Читать книгу Amheida III - Roger S. Bagnall - Страница 10

PART I

INTRODUCTION

1. INTRODUCTION

The ostraka published in this volume come from the excavations of the seasons from 2008 to 2013. Two of those seasons were disrupted by external causes: in 2009, a delay in the issuance of military security clearances, which shortened the season by nearly three weeks; and in 2011, the evacuation of the team during the Egyptian revolution, after only a few days of excavation. The richest finds of ostraka came from the even-numbered years of 2008, 2010, and 2012. The decision to cut off the present volume after the 2013 season reflects not only the considerable quantity of material in hand but the arrival, if not at a conclusion of work in and around Area 2.2 (Building 6) and 2.3 (Building 7), at least at a point beyond which they seem unlikely to yield texts that would significantly change the picture derived from the first two volumes. (The 2014 season in fact found few ostraka.) In addition, work in Area 4.1 (the Temple of Thoth) is essentially complete, at least for the present, with the 2013 season. The 2014 season there yielded more decorated blocks and some Demotic and hieratic ostraka, but few Greek texts. The texts in the present volume have helped to add coherence to our understanding of the texts in volume 1, to which some improvements are given in an appendix at the end of this volume; perhaps more significantly, the analysis of their archaeological contexts has helped sharpen the distinctions between types of material and the contexts that produced them and has improved our grasp of chronology. Although the introduction to volume 1 described the contexts of the ostraka in some detail, and subsequent discoveries have only confirmed the correctness of most of the analysis given there, we now can see the particulars laid out there in a more coherent general framework. We have for this reason organized volume 2 to give priority to categories of archaeological contexts, ordering the texts by type within these categories. We have particularly come to see that our body of texts is dominated by ostraka that found their way to their findspots through ancient processes of dumping and reusing waste.

2. ARCHAEOLOGICAL CONTEXTS

In volume 1, the archaeological contexts have been described under the headings of the three areas of the site from which the ostraka have come: Area 1 (mainly Area 1.3, i.e., House B2), a house with adjacent street and courtyard in the north part of the site, located on the widest street of the city (S1);¹ Area 2 (mainly Area 2.1, House B1, the “House of Serenos”), the central area of the site as we see it today, located to the east of the temple area; and Area 4 (mainly Area 4.1, the temple area), the highest hill of the site, on which was located the precinct of the Temple of Thoth (Fig. 1).

Figure 1. Plan of Amheida.

The ostraka in Area 1 came predominantly from levels connected with the occupation of B2, although some of the contexts are not very secure, as a result of severe deflation caused by wind erosion. There was no excavation in Area 1 since 2007 until the initial seasons of work by the team from the University of Reading (now CUNY), led by Anna L. Boozer, in spring 2012 and 2013. The small groups of ostraka from those seasons are reserved for later publication.

The ostraka from Area 4 come entirely from insecure contexts, because of the extreme degree of mixing of material from different levels that has occurred as a result of treasure-hunting, stone-robbing, sebbakh-digging, and wind erosion. It can in general be said that although in a few places specific contexts with stratigraphy of the Old Kingdom, Second Intermediate Period, and the Late Period have survived, the characterization of find contexts as insecure remains true for the parts of Area 4 excavated in 2008–2013. As a result, all ostraka from Area 4 are presented in a single section of this volume, without attention to the stratigraphic units in which they were found. It may be assumed that none came from a secure context capable of providing useful information.

In Area 2, the situation is far more complex. Already in volume 1 it was possible to distinguish material located under floor levels and thus presumably in place at the time of construction from that accumulated during the period of occupation, particularly during the phase after the partial renovation of B1 around 350. The stratigraphy of B1 and of the courtyard area of Rooms 9 and 10 immediately to its north is described in detail in volume 1. There are just a few texts from this area included in the present volume, as they were found in sorting pottery in years after 2007. As work continued to the north of this zone and in the adjacent streets, and then in Areas 2.2 (Building 6) and 2.3 (the church, B7), it has become increasingly possible to see that the stratigraphic pattern is in large part common to the entire area and to distinguish the ostraka by broad categories of contexts. At the same time, further distinctions can be introduced to take account of the building histories of the individual structures and streets (Fig. 2). This more nuanced picture allows us to describe these contexts under four broad categories:

(1) Windblown sand in the uppermost layers excavated. This sand is not a secure context in any instance, but it is our observation that the material found in such layers is almost always from the period of occupation, and we have therefore included such layers in the broad category of occupational debris. These ostraka have in many cases probably come to their location at the time of excavation as a result of being left on rooftops or other surface levels, although it is not excluded that some sherds used for wall or vault chinking could turn up in such levels. We have in particular kept the possibility of chinking sherds in mind when dealing with apparently anomalous texts in this group.

Figure 2. Plan of the House of Serenos (Building 1).

(2) Identifiable occupation levels, with materials accumulated during the period in which the buildings were lived in or actively used. Here again, there is often the possibility that chinking sherds coming from wall or vault collapse can find their way into the upper parts of occupation debris.

(3) Material dumped in preparation of the site for construction, and always found below the lowest floor (or street) levels. This material will have accumulated over a shorter or longer period before construction and quite possibly at another location at Amheida. It may have varying degrees of internal homogeneity or heterogeneity. In general, it seems that the materials for a general spreading of debris to level the site for the construction of B1, B5, and S2 came from a single large source of debris that included building waste and ash along with ostraka coming from jar stoppers. These ostraka have dates ranging from year 1 to year 21 but not higher. It is thus possible that they could all belong to the regnal years of Diocletian (year 1 = 284/5). Years from 8 up, however, could also belong to other reigns, and we do not see any means of being certain that many of them are not from the reign of Constantine. For this reason we have given a full set of the possible years when publishing these pieces in the present volume, and in the corrections to volume 1 we have listed all of the possible years for such texts in that volume.

