Shine Like the Sun: Chapter 8: Conclusion The conclusions that may be drawn from the above study will be considered in relationship to the three methodological approaches taken by the study. The conclusions themselves may also be divided into three main areas: the effectiveness of the methodologies, the results of the studies, and the relevance of those results. Further comments will also be made on avenues of future research. TYPOLOGY On the whole the methodological approach of the typological study seems to have produced very useful results. In particular it has enabled all the other findings regarding technology and provenance to be put into chronological context, and thus provide them with meaning. The approach of breaking the vessels down into a number of attributes, primarily motifs and forms, and subjecting these attributes to seriation, appears to have been highly effective. However, the chronology developed for the seriated typology is based at times on flimsy evidence, due to the nature of the available data. This may be modified in the light of future work. The ceramic groups defined in this study should nonetheless remain recognizable even if their chronologies are modified. Hence the strategic objective of creating a typology and linking the analytical research to that typology, will mean that the results of that research will always be of use. The detailed study of ceramic forms by accurate measured drawings and by presentation of these drawings in figures of the same scale is a standard approach in archaeology. This enables subtle but significant differences to be brought out that would otherwise be ignored. That this approach has not been applied commonly in the study of art- historically significant pottery is an omission which must be rectified. The overwhelming majority of the pottery studied here was wheel-thrown, and then subsequently turned. To recognize significant change in form it is necessary to recognize significant change in the motor habit pattern of the potter. Motor habit patterns are physiological constraints brought about by customary muscular movements, such as those developed by potters forming standardised wares (Arnold 1985:147). Potters of the status we are dealing with here tend to be innovative, but they will always be constrained by their motor habit patterns. Apparently major differences in forms may be less significant than they seem. In order to assess the significance of the change it is necessary to recreate the process of construction in one's mind and determine what was a major change in the actions of the potter. The breaking down of designs into structural elements, or motifs as they have been called in this study, is also an element of the archaeological study of ceramics. The motifs must be classified strictly, and be as regular as a signature. What is being characterized is, again, the motor habit pattern this time of the painter. Motifs as defined here will be painted in the same way, wherever they are put on the pot. Change in motifs generally reflect small differences in the movements of the painter which accumulate into significant visual change. Different sectors of the decorated surface of the vessel appear to receive different degrees of innovation. In particular, the decoration on the reverse tends to be highly standardized for each typological group. Rim motifs, particularly for flattened rims such as the "camel" rims or those of later Syrian flat-rimmed bowls, also tend to be highly standardized. Motifs in these areas appear to be particularly useful for defining groups. A strength of this study has been in considering wares from all the principle production centres of Lustre-wares and associated pottery. Combined with creating seriated sequences of the ceramics, this has considerably aided any attempt at providing dates for the wares. These parallel sequences reduce the problem of weighing archaeological evidence, which seems such a challenge for some people studying Islamic pottery. By having a sequence that can be compared to a parallel sequence at other centres, it is possible to reduce the influence, for instance, in the dating of Syrian Stonepaste-bodied Group One, a historical reference for ‘Acharneh or Qal‘at Ja‘bar, or a single coin at Tille Hoyuk. A ceramic type or style in isolation cannot be satisfactorily dated by the meagre archaeological evidence that is available. This is especially so when individuals are blinkered by overly stressing the archaeological site or sites with which they are associated. To date these wares adequately, it is necessary to determine what went before, after, and concurrently at other centres. A further conclusion of relevance to the typological study is the significance of characterization and, where possible, the determination of provenance. To create a typological sequence, it is necessary to ensure that one is in fact studying a single continuing tradition, usually in the same place. Hence, pottery from one centre will not necessarily fit into the typological sequence of another. Despite this, the impact of traded wares and regional imitation appear to produce synchronic formal uniformity across the region. TECHNOLOGY The methodological approach of this study to