerable part of the lower exterior wall turned, smoothed upper exterior wall with dense band of blackish-gray patina cenerognola deposition, unfinished interior surface. PEA 33; Acc. # 5738; Pl. 16, no. 91. Dimensions: sd 6,7 x 6,2 cm, ø 4 cm; base height 1,3 cm. Color: ec 2.5YR 5.5/8; ic 2.5YR 6/7; sc 2.5YR 5.5/8; pc 5YR 4/1. Condition: medium; substantially clean vessel surfaces, but worn fracture surfaces and fracture surface edges. Comparanda: Aguarod Otal 1991, p. 350, fig. 90, no. 5 ; Aguarod Otal 1995, p. 137, fig. 9, no. 10; Aquilué Abadías 1995, p. 65, fig. 3, no. 4; Casas & Nolla 1993, p. 209, fig. 7, no. 2; González Villaescusa 1990, fig. 26, no. 76; Rita 1990, p. 32, lower figure.

82

83

84

85

86

90 87

88

89

91

Plate 16. 82 Central Tunisian deep casseroles (82 S.VII); 83-84 Miscellaneous Central Tunisian cooking vessels (83 T.I; 84 T.II); 85-86 Cooking-basins (85 U.I; 86 U.II); 87 Pitcher (V); 88 Handle (W); 89-91 Bases (89 X.I; 90 X.II; 91 X.III). Scale 1:2.