Shine Like the Sun: Chapter 4: Egypt c. 700-1200 INTRODUCTION As in Iraq, the Muslim armies invading Egypt founded a new capital for the region: Fustat. Although Egyptian pottery of this period is found on other sites, Fustat is the chief source of material for study (see Appendices A and C). After abandonment of much of the site, the remains were continually dug over in pursuit of building stone, antiquities for the market, and particularly the phosphatic earth (see Appendix C). Many museums have collections from these diggings; those at the ROM, the Ashmolean Museum, and the MMA were available for detailed study. These collections are heavily weighted in favour of the finest wares, for which the diggers knew there was a market. Among more recent excavations are those by the American Research Centre in Egypt (ARCE), directed by George Scanlon. Material from the ARCE diggings is in the collections of the ROM (see Appendix A), but no stratigraphic information accompanied the finds. Many types that are not of art-historical interest, such as those of the Lead-glazed classes, are represented only in the corpus of material from the ARCE site. Pottery from research in the Yemen (Mason and Keall 1988a) and Aqaba (Whitcomb 1988, 1989) were also available for personal study. Apart from the studied collections, published pieces from Pisa (Berti and Tongiorgi 1981), Qaysariyya/Caesarea (Pringle 1985), Beth-Shan (Fitzgerald 1931), and the Benaki collection (Philon 1980) have also been included, as whole profiles have been difficult to obtain. Although not included in the corpus, other collections have been examined in order to ensure the validity of the typology. They include the collection of the British Museum, the exhibited study collection of the Victoria and Albert Museum, the published collection of the Benaki Museum (Philon 1980), and other published pieces (in particular Bahgat and Massoul 1930; Yusuf 1958; Butler 1926; Mathaf 1922). All of these collections are probably of material obtained through the phosphatic-earth diggings. The corpus will be considered under the headings of typology, technology, and fabric characterization/provenance, with a synthetic conclusion. TYPOLOGY As described in the methodology chapter, the corpus has been divided into a hierarchical taxonomic structure, with major groups or "classes" defined by technology, subdivided into "types" defined by decorative technique, and finally stylistic "groups" defined by form and motif assemblage. The classes defined include the early "Semi-glazed" class; the Opaque-glazed class, including Splashed, Lustre-painted, and Polychrome-painted types; the "Wollastonite-slip" class, including Lustre-painted and Underglaze-painted types; the Lead-glazed class; the Polychrome-glaze class, and the Incised ware class (Figs. 4.1 and 4.2). The classifications are based partly on technical analysis, but in each case they represent obvious visible characteristics. The important Egyptian Lustre-painted pottery is part of the Opaque-glazed (OG) and the "Wollastonite-slip" (WS) classes. Lustre-wares receive most attention because typological analysis of this pottery has provided the chronological framework for the rest of the corpus. By seriation of the ceramic attributes, consideration of inter-group relationships, and reference to available dating evidence, a chronological framework is developed for the corpus. Following each type name heading is the short form used in tables. Semi-glazed class (Semi-glaze) Pottery with lead glazes is thought to have been made in Egypt in pre-Islamic times, largely associated with the widespread Roman Lead-glazed moulded ware industry (Peacock 1982:63-65). The suggestion that these moulded wares were also made in early Islamic Egypt (Lane 1939) is not improbable, but the scale of production does not seem archaeologically significant, nor does there appear to be any good evidence for it. The Semi-glazed class appears to be the earliest specifically Islamic representative of the lead-glazed wares. Pottery of this class primarily comprises a type of bowl with a flat base and discrete bands of glaze (see Fig. 4.1) although distinct designs are also noted, such as a bowl in the Ashmolean Museum with a peacock (Allan 1991:10-11) and other birds from Alexandria (Rodziewicz 1978, 1984). Scanlon (1998) has published a catalogue of wares of this type from Fustat. According to George Scanlon (1966:83-112, 1974:60-78), this pottery is found in association with slip-painted Coptic wares, red-slipped wares derived from Terra Sigillata or Samian pottery, and other evidence dating the assemblage to the eighth century. Given the angular "Roman" look to the form and the possibly indigenous nature of the glaze technology, such an early date does not seem unlikely. The early date is also strongly supported by the evidence from Alexandria where pottery of this description is found (Rodziewicz 1976, 1978, 1984). Similar pieces from Palestine are also considered to be of early date, although the author has observed these only in publication and cannot state conclusively that they have common origins. These include finds of allegedly reliable Umayyad dates from Khirbat al-Karak and Ramla (Delougaz and Haines 1960, pl. 62.1; Rosen-Ayalon and Eitan 1969, fig. 1), and "probably . . . ninth century" finds from Qaysariyya/Caesarea (Holum et al. 1988, fig. 157). Pottery of this type from the excavations at Aqaba, where it is also considered to be of Umayyad date (Whitcomb 1989), have been personally studied by the author and are positively identical to the Semi-glazed wares from the Fustat excavations. Splashed type of Opaque-glazed class (OG-Splash) A few examples of Opaque-glazed class may indicate some production in the ninth or early tenth century, because they reflect forms suitable to that date (Fig. 4.1, compare with Fig. 3.11). Their decoration consists of an opacified glaze splash-painted with cuprous green and manganese-purple, with the paint applied over the glaze in a technique derived from the Blue-painted type of Iraq (see Chapter 3). Scanlon (1974) refers to such pieces as "Fayoumi" wares. One piece of this type in the Benaki collection, obtained in Egypt, is considered not to have been produced in Egypt (Philon 1980/:45), although petrographic analysis of another of this type supports an Egyptian origin (Fig. 4.1: FUS.20). Lustre-painted wares (OG-Lustre and WS-Lustre) The appearance of lustre-painting of pottery in Egypt is usually considered to be a direct transfer of technology from Iraq. There has been some suggestion (Schnyder 1963) of earlier transfer during the Tulunid dynasty (868 905; see Bosworth 1996:60-61). This possibility was lent credence by petrographic analysis of a bichrome (or possibly polychrome) Lustre-painted piece excavated at Fustat, which does not strictly conform to the Basra Petrofabric (Mason and Keall 1990). As considerably more Basra products have been identified at Fustat subsequently, it is considered likely that this is in fact an anomalous Basra fabric, and that Lustre-wares were not produced in Egypt before the mid-tenth century. The bulk of Egyptian production is associated with the Fatimid dynasty (969-1179 in Egypt; see Bosworth 1996:63-65). Traditional arguments for a direct transfer of technology from Iraq to Egypt by the movement of potters comprise the very close stylistic similarities, the effectively simultaneous cessation in Iraq and commencement in Egypt, and the technically demanding lustre-pigment technique. The present study lends considerable corroboration to this hypothesis. Previous comments on the similarities between the terminal Iraqi and initial Egyptian products was based on general impression. The new data suggests that the parallels are so close as to represent a direct continuation of tradition. Copies of the Basra products are widespread in the Islamic world of the ninth and tenth centuries, although these are always made using traditional slip-paint technologies. In each of these cases consideration of the vessel forms illustrate that the regional wares were made in the context of local vessel-forming traditions. Effectively it is impossible for a regional potter to emulate precisely the motor-habit patterns of another potter, at least sufficiently to delude the callipers of a careful archaeological ceramicist. This is not so of the first Fustat products, which are in some cases indistinguishable from the Basra products. The restricted nature of the knowledge concerning the lustre technique is also further enforced in these new results, as the provenance attribution of the wares indicates only one centre at a time in each region: Basra and Fustat. Added to this evidence are new data on technological transfer, as other technical attributes such as adding glass to a ceramic body and making a white slip with crushed quartz were also transferred from Iraq to Egypt at this time (see below). Coupled to this evidence, there is historical support for movement of personnel at this time (Kennedy 1986:207, 223, 227). The corpus of Egyptian Lustre-painted wares assembled from the various sources was studied using the methodology described above, with various attributes of form and motif subjected to seriation. The range of motifs is illustrated in Figure 4.3, their occurrence given in Table 4.1, with forms illustrated in Figures 4.4 to 4.6. The motifs are labelled the FL sequence, signifying Fustat Lustre-painted. Through seriation it was possible to discern a sequence of four principle groups. Although glaze technology was not included in the seriated data, the first two groups were of the Opaque-glazed class, while the second two were mostly of the "Wollastonite-slip" class (see technology section, below). Petrofabric was also not included in the seriated data, but again the first two groups, either of