Alongside, or perhaps better on top of, this general layer of debris there are more restricted places in which we can find pre-construction dumped material, for example in a foundation trench. There are a few instances of such ostraka in this volume. More generally, it looks as if some of the sub-floor and sub-foundation debris in B6 and B7 may have been laid down in a separate operation before building began, and we have concluded that these buildings were either built a little later than B2 and B5, perhaps not until the 350s or, in the case of the south block of rooms in the church, even the early 360s. Alternatively, this material may date from renovation of B6 and removal or renewal of floor layers in the process (Fig. 3).

(4) Material dumped in spaces intended to hold waste or in places after they had ceased to be used, i.e., after the end of occupation; very often, such rooms had been blocked by walls built in doorways before this dumping began. Such material may come from contemporary use, but it may also represent debris found elsewhere and dumped, and such contexts may therefore contain more diverse contents. Rooms 9 and 10 contained some such debris, and Room 30 in B6 contained a couple of stratigraphic units of very mixed character.

The texts in this volume are thus arranged in four parts: (1) Ostraka coming from occupation and post-occupation layers, broadly speaking; (2) Ostraka coming from pre-construction dumped material; (3) Ostraka from the potentially post-occupation or renovation dump layers in Room 30 of Building 6; and (4) Ostraka coming from Area 4.1.

Figure 3. Plan of the baths (Building 6).

Α detailed discussion of the nature of the dumped material in Area 2.1, with more general reflections on the issues involved in working with dumped and recycled material, is given in an article of Rodney Ast and Paola Davoli, to which the reader is referred for a deeper analysis.²

In the following table, we list the stratigraphic units by area, categorize them briefly, and indicate which ostraka came from each of them. The reader is invited to read the full descriptions of the stratigraphic units in the database (www.amheida.com; in the on-line version of this book, links are provided), as they were prepared by the excavation supervisors at the time. These contain full details of their characteristics but are not written from the broader vantage point of the buildings as a whole. We need to point out that it is not always clear how to classify a stratigraphic unit, particularly where floors have not been preserved and it is not obvious whether dumped material belongs to a phase before or after construction; we have tried to preserve degrees of uncertainty in our description, but again the database provides far more detail.

TABLE OF ARCHAEOLOGICAL CONTEXTS OF THE OSTRAKA, WITH DATES

This table represents an expansion of that given in O.Trim. 1, pp. 61–73, so that it includes all ostraka in the two volumes. The reason for this practice is partly that our analysis of the material in the previous volume has become more nuanced over the past five years, and partly that the common characteristics of the dumped and occupational layers across the whole of Area 2 have become clearer as excavation has been extended to B5, B6, and B7.

Ostraka are organized in this table by their archaeological context, Area, Room (where applicable) and S(tratigraphic) U(nit); the SU is a D(eposition) SU unless indicated to be a F(eature). The brief descriptions of context in the fourth column are given for all areas except 4.1 and 4.2. All contexts in those areas were thoroughly disturbed and without stratigraphic value. Room 3 belonged to the house to the south of B1.

Dates are discussed in the description of stratigraphy above, pp. 5–7, and in the notes to individual texts. Precise dates are generally based on regnal years or indictions, coupled with archaeological context. If internal evidence other than years helps indicate a date, this is noted in the edition of the individual texts. Texts without internal evidence of date are dated according to their archaeological context; it will be evident that some texts are probably older than these dates would indicate, particularly in the case of layers of material dumped in preparation for construction. Occupation layers, however, appear to have little or no older material except where a chinking sherd has come out of a vault or wall collapse and mixed with occupational debris.

In the column with text types we note if well tags have the Pmoun (Pm), Moun (M), or Hydreuma-Pmoun (Hyd-Pm) formula.

1. This house and its finds are published in Boozer et al. 2015.

2. Ast and Davoli 2016.

3. CERAMIC FABRICS AND SHAPES

Clementina Caputo

The purpose of this section is to complement the papyrological study resulting from the textual and paleographical analysis of the ostraka of Amheida with information concerning the pottery supports on which they were written.¹

The reuse of ceramic materials for purposes different from their original ones is very common, not only at Amheida but everywhere in Egypt and in the ancient world at large.² One of the main reuses of ceramic sherds in antiquity was in fact as the support material for the production of ostraka.³ At Amheida, pottery sherds were also abundantly used inside walls and especially in vaults covering the rooms of the town’s buildings, prolonging the “life” of ceramic fragments, which became in fact building material.⁴ Additionally, excavations at the site have attested to the reuse of pottery sherds that were already disposed of in larger dump contexts as fillings for the foundation layers of new buildings. In this case, the refuse already accumulated in a dump was laid as a fill for foundation trenches and building yards, and subsequently sealed off by the floors of the rooms of later buildings. In Area 2 at Amheida, ostraka are noticeably abundant in this kind of deposition.

