Shine Like the Sun: Chapter 3: Iraq c. 700-1100 INTRODUCTION Iraq, traditionally the seat of the ancient eastern empires, became the centre of Islamic political power in the middle of the eighth century under the ‘Abbasid caliphs (Bosworth 1996:6-10). Many of the major Islamic cities and palaces were not situated in association with the ancient Mesopotamian centres, but were in new Islamic foundations. Unfortunately for archaeology most of these sites continue to be occupied, but there are a few exceptions. Samarra' is essentially one of the more notable exceptions, although even here part of the vast ruin field continued to be occupied to the present day. Such is the importance of Samarra' that the distinctive Iraqi glazed wares of this period are generally either associated with the name of the caliphal dynasty, or with the site of Samarra'. Pottery made in Iraq at this time had a very wide distribution, and the pottery available for study (see Appendix C) comes from sites not only in Iraq (Hira, Nippur/Niffar, Samarra', Basra), but also from Iran (Susa, Istakhr, Siraf, Nishapur), Syria (Raqqa), Egypt (Fustat), Bahrain, Oman (Lasail), Saudi Arabia (Dhahran), Jordan (Aqaba), Yemen (Zabid), Central Asia (Samarqand), Sri Lanka (Mantai), Thailand (Ko Kho Khao), and South Africa (KwaGandaganda). Published material incorporated into the study extends this distribution to China, North Africa, and Spain. Important findings with regard to the technology of Iraqi wares include the recognition of the first use of tin to deliberately opacify a glaze, and practices prototypical to the development of true stonepaste ceramics (see below). Petrographic analysis is used to characterize the petrographically defined fabric (petrofabric) of pottery, at least to characterize groups, and hopefully to attribute wares to specific production centres (see Chapter 2). To date, thirteen distinct petrofabrics have been defined for Islamic glazed wares from the Tigris-Euphrates catchment. Of these only two sites have been positively associated with identified petrofabrics: Basra and Raqqa. We do, however, have a number of distinct petrofabrics as well as both historical and archaeological evidence of production at specific centres. By consideration of data from sedimentological studies and petrographic analysis of pre-Islamic pottery, it has been possible to develop a model to help suggest possible correlations between the historically and archaeologically attested centres and the petrofabrics (see below). It is therefore possible to attribute particular petrofabrics tentatively to Baghdad, Samarra', and Hira. TYPOLOGY As outlined in Chapter 2, the ceramics in question will be divided into a number of classes, defined chiefly by their glaze technology. Classes are in turn divided into types, based chiefly on the technology of their decoration. Pottery from a single type is divided into a number of styles, defined by form and motif assemblage. Seriation of the stylistic attributes is used to develop a sequence, or chronological succession, which is then tested against available absolute dating information. In all respects this typology is developed without reference to those of previous authors, but comparison with these other typologies is made subsequently. Various published groups of relevant ceramics were examined to ensure adherence to the typology, even if these were not included (e.g., Waagé 1948, Wilkinson 1973, Philon 1980). The ceramic classes defined include the Turquoise-glazed class, the Opaque-glazed class, and the Lead-glazed class including Moulded and Splashed wares. The classifications are based on external expressions of technical characteristics, which in some cases are considered to be less clear cut after technical analysis. The Opaque-glazed class, including the important Iraqi Lustre-painted and Blue-painted types, receive most attention, as they form the basis of the chronological framework for this corpus, and provide the continuing link to subsequent chapters. Turquoise-glazed class The Turquoise-glazed class is defined by a common technology of possessing a transparent turquoise glaze. This has traditionally been considered to be an alkali glaze, but technical analysis appears to present a more complex picture (see below). There are a number of types within this group, including the well-known "Sasanian-Islamic" jar, which will be called here the Hib type (hib being an anglicization of a general Arabic term for jar). Characteristics of these jars comprise heavy horizontal ribbing and three small blind handles, with sprigged and appliqué ornament, including "dot-track" (Fig. 3.1: BHR.08, SRF.10, SRF.50), "flower" (Fig. 3.1: BHR.07), "henge" (roundels of raised dots reminiscent to this observer of a plan of a henge monument: Fig. 3.1: HIR.11, SRF.16), "eye" (Fig. 3.1: LSL.01), "string" (Fig. 3.1: ZBE.09, SRF.50), and "rope" (Fig. 3.1: SRF.13, ZBE.07, SRF.11, STN.02). The decoration of the Hibs could easily provide the basis of a detailed study in motif assemblage which could be linked to a petrographic study to ascertain if particular motifs can be linked to specific centres, but this must await further work which is currently planned. Certainly at the moment it is possible to say that the "henge" sprig on a Basra Petrofabric piece from Siraf is identical to a piece from Hira of the "Samarra 3" Petrofabric, which is tentatively attributed to production at Samarra'. However, the particulars of "rope" decoration may be more easily linked to different petrofabrics (compare STN.02, SRF.11, and SRF.13, Fig. 3.1). Dating for this group may be suggested by archaeological evidence, such as their abundance at Siraf in the make-up of the Friday Mosque (i.e, before 850), and their presence at sites such as Hira (eighth century). Hibs do not seem to be found in intimate association with pottery made after the ninth century. The majority of petrofabrics found in Hib samples are of the group of petrofabrics that contain silt-grade inclusions and in some cases coarse sand grade aeolian grains. This group includes representatives of the Basra Petrofabric. The model developed for ceramic petrographic characteristics in the Tigris-Euphrates basin (see below) would restrict this petrofabric group to the Basra region and the highly fertile drained marshland areas of southern Iraq. However, occasional samples are of the petrofabrics attributed further upstream, including one attributed tentatively to Samarra', and the marshland-margin predominance may be due to sampling biases from studying pottery from trade sites (Siraf, Zabid, Dhahran, Bahrain). A possible exception to this southern-Iraq dominance in Hib production is indicated by fragments of Hib-type jars collected by Aurel Stein in the Makran and/or Kirman provinces, and currently in the British Museum (Fig. 3.1: STN nos.). These have not as yet been made available for sampling but they contain large inclusions of shale. Although they may originate from Iran (see below), it is equally possible that these shale-tempered Hibs come from somewhere such as Ahwaz or Susa, a location that may be more suitable to the typology (a Hib sample from Susa was not of this petrofabric). An interesting point about this pottery is that the forms are more elongate than the more typical Hib (compare to SRF.11, SRF.12, and SRF.50 in Fig. 3.1). As such, they closely resemble the vessels found in the tomb of Liu Hua (who died in about 930) at Tung Shih Mountain in China (Fukien Provincial Museum 1975; Ho 1995). Turquoise-glazed wares are also found at other sites in China, such as Yangzhou (Zhou 1985; Ho 1995), and at several other sites in east and southeast Asia (note unlabelled sites in eastern Asia on the inset map of the Old World on Map 1; all derived from Ho 1995; and Glover, in press). Other types in this class include smaller jars and "lidded" jars with characteristic channelled rims (Fig. 3.2: SRF.39 and SRF.34), and large heavy rimmed bowls (Fig. 3.2: BHR.06; for Siraf examples see Mason and Keall 1991). Clear turquoise glazes can be found on bowl and other vessel forms that can be compared to the sequence established for Opaque-glazed wares, and show continuous production throughout the period of this chapter. Petrofabrics of these other types of the Turquoise-glazed class are again dominated by the southern Iraq group of petrofabrics which includes Basra, although some wares, particularly the heavy-rimmed bowls, may be attributable to sources further upstream than Baghdad. Three non-Hib fragments from Raqqa (see Fig. 3.2) are of the southern Iraq group. An interesting sample is of a fragment from a small jar found at KwaGandaganda, a tenth-century site near Durban, South Africa (Whitelaw 1994). This was too small to be fully diagnostic (c. 1.5 × 1.5 cm) but is probably much like SRF. 35 or NPP.05 in Figure 3.2. Again, this piece was of one of the southern Iraq group. Blue-painted type of Opaque-glazed class In pre-Islamic contexts opaque glazes are often decorated by dribbles and lines of turquoise colour, but perhaps the earliest distinctly Islamic decoration is the use of a cobalt-blue pigment (Figs. 3.3 to 3.5). This Blue-painted type has cobalt- pigmented blue decoration applied over the glaze. In some cases the pigment may stand proud of the surface, while in other examples the pigment sinks into the main opacified glaze, creating an "ink-on-snow" effect. These differences in surface appearance are suggested by some to have a chronological or provenance relevance, but this has not been supported in this study. The difference is more likely due to firing temperature, which may even vary sufficiently within a single kiln firing. In every case of Blue-painted ware that has been thin-sectioned or examined with a binocular microscope it has been of the Basra Petrofabric. A substantial suite of discrete decorative elements or motifs is found in this group (Fig. 3.6), listed as the "BB" series, signifying Basra Blue-painted, and including calligraphically constructed motifs such as signatures and blessings (Tamari 1984:37-51). The most common vessel forms are bowls, of which at least six styles are defined in this study (Fig. 3.7). Seriation of these attributes, shown in Table 3.1, produces a satisfactory seriated succession of examples, with the general movement of attributes to the left of the table, schematically representing a generalized movement through time. Seriation also reveals four distinct groups which are linked by a restricted number of commonly held attributes, indicating four successive phases of ceramic production. As the attributes defined for these groups are equally applicable to other Opaque-glazed class wares from Basra, they are listed as Basra Opaque-glazed Groups One to Four. Basra Opaque-glazed Group One (BOG1) The group of Blue-painted pottery which is suggested as the first phase of production is characterized by a bowl form that exhibits stylistic echoes of pre-Islamic bowls (see Fig. 3.7). It has a simple rim, a segmental overall profile, and a dimpled disc-shaped base. Among the assemblage of distinctive decorative motifs (Fig. 3.6) are the "harp" (motif BB.1) and "sprig" designs (motif BB.4). In this early form these motifs are arranged in a tripartite pattern (Fig. 3.3) with intervening blue and cuprous green wedges ("proto-angel," motif BB.3) all around a central dot (motif BB.2: see Fig 3.4: ASH.26). Basra