stonepastes or of modified Nile, were largely distinct from the second two, which were mostly of a highly calcareous clay (see below). The sequence is labelled FLP1 to 4 for Fustat Lustre-painted. Fustat Lustre-painted Group One (FLP1) This first group of Egyptian Lustre-wares may be divided into two subgroups. The forms and motifs are generally most closely related to the terminal products at Basra, but among the group as a whole there are some that are practically indistinguishable from the Basra wares (FLP1a), while others exhibit forms and motifs for which a step away in stylistic development may be argued (FLP1b). The first group is likely to be generally earlier than the latter group. Vessel forms in this group (see Fig. 4.4) always have variants of the square "ring 2" base, continuing directly from Iraqi production. Rim treatments include debased versions of Iraqi "recurved 2" (e.g., STL.01) and straight rims on conical vessels (e.g., FUL.13). A new development is a flattened rim on hemispherical bowls, called here a "camel" profile, after its similarity to the beast (e.g., FUA.39). The early "camel 1" version typical of FLP1b is flattened at the top by application of a tool. Tooling and turning of the vessel after throwing, presumably while leather-hard, is another characteristic of this group. This is particularly so for those with stonepaste bodies, such as STL.01 (see also Fig. 4.1), which has lotus petals carved on the outside. Group One decoration (Fig. 4.3) includes a number of Iraqi continuities, notably the "ring-and-dash" styles of embellishing the reverse (motif FL.1 series). There is a variety of these, including single rings and double concentric rings. Rather than infer that the Egyptian products echo the development from single to double ring that has been suggested for Iraq, which would imply contemporary production, it is preferable to consider the process to be one of simplification from double (motif FL.1a), to single (FL.1b), to two curvilinear dashes (FL.1c), to a few desultory dashes (FL.1d). Other Iraqi motifs include "V-field" (motif FL.2), "swirl-field" (FL.3), "hook" (FL.4), "bull's-eye" (FL.5), "ladder-band" (FL.6), and "crescent-rim" (FL.7). A second and less common group of reverse motifs is associated with the "palmette" style derived from Iraqi practice. Perhaps the most characteristic Egyptian motif is a stubby rounded palmette, rather like a clover leaf (motif FL.8 group), which is often encompassed by a curling stem, which in turn has characteristic dots or knobs along it. Such "knobby" stems may have a derivation from local tradition, as unglazed painted wares of Umayyad date from the Levant have an identical motif (e.g., Baramki 1942, fig.7.13; Tushingham 1972, fig. 6.11, 6.37; ‘Amr 1986). The most common rim decoration is the "plain band" motif (FL.9), with others in "wedge" motif (FL.10), and also "festoon" (FL.11). Fustat Lustre-painted Group Two (FLP2) Vessels in groups Two to Four generally seem to have no decoration on the exterior, and when it is present the decoration will not be of the standardized nature of the "dash-and-circle" motif. Of course this may not be considered a hard-and-fast rule, as many (possibly later?) Group One pieces had only the occasional desultory dash, which may not be patent on a sherd of the vessel. By cessation of the tradition of exterior decoration the potter would have saved time in execution, and also used less of the expensive lustre-pigment. Group Two bases (Fig. 4.5) are mostly taller than those in Group One; they are generally tapered and rounded, and they splay outward about ten to twenty degrees from the vertical ("acute splayed" foot, see Chapter 2). Rim forms include a continuation of the "camel" rim style; in this case the "camel 2" profile is less angular, partly because of the shallower forms, but also because a tool has not been used to flatten the rim. Using a tool takes more time (you have to pick it up), so abandoning this practice would have saved time in production. Straight rims from conical bowls are also present, as are simple rims from hemispherical bowls. Tooling is unknown on the interior, and is generally restricted to gouging a line or lines on the exterior. The bases have all been turned, and it is probable that this exterior line was executed while the vessel was upside-down on the wheel for base-turning. Decorative schemes include continuation of some motifs from the previous group, and some developments. The broad "camel 2" rim receives a new suite of motifs, including "dot-crescent" (FL.12), often accompanied by the highly characteristic "incised-band" motifs (FL.13). Other rim motifs include "arrow" (FL.14) and "zebra" (FL.15). The "band" motifs are added to with "dot-band" (FL.16), "dot-chain" (FL.17), and "chain" (FL.18), probably all ultimately derived from the "bull's-eye" motif. Philon (1980) publishes a number of pieces that are apparently of this group (without accurate drawings it is impossible to be certain), which have calligraphic decoration consisting of the repetition of the word al-yumn ("good fortune") on the interior (e.g., Philon 1980, figs. 496, 502, 514). The bulk of pieces with the inscription sa‘d, also probably a blessing, would similarly appear to be attributable to this group (Jenkins 1988), although again the attribution to this group must be verified by personal study of sufficiently diagnostic pieces. Fustat Lustre-painted Group Three (FLP3) This and the subsequent group appear to be almost exclusively of the "Wollastonite-slip" class. Technically this will be described in fuller detail below, but suffice it to say that it has a clear glaze, and light-coloured "slip," at variance with the opacified glazes of Groups One and Two. The dominant form in this group is a bowl with a recurved rim and a splayed foot (Fig. 4.6). The recurved rim sees a return to a practice that appeared to have died out in Group One, with only rare examples in Group Two. Most likely the returned predominance of the form may be due to a further influx of influence from China, the commonly supposed home of the recurved rim. The Group Three foot carries on logically from the developments of the previous groups, as it has become even more splayed, usually twenty to thirty degrees from vertical, and is increasingly triangular in section. Motifs generally show debased forms of the previous groups, such as "broken ladder" (motif FL.19), "broken chain" (FL.20), and the ultimately debased "squiggle-band" (FL.21). Fustat Lustre-painted Group Four (FLP4) The last group of Egyptian production of Lustre-painted pottery has a short triangular base, splaying out obliquely, and often has a broad flattened rim (e.g., Fig. 4.6: FUS.03). Generally the potting is very thin and the pink fabric quite hard, an aspect pointed out by Lane (1947:24), who also considered these to be the tail-end of production. Lane also points out the degraded nature of the artistic qualities of this material, which, although not unattractive, does not have the same merits as earlier products. The most common design motif is the "crab-claw palmette" (motif FL.22 group), often accompanied by "discrete tendril" (FL.23). A larger and closely related group is the Underglaze-painted group (see below), which has identical forms, motifs, body, and glaze technology. Fustat Lustre dating Two lustre pieces have been published with inscriptions dedicating the vessel to courtiers of the Fatimid Caliph al-Hakim (r. 996-1021). One of these is in the Islamic Museum, Cairo, and is dedicated to Gaban, referring to titles held from 1011 to 1013 (Jenkins 1988). Illustrations of the front of this piece do not show any standard motifs, and the form is unpublished, but the design is of the palmette style previously noted for late Iraqi production and also made in Egypt. The reverse has the two-ring variant of the "ring-and-dash" motif. The second piece, in the Benaki Museum, is dedicated to another courtier of al-Hakim. Publication of this piece (Philon 1980) includes a sketch-profile, which appears to represent a carinated version of the weakly recurved forms, much like FUR.11 (Fig. 4.1), with a short ring-base, all Group One characters. The decoration includes the two-ring "ring-and-dash" motif on the reverse (Jenkins 1968b, Appendix, no. 9); while the obverse shows lobed stems, small palmettes, and band motifs typical of Group One. This latter piece is also signed by the artist Muslim bin al-Dahhan. Other pieces considered to have been signed by the same individual have been catalogued by Jenkins (1968b), and illustrations of this group all show attributes characteristic of Group One (two forms of vessels signed "Muslim" are included in Figure 4.4: MMA.48 and FUM.20). Hence, this study would support Jenkins's conclusions that these are indeed the work of one potter or workshop. All of these pieces do, however, show some distance between their attributes and those of the Iraqi-Egyptian transition, indicating a later date. This would suggest a date for the beginning of Egyptian Group One of c. 975. Another group of correlative ceramics are the glazed bowls from the wreck of Serce Liman. This material, dated securely to c. 1025 (Bass and Doorninck 1978), has glazed pottery of Splashed-Lead-glazed class. Petrographic analysis indicates a source in the Fatimid sphere in the southern Levant or possibly Egypt (Mason, in press), while isotopic studies of the lead used in the glaze indicates that it is the same as that used in Fatimid Lustre-ware, although its source is probably in Asia (Mason et al. 1992). Jenkins (1992) has already shown the relationships between the decoration on these vessels and general early Fatimid design. The forms of the Serce Liman vessels (for an example of the Serce Liman type, see Fig. 5.1 CSR.03) are segmental bowls with straight rims (parallelled by