In the past, it has been rare for ostraka, widespread as they were, and not only in Greco-Roman Egypt,⁵ to receive much attention for any aspect other than their explicit written contents, and interest in the support, i.e., the type of sherd chosen to write on, has been particularly rare.⁶ Apart from a few exceptions, it was only in the 1990s that a certain interest in ostraka, both as texts and as artifacts associated with an archaeological context, started to develop.⁷

Indeed, when this important category of finds is associated with specific geographic, temporal, and spatial contexts it can inform us about a wide range of activities taking place during all periods of occupation of an ancient site. Their contents add to the archaeological data new evidence for the social and economic interpretation of the settlement and the dwellings or public contexts in which they are found.

In general, ostraka are often dated with far more accuracy than other archaeological evidence, thereby contributing to the dating of archaeological layers and their specific contexts.⁸ When dates can be ascribed to ostraka, they can also provide reference points for the general study of the ceramics, informing us of the timeframe during which particular shapes were in use. In turn, this information can provide dating to contexts in which ostraka without any internal chronological indications occur. Furthermore, knowledge of the different fabrics of ostraka and of the location of pottery workshops not only enables scholars to determine the place of manufacture of the potsherds but also the origins of certain texts and the commodities there mentioned.⁹

This innovative approach to the study of the ceramic support, used on the ostraka from Amheida, allows us to utilize the morphology of potsherds to create a ceramological classification of the ostraka that can then be compared with the groupings resulting from the paleographical study of the texts. The investigation of the array of types these fragments belong to is fundamental in addressing issues of chronology and technical production, as well as the identification of complete vessels from which they derive. Finally, through the combination of vessel types, dating, and texts one can speculate on whether complete vessels of a particular type were still in use at the time when the text was written, or if fragments from the broken vessels were reused after a relatively long period of time, possibly even coming from a dump layer.¹⁰ In general, it must be stressed that the date of the creation of the support provides only a terminus post quem for the writing of the ostrakon, and that in principle centuries could pass between the two acts.

3.1. METHODOLOGY

The direct observation and study of the supports of the ostraka took place during the 2011 and 2013 seasons at the SCA magazine at Ismant (near the site of Kellis), and focused on all fragments found from 2004 to 2013.

Each ostrakon made from pottery has been checked and compared with the remaining pieces comprising the corpus. Additionally, a ceramological study consisting of the classification of the supports according to the morphology, fabrics, and surface treatment was carried out for each ostrakon.

The macroscopic study of all inscribed ceramic fragments followed the criteria below:

• the composition of the clay;

• the color of the surface and fractures;

• the texture and appearance of the fabric;

• the color, appearance, size, frequency, and quality of the inclusions;

• the final processing and surface treatment (i.e., slip, decoration, etc.).

The diagnostic fragments for which a secure vessel type has been recognized have been drawn and compared with similar shapes already present in the Amheida ceramic catalogue.¹¹ The combined data regarding fabric, shape, and function made it possible to establish a typological classification of the fragments analyzed.

3.2. OSTRAKA AND CERAMICS

The corpus of the Amheida ostraka consists of approximately 889 pieces. The Greek ostraka form the largest part of the corpus, with 813 specimens, while Hieratic and Demotic ostraka are less frequent, 13 and 25 examples respectively. 12 pictorial and 26 unidentified ostraka complete the corpus (Table 1).

Table 1. Distribution of the ostraka by language.

The contexts in which the ostraka were found during excavation belong mostly to dump layers and occupational deposits (75%). The dump layers are primarily foundation fills or dumped waste, but ostraka were also found on the surface or embedded in the walls as building material (“chinking sherds”) (Figs. 4–5).¹²

In general, the range of datable ostraka coincides with that of other textual and numismatic evidence dating between the mid-first century and the end of the fourth century CE. The identified texts comprise accounts of different commodities, such as hay, oil, vinegar, wine, cotton, and bread, letters and delivery orders, lists of names, and writing exercises. The largest number of texts, with 344 examples or 36% of the total, are tags, especially well tags. Such small tags were used to label the contents of jars containing wine and other commodities and were placed in mud stoppers, as attested by three examples from the site where the ostrakon is still embedded in the stopper (O.Trim. 1.161, 200, and 204) (Fig. 6).¹³ Comparanda have also been found in 14 complete stoppers, still with their tags, from a Roman building at Kellis.¹⁴

Figure 4. Body sherds embedded in a vault as building material (“chinking sherds”).

Figure 5. Ostraka reused as chinking sherds in a dome covering a room.

Figure 6. Tags still embedded in mud stopper.

Among the items of information written on the well tag sherds were the name of the well or plot, the name of the tenant who has leased the plot, the vessel’s content, and the year of reign from which the contents came and, presumably, that in which the vessel with the tag was delivered.¹⁵ These ostraka, quite well attested within Dakhla, are far less common outside the Oasis.¹⁶

3.2.1. THE MAIN FABRICS OF THE OSTRAKA

When lacking diagnostic fragments, small sherds are difficult to attribute to specific vessel types or to date within a narrow timeframe. In these instances, the identification of distinctive fabrics and surface treatments become fundamental in recognizing some vessel types.

The characteristics of the clays and the classification of the main oasis fabrics and wares have been the subject of a number of studies.¹⁷ Specifically, the analysis of the sherds’ main fabrics in the case of ostraka from Amheida has been based primarily on the Dakhleh Oasis Fabric System’s classification by Colin Hope.¹⁸

Production of ceramics, at least in the region of the Dakhla Oasis,¹⁹ is characterized mainly by two fabric types: red- and brown-firing fabrics (A) and the orange-, green-, yellow-, and white-firing fabrics (B).²⁰ (Table 2)

Table 2. Percentage of vessel fabrics present on the site.