Opaque-glazed Group Two (BOG2) The dominant form of Group Two, considered to represent the second phase, differs radically from that of Group One, although there is some continuity in motifs. The dominant bowl form (Fig. 3.7) has an exaggerated cyma recta profile of recurved rim and wedge-shaped ring base. The characteristic rim, called the "serpent" rim profile, has a distinct swelling to form a "head," and a narrowing to form a "neck" (see Chapter 2). The distinctive "Basra-wedge" base is also highly characteristic. Although there are bowls in the preceding and succeeding groups with simple rims, there seems to be none in this group. It may be suggested that this new profile was made to imitate imports of Far Eastern white stonewares and proto- porcelains. The "harp" motif is not found in this phase, but the "sprig" is highly characteristic. Apart from a continuation of the tripartite arrangement (Fig. 3.4), the "sprig" motif is now commonly found in a quadripartite arrangement (Fig. 3.4), often arranged around three lines of calligraphy forming a box in the centre (motif BB.6), commonly with alternating blue and cuprous green half-moon motifs ("discrete crescent" motif BB.5). Basra Opaque-glazed Group Three (BOG3) There are three dominant bowl forms in Group Three, considered to represent the third phase. There is a simple rim style, effectively a continuation of Group One (which questions its total lack in Group Two). The recurved rim of Group Three is stretched out by the potter, becoming softer than the Group Two model ("recurved 1" rim profile). Similarly, a club-shaped profile, called the "tortoise" rim because it resembles a profile of the beast's head, results when the potter pushes the Group Two "serpent" rim inwards. The forms of these bowls are all united by the base, which is a short ring, with a characteristic thickening of the area inside the ring, making it thicker than the vessel wall between the ring and the rim ("thick-ring" base). The "evolution" of this base from the "Basra-wedge" base may be seen in some of the sherds from Hira in Group Two. Motifs in this group include "track" (motif BB.7), "track-moon" (BB.8), "two-line calligraphy" (BB.9), "angel" (BB.10), "circle" (BB.11), "wheel-blossom" (BB.12), "frond" (BB.13), and "dot-track" (BB.14). The "track" is a structural motif of fine lines used to construct iconographic motifs such as the Solomon's seal. There is little continuity of motifs from Group Two to Three, except for a bowl of Group Three from Istakhr (Fig. 3.4: ISR.01). Basra Opaque-glazed Group Four (BOG4) Forms in this group seem to be dominated by bowls with recurved rims and a simple ring base, which is taller and narrower than that of Group Three, and also lacks the widening within the ring. Although there is some continuity of Group Three motifs, the characteristic Group Four motifs are dominated by thin-line constructs with cross-hatched filling. Motifs include "hatched-track" (motif BB.15), "hatched-field" (BB.16), "flower" (BB.17), and "hatched-leaf" (BB.18). Although there appear to be no distinctive groups of Blue- painted pottery later than this group, occasional pieces indicate that the technology was still known. Evidence of their later date is provided by their sharing the same vessel with later styles of lustre-painting, including Group Six (see below and Fig. 3.5, SSA.22). The dating of the Blue-painted type will be discussed with other types in the Basra Opaque-glazed class, below. Lustre-painted type of Opaque-glazed class Although lustre-pigment use on glass may have precedence, Iraqi wares appear to represent the earliest use of the technology on ceramics (Fig. 3.8). The application of lustre- paint to pottery was an important development in the history of ceramics, and would continue to be an important decorative technique in the Islamic world and also Europe for several centuries (see Chapter 1 for account of technology). The Iraqi Lustre-painted type is often associated with the name of the ‘Abbasid dynasty. These wares form an interrelated sequence of motif assemblages (see Fig. 3.9 and Table 3.2), which continue across their more commonly recognized chromatic divisions (Kuhnel 1934). These motifs are labelled as the Basra Lustre (BL) sequence, as every example of Iraqi Lustre- painted ware that has been thin-sectioned or examined with a binocular microscope, it has been of the Basra Petrofabric. No other centre in the world has been found producing Lustre- painted wares of the same forms or motifs as the Basra wares, excluding the very earliest Egyptian wares which may be dated to the last quarter of the tenth century (see Chapter 4). Although there is clear continuity in the Basra wares, it is possible to arrange the pottery into a number of groups characterized by distinctive forms and motif assemblages. By internal evidence and also by correlation with the Blue-painted groups, it is possible to suggest four successive groups for production of Lustre-painted pottery. In order to correlate with the Blue-painted chronology, the same number sequence will be assumed. The incidence of motif and form is shown in Table 3.2, with individual motifs shown in Figure 3.9, and forms in Figures 3.10 and 3.11. Basra Opaque-glazed Group Three (BOG3) Ceramic forms in this group (Fig. 3.10) are predominantly bowls with a recurved ("recurved 1") or simple upright rim, and a base constructed with a low ring, and a thickening within the ring (a "thick-ring" base). This is essentially identical to Blue-painted Group Three forms, although there seem to be no "tortoise" profile pieces, and there is also a rim with a short recurve at the end (e.g., BRM.10). Of the motifs (Fig. 3.9) the "blossom" (BL.1) is perhaps the earliest, while the "frond" motif (BL.2) is essentially identical to that found in Blue- painted examples of Groups Three and Four. Other motifs are used as the structural basis of the main composition, which may include the well-known "Sasanian wing" design beloved of iconographers (cf. Lane 1947, pl. 10A). Motifs in this structural group which are largely unique to Group Three include "ladder zig-zag" (BL.3), "bar-scale" (BL.4), "star-eye" (BL.5), "dot-ground" (BL.7), and "tartan" (BL.8). The common Group Three "bull's-eye" motif (BL.6) carries on intermittently in later groups. The reverse of these pieces commonly has the "dash-circle" motif assemblage (motif BL.A), consisting of a circle surrounded by dashes. Other irregular reverse motifs include "zig-zag" (BL.C). Basra Opaque-glazed Group Four (BOG4) Group Four comprises two separate groups, which appear to have no relationship with each other except that they may both be shown to derive from Group Three characteristics. The first sub-group (BOG4a) exhibits a barely discernible transition from Group Three, with considerable continuity of motif and form. Recurved rims appear dominant (still "recurved 1"), but bases are more commonly taller, and lack the thickening within the ring. Some motifs carry on from Group Three (see Table 3.2), but there are a number of new motifs such as "V-ground" (BL.9), "discrete-crescent" (BL.10), "V-eye" (BL.11), and "zig-zag" (BL.12), which is a simplification of the Group Three "ladder zig-zag" motif. For the reverse, the "dash-circle" motif continues, with one unusual variant in the "dash-heart" motif (BL.B). The second subgroup within Group Four (BOG4b) is represented by pottery from Susa (Koechlin 1928, pl. XXIII p.160; Kervran 1977, figs. 40.5, 41.1 [SSA.16 and SSA.13 of this study]). Apart from published examples, some pieces were examined personally in the Louvre. The monochrome tiles of Qayrawan have exactly the same motif assemblage (Marcais 1928), and the incidence of motifs on several tiles is combined in Table 3.2. All the Susa pieces and the Qayrawan tiles are monochrome, but a number of polychrome pieces in the Benaki collection exhibit the same singular motifs (Philon 1980, figs. 146, 157, 189, 191). Those motifs used in this study are "hairy-fill" (BL.13), "dot-ladder" (BL.14), and "swirl" (BL.15). The reverse motif (not relevant to the tiles, of course) is a development of the "dash-circle" motif, consisting of circles with three dashes, but no other dashes ("fleur-de- lys," motif BL.D). The forms of the vessels are all bowls with recurved rims and ring feet, of distinctive form but related to the "mainstream" Group Four (BOG4a). Basra Opaque-glazed Group Five (BOG5) Group Five marks a smooth transition between the "mainstream" Group Four (BOG4a) and Group Six. The straightening of the recurved rim, a trend which has been evident since Group Two, is now the "recurved 2" profile, where the amplitude of the curve is not greater than the vessel thickness. Some ring bases also show the continuing trend towards greater height. Lustre colours are generally less diverse than in the previous groups, with some colours such as the ruby red missing altogether. A number of examples are decorated with just two different colours rather than three, and in most cases, the painter avoids laying one colour over another (which would have required a wait to allow the first pigment to dry), as was common in the previous groups. The most characteristic motif of this group is the "V-field" motif (BL.16), consisting of V-shaped marks alone or in combination with dots. The reverses generally carry variants of the "dash-circle" motifs, indicating continuity from the "mainstream" Group Four, but occasionally one finds the "dash-eye" type (BL.E) which has a double ring. Basra Opaque-glazed Group Six (BOG6) The forms of Group Six carry on with little change from Group Five, with weakly recurved rims ("recurved 2" profile), although in some cases the rim has developed into a straight profile, thus forming a cone-shaped bowl. The decoration is now applied in only one colour of lustre pigment, an olive green, which is remarkably consistent in hue when compared to later Egyptian monochrome lustre-pigment. Designs and motifs are also remarkably consistent, with two main stylistic subgroups. The largest subgroup (BOG6a) consists of pieces with a central figure, either human or animal, in a field of V-shaped marks ("V-field" motif), and with the rim decorated with a continuous semi-lunate pattern ("running-crescent," motif BL.17). Other motifs include "dash-field" (BL.18), which is a simplified version of "V-field," "hook" (BL.19), "ladder-band" (BL.20), and "double-L" (BL.21). The reverse is always decorated with the distinctive "dot-line" motif (BL.F), a radical development from the "dash-circle" type. The second subgroup (BOG6b) has bolder decoration in the form of palmettes, while the reverse has bow-shaped designs (Jenkins 1968a). The chronology for Lustre-painted pottery will be discussed collectively below with other Opaque-glazed types. Plain type of Opaque-glazed class Those pieces with no decoration at all over the (usually white) plain glaze are of the Plain type. Production of these wares may be attributed to three centres: Basra, which is by far the largest, another important centre tentatively identified as Baghdad, and Raqqa. For the Basra wares it is possible to suggest dating based on form. For instance, the Plain piece from Nippur/Niffar (see Fig. 3.12: NPP.03) is clearly identical to those predominant in Group Two, while those from Siraf (Fig. 3.12, SRF numbers) can be attributed to Group Six. The wares tentatively