FUA.40 and FUA.32, Fig. 4.4, and short, slightly splayed bases (cf. FUS.05, Fig. 4.4, and FUA.52, Fig. 4.5). Such a form would be suitable to the transition between Groups One and Two. Although one must consider it possible that the regional products may have lagged in style somewhat, a tentative date of c. 1025 for the transition between Group One and Group Two would be tenable. Vessels incorporated into European public buildings, which have known construction dates, also provide some dating evidence, although one must be aware that they may have been introduced at any time after their production, and in some cases may have been placed after construction. One of the largest and best published of these corpora are a group of lustre-painted bowls incorporated into Pisan churches (Berti and Tongiorgi 1981). Chemical analysis of the Pisan bowls (Berti and Tongiorgi 1981:287-88) reveals a number of groups, of which two are comparable to Egyptian clay bodies. Among these wares with probably Egyptian chemistry, one vessel (no. 3) is typologically attributable to Fustat Group Two and has a body commensurate with this group. A further vessel is typologically attributable to Group Three although it has a body like those of Group Two (perhaps of transitional date). A third vessel has the style and body typical of Group Three. Based on the date of their host buildings, Berti and Tongiorgi (1981:210-11) attribute the first two bowls to the first half of the eleventh century, and the third to the last quarter of the eleventh century (Berti and Tongiorgi 1981:259). This would indicate a transition from Egyptian Group Two to Group Three in c. 1075. The remainder of the Pisan Lustre-wares differ in chemistry from the Egyptian clays analysed in this study, and pieces with precise analogues of their decorative motifs have not been found on any proven Egyptian vessel, suggesting that they originate elsewhere, most probably in Spain. This would stand as a caution to those who would use these vessels to say too much about Egyptian pottery. However, the forms and motifs are such that it could be argued that they would be a direct continuation of practice at Fustat towards the end of production of Group Two. Such precise dating is due to some elements reflecting transitional phenomena between Fustat Lustre-painted Groups Two and Three. Berti and Tongiorgi (1981) date the earliest of this group to the last quarter of the eleventh century, which again would put the transition between Egyptian Groups Two and Three at 1075. Similar migrations of potters from Fustat will be suggested for roughly this period when discussing the origins of Lustre-wares in Iran and Syria (see below), and these Pisan bowls appear to indicate that potters left for Spain at this time also. Other possible correlative ceramics from European buildings include those in the Civic Tower in Pavia, built in the second half of the eleventh century and including one probable Group Two or possible Group One piece (Aguzzi 1973-75, fig. 2). Excavations at Corinth produced significant amounts of Islamic pottery dated collectively to the eleventh and twelfth century—the heyday of medieval Corinth (Stevenson 1947, figs. 147-50). These are consistently misidentified as "Persian," but include at least one Basra Group Six Lustre-ware, some Fustat Group One Lustre-wares, Underglaze-painted wares equal to Fustat Lustre Group Four, and Incised wares which may be of either Egyptian or Syrian origin. Comparative pieces from Qaysariyya (Caesarea) and Beth Shan (Fig. 4.6) represent a number of examples from the Levant. Here a useful archaeological horizon is often found at the first arrival of Crusaders in the region (1098-1099). Pottery from pre-Crusader levels includes a number of Group One and Group Two examples, including pieces from Qaysariyya (Brosh 1986, pl. V.3), Yoqne‘am (Ben-Tor and Rosenthal 1978, fig. 5.2), and Aqaba (Whitcomb 1988, fig. 8.a-h, includes Iraqi wares; a much larger corpus of this material from Aqaba has been studied by the author). Pottery from Crusader-period levels includes a large amount of Group Three and particularly Group Four pottery, the latter dominated by Underglaze-painted wares. Crusader-period pottery includes the illustrated profiles from Qaysariyya (Pringle 1985), and numerous others from this site (Holum et al. 1988, fig.163 centre; Brosh 1986, figs. 3.12, 3.13, pls. 2a-b, 3-4; Blakeley 1987, fig. 46). Other pieces include the illustrated profile from Beth-Shan (Fitzgerald 1931, pl. XXVII.3). Brosh (1986:70) makes reference to a large body of unpublished examples of Groups Three and Four Underglaze-painted wares in the Levant, and argues for a local origin. As the author has not inspected all of the Levantine material in person, a local origin for this material can not be ruled out, but the precise parallel in form and the singular characteristics of the motifs argue against this, and two fragments from ‘Acharneh are certainly Egyptian (see Appendix C). Arthur Lane (1947:22-23) considered all Lustre-painted pottery inscribed with the word Sa‘d, to be the product of a single potter working in the late Fatimid period, so this group should also be considered as possible dating evidence. Published pieces with this inscription have unfortunately not included profile drawings, so the full information is not available. However, when considered in relation to the chronology of this study, these pieces would appear to represent quite a limited time span, effectively representing examples of Groups Two and Three. Although production from one group to another can be shown for the Persian lustre-potter Abu Zaid (see Chapter 6), the usage of the Sa‘d term can be more closely compared to the usage of the term al-yumn noted previously for pottery of Group Two, often, indeed, on the same vessel. Hence, although this study would certainly agree with Jenkins (1988) that the inscription would certainly not appear to be the signature of a single artist, it may in the majority of cases represent a peculiarity attributable to a single potter or workshop, and is worthy of further study, particularly involving accurate profile drawings. It has often been suggested that production of Egyptian Lustre-painted pottery ended with the unrest at the end of the Fatimid dynasty. Porter and Watson (1987) have suggested that the production ended in Fustat in the early to mid twelfth century, and then commenced in Syria with the "Tell Minis" style. The abundance of Lustre-painted pottery is such that it would seem highly unlikely that it was produced by a single individual in any one generation, although the similarities in motif and form would suggest a single workshop or small group of professionally related workshops or individuals. Hence, although it may be argued that lustre production ends in Basra as it commences in Fustat, theoretically an individual or group of individuals could have left Fustat at any time, with production being continued by others. Indeed, movements of potters from Fustat have already been suggested in two separate periods in the eleventh century; however, the evidence from Crusader sites would indicate that Lustre-painted pottery production in Egypt continued into the twelfth century. Egyptian Lustre-wares are not found in those forms common for wares in Syria and Iran at the terminus of the twelfth century and through the first half of the thirteenth century, in particular the fully developed biconical bowl form (discussed more fully in Chapter 5). Such commonality of forms can be argued for the earlier twelfth century wares, and biconical bowls may be noted in Egyptian Incised wares presumably of the terminal twelfth/early thirteenth century (see below). Hence, the cessation of Lustre-painted pottery production at roughly the end of the Fatimid dynasty (1179) would be about right. It is possible that Underglaze-painted wares continued in production in Fustat under Ayyubid suzerainty (see below), but Lustre-wares of Group Four production certainly appear to represent the last Egyptian production of this type. Although no evidence for date of the transition from Group Three to Group Four has been found, given the evidence for the earlier groups and comparison with Syrian forms (see Chapter 5), it would seem tenable to propose the following chronology for Egyptian Lustre-painted wares: Group One: c. 975-1025; Group Two: c. 1025-1075; Group Three: c. 1075-1125; and Group Four: c. 1125-1175. Once again these are clearly generalized time-ranges, and some degree of overlap may occur. However, the evidence argues for the general succession, the dating is a good basis for further study, and it is reliable enough to provide important insights into the beginnings of Syrian and Iranian production. Underglaze-painted of "Wollastonite-slip" class (WS-UGP) Reference has been made to this group in the section on Lustre-painted pottery, as the forms, motifs, body, and glaze technology are precisely identical to those of Group Three and particularly of Group Four Lustre-painted production, which also form part of the "Wollastonite-slip" class. The type can be defined as having a reddish pink body, a light-coloured "slip," a black-pigmented paint, and a clear turquoise or occasionally colourless glaze. All of the pieces studied in person, and most of the published pieces, can be attributed to Group Three (1075-1125) and Group Four (1125-1175) of Fustat Lustre-painted pottery production, predominantly the latter. Lane (1937, fig. 9) publishes a piece from al-Mina' that he suggests is of Egyptian origin and of Ayyubid date, and there are a number of other published pieces that may be related to the al-Mina example (e.g., from Qaysariyya/Caesarea, Brosh 1986, pl.VI4, 6), perhaps characterized by heavier brushwork than the Fatimid versions. These characteristics may indeed be appropriate