technological analysis utilizes what is already recognized as the most effective technique: the SEM with attached X-ray spectrometer. Although this technique requires no further endorsement, it may be pointed out that numerous glazes were found to be highly weathered and devitrified on the surface. This was true even of many glazes that looked in good condition to the naked eye. Hence, the results of any non- destructive technique, such as X-ray fluorescence analysis of the surface, would necessarily have produced spurious results. It might further be pointed out that in some cases the surface of the glaze, even if in pristine condition, may not fully represent the glaze chemistry itself. For example, in the very early Iraqi glazes with tin-opacification, the tin-oxide grains were actually near the glaze-body interface (see Chapter 3). This phenomenon could have been revealed only by a technique in which examination is made of a section cut from the ceramic. Apart from confirming the general validity of the technique, there were a number of technological questions raised Chapter 1 which have been answered by this study. As the pertinent findings have been spread across several chapters, it seems appropriate to bring them together here. Stonepaste and its origins One of the findings that may be attributed to the approach of studying a chronologically and geographically diverse group of ceramics has been the determination of the development of the stonepaste technology. A number of analytical studies had been undertaken on stonepaste bodies previously, but these were on material post-dating the earliest dated pieces and therefore reflect the fully matured technology (Allan et al. 1973; Kingery and Vandiver 1986; Tite 1989; Henderson and Raby 1989). Stonepastes appear to have their origins in Iraq. There are two phenomena that indicate this. The first is found in the body of the "Samarra 2" Petrofabric, tentatively attributed to Baghdad (see Chapter 3). This ware, which has a tin-opacified glaze, is found in archaeological contexts dating from the late eighth to the mid-ninth century. Along with the range of inclusions typical of the region were found fragments of an alkali-lime-lead glass, in some cases also containing tin. Why fragments of glass were added is an important question. No inclusions of crushed pottery, or grog, were seen, so the glass cannot have been glaze adhering to included crushed pottery. Theoretically, there are a number of practical advantages to adding glass fragments. If the alkalis were efficiently taken into the clay matrix, they might accelerate vitrification at a relatively low firing temperature, and thus increase the hardness and density of the body. The second phenomenon is the application of a quartz-based slip to certain types of lead- glazed wares. This seems very similar in appearance to true stonepaste, but analysis indicates that the slip was applied without added glass fragments, so it does not seem likely that the two techniques were closely related at this stage. At some time in the later tenth century potters left Basra and set up potteries to make Lustre-painted wares at Fustat, the economic and industrial capital of Egypt. Although the evidence for the movement of the Baghdad potters is not as strong as that for the Basra potters, to explain the next development in stonepaste it is necessary either that the Baghdad potters did indeed move, or that their technology was known to the Basra potters. This next development was the introduction of the Fustat proto-stonepaste body, comprising quartz and inclusions of relict glass additives in a matrix of highly vitrified clay. This is a logical development from the "Samarra 2" body, as it is still predominantly formed from clay with added glass fragments. But whereas the "Samarra 2" body was formed from an ordinary central Iraqi clay, all the components of the proto-stonepaste were prepared separately and brought together deliberately. Chemically, the clay does not appear to resemble any of the other Fustat clays, and has no natural inclusions, unless these have been removed by levigation. The quartz is the only inert temper, and does not appear to be a natural component of this clay. The glass appears to be specifically constituted for its purpose, and is not a by-product of glaze-making. Hence, the Fustat proto- stonepaste body has all the attributes of the standard stonepastes, but in different proportions. Knowledge of proto-stonepaste in Iraq must also be considered possible. We have already seen in Chapter 6 that an early ninth century Opaque-glazed bowl from the Tell Aswad excavations at Raqqa in Syria had what must be considered a proto-stonepaste body. In discussion of this piece it was hypothesized also that it was made by Basra potters encouraged into the area. Another relevant point is the eleventh-century text by al-Biruni, writing at Ghazna in Afghanistan, which states: "One may make [Chinese bowls] here from pure Marwa, . . . mixed with clays, though these are half-bred, impure Nabati" (Allan et al. 1973). Al-Biruni makes clear elsewhere in the text that Marwa is quartz pebbles collected from stream beds. Nabati is a somewhat abusive term for inhabitants of the Sawad of