attributed to Baghdad may be dated by archaeology; for instance, pieces from Hira and Nippur/Niffar (Fig. 3.13) should date to around 800 (Wilkinson [1973:184, first analogy to 8] published an identical piece from Nishapur); the piece from Siraf (Fig. 3.13: SRF.43) is from the collapse deposits of a tenth-century building at Site M (see Appendix C); and it might be argued that examples from Samarra' are of the ninth century (see below). Four discernible petrofabrics were identified in pottery with tin-opacified glazes from the excavations at the site of Tell Aswad in Raqqa (not including petrofabrics from Basra and elsewhere in southern Iraq), all of which may be compatible with the site. Pottery production is known at this site, including large storage vessels, unglazed moulded vessels, and lead-glazed wares, but no actual wasters were included for analysis. One of these petrofabrics ("Tell Aswad 4"), mostly comprising slip-painted, lead-glazed, and unglazed wares of what might be called a Syrian derivation (including the "mustard and cress" glazed type known at Afamiya/Apamea [Rogers 1972, 1984]), essentially represents Euphrates sediments. Two others may be variants or may represent material from a source perhaps more in keeping with the basalt-dominated local terrain. One of these ("Tell Aswad 5") again represents "Syrian" lead-glazed wares and unglazed moulded water jars, while the other ("Tell Aswad 3") is solely represented by opacified wares that show close stylistic similarities to the contemporary Basra wares. The last petrofabric from this group is a proto-stonepaste, with clay and quartz modified by a high content of added glass (see below). This actually represents a single sample, a bowl of contemporary Basra form (Group Three) with a white opaque glaze (see Fig. 3.13). It may not as yet be proven that any of these wares were actually made at Tell Aswad, but all, hypothetically even the proto-stonepaste piece, would be compatible with production in the area. The "Tell Aswad 3" wares, and also the proto-stonepaste piece if locally made, may suggest a transference of personnel to serve the new palaces at Raqqa, just as recorded for the establishment of the palace site at Samarra' (see below). Splashed type of Opaque-glazed class The Splashed type has one or a number of oxide pigments splashed informally onto the glaze, usually copper green or turquoise, iron black, and manganese purple (see Figs. 3.12 and 3.13). Informal decoration of this nature goes back to Parthian times, as mentioned above, and specific styles of application and forms of vessel must be used to date the pieces. Many wares of an Iraqi origin that have been sampled to date are of the Basra Petrofabric, and so may be reliably related to the sequence developed for the Blue-painted and Lustre-painted wares. Two other groups, both associated with Raqqa (see above), would also appear to be typologically linked to the Basra wares. For instance, green pigment, perhaps meant to be a semi-lunate rim decoration but generally running in long fingers, is a common decoration on Group Ones and Two of the Basra Blue-painted type. Hence, pottery of the same form and decorated in this manner alone are contemporary to the Blue-painted wares. A large corpus of turquoise-splashed pieces, particularly from Siraf, with radial ridges and rings at the base of the interior, most closely resemble in form the products of Groups Five and Six. Although radial ridges emulating lotus petals may be traced back to c. 800, and rings within the base of the interior are found in ninth-century lead-glazed wares, they are generally not characteristics found in Lustre- or Blue-painted pottery, while the thick ring of the foot (e.g., Fig. 3.12: BSR.08) also appears to be unknown in the datable Basra wares. It is possible, therefore, that many of the Splashed wares may be later than the production of the decorated wares. Opaque-glazed chronology By reference to the published excavation reports and consideration of some unpublished data it is possible to suggest dates for the seriated ceramic sequence that has been developed in this study. At Susa (Kervran 1977) Blue-painted pottery is considered to have been introduced shortly after 700. Although this explicitly includes pottery of Group One, some examples of Group One (with green paint as well) are considered to have been introduced in about 830, along with pottery equivalent to Group Three, of both Blue- and Lustre-painted types. However, it would be acceptable to use this first Susa date as the introduction of the Blue-painted type (i.e., Group One). Justification for this includes the already mentioned similarity to forms of terminal Sasanian date from Nippur (unpublished). The absence of the type from other early sites, for instance, the c. 800 occupation at Raqqa, is easily explained by the distribution of the first Blue-painted groups, which are restricted to trade routes (Bahrain, Sri Lanka) or important sites moderately close to Basra (Susa, Istakhr). At Nippur/Niffar, late occupation of the site with Group Two and some possibly early Group Three pottery is considered to have ended in the first years of the ninth century, because of the total lack of coins collected from the site post- dating this period (M. Gibson, personal communication). This would suggest a termination date for Group Two of c. 800. The relationship between the Hira pottery and the coin evidence is not well recorded (see Appendix C), but may suggest an end date for Group Two of c. 780. The Hira pottery is especially pertinent as it seems to reflect a transition to Group Three, with obvious Group Two motifs, but in forms that begin to resemble the Group Three examples. At Nishapur a Group Three vessel was found "in the lowest level of a ninth- century building" (Wilkinson 1973:183, no. 1). At Ko Kho Khao the occupation is considered to have lasted from c. 800 to c. 850 (Bennet Bronson, personal communication; see also Bronson and Ho 1991), and a few diagnostic sherds of Group Two and Three were among the assemblage. Collectively this evidence suggests that the transition from Group Two to Three should be put at about 800, which would in turn support the hypothesis that Group One must have commenced some considerable time beforehand. The Lustre-painted tiles of Qayrawan have been dated to 862-863 by reference to a historical document (Marcais 1928:9-13; Creswell 1969:313-14). Although ambiguity within the text and its lateness (the writer, Ibn Naji, died in 1494) suggest caution, it is still a date worth considering. The document states that some tiles were brought from Baghdad, and that other tiles were made by a man from Baghdad, presumably at the site. The designs and motif assemblages of the polychrome tiles are suitable to Group Three of this study. The monochrome tiles from Qayrawan are Group Four (BOG4b), although they are in the second subgroup, that which does not show "ancestry" to Group Five. If the Qayrawan date reflects a stage when Group Three material was still available, but Group Four was current, then this would suggest a transition date from Group Three to Group Four of c. 850. The site of Siraf has seen some of the most extensive excavations of any Islamic site because of its importance in the trade with east Asia and Africa, attested by documentary evidence and initial survey. Some authors have attempted to use Siraf in isolation to develop general theories concerning Islamic pottery. The broader approach of the current study questions the conclusions of some of these studies. The complex history of abandonment and rebuilding coupled with otherwise unexplainable gaps in the ceramic sequence would suggest that Siraf's position as a merchant's settlement, with hardly any supporting hinterland, led to gaps in the archaeological record (Mason and Keall 1991; Mason 1994; Appendix C). Thus it cannot be expected to provide the continuity of those settlements with more secure bases. At Siraf it is claimed that Lustre-painted pottery did not appear until after 850 (Whitehouse 1979). Compared to the abundance of early styles at Susa, the dearth of useful pieces at Siraf and their general lateness must count against their relevance in the dating of Lustre-wares generally. However, the earliest pieces may all be considered to date to Group Five, the third phase of Basra Lustre-painted production. Although these are abraded sherds, they do make the Siraf date acceptable. The bulk of pieces, including those of better preservation (Tampoe 1989), are all of Group Six. A general tenth-century date for Group Six is indicated by short-occupation sites such as Madinat al-Zahra (Ettinghausen 1954). The end of Group Six may be equated with the end of Lustre-painted pottery production in Iraq. As the earliest Egyptian products show marked similarity to the last Iraqi products, it is likely that there was no gap in production (arguments for the transfer of personnel are provided below). Among this Iraqi-Egyptian transitional group are pieces in the figure and "V-field" style and a number of pieces with boldly painted palmette motifs (Jenkins 1968a). Among some of the more "developed" Egyptian (i.e., clearly not Iraqi) products are two pieces with inscriptions datable to c. 1000 (Yusuf 1958, pl. 1a-b; Jenkins 1968b). Given the rate of "development" shown so far, it is feasible to suggest about a generation's difference between the last Iraqi product and these pieces. Hence, the end of Group Six might be put at c. 975. Collectively, the foregoing suggests a time range of about fifty years each for Groups One, Two, and Three. The remaining groups fill the time between about 850 and 975, and of these, Group Five may be suggested to have been short- lived, judging by the number of examples and degree of "development" within it. Hence, dates may be suggested as follows: Group One, 700-750; Group Two, 750-800; Group Three, 800-850; Group Four, 850-900; Group Five, 900-925; Group Six, 925-975. Although this is clearly something of a simplification, and some of these groups may overlap in time, these dates provide as close a chronology as is possible, and are as close as is necessary for our purposes. The relevance of material from Sarre's excavations at Samarra' has been questioned by many in recent years (including the author, see Mason and Keall 1991), and this has caused it to be considered after the development of the tentative chronology. The ceramics from this site that have been considered in the present study consist of those published by Sarre (1923), and the holdings of the ROM, the BM, and the MMA. These include rarely occurring pieces of Group Two, a preponderance of Group Three, some Group Four, no Group Five, and very rarely occurring Group Six. The latter group is represented by only one piece known to the author (Sarre 1923, Table XV.2), recent surveys have also found them very rare (Northedge and Kennet 1994). This distribution would suggest the bulk of activity between 800 and 850, with effective cessation by c. 900. Given that this material is from Sarre's excavations, it would appear to vindicate his assertion that whatever the true length of occupation at Samarra', his excavations were in levels deposited during the caliphal occupation (836-892). One other historical source that may be considered is the somewhat apocryphal story given by Lane (1947:10) of the introduction of Chinese