to a group derived from Fatimid Group Four practice and produced in the early Ayyubid period (1179-1250). Polychrome-painted Opaque-glazed (OG-Poly) The Opaque-polychrome group again has an opaque white glaze and has decoration applied as coloured glazes on the surface in an "enamel" technique related to the Basra Blue-painted type. It is distinguished from the earlier Splashed style of Opaque-glazed ware by form and the more deliberate application of decoration. The distinct, bright, and opacified colours of these overglaze paints include green, turquoise, and purple, and are applied as distinct designs of dots and lines. At times the paints of the Opaque-polychrome wares are so dominant that it is hard to determine whether or not there is indeed an overall white glaze beneath, so these pieces may be confused with the Polychrome-glaze class. This is one of a number of groups defined in this study that are collectively called "Fayoumi" ware by others. Those pieces available for study are of a "camel" rim profile, or some variant of it, and of simple ring or "acute-splayed" base profiles (see Fig. 4.7); this group is therefore considered to be contemporary with Groups One and Two of Lustre-painted production (975-1075). This dating is supported by the excavations at Aqaba, where this ware is found in pre-Crusader levels (Whitcomb 1991, fig. 3a-c; this material has been examined by the author). Lead-glazed class (LG Splash, LG Slip-inc) This class is as described in Chapter 3, essentially comprising a light-coloured slip, often over a darker red clay body, with an overall lead glaze. This ware is usually decorated with "splashes" of colour, commonly cuprous green and ferrous brown-yellow and less often manganese purple and chromium black, and quite often it has incised decoration (Slip-incised wares). Although Basra Lustre-painted types were imported into Fustat in large numbers, and seem to run the full range of production from the beginning of the ninth century until the end of the tenth (see Chapter 3), contemporary Iraqi products of the Lead-glazed class do not seem to have been imported in such numbers (this estimate is based on observation of excavated assemblages, not early collections). In this study only one piece from Fustat has been attributed to an Iraqi origin (see Chapter 3). Given the dearth of Iraqi examples in Egypt, the origins of the Egyptian Splashed-lead ware production appear to be a mystery. It is possible that Egypt was responding to the same Chinese sancai influence as the Iraqi potters, because the Chinese originals have been discovered at the site (Rawson et al. 1987-88). However, the Egyptian products have the same slip of stonepaste as the Iraqi products, so a direct transfer of technology from Iraq appears to be more likely. Forms are varied in this class (Fig. 4.7) and include some of direct Iraqi influence, which are perhaps contemporary with the first production of Lustre-wares in Egypt. Others are in forms contemporary with Groups One and Two of Lustre-painted production. A particular type within this class is Egyptian Hatched Slip-incised ware (Fig. 4.2). Philon (1980:285-89) dates this type to the late tenth to twelfth centuries. The distinctive cross-hatching has a stylistic effect similar to that of Iranian Hatched Slip-incised ware dated by Whitehouse (1979, 1983) to between 1050 and 1150. The forms of Egyptian Hatched Slip-incised ware are again comparable to Lustre Groups One and Two (975-1075). Polychrome-glaze class (PG or Poly-Glaze) Most ceramic classes have an overall uniform glaze: the vessel has been covered in glaze-frit particles in suspension in water, by dipping the vessel into the fluid or pouring the fluid over the vessel. In some classes the surface is decorated subsequently by "splashing" or painting it with coloured glazes, but the nature of the overall glaze, forming the major colour and clearly underlying the decoration, is evident. In the case of the Polychrome-glaze class, the glaze itself is applied to localized areas on the pot in different colours and this itself forms the decoration. In this class there is never a dominant colour, and the different colours abut each other. Very occasionally pieces occur with an overall glaze, but it is never opaque, and the form and other attributes put such pieces firmly in this group. The decoration is quite often applied as radial lines (Fig. 4.2). Along with other types, such as the Opaque-polychrome and Opaque-splashed wares, the Polychrome-glaze group is also called "Fayoumi" ware by others. Forms include rim profiles of "camel" style (Fig. 4.7), perhaps nearer to those of Fustat Lustre Group Two, while bases are of the "ring-2" form (e.g., FUS.25) more closely associated with Fustat Lustre Group One. In reference to the Lustre-ware chronology established above, this would suggest a date range of 975-1075, but with a larger number of representatives typical of the transitional period (i.e., c. 1025). Incised-ware type of Monochrome-relief class (IW) This class generally has a pale buff to white body with decoration executed in relief, whether through incision of a line or more rarely of carving or excision of areas. The glaze is transparent and monochrome, including cobalt-blue, manganese-purple, copper-green, copper-turquoise, and colourless, the latter often further decorated with cobalt-blue splashes. Although the name of the Incised ware group may possibly be confused with the Slip-incised type, it is considerably less confusing than calling them both "sgraffito" (along with unglazed incised wares), as is done by others. The generally received wisdom concerning this ware is that it was inspired by imports of Sung pottery (Scanlon 1971a). At Fustat this class has been referred to as "Fustat Fatimid Sgraffito" or "FFS," and wasters have been found in excavations (Scanlon 1967), although the exact contexts and descriptions are not given. Forms and motifs may be used to divide the Incised wares into four groups, two of which may be related to the Lustre chronology. The first of these may be considered to have the forms of Lustre Group Two, including "camel" rims and slightly splayed bases (Fig. 4.8). Decoration in this group includes typical early Fatimid designs, notably the stumpy palmette encompassed within a lobed stem (Figs. 4.2, 4.3, motif 8 group). Another motif is a rosette that also occurs on the Sung pottery considered to be the prototype for this ware (Fig. 4.3). The second Incised-ware group has wedge-shaped bases which parallel those of Lustre Groups Three and Four (Fig. 4.6), but which are at times sufficiently different to suggest a different workshop, or perhaps a subsequent phase of production. Motifs on this group reflect continuity of the styles used in Lustre Groups One and Two and also motifs are likely to have east Asian origins. The third Incised-ware group may be referred to as the "circle" group, named after a circle or broken circle incised into the body at the centre of the underside, an affectation perhaps meant as a signature. The bases of this group are similar to some of the tall slightly splayed bases found on the first group of Incised-ware pieces, equal to Lustre Group Two, but in this case they are vertical. This form may indicate contemporaneity with or development from the Group Two practice, but is also a suitable form to have been copied from the Chinese imports. Certainly the motifs show stronger Chinese influence than the previous Incised ware groups, with central blossoms or complex foliage. The fourth Incised ware group consists of biconical bowls with bases identical to those of the third group (Fig. 4.8). Decoration includes debased versions of the Sung rosette of the previous group, and designs based on a rosette with pointed petals, which may be cognate. The biconical bowl of this group is more commonly associated with pottery made after about 1200 in Syria and Iran, although I would prefer to see the precise "proto-biconical" bowl form of some of these examples in the second half of the twelfth century (see Chapters 5 and 6). Archaeological evidence of dating is scanty. At Athens a fragment of "imitation porcelain" was found in an eleventh to twelfth century context but the publication does not provide much that is diagnostic (Frantz 1938; see Appendix C). The fragment does, however, have a white body and "white glaze" (whether opaque or transparent is not clear), and a footring. The piece was identified as Egyptian by Ernst Kuhnel and the attributes do appear to be more Egyptian than Syrian. Blake (1978) describes a number of monochrome-glazed stonepaste-bodied pieces from Italy attributed to the eleventh or early twelfth century on the basis either of numismatic association or of incorporation into buildings, and at least some of these may be Egyptian. Certainly a piece from a late eleventh century context at Pavia contains lead in the glaze, and must therefore be either Egyptian or Syrian Group One. Pieces from Genoa that have been attributed to Syria are more likely Egyptian, with a "friable" body, opaque glaze, and forms typical of Group Two Incised wares (Pringle 1977). These wares are from a context dated to the late eleventh century by the excavators (Ward-Perkins 1978). Promising pieces from excavations at Sarachane in Istanbul (Hayes 1992) include find No. 55.1, which has a "rather rough-textured light yellow fabric" compared to the fine white stonepastes of most of the Syrian wares Hayes found at the site. It also has a thick dark leaf-green glaze, is incised, and has very close parallels to some Egyptian Group One Lustre-ware forms although I haven't seen any Incised wares. A short flat rim fragment may belong to this piece (Hayes 1992, pl. 9f) and would also appear suitable to Fustat Group One or Two. Unfortunately the context, Deposit 55, is dated to c. 1200 