The first, and by far the most prominent, fabric (93.06% of the sherds used for ostraka at Amheida) is made of clays rich in iron oxides (iron-rich clays), ranging in color from red-orange (A1a/A2a) to gray-black (A1b/A2b). The distinction between A1 and A2 fabrics is not based on the clay’s composition, which is indeed identical, but on the addition of inclusions. In the A1 fabric, sand temper, quartz grains, calcareous inclusions, and red or black particles are very numerous. Conversely, the clay of group A2 is finer, resulting in a thinner and denser ceramic core. A1 and A2 groups are predominant both in the overall amount of sherds found on site as well as among the ostraka. This is in part due to the fact that these fabrics were used in the production of the most common ceramic types (i.e., storage and transport containers, cooking wares, and table wares). The existence of pottery workshops at Amheida during the Roman period (at least in Area 1)²¹ and at other sites in the oasis reinforces the hypothesis of a regional and local ceramic production of groups A1 and A2.²² As far as chronology goes, A1a and A1b were produced from the early second century until the mid- to late-fourth century CE, a timeframe confirmed by the ostraka found at the site. A2a and A2b are instead dated more narrowly to the third–fourth centuries.²³

A variant of these groups is fabric A5, similar in composition but over-fired, with a coarser texture, more porous, and with a larger quantity of white, green and yellow particles. The inherent qualities of this fabric make it the ideal choice for containers for liquids (i.e., jugs, kegs, and storage jars).

Type A11, also known as Christian Brittle Ware, is a kaolinitic brittle fabric characterized by a hard core and fine texture. The breaks differ according to the firing: lighter surfaces have usually bluish-gray cores, whereas darker surfaces correspond to cores with orange-pink zones. The inclusions consist of several red and black particles of various sizes (silicified clay platelets), quartz grains, and some small white particles of calcite. Quite commonly the external surface is coated in a red slip, which turns gray after firing. Used in flasks of the Third Intermediate Period/Late Period, this type was resumed from the late third and fourth century CE for thin-walled cooking jars, casseroles, and bowls. With a red and cream coating, these vessels are sometimes decorated in red dots on cream bands, a characteristic of Late Antique productions.²⁴

The B10 type is a marl clay with a porous texture; it is light in color and moderately coarse. The breaks are usually greenish-gray, with several grains of quartz, some limestone, and red and black particles of various sizes. This type was used almost exclusively for closed vessels (i.e., lids used also as footed bowls, jugs, and costrels). More specifically, water jugs are dated mainly, but not exclusively, to the middle of the third century/end of the fourth century CE.²⁵

The B3b type is characterized by an orange kaolinitic clay containing mostly red hematite inclusions of medium and small size, quartz, and platey fragments of shale and mudstone.²⁶ Usually this fabric is attested at Amheida in flasks and small bowls with yellow slip. Its production has been linked to sites located in the North of the Kharga Oasis.²⁷ The double-handled flasks, designed to contain and transport wine from Kharga, are small containers similar to lagynoi of the Hellenistic world (Fig. 10). They are characterized by a yellow slip on the exterior surface, sometimes decorated in a monochrome or bichrome painted motif. They can also present thick blackened traces on the inner surfaces, on the rim, and on the outside of the neck. These are traces of resins that covered the whole internal surface to seal them so that the liquid content would not be spoiled. The small convex bowls with yellow slip are characterized by a simple rim with spiral decoration on the inner surface. They are usually found in the same contexts as the flasks. This production appears in the Kharga Oasis in small amount at the beginning of what is called Phase III, dated to the 3rd century CE, and increases progressively during the 4th or 5th century CE, at the end of Phase III.²⁸

3.2.2. CERAMIC VESSELS

The morphological repertoire of Amheida’s pottery vessels includes a great variety of shapes, mainly of local or regional manufacture, as already mentioned above. Most of the fragments come from common wares, especially iron-rich fabric A1, and can be divided into closed forms (jars, kegs, water jugs, and cooking pots) and open forms (craters, basins, and small bowls).

Kegs constitute the main import from the oases of the Libyan desert. The manufacture of this type of container begins at the end of the Third Intermediate Period (25th Dynasty) in the oases of Bahariya (Qasr Allam),²⁹ Kharga (27th Dynasty at Ain Manâwîr), and Dakhla during the Late Period (fifth century BCE).³⁰ During the Roman Period and Late Antiquity, kegs have a large oval shape, short neck, cylindrical or slightly tapered, and a rounded triangular rim.³¹ Today, these vessels are still used, particularly in the Dakhla Oasis (El-Qasr), as churners, for storing cheese, or for drawing water in shallow channels.³²

Storage containers at Amheida include several types of jars classified according to the rim shapes. They usually present short necks and rims with quadrangular sections and sometimes grooves. Roman jars have usually a projecting rib at the junction between the neck and shoulder, a short neck, and a rim of quadrangular section,³³ or the surface right below the neck can be decorated with grooves.³⁴ Containers with rounded rims are characteristic of Late Antiquity.³⁵

Water jugs characteristic of the Late Roman Period and Late Antiquity have usually a pinched rim, narrow cylindrical neck provided with a filter, large carinated body, and flat base. The handle is attached to the lip and the upper part of the body. Generally, these vessel types are made with a marl coarse clay favorable to the keeping of water and its freshness (B10 fabric), but they can also be produced in a ferruginous clay rich in calcite particles (A5 fabric). In both cases, the outer surface is covered with a white slip.