ceramics during the reign of Harun al- Rashid (r. 786-809). He suggests that this dates the beginning of Chinese influence in the production of Opaque-glazed wares. It may be argued from the present study that the transition from Group One to Group Two is most likely to represent a change from a form that can be considered to be a local development, to a form that would seem to imitate Chinese imports. The date suggested for this transition in the present study is 750, which is thirty-six to fifty-nine years before the reign of al-Rashid. Were this story of Lane's a hard historical fact, rather than a story associated with Harun al- Rashid, to whom many another fable is attached, it may have necessitated modifying the chronology. Given that Islamic merchants were well established in Canton before 750 (Hourani 1995:62-63), it is likely that some Chinese pottery had reached Iraqi markets before the reign of al-Rashid. Given the distribution of these wares along the trade route to China, imperial diplomacy hardly seems necessary. More amenable to our chronology is the date of the battle between Islamic and Chinese armies near Samarqand in 751 (Fehérvári 1973:27- 28), at which craftsmen and possibly booty from the defeated Chinese army are said to have been brought to Islamic lands. As the author may claim no knowledge of the number of potters enlisted in the Imperial Chinese forces, this point can be taken no further. Tite (1988) could find no technical relationship between contemporary Islamic and Chinese white wares (see also Chapter 8). Moulded type of Lead-glazed class Vessels in this group are formed from a mould, and covered with a green or yellow lead glaze (Lane 1939). It may be argued that the origins of this technology lay in the Romano- Byzantine Mediterranean, where small cups imitating bronze vessels were made. The wares I am most familiar with on a first-hand basis are generally small vessels with flat bases and vertical walls, and these are found in Islamic contexts associated with Opaque-glazed pottery of Groups Three and Four at Samarra' and Dhahran. Other pieces found as flat plates, often with three feet, have designs that may be related to Group Four Blue-painted motifs, while some of these footed plates are also said to bear lustre pigment (Lane 1939; Lane 1947, pl. 5B; I have never seen lustre-paint on wares of this type and would find its presence on a simple lead glaze unique and probably problematic). The above would suggest a date range between 750 and 900, and indicate that the transmission of the moulded and lead-glazed technology from the Mediterranean to Iraq may not have been immediate. Splashed and Slip-incised types of Lead-Glazed class This class of ware is defined by a lead glaze over a white slip, usually splashed with oxide pigments. In the case of the Iraqi products, the white slip is always of crushed quartz (see technology section, below). The Lead-glazed class includes various types of Slip-incised ware, in which the slip is incised to reveal the underlying dark body (often called "sgraffito"). The splashed lead-glaze phenomenon is often suggested as originating in China, where it was important in the T'ang Dynasty (618-906), and Chinese wares have been found in Islamic sites (Rawson et al. 1987-88). The incised-slip technique appears to be found only on Islamic pottery. Broad divisions of this class by previous authors (Sarre 1923; Reitlinger 1935; Adams 1965, 1970), based upon their respective field studies, seem to produce conflicting opinions. This is partly because the sites covered different periods, but perhaps also because these wares were produced in more than one centre and so do not from a continuous sequence in one tradition (see provenance section, below). This diversity of production has led to difficulties in developing a sequence of forms and motif assemblages for the splashed Lead-glazed wares in this study, compared to that developed for the Opaque-glazed class. However, by reference to archaeological data and correlations with the form typology presented for the Opaque-glazed wares, it is possible to suggest some generalizations, although only considerable further work will determine if these generalizations are valid. The earliest group of this class may be dated roughly to the eighth century. Although unincised styles continue later, they appear to predominate in this period, with the main design effect carried by the splash-painted colours. Incised designs are rather general, often made up of thin incisions executed in broad sweeping curvilinear strokes. In the following group (ninth century?) the pieces are less dominated by the colours, although these are still important, and the incised designs are more conspicuous, with recognizable motifs, such as the acanthus scroll, and patterns of manganese-purple dots. This style is probably equal to Sarre's Uberlaufglasuren style, and it is well represented at Samarra'. Forms include the distinctive broad-rimmed bowl ("bird-beak" rim, see Fig. 2.1), which is usually not found in the other classes (for an exception see Waagé 1948: pl. XV 1228). The next group (tenth century?) seems more often to favour geometric patterns, often with bands of weakly incised decoration. The thickness of the line is not as consistent, however, and some parts of the design approach the slip-excised technique (often called champlevéelsewhere). Colours are often weak, and splashed in a desultory manner, although some centres seem to favour dots of varying size. Later wares (eleventh century?) include pieces with the slip-excised technique, often with figural devices, and some are well made, while the pigment is very weak, often with few colours (a group of this period is defined in the upper Euphrates region as "Tell Shahin ware" [Tonghini 1995a], although, oddly enough, her rational for naming it after this site was finding evidence of production at Raqqa [Tonghini and Henderson 1998]). More coarsely made wares of forms not related to any elite ware group, often with decoration of large dots, may be contemporary or later (eleventh-twelfth century; a useful assemblage is published from Ana [Northedge et al. 1988]). These later wares can often be distinguished from earlier dotty styles by the tendency of the painted decoration to remain as discrete dots, and not flow into the glaze. This may suggest that the pigments were applied as coloured glasses, which was an important decorative technique in Egypt in the eleventh century (see Chapter 4). Wares with coloured decoration delimited by incisions can also be put in this later period (twelfth century and later). This study of the splashed lead-glazed wares must be considered very tentative, and is probably only specific to Iraq. A fuller study, in which considerably more examples could be sampled to ascertain provenance or at least groupings, would certainly be worthwhile. TECHNOLOGY Technology is divided into two subsections: that of the bodies, and that of glazes, slips, and pigments. Chemical data for bodies is provided in Table 3.3, and for glazes in Table 3.4. Body technology The ceramics discussed in this chapter primarily have fired- clay bodies, although some are modified by the addition of glass. Bulk analysis of some of these bodies is provided in Table 3.3. All of the Iraqi pottery sampled (see Table 3.3) has a similar bulk chemical composition, although it comes from a number of different petrographic groups. These are all calcareous clays (16-22% CaO), and are also similar to each other in the remaining major oxides (10-12% alumina, 5-7% magnesia, and 5-6% iron oxide). In a study of Basra Blue- painted wares, Tite (1988) suggests a firing temperature in the range of 850-1050 degrees C. Two ceramic bodies are notable for their technical modification. The first of these is defined by petrographic analysis as the "Samarra 2" Petrofabric and tentatively attributed to Baghdad (see below). Pottery of this group differs considerably from normal clay bodies in containing angular fragments of glass. Plate 3.1 shows an SEM photomicrograph of a sample of this pottery viewed in back-scattered electron (BSE) mode. Plate 3.1 also shows a view of one of these inclusions. They are primarily composed of glass, with some crystallization, particularly near the margins, and often have voids at the centre. Microprobe analysis of these inclusions, undertaken so as to obtain the bulk of the entire inclusion, produces values of 50-60% silica, 8-13% lime, 8-10% lead oxide, about 5% magnesia, about 3% potash, 2-5% soda, with smaller amounts of tin oxide, manganese oxide, alumina, and iron oxide. Of these elements, calcium, aluminium, magnesium, and iron may be from the body, which is higher in all these elements. The origin of this technology is unknown, but could be a continuation of pre-Islamic practice. Previously it has been suggested that this phenomenon was ancestral to the development of stonepaste (Mason and Tite 1994a). However, the second example of modified bodies may have chronological precedence. The second example of a technologically modified body from the Tigris-Euphrates basin is represented by a single example from the Tell Aswad excavations at Raqqa (Fig. 3.13: RQA.22). This is actually an example of what has been defined as a proto-stonepaste, being comprised of the components of a stonepaste (quartz, glass, and clay) but not in the proportions of the fully developed technology (Mason and Tite 1994a; see also Chapter 4). In the micrograph of the thin-section of this piece (Fig. 3.1:, RQA.22) can be seen quartz and glass in a matrix of clay, together with further inclusions of clay nodules. This piece can be dated to c. 800 by both style and archaeological context, so it remains to be seen if the Baghdad body is the earlier. More likely the previously hypothesized situation (Mason and Tite 1994a), that this was a technique generally known among the elite of Iraq's potters, is still valid. The transition to full stonepaste appears to have taken place in Egypt by potters who had relocated from Iraq (see Chapter 4). Glazes and surface zone Glazes of pottery of the "Samarra' assemblage" have received more attention than perhaps any other Islamic ceramic glaze, with analytical SEM studies by Kleinmann (1986, 1991) and Tite (1988), and with other studies by Hedges and Moorey (1975), Jones (1980), and Crowe (1977). None of these studies attempted to consider the full range of types, nor were they able to use a chronological sequence to explain the results. The most useful technological studies often do not publish illustrations of the pottery being studied, and when they do they are typologically non-diagnostic pieces. An important consideration in these wares is the creation of opacity in glazes. Theoretically, there are a number of mechanisms that have an effect upon the opacity of a glaze—its ability to prevent a light reaching the body and being reflected back to an observer. First, the refractive index (R.I.) of a substance is represented by the light-wave velocity in a vacuum divided by the wave velocity of that substance. If the R.I. of the substance is substantially different (higher or lower) than the R.I. of the mounting medium (in this case the glaze), the wave velocity will slow accordingly, possibly even having the effect of blocking the light entirely. Second, the light may be reflected if the angle of light