or slightly later, and includes survival material from the eleventh century and some from the sixth to seventh century. These observations would suggest that Incised wares were first made in the period contemporary with Lustre Group Two (1025-1075). If the influence was indeed the import of Chinese incised wares, then it would appear that there were insufficient examples for the potters to imitate at this time, as forms and motifs are entirely Islamic in derivation. That is, the "idea" of monochrome incised wares may have been introduced, but no more. The second Incised-ware group appears to represent locally inspired forms with an amalgam of Chinese and Islamic ornament, and represents production perhaps as late as Ayyubid times (after 1179). The third group would appear to most closely represent direct copies of Chinese originals, perhaps introduced around 1100. This group may have continued after 1175. Certainly the fourth group, with its more debased motifs and biconical forms, is likely to represent the continuation of the Incised-ware class into Ayyubid times. TECHNOLOGY This section is divided into two subsections: the technology of the bodies, and the technology of the glazes, slips, and pigments. Chemical data for bodies is given in Table 4.2, and for glazes in Table 4.3. Body technology The pottery discussed in this chapter may have either fired-clay or stonepaste bodies. The clay bodies are more properly dealt with in the provenance section, below, but it may be said here that they essentially consist of two main types: the "Ca-Nile" type, with about 11% alumina and 21% lime; and the "High-Ca" type, with about 8% alumina and 36% lime (see Table 4.2). A third fired-clay body is found only in Semi-glazed ware and contains about 21% alumina and 1% lime. Nile alluvium does not appear to have been utilized in this period in an unmodified state, but contains about 15% alumina and about 6% lime. Stonepaste bodies are not as clearly defined in this chapter as they will be in subsequent chapters. As is described more fully in Chapter 2, standard recipes for stonepastes describe a mix of about 80% quartz, 10% fine clay, and 10% of a glass additive. This describes the fully developed technique as practised in central Iran and described in sources of the fourteenth, nineteenth, and twentieth centuries (see Chapter 2). Whatever the precise recipe, the mix should include quartz to make up the bulk, clay to keep the material together during forming, and glass particles to bond the body together during and subsequent to firing. We have seen in Chapter 3 that Iraqi potters had been adding glass to clay bodies, presumably to accelerate vitrification and harden the body. Also, Slip-incised wares of the Lead-glazed class in Iraq generally had a white slip of quartz, although this was applied without glass in the mixture. Among the earliest examples of Group One (975-1025) of the Egyptian Lustre-painted pottery are a number of pieces that show the first use of stonepaste technology in Egypt (Mason and Tite 1994a). As stonepastes they are unusual in that in thin-section they include only about 10% quartz, 15-20% of relict glass additive, about 40% matrix, and the remainder void (see Table 4.4 and Pl. 4.1). As these fabrics are sufficiently different from, yet considered ancestral to, true stonepastes, they will be referred to here as proto-stonepaste. The overall bulk chemical composition is high in alumina, calcium, and sodium (see Table 4.3). The body consists of vesicular fired-clay zones, glassy matrix zones, inclusions of relict glass fragments, and quartz. The vesicular zones are high in alumina and calcium (up to 20% Al2O3 and 15% CaO), and are considered to most closely represent the original chemical composition of the clay component. The glassy matrix zones are considered to be areas of clay matrix that have received more of the component elements of the glass fragments, as they contain slightly less alumina and calcium, but higher concentrations of sodium and potassium (up to 9% Na2O and 4% K2O), which they have received from the glass inclusions. The relict glass inclusions retain distinct boundaries, and probably represent the original dimensions and shape of the glass additive. Bulk analyses of these relict glass inclusions are depleted in fluxes compared to the glassy matrix zones, a result of crystallization from the glass, initiated at the contact with the matrix. Crystallization consists predominantly of acicular calcium silicates (probably wollastonite), stumpy calcium magnesium silicates (probably diopside), and occasionally silica (probably cristobalite), predominantly arranged around the margin. The remaining glassy phase of the relict glass inclusions contain higher amounts of sodium, potassium, and lead than the overall bulk analysis for each fragments. Lead probably also disseminates into the glassy matrix zones of the body, but is below detection limit. Voids are found within the relict glass fragment. In some cases the voids are restricted to the glassy phase, but in other cases they occupy the bulk of the space formerly represented by the glass additive, leaving only a rim of glass covering the crystallized zone. These observations relating to the relict glass inclusions are largely identical to those for typical later stonepastes. During firing the chemical mix within the body tries to come into equilibrium. Highly mobile elements in the glass additive would escape into the clay matrix, accelerating vitrification and bonding the body. Within the relict glass fragment this loss of alkalis would result in saturation in silica, calcium, and magnesium. These elements would crystallize out of the molten glass, recreating the mix of the original glass additive by enriching the remaining glassy phase with alkalis. Thus, crystallization would augment the capacity of the glass inclusions to transfer alkalis to the body. The crystallization around the perimeter may be the result of a gradient of saturation where alkali loss is at its greatest, or perhaps initial crystallization here is caused by disequilibrium between the glass inclusion and the matrix, which is high in calcium, aluminium, and, locally, silica. A true stonepaste, after firing, consists of about 60-70% quartz and 10-20% interstitial glass, the remainder being void. In most cases there will be some relict of the glass additive ranging in degree of preservation from angular inclusions of partly crystallized glass, to voids with characteristic crystallization around the rim. In most extreme cases no structure of the glass additive remains, and the rim of crystallization is spread around a bloated sub-spheroidal void. Modal expressions of abundance of relict glass inclusions, normally about 10%, refer to the area formerly taken by the inclusion, including area now void, and are not a measure of how much glass additive has remained as relicts, having failed to transfer to the matrix. The interstitial glass will appear in thin-section and SEM as a thin film covering the quartz grains, and cementing them together; at times this is so insubstantial that one wonders how the ceramic stays together. In other cases the quartz appears to be within a vesicular glassy matrix. Chemically, a typical stonepaste will consist of 80-90% silica and less than 5% each of alumina, soda, and lime. During production of Group Two (c. 1025-1075) the stonepaste technology approaches these later true stonepaste standards (see Table 4.2). In thin-section the Group Two stonepastes comprise about 50-60% quartz, and 8-12% relict glass additive (although about 5-7% relict glass is closer to the norm), the remainder being interstitial glass and void. Bulk chemistry of the body is also closer to the standard, over 80% silica with generally less than 6% alumina. In Plate 4.2 a BSE image of one of these (FUA.48) shows the well-developed glass bonds of the fabric. Incised-wares of Group Two, as in all periods, may be considered to consist of stonepaste, but it is not as high a quality of material as in the lustre bodies. There is an appreciable amount of relict glass inclusions (about 10% in thin-section), but there is a very poor degree of bonding between quartz grains, which would explain the friable fabric. In Groups Three and Four (1075-1175) there appear to be no Lustre-painted samples of typical Egyptian design with stonepaste bodies. Incised-wares continue to be made of what may be termed stonepaste, although in thin-section the observable remnant glass has fallen to 3%, and the quartz to about 30%. In unaided observation, pottery in this group often appears to resemble more closely a clay body that has been excessively tempered with sand, than a stonepaste body. Glazes and surface zone Although the continuum from one region to another is discussed more fully in Chapter 8, it may be said here that the glazes on Fustat Group One Lustre-painted pottery follow Iraqi practice of applying tin-opacified lead-fluxed glazes (Pl. 4.3), but there are important differences (see Table 4.3). The glaze of the stonepaste piece sampled is closer in composition to the Basra glazes, but those with fired-clay bodies have higher concentrations of lead and tin and lower concentrations of sodium (compare to Table 3.4). All the Egyptian glazes are considerably lower in magnesium than the Iraqi glazes (often below detection limit in EDS analyses). The glazes in both stonepaste and clay bodies are usually about 300 microns thick on the interior, and about 150 microns thick on the exterior. The clay-bodied samples have a narrow glaze-body transition zone of under 100 microns, while the glaze of FUA.04 fades into the glassy matrix over about 300 microns. None of the samples possesses a slip. The glazes of Group Two Lustre-painted samples essentially continue this glaze technology, with roughly the same