All these containers are made in A1, A2, and A5 fabrics, attesting a local or regional production. They were intended to store water and local wine, but also dairy products (Fig. 7).

The large family of cooking vessels consists mainly of globular pots with rims of different shapes. They are usually short necked, slightly flared, and sometimes molded, or with thickened flared rims, or even neck-less with flat and flared rims. The fabric used for this types of vessel is A1, A2 or, from the late third and fourth century CE, A11 (Fig. 8).

Open forms used for food consumption and serving have many variants that are sometimes difficult to classify because of their very slow and minimal changes between the Imperial period and Late Antiquity (Fig. 9). Large bowls used in food preparation include craters and basins. Craters have usually outward everted rims, deep carinated bodies, and rounded bases. Most of the examples found at Amheida have red-purple wavy lines or spiral patterns painted over cream-colored bands that decorate the inner side of the rims. Very common at Amheida are also basins or deep bowls used in food preparation. They typically have slightly flattened rims with triangular sections, convex walls, and ring bases. Although these shapes are usually dated to the Early Roman Period (1st–2nd c. CE),³⁶ the comparative materials from houses at Kellis are dated to the Late Roman Period.³⁷

Small and medium-sized bowls, part of the table wares, can be divided into three main types, all very common at Amheida.

Figure 7. Transportation and storage vessels.

Figure 8. Cooking vessels.

Figure 9. Food consumption and serving vessels.

Small bowls with convex profiles and slightly domed or flat bases were used for the consumption and the preparation of food, as evidenced by the presence of black soot on the external surfaces. However, they were more often utilized as lids for storage jars, as proven by the presence of residues of gypsum plaster attached to their inner walls. Convex bowls were produced as early as the beginning of the Hellenistic period, both in the oases and in the Nile valley.³⁸ Parallels from Kellis date primarily to the second or third century CE.³⁹

A second type of bowl, with straight-sided rim, is characterized by slightly raised bottoms and sloping walls. Many of them have an internal ledge where the lid sat and present heavily blackened surfaces caused by exposure to fire during the cooking of food. This type was found at Kellis in contexts dated to the second and third centuries CE.⁴⁰

Finally, small bowls with high carinations on the upper body have small flared or tapered rims and convex bases. They are sometimes heavily blackened on the exterior surfaces, as they were used also as cooking vessels. The inner side of their rims is usually decorated with painted red dots on a cream-colored band. This decoration is characteristic of the end of the third century and fourth century CE.⁴¹

All these containers are made in A1 and A2 fabrics; they often present white drips and spots on both rims and bodies, and they are frequently decorated with red dots or thick marks on the rims, a decorative motif which is characteristic of the oasis during the Late Roman Period.⁴²

3.2.3. CHARACTERISTICS OF THE OSTRAKA

In general, all ostraka at Amheida are in a good or fair state of preservation. Most of the texts were written on the potsherds’ convex faces, though some examples were written on the concave sides. In a few instances the same text stretched over both sides of the sherd, or the same sherd was used for two separate texts, each written on one of the two sides. The texts were written for the most part in black ink, but there are some examples in brown ink.

The fragments were originally part of the vessels’ bodies, more commonly from the shoulders or the main body of the containers, areas that are generally straighter and have smooth surfaces and lack external ribs (98%). There are only a few instances where the texts are written on diagnostic parts of the vessel (i.e., rims, handles, bases); in fact, only 17 out of 893 Greek ostraka (2%) are diagnostic sherds (Figs. 11–14). Texts identified on these sherds are mostly memoranda, lists or uncertain (i.e., 1.77, 258, 355, 380, 402, 419, 2.504, 569, 644, 737, 806, 815, 838). None of these refers to trading or administrative transactions.

Most of the ostraka analyzed are from locally produced vessels: 122 from jars, 30 from kegs, 24 from jugs, 23 from basins, craters, and large bowls, 17 from small bowls, 6 from cooking pots, 6 from lids. In addition there are 318 ostraka also made from locally produced vessels, but for which it was not possible to define the shape. The highest percentage (96.50%) belongs to fragments in Group A fabrics (i.e., A1a, A1b, A2a, A2b, and A5), the same used for the production of the majority of containers present on the site. 1.60% is kaolinitic clay (A11), used for the production of containers, known as Christian Brittle Ware, 1.04% is in calcareous local clay (B10) used mainly for the production of jugs. Only 0.86% is in B3b, the fabric associated with the yellow slipped productions of the Kharga Oasis.

Table 3. Distribution of ostraka according to ceramic group (fabric).

No ostrakon made with fragments of Oasis Red Slip Ware or imports from other areas of Egypt and the Mediterranean was found (Table 3). The majority have irregular quadrangular shapes; however, ostraka used as tags seem to maintain a quite constant shape and size, with rectangular and triangular outlines. Very few ostraka are circular in shape (Fig. 15). These circular examples are the result of a secondary reuse of the sherd: indeed, the sherds were first cut to become circular lids used in stoppers to seal containers such as jars or amphorae, and only after the jars were opened and the lids disposed of did they become supports for writing. Interestingly, at Amheida one example shows that also the opposite process was possible. In this case, the sherd first was used as an ostrakon and only later was recut, partially obliterating the written text, to become a lid (Fig. 16).