propagation is greater than the critical angle of incidence (calculated by the R.I. of the substances). The amount of reflection is a function of the reflectance of the material. Third, light may be absorbed by the substance. This may be complete, as in the case of opaque materials such as most metals which have zero transmittance, or in other cases there is preferential absorption of particular wavelengths of light (absorption). Theoretical considerations aside, the realities of opacification are best examined by observation. With the aid of a 10× to 40× zoom microscope the various glazes were examined to determine exactly what was preventing the light from reaching the body-glaze interface. One agent was weathering of the glaze, a post-depositional product which was not intended by the potter, but nonetheless important to distinguish. Weathering occurs through loss of flux elements and crystallization of the remaining silica (as graphically interpreted by Brill 1963:128), producing restructured layers of silica. A weathered glaze has an opalescent appearance when viewed through the microscope, and is also highly distinctive in thin-section. Hence, decayed glazes were omitted from the SEM study. Typical agents of deliberate opacification appear to include bubbles, undissolved grains of quartz and feldspar, silicates crystallized from the molten glaze, and tin oxides. Theoretically, bubbles make poor opacifiers, but examination reveals that they appear to have a high reflectance, with most of the light being sent back to the surface. Quartz and feldspar grains are also theoretically poor opacifiers, but when large enough they also appear to absorb light, in the case of quartz perhaps because of fluid inclusions within the grains themselves. Silicates such as wollastonite (calcium silicate) and diopside (calcium magnesium silicate) probably crystallize from the molten glaze during cooling. These often form with numerous crystals radiating from a single point; hence, the alignment of some crystals will always be above the critical angle of incidence, and thereby reflect light away from the body. In some glazes these crystals are found exclusively along the glaze-body interface, owing to the higher lime content near high-lime bodies, so that although the glaze is clear, the last few microns next to the body are in effect opaque. Tin oxide may be present as discrete white crystal agglomerations, or be so dense that it is impossible to see into the glaze at all. Tin oxide has a large difference of R.I. between it and any glaze, and probably is also capable of absorbing light passing through it. The crystalline habit of these inclusions would suggest that at least some crystal growth occurred during firing or early cooling. For some early Basra tin-opacified glazes there appear to be remnants of coarse angular tin inclusions, added in the mix, which had melted and recrystallized, but the glaze had not matured sufficiently to disperse the tin throughout the glaze, and so maintained the general shape of the original tin additive. One opacifier observed in a Parthian sherd with the binocular microscope rendered the dark-coloured glaze totally opaque under the microscope, and was only visible as vague dark clouds in thin-section. In the SEM this was identified as iron sulphide, which would of course be simply opaque, regardless of refractive index or anything else. The present study has effectively been able to divide ceramic glazes of this period into three basic types: predominantly alkali turquoise glazes (Turquoise-glazed class), lead-alkali opacified glazes (Opaque-glazed class), and simple lead glazes (Lead-glazed class). Evidence suggests that the distinction between the alkali turquoise glazes and the opaque glazes was not as clear cut at the beginning of the period as might have been previously thought. Such a finding would only be possible in a study that integrated chronology and technology. The data for glaze chemistry is provided in Table 3.3, and is also discussed further in Chapter 8. First we shall deal with alkali and opaque glazes. The relevant pottery of the eighth century (Groups One and Two) consisted primarily of Turquoise-glazed types and types of the Opaque-glazed class, including early examples of the Blue- painted type. An important point to be taken into consideration when discussing the glaze chemistry of this period, is that the glaze that was to be seen (the "face" glaze) was different in chemistry from that of the glaze that was not to be seen (the "rear" glaze). In the case of jars such as the great turquoise Hibs, the "face" glaze was the exterior, while the "rear" glaze was the interior. Conversely, in the case of Blue-painted bowls, the "face" was the interior, while the "rear" was the exterior of the vessel. Plate 3.2 shows a photomicrograph of what is considered to be a typical pre-Islamic Iraqi glaze. It is of Achaemenid date, and is from the excavations at Nippur (see above). Although only one other similar pre-Islamic glaze, of Parthian date, was subjected to SEM analysis, the author may also include optical microscopic observation of a much larger body of Parthian and Sasanian wares from the same site. Both analysed pieces are alkali, with about 10-16% soda, 5% potash, 7% lime, and 3-4% magnesia. These results largely agree with those of earlier studies (Hedges and Moorey 1975, Hedges 1976; but see also Chapter 8). The glaze appears semi- opaque, and when examined with a binocular microscope it can be seen that this appearance is caused by colourless acicular crystals, gas bubbles, and undissolved quartz and feldspars. An identical observation was made on the large corpus of Parthian wares with opaque white glazes, using a polarizing microscope. Examination in the SEM (Pl. 3.2) reveals these crystals to be acicular calcium-silicate (wollastonite) and more equant magnesium-calcium-silicates (diopside). Examination of two further pre-Islamic glazes from Nippur, both with opaque black glazes, showed that they were opacified by the inclusion of iron-sulphide particles (Mason and Tite 1997). The general glaze chemistry was, however, comparable to the other two samples (see Table 3.4). The earliest Islamic Opaque-glazed pottery is represented by Group One Blue-painted pieces from Susa (Pl. 3.3) and Bahrain. Here crystals of the same nature as observed in the pre-Islamic glazes may be found, but there are also other inclusions. These are grains of tin oxide (white inclusions), found at the interface between the glaze and body. Presumably this was an attempt at a "tin-slip," perhaps by application of powdered tin oxide to the body, prior to application of glaze. Similar inclusions were noted in thin-sections of other examples of this style from Susa and Mantai. Examination of examples of the next group of Opaque- glazed wares reveals the next stage in the development of tin- opacification (Pl. 3.4). Here may be seen essentially the same range of opacifying agents that existed in the pre-Islamic glazes. However, here the tin-oxide grains are spread throughout the glaze, not just on the interface between the body and glaze. By consideration of Plate 3.4, it may be seen that light entering the glaze would be blocked, reflected, diffused, or diffracted by one of a number of agents. No one agent alone would be sufficient to make the glaze opaque. Later opaque glazes (Pl. 3.5) show an increasing reliance on tin as the opacifying agent, a trend that would continue with later opaque glazes in Egypt and elsewhere. The Blue-painted type is found either with the blue decoration standing proud of the glaze surface, or sunk into the opaque glaze to create an "ink-on-snow" effect. Examination of the former variant in thin-section shows that the pigment consists of visible opaque grains in a separate glaze layer applied over the main glaze. In the "ink-on-snow" variant, this glaze layer is not as distinct, and it may be that the pigment was applied alone to the glaze surface. This may, however, be simply the result of a higher firing temperature. Analysis of the metallic oxide pigment of the Blue-painted type indicates a high nickel and iron content, together with cobalt, the real colorant. X-ray diffraction by Kleinmann (1991) indicates that the colorant comprises grains of spinel. Kleinmann suggests that this is an artificially produced material, derived from a Co- Ni-Fe ore (linneite). Both Kleinmann (1991) and Tite (1988) consider the pigment to have been applied over the glaze prior to firing. All of the opaque glazes described above were the products of Basra. However, there is also a group of opaque- glazed pottery from another production centre. This centre is as yet unidentified, but the wares have been grouped into a common petrofabric ("Samarra 2," see below) which is tentatively attributed to Baghdad. This is the same pottery that had the glass added to the body (see above). The pieces examined in this study are from Samarra' (ninth century) and Nippur/Niffar (c. 800). Compared to the Basra glazes, these are very high in lead (34-44% PbO compared to <14% for the Basra glazes; see samples SMR.01 and NPP.07 in Table 3.4). In these glazes the opacity relies entirely upon the tin-oxide crystals, there are no relict inclusions of quartz or feldspars, and no crystallized wollastonite or other silicates. The dating of the Nippur/Niffar sherd and similar pieces from Hira suggests that the high lead and tin content of this glaze may have been contemporary with the low lead and tin products of Group Two Basra (i.e., 750-800). Combined with the glass additive in the body, this may indicate that the "Samarra 2" centre was a technologically more "progressive" centre than Basra. However, all the art-historically important wares were made at Basra, and the gradual development of tin opacification appears valid in these wares. One characteristic that easily separates the opaque glazes of Iraq from those of Egypt, apart from variations in relative abundance of essential ingredients, such as lead, tin, and alkalis, is the presence of about 2-4% magnesia in the Basra glazes (1-1.5% in the "Samarra 2" glazes), and its much lower presence in the Egyptian glazes, often as low as 0.2%. Both Hedges (1976) and Kleinmann (1986) note the magnesia content of the Iraqi glazes and attribute it to the use of plant ash. If so this would indicate that the Egyptian potters utilized a different source of alkalis. This is discussed further in Chapter 8. Turquoise-glazed wares are generally considered to represent a continuation of the pre-Islamic alkali glaze tradition. Although this appears to corroborate the situation as revealed in this study, it is equally true with regard to the opaque wares. Apart from the lack of tin and the lower lead content, these glazes are essentially the same as the contemporary Groups One and Two opaque glazes. These glazes contain undissolved quartz and silicates which have crystallized during the firing (wollastonite and diopside) at the glaze-body interface. This phenomenon was also observed by Kleinmann (1986, fig. 7.2). The effect of this silicate layer is quite significant, for without it the glaze would be considerably darker, as occurs elsewhere (for example, as observed by the author on some Turquoise-glazed Yemeni wares). The glaze is coloured by about 2% of copper. These results compare well with the earlier studies on alkali and opaque