southern Iraq. Allan uses this text to argue for an origin for stonepaste in Iran, but as al- Biruni is contrasting with China, it seems reasonable to suggest that "here" refers to the Islamic world. In that case the reference may actually be indicating products of southern Iraq (i.e., Basra). If the technique was known in Basra, but not utilized, we should not be surprised to find it on high-status or industrial sites (the Tell Aswad find was in an area dominated by the waste of pottery production and is contemporary with nearby caliphal palaces), or the knowledge of the technique being known among the scientific community (al-Biruni was writing a technical treatise). However, it should be reiterated that it is in Egypt that proto-stonepaste first becomes a significant part of the archaeological record, and leads on rapidly to true stonepaste. As it stands there is no evidence for the use of stonepaste outside Egypt earlier than the date at which it would have been introduced from Egypt. In the second phase of Egyptian Lustre-ware production (1025-1075), we see a rapid development to true stonepastes, resulting in essentially the same technology as that used for fine Islamic pottery to the present day. This development does not readily appear to be a natural and obvious progression, since it involves a change from a body predominantly of clay to one consisting predominantly of quartz. However, the resulting true stonepaste in Lustre pottery essentially fills the same niche as that previously occupied by proto-stonepaste; that is, it was used for the better quality products, perhaps for a clientele that considered itself more discerning. The stonepaste wares in this period represent a larger proportion of total Lustre-wares than did the proto-stonepaste. Given this continuity, it is conceivable that stonepastes did develop naturally, perhaps to solve problems that may have existed in the prototypical practice. During the final periods of Lustre-production in Fustat (1075-1175), Lustre-ware was not made using stonepaste at all, but was found with the hard red-pink clay "High-Ca" petrofabric body. However, the typological study indicates that it was in about 1075 that potters moved on again to Iran and Syria. In these cases it was not only the technologically demanding lustre-painting technology that was introduced into these new areas, but also the stonepaste technology as well, which also required special expertise. None of these new manufactories used clay-bodied ceramics; instead, in both Iran and Syria Lustre-ware and all other fine wares were made only of stonepaste. Stonepastes in Iran and Syria conform to the classic descriptions of the technology derived from the treatise of Abu ‘l-Qasim and later accounts. Most Iranian petrofabrics consist entirely of angular grains, indicating that the quartz was obtained by collection of pebbles in stream beds. The only exception to this in Iran is the petrofabric attributed to Rayy, which appears to represent a source of carbonate- cemented sandstone (see Chapter 5). In Syria the use of cobbles also seems indicated at most centres, but at Damascus an exceptionally clean sand was used. In some cases the quartz is ground so fine, such as in the Syrian "Tell Minis" and Iranian "Rayy 3" Petrofabrics, that it is probably impossible to be certain whether the source was sand or pebbles. The body attributed to Kashan is unique in having the quartz grains almost entirely represented by chert. The clay in the stonepaste mixture is indicated by the amount of alumina. This is generally around 2%, but this differs considerably in two examples. The first represents the bulk of Syrian wares. Although at first the Syrian potters appear to have used a highly aluminous clay (see Chapter 6), the bulk of wares have very little alumina and high amounts of lime, suggesting that a highly calcareous clay was used. The second anomaly are the products of Kashan, in Iran (see Chapter Five). These are even more highly aluminous than the standard stonepaste bodies, and it is clear from the extensive interstitial matrix that considerably more clay went into the mix. Unfortunately, no contribution has been made in this study towards estimating the firing temperatures of these wares. According to ethnographic study firing stonepaste requires twice as much fuel as firing clay; it is unknown if this is due to higher temperature or more sustained firing. This would be a useful avenue of future research, and could perhaps be initiated by a series of experimental firings to see what phenomena, for example, microstructure observed with the SEM or mineralogy as determined by X-ray diffraction, may be related to temperature. The ultimate origins of the Iraqi technologies that led to stonepaste remains a puzzle. Siliceous ceramics certainly existed in pre-Islamic times in Egypt and Mesopotamia, with production of so-called Egyptian faience or Egyptian blue. This technology persisted in Iran until recent times, presumably as a continuing tradition, for making beads (Wulff et al. 1968). A number of analytical studies have been made of this ancient siliceous body (Tite et al. 1983; Kaczmarczyk and Hedges 1983; Tite and Bimson 1986; Kingery and Vandiver 1986:51-68; Vandiver