compositions and glaze thicknesses, and they similarly lack a slip. As in Group One, the stonepaste samples have slightly lower tin and higher sodium contents, but there are no significant differences in the lead content. The stonepaste bodies generally have a thicker glaze-body transition zone of about 500 microns. Groups Three and Four Lustre-ware glazes each have essentially the same glaze technology, which differs markedly from that of Groups One and Two. Although the primary flux is still lead, its concentration has been halved, while that for soda has doubled. Of more obvious effect is the total lack of tin, making these all transparent glazes. However, between the 250 thick glaze layer and the body, there is a separate layer, about 50 microns thick. In unaided examination this layer gives the impression of a thin white slip. In thin-section this entire zone between the glaze and the body appears as an anisotropic area of fine acicular crystals. The BSE image (Pl. 4.4) also indicates a crystallized area, with chemical analyses indicating calcium silicate (wollastonite). In Chapter 3 we have already seen how crystals such as these were found on the interface between glaze and body on Turquoise-glazed wares. Further, we have seen how in very early Islamic opaque glazes crystals were dispersed throughout the glaze in conjunction with small amounts of tin oxide and undissolved quartz to aid opacification. This technology is always found on the "High-Ca" body (which contains about 30% CaO) and so may arise simply through a reaction between the glaze and the body. However, this same technology exists in the Underglaze-painted group, along with forms and decoration equivalent to Lustre Groups Three and Four. In the Underglaze-painted group it appears that the pigment has been applied to a pre-existing slip, as the manganese pigment grains seem to be on top of the wollastonite layer (Pl. 4.4). Perhaps in some way this is related to the surface "whitening" or "false-slip" noted by Matson (1971), and further discussed by Peacock (1984), on pottery containing more than 15-20% lime, and is a result of salts efflorescing on the surface of the unfired vessel, which gives the appearance of a light-coloured slip. The Incised-wares seem to form a coherent technological group throughout their period of production. They generally have lead-alkali glazes (see Table 4.3), usually over 500 microns thick. The glaze-body interface tends to have a deeper zone of glaze penetration in the case of the earlier stonepaste bodies than for the later stonepaste bodies that resemble a sandy clay. The glaze chemistry does not show the trend in gradually lower lead and higher sodium noted in the Lustre-and Underglaze-painted glazes. Instead the Incised wares generally possess the high lead content of the Group One and Two glazes, combined with the higher sodium content of Groups Three and Four. Copper may be noted in the green and turquoise glazes (FUR.20 and FUS.37), and manganese in the purple-coloured glaze (FUS.44). The early Semi-glazed class has a very high lead glaze (sample FUS.11, EDS analysis of 55% PbO), which is largely featureless and about 150 microns thick. The later Lead-glazed group also has very high-lead glazes. This group includes a sample of the simple unincised type (FUR.18, EDS analysis of 61% PbO), and the Egyptian Hatched Slip-incised ware (FUS.17, WDS analysis of 63% PbO). The glaze of the Hatched ware is about 100-150 microns thick, with an underlying slip of stonepaste. The interstitial glass of the slip has considerably less lead than the glaze (18% PbO), and is correspondingly higher in alumina and potassium (25% compared to 2.4% Al2O3; 6.3% compared to 0.5% K2O). This would seem to indicate use of clay in the slip-mix, possibly representing true stonepaste, and at variance with the Iraqi practice of fixing the quartz to the surface of the vessel without glass and clay in the slip-mix. The Polychrome-painted type of the Opaque-glazed class (OG-Poly) is generally comparable to contemporary opaque glazes on Lustre-painted wares, but one has a high-lead glaze (see Table 4.3). The high lead content of this piece may be related to the practice of the unknown production centre for the Iraqi "Samarra 2" Petrofabric tentatively attributed to Baghdad (see Chapter 3). However, textural and chemical inhomogeneity in the glaze resulting from the application of the enamel paints may produce anomalous analyses. The glaze, which is about 400 microns thick, is poorly integrated with the clay body . CHARACTERIZATION AND PROVENANCE This section is more concerned with characterization than with provenance as there are no samples of wasters included in this part of the study, except those of Mamluk date (1250-1517) included as the only comparative material available. Production at Fustat is assumed for most petrofabrics largely on the basis of the scant historical and weighty but inconclusive archaeological evidence for production of the sampled ceramic types at Fustat. The use of Nile alluvium, although apparently modified, appears to be predominant, and this has been characterized by analysis of the wasters of Mamluk date ("Average Nile" and Nile Petrofabric in Tables 4.2 and 4.5). Attribution of this material to Egypt can thus be made with confidence, even if attribution to Fustat is tentative. Historical references to the production of pottery in Fustat are rare. A description of ceramics by the fifteenth-century historian Maqrizi has been interpreted as a reference to pottery made in Egypt, dating to 1056 (Grabar 1972). In 1050 Nasir-i Khusraw described vessels which he claimed to be Egyptian; this text has been interpreted as a reference to Lustre-painted pottery. However, the vessels are described as being so translucent that it is possible to see one's hand through them, which is hardly a description of any product of an Egyptian kiln. According to E. J. Keall (Mason and Keall 1990), the term translated to describe lustre, sofalieh, could be interpreted as "glassware." Inscriptions on vessels are slightly more helpful. One bowl claims to have been made in Misr, which has been interpreted as referring to the capital, but may conceivably mean anywhere in Egypt (Lane 1947:21). Two further Lustre-painted vessels are dedicated to Egyptian notables (see above), but if compared with certain other dedicated vessels, for instance, Egyptian glass for Yemeni Rasulid sultans (Atil 1981:131), then it can be seen that such inscriptions may have no relevance to provenance. More importantly, ceramic production in Fustat is indicated by archaeological evidence. Wasters have been known from the ruins for years, and excavations by Bahgat (1914), and more recently by the American Research Centre in Egypt, have revealed a number of kilns. Unfortunately it is usually not possible to say precisely which styles, which types, or even at times which classes were found as wasters. For instance Scanlon (1974) mentions wasters of "Fayoumi" ware found at Fustat, but does not describe either the finds or their context any further. Pieces that would be classified in this study as Opaque-glazed class of Splashed and Polychrome-painted types, and others of Polychrome-glaze class, each of somewhat different petrofabrics, would all be classified as "Fayoumi" wares in the Fustat excavation reports (e.g., Scanlon 1974). Ceramic production continued at Fustat after its decline, and potters still work in the vicinity (Abu-Lughod 1971:201; Matson 1973). Chemical data for bodies is given in Table 4.2, petrographic data is reported in Table 4.4 for wares with stonepaste bodies, and in Table 4.5 for wares with clay bodies. Stonepaste bodies Fustat stonepastes are dealt with in greater detail in the technology section, above, but it is still necessary to reiterate features pertinent to characterization. The Group One proto-stonepastes are considered to be the earliest whole vessels constructed using this technology. These consist in thin-section of about 10% quartz, 15-20% of remnants of the glass additive, and some 50-60% matrix (in this and later modal percentages for stonepaste, the remainder is void). The matrix is glassy and locally vesicular, and would appear to be mostly clay. True stonepastes are primarily found in the Lustre-wares of Group Two (1025-1075) and in the Incised-ware pieces equivalent in date to Lustre Groups Two to Four (1025-1175) and possibly also early Ayyubid date (c. 1175-1250). The bulk of these stonepastes comprise 50-60% quartz, and about 10% glass, with about 10% of the thin-section comprising glass, which bonds the material together. This latter description is much nearer the standard composition as defined for later products. Some of the bodies of Incised-wares are poorly bonded, with about 30% quartz, as little as 3% glass remnants, and larger amounts of clay matrix with perhaps as much as 40% void. In some of these cases one wonders what is keeping the material together. The quartz in all of these stonepaste bodies is predominantly slightly cloudy (see Chapter 2), with as much as an equal amount of clear quartz, and some cloudy quartz (see Table 4.4). Feldspars are generally present up to 1% and some samples also contain carbonate and chert. Grains are angular, generally ground to very fine sand grade, with some to coarse and medium silt grade. There are no obvious rounded grains, except those partly dissolved by the lead-rich glass, but the maximum grain size is 0.6 mm in diameter. The source material used for Mamluk Blue and White pottery is considered to have been in rounded grains about 0.9 mm in diameter, although in many examples the quartz had been crushed so that no rounded grains remained (see Appendix B). In some Fustat clay-bodied pottery (see below), quartz grains of a similar nature have been interpreted as being added, perhaps