glazes. Kleinmann (1986) effectively reports the same phenomena but does not have the typological criteria that enable the phenomena to be placed in a progressive order, and instead averages two hundred years of production and development. Kleinmann (1986) also notes a high-lead tin- opacified glaze such as that attributable to our "Samarra 2" production centre. The existence of a glaze that had been deliberately made "grey," as described by Crowe (1977), has not been confirmed in this study. All glazes that approximate the macroscopic description of the "grey glaze" were found to be severely devitrified upon microscopic examination. Lead glazes in this region are essentially found on two types, the Moulded wares, and the Splashed or Slip-incised wares (Fig. 3.14). Both represent basic high-lead glazes, with a content of about 50% lead oxide. This may be related to the pre-Islamic lead-glaze technology of the Mediterranean, but no samples of this ware were available for this study to make sure. The Slip-incised ware has a slip primarily composed of quartz (Pl. 3.6). Superficially this resembles the stonepaste body of later Islamic ceramics, but no glass additive appears to have been present in the slip application, so it probably simply represents an application of quartz kept together by clay and/or gum. Any glass in the matrix is considered to have come from the glaze, but this white quartz-slip technology may have contributed to the body of knowledge that would eventually lead to the full development of stonepaste. CHARACTERIZATION AND PROVENANCE Documentary evidence attests to a number of production centres in Iraq during the period in question, the sources for which are discussed in greater detail by Tamari (1984). According to the medieval historian al-Ya‘qubi, potters from Kufa and Basra were brought to Samarra', which would implicate all three sites as possible potteries. Other references also attest to Basra production. Al-Azdi describes the raw material for "vessels" at Basra to be "a white layer," which actually sounds more like gypsum (alabaster). Al-Jahiz appears to be more specific about clay vessels from Basra, whose body "colour seems as if it has been moulded out of an egg-yolk." Yaqut also mentions pottery production in the area. The pottery produced at Baghdad gets honourable mention by both al-Azdi and ibn al-Faqih, and some of the lustre-tiles at Qayrawan were supposedly made by a potter from Baghdad (Marcais 1928). A further reference by ibn al-Faqih concerning Samarra' is considered by Tamari (1984:16-17) to allude to production of ceramic tiles or writing tablets. Further "documents" of pertinence are two vessels from Raqqa currently in the Damascus National Museum, which are inscribed with their places of manufacture: Hira and Basra (accession numbers A.16476 and A.17261, respectively [Michael Meinecke, personal communication]). Archaeological evidence is no more informative. Raphael (1925-26:27) makes claims to a kiln site at Samarra'. Unvala (1935:79) describes wasters and kiln furniture from Susa, and even claims Lustre-ware production at the site, but gives inadequate evidence for this. The only kiln material from any of these sites presently known and available is from Basra (see below), although the clay-bodied wares from the Tell Aswad site at Raqqa are also within the geological boundaries with which we are presently concerned. Although this list does not include some important towns of the period from the region, notably Wasit and Ahwaz, we may for now use the historical data in conjunction with a petrographic model derived from sedimentological studies and analyses of pre-Islamic pottery to suggest origins for some groups of pottery. The petrofabrics To date, thirteen distinct petrofabrics have been defined for Islamic glazed wares from the Tigris-Euphrates catchment. Of these only two sites have been positively associated with identified petrofabrics: Basra and Raqqa. We do, however, have a number of distinct petrofabrics and both historical and archaeological evidence of production at specific centres. In order to provide some basis for indicating a possible relationship between petrofabrics and centres of production, it is necessary to develop a predictive model based on the known geology of the Tigris-Euphrates basin,and standard models of sedimentary basin interpretation, and also to refer to a body of thin-sections of pre-Islamic pottery from sites across the basin. Figure 3.15 shows the Tigris-Euphrates basin as closely as possible to how it would have been in the eighth to eleventh centuries AD (geology and overall drainage from Aghanabati 1986; modifications from Naqib 1967; period modifications from Le Strange 1905, Adams 1965, 1981, Hansman 1967; pre-Islamic site locations from Roaf 1990). Structurally, the region represents a basin formed between the stable Afro- Arabian shield to the southwest (Bowen and Jux 1987) and the Zagros-Taurus mobile orogenic belt to the north and east (Haynes and McQuillan 1974). With regard to sedimentary rocks, the geology of the bulk of the region, specifically to the west and in the lower lands generally, represents deposits made on the stable margin of a continent, with shelf limestones and mature sandstones (Naqib 1967). The Zagros Mountains to the east and the mountains of the headwaters of the Tigris and Euphrates are generally made up of deeper sea deposits of the Tethys Sea, including shales and limestones. The geology of the mountainous areas is made considerably more complicated by the orogenic activity associated with the closure of the Tethys Sea, including high-grade metamorphic Precambrian terranes, and ophiolite sequences including basalt, gabbro, and ultramafic rocks (the scale of Fig. 3.15 necessarily makes the geology of the central Zagros appear less complicated than it truly is). Post-orogenic, and indeed still active, volcanism is widespread in the northern mountains, where it is predominantly of felsic chemistry, and in the lower lands of the north, where it represents flows of plateau basalt. Rainfall and relief of the region dictate that the bulk of recent sediments deposited in the greater Tigris-Euphrates basin would be derived from the highland areas, as indicated by studies of recent sediment load (Philip;1968, Ali 1976). The effect of the sediment load of the Zagros rivers may be seen in the wide curve of the Tigris around the fan of the Diyala, and the placement of the Shatt al-Arab near the western margin of the plain, away from the Karun-Kerkha fan. Cooke (1987) has suggested that the present coastline was in effect created two thousand years ago by the progradation of the Karun-Kerkha delta across an embayment which at that time reached almost 300 km inland, thereby isolating the inland waters which gradually were replaced with fresh water. Period-specific constraints are dominated by the presence of the Great Swamp, apparently formed in the fifth century AD by the switch in course by the Tigris from the ancient and also modern course, to a course which went through Wasit (Le Strange 1905). The Euphrates appears to have been unnavigable once it disgorged into a number of pools in the swamp, while the Tigris was navigable through a series of interconnected pools. These swamps would have acted as sediment traps and stopped any material from the Tigris and Euphrates headwaters from going further downstream during the early Islamic period. The smaller present-day marshlands and lakes, collectively called Ahwar, also block water and sediment flow (Philip 196;, Berry et al. 1970), with nothing of silt grade or coarser being deposited by the Tigris or Euphrates lower than about 60 km upstream from their confluence. Presently existing sands of the Shatt al-Arab are derived from the Kerkha and Karun rivers (Philip 1968). Even suspended loads are predominantly deposited on the plain (Hassan 1970), and beyond their alluvial fans in the marshlands the Tigris and Euphrates are effectively clear (Baltzer and Purser 1990). Core studies of the present Ahwar sediments indicate organic, carbon-rich, predominantly silt-grade material with some fine sandy deposits in recent times, but finer shelly deposits predominate before four hundred years ago (Aqrawi and Evans 1994), the same time at which the Great Swamp disappeared from the historical record (Le Strange 1905). Deposits derived from ephemeral streams, difficult to measure by regular methods of measuring sediment load, will also be important locally, and would bring in limestones and mature clastic sediments from the western regions (Hassan 1970; Baltzer and Purser 1990). Mineral analyses of the sediments (Philip 1968; Ali 1976; neither study contains findings regarding granulometry, texture, or overall abundance of minerals other than within heavy or light fraction, and both say little about lithic grains; see Baltzer and Purser 1990 for estuarine regions) indicate a light fraction comprised of quartz, carbonate, chert, chlorite, biotite, muscovite (mica more common in the Euphrates), and plagioclase feldspar of basic to intermediate chemistry, with a heavy mineral fraction dominated by iron oxides, epidote (more common in the Euphrates), amphibole (mostly hornblende, more common in the Tigris, especially downstream from the Greater Zab confluence), and pyroxenes (mostly augite). Mineral analyses of older sediments in the basin, specifically Pleistocene deposits found in fluviatile terraces and Pliocene fluviatile deposits, correlate closely with adjacent Tigris-Euphrates sediments, indicating an identical source in the then rapidly developing Taurus-Zagros orogenies. Grainsizes generally decrease downstream, particularly as noted by Baltzer and Purser (1990), until influx from the Karun increases grain size again. Natural levees are composed of finely bedded silts, generally coarser than the surrounding massively bedded floodplain deposits. Apart from the alluvial sediments, aeolian deposits would also be of importance in certain areas, particularly the southern regions. Airborne dust fallout over the northern Gulf region has been calculated to accumulate at 0.8 mm a year (Foda et al. 1985). Pre-Islamic pottery from the Tigris-Euphrates catchment has been sampled for a number of research projects, none of which has been published, but which collectively range in date from the Neolithic (Qalat Jarmo) to the Sasanian (Nippur) and Hellenistic (Dura Europos). Numbers of samples for each site vary widely. For example, the samples taken at Godin Tepe, include over 100 thin-sections and hundreds of sherds examined with a 10× to 40× light microscope. For a number of sites only a handful of representative pieces were analysed (e.g., Tell al-Swayhat, Ur); while a few sites were represented by as little as a single sample (e.g. ,a single brick from Ashur). In some cases pieces were examined that appear to have been wasters and were from kiln-associated contexts (Tell Hadidi; cf. Mason and Cooper, in press, "a"), or appear to have been wasters from unknown contexts (Swayhat), or were of styles for which precise parallels were known to have been produced on the site (Nineveh). In some cases the samples were of pieces for which trade is not considered likely (e.g., a bevel- rim bowl from Ur), but on the whole the possibility that the pottery was not made at the site remains a distinct possibility. However, for our present