The thickness of the sherds used as support for writing ranges between 0.5 and 1 cm (Table 4), dimensions that are characteristic of vessel types such as medium-sized bowls, jars, kegs, and cooking pots. Therefore, fragments of basins and craters were rarely used as support for writing, apart from a few notable exceptions. The size of the fragments is between 6 (min.) and 18.5 (max.) cm in the case of accounts or letters, while tags vary between 1.4 (min.) and 7 (max.) cm.

Figure 10. Wine flasks from the Kharga Oasis.

Table 4. Thickness range and writing surface of fragments in cm² of ostraka.

Contrary to what has been previously argued, some categories of texts, primarily well tags, were not written on randomly chosen sherds,⁴³ but rather they were specifically manufactured in large quantities by cutting body sherds into similar shapes meant to accommodate formulaic texts.⁴⁴

The process was probably as follows: once a ceramic vessel broke, as commonly happened because of heavy daily usage, the larger body sherds were selected and collected for reuse as ostraka. This could have happened inside each dwelling right after the vessel broke, or the selection could have happened after the sherds were discarded in an open area dump. At present there is no evidence that could exclude or confirm one hypothesis over the other. An attendant, possibly trained in this skill, would break these larger fragments using a sharp tool, most likely a flint or a hammer, by hitting them. Such a blunt stroke would produce a series of fragments differently shaped.⁴⁵ Only those fitting a predetermined size would be kept, while the remaining ones would be discarded.

The thickness of the sherds used was crucial, since thicker walls did not break easily; additionally, body sherds must have been preferred to rims, bases, or handles, as these thicker portions of a vessel did not allow for a regularly shaped break.

This process seems to be confirmed especially in the case of well tag ostraka, which, as said above, present roughly similar shapes and sizes, since they needed to fit on top of mud stoppers (Figs. 17–18). In the case of longer texts, such as letters or lists, the body sherds needed to be larger to fit more text, and to come from a rather straight portion of the vessel to facilitate the process of writing (Fig. 19). Larger vessels, like jars or cooking pots as well as kegs, seem to be preferred because of their larger body shapes, which provided a wider field for the text.⁴⁶

Several scholars have argued in the past that the color of the sherd’s surface and the texture of its fabric were the main discriminants behind the choice of specific sherds versus others. I believe, however, that fragments from Group A vessels were preferred at Amheida because of the intrinsic qualities of the fabric, and not only because this ceramic type was the most common at the site and thus readily available. Indeed, Group A sherds present a degree of hardness that allows for neat breaks.⁴⁷ This is not the case with B10 and A11 fabrics: during the experiment, B10 sherds were almost pulverized when they were hit by a flint tool, whereas A11 sherds shattered, and the fragments became unusable as writing surfaces. The choice of fabric was thus not at all random, nor was it dictated by qualities affecting its suitability as a writing surface. Those concerns, rather, drove the choice of vessel shape.

Figure 11. Diagnostic ostraka.

Figure 12. Diagnostic ostraka.

Figure 13. Diagnostic ostraka.

Figure 14. Diagnostic ostraka.

Figure 15. Circular lids reused as ostraka.

Figure 16. Ostrakon reused as a lid: O.Trim. 1.60.

3.2.4. CATALOG OF DIAGNOSTIC OSTRAKA

Fig. 11 a. O.Trim. 1.258

Text: Tag or Memorandum (275–350 CE)

Support: bowl (second–third century CE)

Fig. 11 b. O.Trim. 1.419

Text: Uncertain (275–350 CE)

Support: bowl (third–fourth century CE)

Fig. 11 c. O.Trim. 1.446

Text: Uncertain (275–350 CE)

Support: dish/bowl (second–fourth century CE)

Fig. 11 d. O.Trim. 1.355

Text: Tag ? (345–370 CE)

Support: dish/bowl (mid second–mid fourth century CE)

Fig. 11 e. O.Trim. 1.402

Text: Uncertain (275–350 CE)

Support: Knob of lid/footed bowl (second–third century CE)

Fig. 12 f. O.Trim. 2.644

Text: Tag Pmoun (286/7 CE)

Support: lid (second–third century CE)

Figure 17. Some Well Tags with Pmoun formula.

Figure 18. Some Well Tags with Hydreuma-Pmoun formula.

Figure 19. Letter and some receipts of Serenos.

Fig. 12 g. O.Trim. 1.381

Text: Uncertain (?)

Support: lid (second–third century CE)

Fig. 12 h. O.Trim. 1.77

Text: List ? (350–370 CE)

Support: lid (fourth century CE)

Fig. 12 i. O.Trim. 1.225

Text: Tag (350–370 CE)

Support: lid (fourth century CE)

Fig. 12 j. O.Trim. 2.815

Text: Miscellanea (350–370 CE)

Support: Large bowl (fourth century CE)

Fig. 13 k. O.Trim. 2.504

Text: Memorandum (350–370 CE)

Support: Large bowl (fourth century CE)

Fig. 13 l. O.Trim. 2.806 + 807

Text: Memorandum (Philippos) (350–370 CE)

Support: Large bowl (fourth century CE)

Fig. 13 m. O.Trim. 2.838

Text: Letter (350–370 CE)

Support: Large bowl (third–fourth century CE)

Fig. 13 n. O.Trim. 2.737

Text: Memorandum (275–350 CE)

Support: Large bowl (first–third century CE)

Fig. 13 o. O.Trim. 1.380

Text: Uncertain (350–370 CE)

Support: Jar (fourth century CE)

Fig. 14 p. O.Trim. 2.569

Text: Uncertain (second half of fourth century CE)

Support: Base of bowl (second–fourth century CE)