and Kingery 1986; Tite 1987). Normally the ground quartz bodies were probably held together initially with gum, and the glaze subsequently stabilized the material during firing. There appears to have been three glazing methods. The first is direct application with powdered glaze materials (i.e., silica, lime, alkali, and copper) suspended in water. The second is the cementation method, in which bodies are fired while buried in the glazing mixture. The third technique is efflorescence, in which glazing components are mixed with the body and brought to the surface during subsequent drying. In addition, there does appear to be some evidence that ground glass or frit, possibly with clay, was very occasionally mixed with the quartz body (Tite 1987). This limited and rare use of added glass fragments and clay, linked with the apparent absence of siliceous vessels for late pre-Islamic periods, would argue that Islamic stonepaste is an independent invention. A parallel phenomenon to the added glass particles found in the early Islamic wares may be observed in England during the seventeenth and eighteenth centuries, where conscious research and development was aimed at producing porcelains (Freestone 1993). Here bottle glass was added to a clay body to accelerate vitrification, and produced similar textures to those subsequently observed for the Fatimid proto-stonepaste. This conscious experimentation with raw materials may also apply with the Islamic potters, particularly given that Baghdad was among the foremost scientific centres at this time. It may be misguided to ignore the possibility that the effect of adding glass was expected. Conceivably the entire siliceous ceramic concept was understood in early Abbasid Iraq, but there was no need to apply the technology to making whole bodies as the potters already had full control over suitable clay bodies. The existence of this knowledge may explain the proto-stonepaste body found at Tell Aswad, which by its very close similarities to Basra forms may be dated to the first half of the ninth century (see Chapter 6). There are probably a number of reasons for the development of stonepaste. One of these is raw material procurement. When the Iraqi potters arrived in Egypt they would have been unfamiliar with local clays and the other raw materials that would need to be added to render the clays suitable to their purposes. Indeed, these resources would most likely have been already claimed by Egyptian potters. About 80% of the first group of Egyptian Lustre pottery was produced using local clays that appear to have been little utilized previously, and that appear to have needed considerable processing to produce the required standard. Proto-stonepaste would have been another way of making a body of the correct type. During production of Egyptian Group Two, the clays seem to deteriorate, becoming coarser and redder. If this was also the result of problems with raw materials, then it would explain the switch to stonepaste. The clay used for proto-stonepaste would certainly appear to differ from the standard clays used at Fustat, and so may have been in short supply. Hence, true stonepaste could represent a simple move from a high content of costly glass and rare clay, to a high content of cheap abundant quartz sand. The explanation for the final and complete reliance on stonepaste bodies by the potters who went to Syria and Iran may also relate to raw material procurement. Problems in obtaining the right clays encountered by Iraqi potters coming to Egypt would have been even worse for the Egyptian potters in Iran and Syria. Iraq and Egypt are regions of large mature rivers with uniform and broad alluviation, while Iran and to some extent Syria are high-relief areas with immature and active rivers, which deposit less suitable clays. But, by using stonepaste it would not be necessary to first find suitable clays which were not already spoken for, and then spend considerable effort developing new processing, forming, and glazing practices suitable to the new clay. Instead, the stonepaste technology would enable production of a dependable ware, using materials not already being used. Quartz is effectively the same, at least as far as the potter is concerned, wherever it is found. Glass is an industrial product, made to order. The clay needed is of a high quality, but only small amounts are required, and it is possible to transport this some distance. That this predictability of product is important is supported by the fact that the whiteness of the body was not important, as the earliest Iranian Lustre-wares have a tin- opacified glaze essentially the same as the previous Egyptian products. One other reason for the development of stonepaste is obviously the oft-cited fact that it is white. This would become very important for later pottery, particularly the Underglaze- painted wares, but all the lustre-painted pottery of the period of the first development of stonepaste had a very effective tin- opacified glaze. However, whiteness would have been a very important consideration for the Monochrome Incised wares which were introduced in this period, and are often said to be made in imitation of imports of Sung carved wares from China. These are