purposes we are only trying to create a model to predict where we should tentatively attribute Islamic petrofabrics, so if any pre-Islamic petrofabric appears not to fit the model, trade would be a good excuse to disregard it. Generally, with regard to the pre-Islamic pottery, choice of raw materials appears to be just as important as geographical restraints. This is particularly evidenced in early pottery (Neolithic, Chalcolithic), which generally appears to have a body containing fine inclusions (fine sand grade), which is then heavily tempered with vegetable fibre from various sources, including short chaff-like, long grass-like, and tubular-structured fibres (reeds?, hay?). This early material is therefore of limited use, as glazed wares, including those of pre-Islamic date, are generally not constrained by such choices, and tend to be relatively coarse. Other choices of raw materials may, however, provide supplementary information regarding the availability of raw materials. For instance, the tempering of cooking pots with inclusions suitable to that function, such as calcite at Tell Hadidi and Godin (Mason and Cooper, in press, "a", "b"), indicates its availability in the region of the vessel's manufacture. The types of aplastic material available from the central Zagros are exemplified by a number of sites in the region. Chalcolithic pottery from Godin, Sarab, and Giyan is of limited use, but later (Bronze and Iron Age) pottery from Godin is most often characterized by the presence of fragments of a fissile argillite (shale) which is sometimes silty, and sometimes metamorphosed to a low grade (Mason and Cooper, in press "b"); together with minerals from a granitic source. However, cooking pots are often characterized solely by granitic inclusions, presumably from sources nearer granitic outcrops in the region. A number of "exotic" pieces, particularly in later periods (Iron Age), point to trade with areas of more highly metamorphosed rock and areas of volcanic rocks. Such an area is exemplified by Iron Age pottery from Baba Jan, which is characterized by inclusions of basalt. Other sources on the eastern flank of the basin are illustrated by sites such as Arpachiyah, where coarse micritic carbonate and chert are associated with coarsely crystalline calcite and muscovite schists; and Gawra, where pottery containing coarse inclusions of a fissile argillite with incipient low-grade metamorphism appear. Other localized sources are exemplified by the sites in the Khabur headwaters. Here the diversity that may be found among sites such as Halaf, Aqab, Chagar Bazar, and Brak may be generalized in that the pottery contains generally coarse inclusions of sedimentary rock, often with basalt fragments. One exception to this may be pottery from Taya, which includes rounded grains typical of the Tigris alluvium, but they are from a calcite-cemented sandstone, perhaps from an older fluviatile deposit as described by Philip (1968). The type of material available from the headwaters of the Tigris and Euphrates is exemplified by pottery from sites such as Hadidi and Sweyhat. Here, coarse poorly sorted or bimodal sand-grade fragments of a diverse lithology include fragments of volcanic rocks, metamorphic rocks, minerals from granitic rocks, and limestones. Volcanics run the gamut from glassy and felsitic types to basalts, with various textures of each. Metamorphic rocks including amphibole schists and biotite schists. Inclusions from limestones include not only sparry and micritic calcite but also chert, some exhibiting radiolarian structures. Other samples from these sites provide further information on local availability, as locally derived deposits are characterized by basalts and limestones, but the above diverse assemblage seems to characterize the Euphrates at this point. At Raqqa, the next point downstream where pottery has been sampled, what is suggested as the Euphrates-derived material ("Tell Aswad 4") has the same general range of inclusions, but greater uniformity. From consideration of these sites and sites downstream it is possible to make further general statements concerning the compliance of the region to general sedimentological models. Increasing mineralogical maturity, the weathering-out of some minerals, and the relative increase in abundance of others, seems valid. Grain size also seems to decrease downstream, sotedness increases, as does increasing roundness. Hence, the coarse (c. 0.4-0.5 mm diameter) subangular to subrounded assemblage of diverse mineralogies and lithologies at Hadidi changes to a finer (< 0.1 mm diameter) assemblage of rounded grains of quartz, feldspars, and felsic volcanics at Ur and al-Ubaid. The petrofabrics defined for Islamic pottery may be found fully reported in Appendix B, but the context and general nature of the petrofabrics will be covered here. Generally, the petrofabrics found in Islamic pottery attributed to the Tigris-Euphrates basin may be divided into four main groups: a fine silt group (Basra group); a coarser, less mature group (deltaic group); a yet coarser even less mature group (Tigris group); and a yet coarser and still less mature group (Euphrates group). A further group, tempered with shale, may also originate in the region. Basra group The Basra group is dominated by fine silt (grainsize <0.05 mm diameter) with or without aeolian sand grains (0.25 to 1 mm diameter), and may be attributed to the sediment-starved region of the southwest. This group includes the petrofabric attributed to Basra, plus four more petrofabrics ("Siraf 2," "Siraf 3," "Siraf 4," and "Yemen 3"). The Basra Petrofabric itself is defined by analysis of a group of sherds and rod-shaped kiln furniture from Basra in the collection of the MMA (see Appendices A and C). Quartz content is generally about 2% while the content of ferromagnesian minerals (including hematite) rarely rises to 1% or more. The coarse population consists of rounded grains of trace to 10% quartz, trace to 2% cloudy untwinned feldspar, trace to 1% clear plagioclase, and occasional felsite. Pottery of this petrofabric comprises all Blue- and Lustre-painted wares compatible with the above typology, wherever found, most Opaque-glazed of Plain and Splashed types, all the Moulded wares analysed to date, some Turquoise glazed wares (particularly Hibs), and some pieces of Slip-incised ware (see Table 3.5). The "Siraf 2" Petrofabric includes 3-5% quartz, 1-2% feldspars, trace to 2% platy hematized inclusions (suggested as "biotite" in Table 3.6, and in many cases positively identified as such), and occasional amphibole, basaltic and felsic volcanic, clinopyroxene, and what may be a fine biotite schist. This last inclusion is represented by fine, polyminerallic, highly hematized, platy fragments with quartz and/or feldspars. Given the fineness, other identifications may be possible, such as highly fragmentary hematized basalt. This petrofabric is represented only by Turquoise-glazed pottery found in Siraf, including Hibs, so it is possible that "Siraf 2" represents a distinct workshop in Basra, specializing in these wares, or perhaps it is from another centre in the region, receiving slightly more silt from the Tigris and Euphrates. The "Siraf 3" Petrofabric includes about 3-4% "biotite," 2% quartz, and trace feldspars. This is fine grained, resembling the fine population of Basra and "Siraf 2," but totally lacks the coarse rounded grains, and also differs in abundance, being high in biotite (see above regarding identification of "biotite"), but lacking other ferromagnesians. Again this is found only in Turquoise-glazed jars from Siraf, and is possibly a variant of Basra or "Siraf 2." The "Siraf 4" Petrofabric includes 5% quartz, 2% each of biotite, felsic volcanic, and biotite schist, 1% untwinned feldspar, and trace amounts of plagioclase, potassic feldspar, amphibole, muscovite, clinopyroxene, and opaques. This also has a bimodal grainsize distribution, with grain counts dominated by fine grains, and a population of larger rounded grains primarily of quartz, feldspars, and felsic volcanics. Again, identifications of "biotite" and "biotite schist" may be looked upon with some caution (see above). Although the high content of felsic volcanic and biotite schist distinguish this from the similar petrofabrics, it is only represented by a single Turquoise-glazed piece from Siraf, so it could conceivably be an anomaly of Basra, or more likely of some other petrofabrics ("Siraf 2," "Siraf 3 "). The "Yemen 3 " Petrofabric includes 3-4% quartz, 1% each of untwinned feldspar, opaque, and felsic volcanic, 1% to trace plagioclase, and trace amounts of amphibole, biotite, clinopyroxene, and muscovite. There are no coarse rounded grains. Once again this petrofabric is represented solely by Turquoise-glazed pottery, consisting of a Hib from Zabid and a jar from Nippur/Niffar. Deltaic group The second group is consistently coarser (c. 0.1-0.2 mm diameter), and generally has higher total contents of inclusions, although it is generally similarly low in ferromagnesian mineral inclusions ("Siraf 7," "Yemen 2," "Hira 2"). This would suggest that these petrofabrics originate from the deltaic regions of the Tigris and Euphrates. The "Hira 2" Petrofabric is distinctly higher in carbonate, which may suggest that it originates from somewhere that could receive sediment from the west, such as Hira or Kufa. The "Siraf 7 " Petrofabric includes 8-10% quartz, 1-3% biotite, 1-2% each of untwinned feldspar and opaque, trace to 2% felsic volcanics, trace to 1% plagioclase, and up to 1% amphibole. This petrofabric too is solely represented by Turquoise-glazed pottery from Siraf, but is coarser than the petrofabric of Basra and its affines, so a different region may be postulated. It should also be noted that with the coarser grainsizes of this and subsequent petrofabrics, identifications of platy hematized ferromagnesian minerals may be considered more reliable. The "Yemen 2" Petrofabric includes 4-5% quartz, 3-4% untwinned feldspar, 1% clinopyroxene, trace to 1% amphibole, and trace amounts of plagioclase, potassic feldspar, and opaque. Again we have a petrofabric represented solely by Turquoise-glazed wares, this time Hibs from Lasail in Oman, and Zabid, but as with "Siraf 7" the grainsize distribution is distinctly different from the "Basra group." The "Hira 2 " Petrofabric includes 3-10% carbonate, 3- 5% quartz, trace to 1% untwinned feldspar, trace amounts of plagioclase, biotite, and opaque, and up to trace amphibole and clinopyroxene. This petrofabric is represented by pottery from Hira and a Hib from Dhahran. Although there is no special significance to this petrofabric's presence in Hira, the distinctive feature of this petrofabric, the high carbonate content, would be explained by Hira or Kufa's relative adjacency to outcrops of the Euphrates limestone west of the alluvial plain (al-Naqib, 1967). Tigris group The third group is not significantly coarser than the previous group but does contain higher abundances of ferromagnesians, such as amphiboles which are characteristic of the Tigris below its confluence with the Greater Zab river. This group includes two petrofabrics first defined from analysis of pottery from Samarra': "Samarra 2" and "Samarra 3" Petrofabrics. Of these the "Samarra 3" contains proportionally even more ferromagnesian minerals, suggesting that it originates further