Fig. 14 q. O.Trim. inv. 11060

Text: Uncertain (not published)

Support: Base of jug (second–fourth century CE)⁴⁸

3.2.5. FABRICS

A1a	- Iron rich clay with considerable amounts of sand temper, relatively coarse and poorly fired. It contains frequent medium-to-fine quartz grains and less frequent coarse ones. White calcareous inclusions of varying sizes, some fine-to-medium red and occasional black particles are visible. Medium hard fabric.48 - Generally firing to a pale red, red or red-brown color (oxidizing atmospheres). Gray core can be present especially in thicker sections. Surface can be pale red, brown-red or red in color. - Storage and transport containers, cooking ware, and table ware. - Early second century until the mid- to late-fourth century. Nr. Ostraka: 512
A2a	- A fine version of A1a. The fabric has abundant mineral inclusions, with quartz grain particles, smaller than A1a fabric. Texture is denser and the hardness is slightly higher than A1a. - Fired colors range from light red and brown to reddish-brown, sometimes with a gray to brown-gray core. Surface can be pale pink-beige, brown-red or red in color. - Small bowls, large jars and kegs. - Third–fourth centuries. Nr. Ostraka: 21
A1b	- Similar in composition to A1a with rounded quartz and white calcareous inclusions which are usually numerous and conspicuous. The reducing atmosphere increases the hardness. - Fired black, gray or sometimes dark reddish-gray colors in reducing atmospheres. Surface is gray in color. - Large jars and kegs. - Early second century until the mid- to late-fourth century. Nr. Ostraka: 409
A2b	- A fine version of A1b. Same characteristic of A1b but with smaller size temper. - Fired gray in a heavily reducing atmosphere. Surface is gray in color. - Large jars and kegs. - Third–fourth centuries. Nr. Ostraka: 19

A5

- Variant of group A. Iron rich clay, with a larger quantity of white, green, and yellow particles. Coarser texture, more porous. - Overfired purple-brown and gray in color. The surface is purple usually with a thick white or cream slip. - Jugs, kegs, and storage jars. - Third–fourth centuries. Nr. Ostraka: 62

A11	- Dense kaolinitic brittle fabric characterized by hard core and fine texture. Medium scatters of small calcareous inclusions, dark red and sometimes black particles are visible. Known also as Christian Brittle Ware. - It fires a range of colors including pale, beige, apricot, and pale gray. - Thin-walled cooking jars, casseroles, and bowls, with a red and cream coating, sometimes decorated in red dots on cream bands. - Late third and fourth century CE. Ostraka: 17
B10	- Marl clay. The fabric is lightweight, extremely porous texture and moderately coarse. Several grains of quartz, some limestone, and red and black particles of various size. - Fires a luminous pale, gray-green color. - Lids used also as footed bowls, water jugs, and costrels. - Mid third century to the end of the fourth century CE. Ostraka: 11
B3b	- Kaolinitic clay with red hematite inclusions of medium and small size, quartz, and platey fragments of shale and mudstone. Calcium rich clay with very few inclusions. - Fired an orange-red in color. - Flasks and small bowls with yellow slip. (Kharga Red and Yellow Slip). - Third to fifth century CE. Ostraka: 9

1. I would like to express my gratitude to Professors R. S. Bagnall (Director, ISAW) and P. Davoli (Field Director, University of Salento) for entrusting me with the study of ceramics of the ostraka found at Amheida, and for giving me the opportunity to present the results of this work. Special thanks go to Professor P. Ballet (University of Paris Nanterre) for her teaching and supervision, and D. Dixneuf (CEAlex) for advice and help given to me in the early years of fieldwork. I also thank my colleagues J. Marchand (PhD, University of Poitiers) and I. Soto (PhD student, ISAW). Additional thanks to B. Bazzani, R. Casagrande-Kim, and V. Liuzzi for their assistance in the writing of this contribution. The study of ceramic supports used for the ostraka at Amheida and at Dime es-Seba (Fayoum) is also the subject of my PhD dissertation defended in July 2014 both at the University of Salento (Lecce) and at the University of Poitiers. Advisors for this research are, respectively, Prof. P. Davoli and Prof. P. Ballet.

2. Lister and Lister 1981; Dupré Raventos and Remolà (eds.) 2000; Ballet, Cordier, and Dieudonné-Glad (eds.), 2003; Peña 2007.

3. Peña 2007: 119–192.

4. According to Bonnet, the potsherds reused in the walls were contemporary to the building, while those present in the fillings were collected in dumps: Bonnet 1994; Henein and Wuttmann (eds.) 2000: 76. See also Ast and Davoli 2016.

5. Peña 2007: 160.

6. In 1899 the papyrologist U. Wilcken showed interest in the ceramic supports used for the ostraka, but it remained an isolated case: Wilcken 1899: 4. Publications from 1902 to 1986 provide almost only philological information. In fact, there are just a few works in which short descriptions of the color of the fragments used are given, but this information is insufficient for identifying the pottery supports: see Amundsen 1935; Gascou 1979; Devauchelle 1983.

7. A substantial breakthrough in the study of these objects occurred in 1994 with a publication of S. P. Vleeming, who, with the cooperation of R. Van Walsen, gave an accurate ceramic description of the ostraka examined: Vleeming 1994: 2–3, 149–156. Two works that are certainly the best example of interaction between philologists and ceramicists are those of Worp (2004), in which the analysis of the supports is managed by Colin A. Hope, and Lozachmeur (2006), in which the analysis of the ceramic supports is by P. Ballet: Hope 2004a: 5–28; Ballet 2006: 106–133.