deliberately. The quartz in the Fatimid and early Ayyubid stonepastes is essentially similar to the Mamluk period sand in its degree of cloudiness, and it is possible that the sand was crushed so that no rounded grains exist. The source of quartz may therefore be the same. The range of values is certainly not as standardized as for the Mamluk production, but this is understandable given the formative nature of the technology in this period, and the acceptability of relatively unreliable raw materials. These characteristics are certainly enough to enable distinction between Fustat and other early stonepaste petrofabrics such as Raqqa-2 (cloudier, sheared quartz), Damascus(?) (clear quartz), "Tell Minis" (clearer quartz), and Kashan(?) (dominantly chert). Clay bodies Mineral abundances for the clay-bodied pottery are presented in Table 4.5, with grainsize-distribution histograms presented in Figures 4.9 and 4.10, and ternary diagrams of the mineral data presented in Figure 4.11. Petrography has revealed a great deal of diversity and variation among the petrofabrics of pottery attributed to Fustat, although specific pottery types often have quite well defined petrofabrics. The pottery considered to be of Egyptian origin forms three main groups: a petrofabric rich in volcanics found in the early Semi-glazed ware; a highly calcareous petrofabric found in later Lustre-ware ("High-Ca"); and a complex group of petrofabrics, consisting of at least two main subgroups, which appear to be closely related to Nile alluvium, but differing somewhat. These latter petrofabrics include most of the early Fatimid pottery. The distinctive nature of the Nile mineral assemblage, running as a theme throughout most of this pottery, confirms that they were made in Egypt, even if attribution to Fustat is tentative. "Semi-glazed" Petrofabric As noted above, the earliest glazed ware sampled from Egypt, the Semi-glazed ware, has a unique petrofabric among the Egyptian wares of this study. Important features of this petrofabric include the coarse grainsize distribution (Fig. 4.10) together with the abundance and nature of the volcanics (see Table 4.5). Although the general proportions of volcanics and ferromagnesian minerals are not far removed from extreme Nile examples (see Fig. 4.11), the exact nature of the inclusions is significantly different. Felsic volcanics are occasionally found as trace amounts in other petrofabrics, but never of this abundance. Also the given figure of 1-2% basalt (Table 4.5) requires clarification. The entries in this column for the other petrofabrics represent simple basalts derived immediately from Nile alluvium and ultimately from the Ethiopian plateau. In the "Semi-glazed" Petrofabric "basalt" represents an ultramafic rock consisting primarily of clinopyroxene with some plagioclase, which is not found in other petrofabrics from Fustat or in Nile alluvium wherever the author has sampled it (see below, and Mason 1987). As there is evidence that certain later ceramic types were made in Fustat, and as the petrofabrics of these types are considerably different from the "Semi-glazed" Petrofabric, it may be interpreted either that Semi-glazed ware was not produced in Fustat, or that the potters were utilizing a clay source not used subsequently. Whitcomb (1989) has suggested that this ware was made at Alexandria, but this does not explain the observed petrofabric since it must be assumed that only deltaic deposits of Nile alluvium would have been available here. Fustat(?) "High-Ca" Petrofabric In thin-section this petrofabric comprises 4-7% quartz, generally 3-5% but up to 16% carbonate grains, and occasionally trace feldspars and opaques. The carbonate grains range in size from fine sand grade up to about 0.5 mm in diameter (see Fig. 4.10). They have mostly lost structural integrity but in one case comprise fine micritic material with occasional foraminifera. The boundaries between the grains and the surrounding ceramic matrix are poorly defined, which would suggest that they do not represent indurated limestone fragments. The carbonate inclusions are considered to be natural inclusions in the clay. The quartz and feldspar are often found as large rounded grains (average diameter about 0.4 mm), although finer angular fragments are numerically more dominant. This would suggest that the sand was processed (crushed) and added to the clay mix deliberately. Chemically this body is highly calcareous (Table 4.2) and quite low in alumina (36% CaO; 8% Al203). In unaided visual observation the body is generally pinkish red (Munsell 5YR 6/6 appears dominant), and is quite hard with an even fracture. The large rounded quartz grains and carbonate grains are clearly visible, particularly with a 10× hand lens. Fustat "Nile" Petrofabric Although not found used alone in pottery of the period relevant to this chapter, it is pertinent to describe this petrofabric here. It is characterized by the high abundance of a wide range of inclusions of fine sand grade, with the largest grains in the 120-150 range being subrounded-subangular. This includes 9-15% quartz, 4-7% plagioclase, 2-5% clinopyroxenes, 1-2% amphiboles, trace to 1% each of untwinned feldspars, biotite, and basalt, with occasional traces of felsic volcanics, volcanic glass, metamorphic schists, phytoliths, microcline, epidote, and muscovite. Each mineral is commonly represented by more than one variety or texture. For example, plagioclase includes cloudy and clear grains, and clinopyroxenes are green or clear. The phytoliths may have been added as ash. Matson (1973) describes the addition of ash to a clay in Egypt, yet phytoliths appear to be restricted solely to the Nile Petrofabric, or derivatives of it. This description generally fits other descriptions of Nile alluvium, including those given in studies by this author on Nubian pottery (Mason 1987) and Egyptian Pharaonic pottery (unpublished); and also studies by others (Nicholson and Rose 1985, Nordstrom 1985, Porat 1989, Hays and Hassan 1974; Alexander Smith undertook a major study of the petrography of Egyptian clays at Oxford University, but this work is not yet published). Data from the author's Nubian studies have been included in the ternary diagrams (smaller dots), to show the sort of variability that may be found in a chronologically wider sampling (Neolithic to medieval Christian periods). Chemical analysis of the Mamluk pieces of this petrofabric (average Nile alluvium in Table 4.2) shows a moderately high aluminous content, low calcium content, and high iron content (15% Al2O3, 5.6% CaO, 9% FeO). In unaided visual observation, the body is red (Munsell 5YR 5/6 is common), and quite soft with a porous and open structure. The range of inclusions, particularly the ferromagnesian minerals, is clearly present with a 10× hand lens. Fustat(?) "Ca-Nile" Petrofabric group Generally the mineralogy of this group resembles Nile alluvium, with grainsize dominantly of very fine sand grade, and including subrounded-subangular grains of 120-150 diameter. The body is porous and open, and ranges in colour from buff to pink-red. The individual characteristics of the two subgroups will be considered in turn. Fustat(?) "Ca-Nile 1 " (buff) Petrofabric This petrofabric is distinguished from "Ca-Nile 2" by the lack of carbonate inclusions, the generally higher quartz content, and the lower content of ferromagnesian minerals (Fig. 4.11). There are also minor chemical differences, such as lower potassium and, on average, higher magnesium. This body is porous and usually grey-buff (e.g., 10YR 7/3), with some pinky-buff bodies (e.g., 7.5YR 7/4). Although there are no visible carbonate grains, the lime content is relatively high and is exactly the same as for the carbonate-bearing "Ca-Nile 2" Petrofabric. The carbonate grains may have been crushed to a finer consistency and thus dispersed throughout the clay. This would be consistent with the buff colour of the body, as the finely dispersed lime could then react with the iron. The higher quartz content is probably due to the addition of quartz, as coarse rounded grains and crushed angular grains of sizes larger than Nile inclusions are often found (see Fig. 4.9). This body was formerly considered to represent an original, unmodified raw material (Mason and Keall 1990), but the larger sample size provided by this study appears to show it as part of a continuum with the other Nile-variant group. Fustat(?) "Ca-Nile 2" (pink) Petrofabric This petrofabric most closely resembles Nile alluvium, and in some cases is barely distinguishable from it, except for the presence of carbonate inclusions (see Fig. 4.11). Although there is considerable variability within this group, each particular ceramic type will generally have a more restricted variability. For instance, the Polychrome-glaze group appears closer to Nile alluvium in thin-section, while Egyptian Hatched Slip-incised ware generally has relatively low total mineral abundances. It is interesting to note that this depletion does not appear to be in inverse proportion to carbonate abundance, which is contrary to expectations if a clay-mixing hypothesis was valid. In unaided visual observation this body is porous and pink (Munsell 5YR 6/6 is common). Clay-bodied Petrography discussion Consideration of the complete data (Table 4.5, Figs. 4.9 to 4.11) leads one to suggest a number of general conclusions. First, the "Semi-glazed ware" and "High-Ca" Petrofabrics are clearly and distinctly different from the other petrofabrics. Second, it would seem that the petrography of the remainder of Fustat pottery represents a continuous spectrum of diversity with particular ceramic types accommodating restricted portions of this diversity. Nile alluvium appears to be the one consistent component of this continuum, particularly if one includes the chronologically diverse data