upstream. This would suggest that the "Samarra 2" could be attributed to Baghdad, and the "Samarra 3" to Samarra'. The Baghdad(?) or "Samarra 2" Petrofabric ("Abbasid B" of Mason and Keall 1988a; "Siraf 8" of Mason and Keall 1991) includes 3-7% quartz, up to 4% felsic volcanic, up to 3% each of biotite and untwinned feldspar, up to 2% each of amphibole and basalt, trace to 1% plagioclase, up to 1% potassic feldspar, clinopyroxene and biotite-quartz schist, and up to trace amounts of muscovite and shale. The most distinctive features of this petrofabric are the high content of ferromagnesian minerals and volcanic rock fragments. Grains are subrounded to subangular, and are mostly around 0.1 mm in diameter. A distinctive inclusion in this material which does not appear in Table 3.7 (in the interests of standardized tables, and also because it is a technological feature) is a glassy inclusion, usually present in the region of 1-2%. This glass has not been analysed in all cases, and so in principle it may include volcanogenic glass. However, microprobe analysis of some of these inclusions (see above) has revealed them to be a lead-alkali glass, probably added to aid vitrification of the body. Pottery of the "Samarra 2" body includes the only pieces with tin-opacified glazes other than the products of Basra and Raqqa. These are always plain and undecorated. Opaque- glazed pieces have been sampled from Samarra', Nippur/Niffar, Hira, and Siraf. This body has also been noted in Splashed Lead-glazed pieces, including Slip-incised ware, from Samarra' and Siraf, and a Plain Lead-glazed tile from Samarra'. The Samarra'(?) or "Samarra 3" Petrofabric ("Abbasid C" of Mason and Keall 1988a) includes 2-4% quartz, 1-4% amphibole, up to 5% carbonate, 2-3% biotite, 1-3% each of untwinned feldspar and opaque, trace to 3% felsic volcanic, up to 3% muscovite, up to 2% each of clinopyroxene and basalt, and up to 1% each of potassic feldspar and biotite-quartz schist. Of these, the large amount of coarse amphibole is particularly distinctive, as is the presence of carbonate. Grains are subrounded to subangular, and average about 0.1 mm in diameter. The most significant differences between this petrofabric and "Samarra 2" are the presence of carbonate and the higher content of ferromagnesian minerals. Pottery in this petrofabric comprises Slip-incised ware from Samarra', Susa, Hira, and Fustat, and a Hib from Hira. Raqqa group The fourth group comprises the four petrofabrics defined by analysis of pottery from Tell Aswad at Raqqa on the Euphrates (not including petrofabrics from Basra and elsewhere in southern Iraq; Mason and Keall, in press). Pottery production is known at this site and includes large storage vessels, unglazed moulded vessels, and Lead-glazed wares, but no actual wasters were included for analysis. One of these petrofabrics ("Tell Aswad 4"), mostly comprising slip-painted, Lead-glazed, and unglazed wares of what might be called a Syrian derivation (including the "mustard and cress" glazed type known at Afamiya/Apamea [Rogers 1972, 1984]), essentially represents Euphrates sediments. Two others represent material from a source more in keeping with the basalt-dominated local terrain, one of which ("Tell Aswad 5") again represents "Syrian" Lead-glazed wares and unglazed moulded water jars, while the other ("Tell Aswad 3") is represented solely by opacified wares that show close stylistic similarities to the contemporary Basra wares. The last petrofabric from this group is a proto-stonepaste, with clay and quartz modified by a high content of added glass (see technology section; see Mason and Tite 1994a). This actually represents a single sample, a bowl of contemporary Basra form with a white opaque glaze (see Fig. 3.13: RQA.22). It may not as yet be proven that any of these wares were actually made at Tell Aswad, but all, even the proto-stonepaste piece, would be compatible with production in the area. Shaly ware This fabric was first noted in fragments of Hib-type jars collected by Aurel Stein in the Makran and/or Kirman provinces, and currently in the British Museum (Fig. 3.1: STN nos.). At least one of these is probably from the area of Shahr- i Daqianus. Although these have not as yet been made available for sampling, they contain large (<3 mm in diameter) inclusions of shale. Shale-tempered petrofabrics have been identified for a number of Lead-glazed wares at Siraf, including the locally produced pottery (Mason and Keall 1991), and it is conceivable that the Makran geology would also be suitable. However, shales are also well attested in the pottery of the Zagros within the catchments of the Karun and Kerkha rivers, notably including the extensively sampled Godin Tepe pottery. Hence, it is equally possible that these shale-tempered Hibs come from somewhere such as Ahwaz or Susa, a location that may be more suitable to the typology (a Hib sample from Susa was not of this petrofabric). SYNTHESIS The earliest Islamic glazed wares in Iraq appear to follow the practice established in the region in pre-Islamic times. The exact relationship with these earlier practices, involving body technology and forming, will have to await further study of the pre-Islamic wares. Nonetheless it has been demonstrated that there is a considerable degree of continuity. Among the early Islamic white opacified wares, forms would seem to be a continuation of pre-Islamic practice. The Turquoise-glazed wares continue practices of Sasanian date, including details such as appliqué decoration. Traditionally the two glaze types have been considered to be radically different in composition, representing an alkali turquoise glaze, and a tin-opacified white glaze. The results of this study would seem to indicate that the glazes were in fact practically identical in chemistry, with only very small amounts of tin in the Opaque-glazed wares. In the opacified glaze the tin oxides combined with crystals of diopside and wollastonite that had crystallized from the molten glaze, undissolved quartz and feldspars, and gas bubbles. Collectively these provided the opacification for the glaze. In the case of the turquoise glaze, although undissolved quartz is also found, the calcium silicates are found only at the glaze-body interface, forming a transparent glaze which is rendered more brightly coloured. The actual coloration of this glaze is due to the addition of about 2% copper. The similarity of glazes would suggest that control over the appearance (apart from colour) may have been controlled primarily in the firing process. Among the important types produced in this initial phase of production is the Blue-painted type, apparently an Islamic innovation. Although decoration with copper-turquoise was known previously on white opaque glazes, this was the first such use of cobalt. Together with the vessel form, it is the motifs of this pottery that provide the chronological framework for the earlier part of the period covered in this chapter. Production centres for glazed wares in this early period are dominated by Basra, a situation which would continue throughout the period covered by this chapter. Also in this early period we see the introduction of tin, apparently at this stage conceptually related to a slip. In about 750 (according to the chronology suggested in this study) the forms of the Blue-painted wares change significantly, from bowls continuing a pre-Islamic tradition, to a radical new form, with a wedge-profiled foot-ring and a recurved rim, possibly influenced by the first imports of Chinese white wares. The motifs show a continuation of tradition, although some motifs had gone out of use by this time. Glaze chemistry also exhibits a continuing tradition, although progress is made with tin-opacification. In these glazes the tin is now spread throughout the glaze, but is still not the sole agent of opacification. Also at this time we see the products of a production centre to rival Basra. This centre, possibly Baghdad ("Samarra 2" Petrofabric), was considerably more "advanced" in its use of tin-opacification (i.e., closer to later practices), and also may have introduced the practice of adding glass to the body, which would facilitate vitrification and lead to the stonepaste technology of later periods. If this centre was Baghdad, then the potters may have been benefitting from its status as a centre of technological scholarship and research. At the transition with the succeeding group, put at about 800, we see significant but not radical changes. Bowl forms develop more gradually recurved rims, and ring-feet become square. The glaze chemistry switches from a multi-component opacified glaze to a primarily tin-opacified glaze. Motifs in the Blue-painted type show little transition from the Group Two suite to that of Group Three, but now Lustre-painted wares are introduced. Although Lead-glazed Moulded wares may represent a pre-Islamic tradition from the Mediterranean, in Iraq there does not seem to be any evidence for these wares before this time. This period marks the beginning of the floruitof the Basra kilns, with products reaching the furthest limits of the civilized world. The division between the Lustre-painted pottery of Groups Three and Four (c. 850) is somewhat arbitrary, with no radical changes in practice. It is at this time that we may have evidence for a centre to rival Basra in the production of Lustre-painted pottery. Some pottery from Susa and tiles from Qayrawan exhibit a motif assemblage clearly derived from Group Three practice, but has no relationship with other Group Four pottery, and shows no ancestral relationship to any Group Five wares. The Susa wares are of the Basra Petrofabric, and documentary evidence suggests that the Qayrawan tiles may have been made there. By the commencement of Group Five (c. 900) the production of Blue-painted pottery ceases almost entirely. The poly- and bi-chrome lustre-painted pottery of this short-lived group effectively represents the transition to the monochrome lustre group, and has essentially the same range of forms and motifs. The beginning of Group Six (c. 925) is marked by a radical change to monochrome lustre-painting. Motifs generally continue without major change at the transition, but the decoration on the reverse changes abruptly from the "dash and circle" style to the "dot-line" style. The Lustre-ware products of this group are remarkably uniform, which probably indicates a highly standardized production, perhaps suggesting a larger volume. The end of Group Six (c. 975) would seem to coincide with the end of Lustre-painted pottery production in Iraq. Pottery made subsequently would appear to be represented by the Splashed type of Opaque-glazed pottery, and various types of Lead-glazed pottery, particularly Slip-incised ware. Some of these would be of high-quality, but are nonetheless in pale imitation of the creative centres of Egypt and elsewhere. Consideration of the archaeological evidence from the various sites examined for this study would seem to indicate that Iraq was never again to become an important region for the manufacture of pottery. Why pottery production in Iraq, and particularly Basra, went from world-wide to purely local status, is a problem we will discuss in a later chapter. For the continuation of our immediate concerns, we must turn to the next great centre of ceramic production: Egypt. Return to Table of Contents