8. Bingen 1996: 31.

9. Hope 2004a: 5.

10. Bingen 1996; Cuvigny 2006.

11. Since 2009 I have been entrusted with the cataloguing and study of ceramic vessels found during the excavations conducted on the site. This experience has led to better knowledge of the Dakhla Oasis productions and shapes.

12. Ast and Davoli 2016.

13. Bagnall and Ruffini 2012: 13.

14. Bagnall and Ruffini 2012: 13. One of these tags (O.Kell. 270) was published in the first volume of Greek ostraka from Kellis, while the others are still unpublished, Worp 2004: 169.

15. Bagnall 2011: 132–133; Bagnall and Ruffini 2012: 13–14, 31–37.

16. Similar tags have been identified at Bakchias, Karanis, and Tebtynis (Fayyum). The text on these tags consists of the name of the owner or a date, referring to the delivery of the commodities or the year of harvest. H.C. Youtie was the first to suggest that these ostraka probably were located on wheat bags. However, the finds from Amheida and Kellis suggest another use for the labels: H. C. Youtie, Scriptiunculae posteriores I (Bonn 1981) 122–126; Reiter 2007: 266 and 274–277; Reiter 2005, especially 132.

17. For the characteristics of the clay and ceramic materials of the oases, see: Soukiassian, Wuttmann, Pantalacci, Ballet, and Picon 1990: 75–85; Marchand and Tallet 1999; Hope 1999; Patten 2000: 87–104. See also Nordström and Bourriau 1993.

18. For the fabric descriptions see Hope et al. 2000; Hope 2004a: 7–9. Hope has worked extensively on the production of pottery artifacts in the Dakhla Oasis for all historical periods. Most of the fabric descriptions here are taken verbatim from conversations with Colin Hope and Pascale Ballet.

19. Concerning the clay, according to Ballet, the Amheida production differs from the Kharga wares, which are mainly made with kaolinitic material (with red inclusions, probably hematite, and “plaquettes,” which are silicified clay inclusions).

20. Hope et al. 2000: 194.

21. The study of ceramic materials from B2 (Area 1) by D. Dixneuf (CEAlex) appears in Boozer 2015.

22. Hope 1993.

23. Hope 2004a: 9. The standard fabric was in use from the fourth millennium (Sheikh Muf-tah Cultural Unit) and occurs throughout all subsequent periods: Hope et al. 2000: 194. According to P. Ballet, the red clay of the Mut Formation could have been used for making pottery in Amheida. The clay used at the end of the Old Kingdom in Balat seems to be the same used for the Roman production in Amheida. The ceramic types of Amheida production are close to the Old Kingdom material from Balat, especially for the slight or strong over-firing process. For the production of Balat see: Ballet and Picon 1990: 83–84.

24. Hope 1999: 235; Hope et al. 2000: 194; Dunsmore 2002: 131; Hope 2004a: 9; Dixneuf 2012: 459.

25. Hope 2004a: 9.

26. The fabric description has been defined with P. Ballet during a workshop in January 2013 and February 2014 in Dakhla.

27. Ballet and Vichy 1992: 116–119, Fig.13 (g–h); Ballet 2004: 224–225, 237 (Fig. 220, nos. 48–50).

28. Ballet and Vichy 1992: 119; Ballet 2004: 221–225.

29. Rougeulle and Marchand 2011: 443, n. 13.

30. Marchand 2000 : 221.

31. Patten 2000: 233 and pl. 79 (SS160k and SS91e; Form 157, Phase 4+); Hope et al. 2006: 189.

32. Henein 1997: 161–166.

33. Hope 1999: 232; Hope 1987: 171, fig. 5 (o: 2nd–3rd c. AD); Patten 2000: pl. 68 (Form 123/1, H7-1/82/18a); Bowen et al. 2007: 26 and 42, fig. 5 (c: end of 1st–2nd c. AD).

34. Hope 1987: 171, fig. 5 (p: 2nd–3rd c. AD); Patten 2000: 228 and pl. 77 (Form 138, Phase 4); Hope et al. 2006: 27 and 48, fig. 3 (d.69: end of 1st –beginning of 3rd c. AD); Patten 2000: 228 and pl. 77 (SS11 1j: Form 138, Phase 4).

35. Patten 2000: 227 and pl. 75 (SS18ab: similar to Form 136, Phase 4).

36. Patten 2000: 177–179, Pl. 52 (Form 62/2); Hope et al. 2006: 27, 49 fig. 4 (a).

37. Dunsmore 2002: 137, fig. 5 (a 97/106a).

38. Hope 1987: pl. 171, fig. 5 (n–d) and fig. 5 (l); Patten 1999: 85, fig. 2.12; Patten 2000, Form 10/1 and p. 42; Patten 2000: pl. 46 (Form 36); Hope 2003: fig. 18 (f), p. 269, fig. 20 (b–c); Hope 2004b: 40, fig. 7 (a, c, f) and 41, fig. 9 (d, e, h).

39. Hope 2003: 213, fig. 5 (a), 269, fig. 20 (b–c) and 272; Dunand et al. 1992: 48–57, Pl. 79 (3).

40. Hope 1987: 171, fig. 5 (c); Patten 1999: 85, fig. 2 (18); Patten 2000: pl. 41 (Form 7 R-F) and pl. 41 (Form 6/1).