from the Sudan (see Fig. 4.11). But the Nile alluvium itself does not appear to cover all of this diversity. The "Ca-Nile 2" group has more carbonate, clearly more than the Fustat Nile samples, which contain none at all. The "Ca-Nile 1" group are generally more quartzose (here the Sudanese data may be deceptive, as quartz was added here also, although possibly by nature), and the precise mineralogical differences are more distinctive, with considerably more plagioclase and pyroxene in the Nile pottery (see Table 4.5). It might be possible to explain this as natural diversity, with the River Nile itself mixing the different varieties of its own deposits with older deposits in the region. However, given the degree of raw material manipulation noted at Fustat by Matson (1973), it may be preferable to interpret this variability as being due to the mixing of raw materials. By combining Nile alluvium with the "High-Ca" clay, it would be possible to "create" the "Ca-Nile 2" clay (excluding the quartz added to make the "High-Ca" Petrofabric). This may in particular be seen in the Polychrome-glaze ware examples (see Table 4.5), which in some cases are barely distinguishable from the Nile wares except for the presence of carbonate inclusions. Chemically, "Ca-Nile 1" is practically identical to "Ca-Nile 2" and has therefore also been interpreted as a mixture of Nile with the "High-Ca" clay. In this case, however, there is no visible carbonate, suggesting that the "High-Ca" clay would have to be crushed or sieved to remove the carbonate grains. Further, there is proportionally more quartz, making the entire grainsize profile coarser (see Fig. 4.9). Since this quartz consists of larger rounded grains and finer angular grains, its source is a sand which has been crushed and then added to the mix. SYNTHESIS The beginning of glazed Islamic pottery production in Egypt is represented by an unusual class of ceramics, apparently unique to Egypt. This Semi-glazed ware has a glaze with a very high lead content, perhaps technically descended from Romano-Byzantine glazing practice, although the glaze is applied only to certain areas of the interior. The provenance of this pottery is unknown, but its petrofabric does not seem compatible with production at Fustat, and may reflect pre-Islamic patterns of ceramic production in Egypt. During the ninth and early tenth centuries Egyptian pottery is stylistically and technically more in the mainstream of Islamic ceramics, producing copies of pottery imported from China and ‘Abbasid Iraq and including Splashed Opaque-glazed wares and Splashed Lead-glazed wares. Generally, this pottery is made of the "Ca-Nile 2" body, which" has the appearance of a mixture of Nile alluvium and a calcareous clay. The bulk of fine wares in this period are imports, the products of Basra, in Iraq (see Chapter 3). During the later part of the tenth century we see a major shift in emphasis with the introduction of Lustre-painted and associated wares. Given the nature of the lustre technology, which was ever a prized and guarded secret, and the close parallels in form and decorative motif between the last Basra Lustre-painted pottery and the first of that from Fustat, it may be taken that the Fatimid potters came from Iraq, including specifically Basra. As it appears that Lustre-painted pottery production ceased in Basra at this time, it would seem that there was more to this movement than just a few potters taking advantage of the wealth of Egypt. When the Iraqi potters came to Fustat, they would have needed to use the local materials to produce their pottery. According to analysis of form and decorative motifs, the earliest pieces have a buff-coloured body, visually as close as possible to the clay of Basra. This was either made of a modified Nile alluvium ("Ca-Nile 1" Petrofabric) or proto-stonepaste. This early appearance of the stonepaste technology is of fundamental importance in the history of ceramic technology. Evidence from Iraq, such as glass-fortified clay bodies, quartz slips, and rare examples of Iraqi proto-stonepastes (see above and Chapter 3), clearly indicate that aspects of this technology were known there, but it is in Egypt that it was first used on a significant scale. By the end of the period of production of Lustre Group One (c. 975-1025), practice in clay preparation would relax, and the pink-red "Ca-Nile 2" body is also used in Lustre-wares. In the case of Lustre-painted pottery, the opacified glaze would continue to conceal this body colour adequately. The "Ca-Nile 2" petrofabric was also the dominant body used for the coarser glazed wares of this period, such as the Polychrome-painted Opaque-glazed type, the Polychrome-glaze wares, and the various continuants of the Lead-glazed class. This does not necessarily mean that these were all produced in the same workshop. The more casual throwing and shaping of the coarser wares, and also the high lead content of the opaque glaze of Polychrome-splashed wares would in fact argue against this. At the end of this first phase of production ceramic styles appear to change, indicating the beginning of a new phase. According to the chronology suggested in this study, this change occurred at about 1025. In Lustre Group Two (c. 1025-1075), stonepaste and the "Ca-Nile 2" clay body would continue to be dominant, while the tin-opacified lead-soda glaze technology would also continue but with minor changes. These changes involved a gradual trend of reduction in lead and tin with an increase in soda that would continue subsequently. The stonepaste in this phase more closely resembles the standard descriptions, with less than 10% glass and over 50% quartz as observed in thin-section. During this second phase of Lustre-ware production, lower-status wares continued to be produced. We also see the beginning of Incised-ware production. These wares are traditionally attributed to influence by Chinese imports, yet forms and decoration more often reflect contemporary Lustre production than possible imports. Incised-wares of this period are invariably made of stonepaste and have a lead-soda glaze, generally higher in soda than the contemporary Lustre-painted pottery. At the end of the period of production of Lustre Group Two we see a more radical change, not only in ceramic style, but also in technology, including the apparent disappearance of stonepaste except for Incised wares. The Lustre-painted pottery characteristic of Group Three (c. 1075-1125) is invariably found produced on a hard pinky-red body with visible inclusions of carbonate. This "High-Ca" body was not formerly used by itself in Fustat, although it is thought that this clay that was mixed with Nile alluvium to produce the "Ca-Nile" Petrofabrics. By tempering the clay with crushed sand, it was also usable on its own. The glazes of this period are not opacified, and are of the lead-alkali type with considerably less lead and higher sodium than the earlier glazes. To create the white ground needed to conceal the pink-red body, an unusual "Wollastonite-slip" technology, involving the crystallization of calcium-silicate crystals, was used. Production of Incised-wares also continues in this period, using the same lead-soda glaze technology of the preceding period. Forms would seem to have developed from Group Two prototypes with no reference to contemporary Lustre production, perhaps under the influence of Chinese imports. The bodies of Incised wares in this period in some cases are typical stonepastes, but in many cases the body is friable and sandy, indicating a relaxed approach to stonepaste. In Group Four (c. 1125-1175) the potters continued to use the "High-Ca" body, the lead-alkali glaze, and "Wollastonite-slip." Lustre-painted pottery continued to be produced, but Underglaze-painted pottery was also produced, probably in greater numbers. Unlike imitations of Lustre-painted pottery found in regional centres (e.g., slip-painted wares of Nishapur [Wilkinson 1973:190-93]), this pottery was made by the lustre-potters themselves, with precisely the same forms, body, slip-technology, and motifs, but using a different decorative medium. It has been suggested that lustre-painters do not actually make pottery, but paint on pottery already made and fired by others. One argument against this hypothesis has always been the dearth of vast volumes of un-painted white pottery, but an argument put forward in its favour has been the variability found in vessel forms. In fact the variability of forms for Egyptian Lustre-painted wares is reflected in variability in decoration, a symptom of lack of standardization of production when compared, for example, to later production at Basra, where forms, motifs, and decorative schemes varied little. Strong evidence against the separation of the lustre-painter from the potter comes from the Underglaze-painted ware. As the decoration is painted before glazing and firing, obviously the painter must be included in the pottery-production process. As the Underglaze motifs are precisely identical to the Lustre motifs in this period, by extension, the lustre-painter, being the same individual as the underglaze painter, must also be so incorporated. Incised-wares also continue to be produced in this period, mostly with poorly bonded stonepaste bodies, which are more like fortified sandy clay bodies. Again, their lead-soda glaze chemistry continues the tradition established in the period of production of Lustre Group Two. Some of the Incised wares may have continued to be produced at Fustat in the succeeding Ayyubid period, but production of most other types, particularly Lustre-painted wares, appears to have ceased entirely by the close of Fatimid rule. Many of the potters had already left, perhaps a century earlier, to establish potteries in Syria, Iran, and possibly Europe. Return to Table of Contents