always made with a stonepaste body, upon which designs are incised or carved, the whole then covered with a monochrome lead-alkali glaze which is not opaque. Glazes The Islamic potters have often been said to be inheritors of two glazing traditions, the alkali glazes of Mesopotamia and the high-lead glazes of the Romano-Byzantine Mediterranean. No analyses have been made of the second group in this study, but continuity with pre-Islamic Iraq does appear patent. The origins and developments of Islamic ceramic glazes will be considered in three main threads: opaque and lead-alkali glazes, alkali glazes, and lead glazes. Opacified and lead-alkali glazes Pre-Islamic Iraqi glazes include a number of examples that may be called opaque. The opacity is rendered by the presence of crystals of calcium silicates and magnesium-calcium silicates; bubbles; and undissolved grains of quartz and other inclusions. Opacity may be an extreme term for these glazes, but when you look at them you are looking at the glaze, not the clay body. The first significant development in Islamic glaze technology was the tin-opacified glaze (see Chapter 3). During the first century of existence of this glaze type it can effectively be described as a standard pre-Islamic glaze with tin in it. The very first of these glazes, found in pottery of Basra Group One (c. 700-750), consists of a typical pre- Islamic glaze with tin-oxide grains found on the interface between the glaze and the body. This may be called a tin- oxide slip, as it appears to be applied by covering the vessel with tin-oxide powder, and then glazing it. Presumably this was linked in some way to other slip techniques. In the next development, during Basra Group Two (c. 750-800), the glaze again resembles a typical pre-Islamic semi-opaque glaze, but this time there are grains of tin oxide spread throughout the glaze. There is not enough tin oxide to fully opacify the glaze by itself, and without the presence of the agents of pre-Islamic opacification it would not be opaque. However, the tin was added to the glaze mixture itself and is hence a development in the direction of more typical tin-opacified glazes. In pottery of the next group, Basra Group Three (c. 800-850), the tin is the predominant source of opacification, and the tin content would increase with later wares in Iraq, Egypt, and Iran (see Fig. 8.1). Lead is always found with tin in these early Basra glazes, but usually in small amounts, perhaps because it was required in the tin-processing (Allan 1973a), although similarly small amounts of lead are also found in contemporary turquoise glazes. Some later tin-opacified glazes have quite high levels of lead in them, over 30% (see Chapter 3 and 4). Tin-opacified glazes continued to be made in Egypt during the first century of Egyptian Lustre-painted production (c. 975-1075), and when potters left Egypt c. 1075 for Syria, Iran, and Europe, they took the technology with them. Tin- opacified glazes became very important in Europe, but less so in Syria and Iran. In Syria they were rapidly dropped, and during the production of Syrian Group One (c. 1075-1125) a lead-alkali glaze was dominant. This glaze type may have been derived from practice in Egypt for use in Incised wares (see Chapter 4). It was certainly also used in Egyptian Lustre- painted wares, very rarely in Group Two and entirely in Groups Three and Four. For the production of these last two groups the potters used a red clay body; therefore, the clear lead-alkali glaze required an underlying slip of wollastonite crystals. In Syria the potters relied solely on stonepaste bodies, so such a slip was unnecessary. This glaze technology was not to last. By the period of production of Syrian Group Three (c. 1125-1150), the Syrian potters had switched to an alkali glaze. In Iran tin-opacified and alkali glazes appear to be used throughout the period of Lustre-ware production. In some cases there are even bowls with tin-opacified glazes on the inside and dark-blue alkali glazes on the outside (see Chapter 5). Alkali glazes The aim of this section is not only to consider the history of alkali glazes, but to examine the interrelationship between the alkali elements found in all the Islamic glazes where they are found. Further, we will consider what changes in use occurred through time, what cause there may have been for these changes, and how these changes may reflect use of raw materials. As varying overall abundances of these elements in the glazes may at times obscure the true relationships, the focus of discussion will be a series of ternary diagrams (see Fig. 8.2), one sequence relating soda (Na2O), potash (K2O), and lime (CaO); the other relating soda and potash to magnesia (MgO). Data used in these diagrams comprise WDS microanalyses of various glazes used in this study and also some further analyses derived for Chinese ceramics from Guo (1987), and supplementary pre-Islamic data from McCarthy et al. (1995). As explained in Chapter 2, there were a number of problems with the EDS data; for instance, they were consistently lower in MgO than the WDS analyses, so the EDS data were not used, although they plotted within the vicinity of WDS data in the soda-potash-lime plot. The sources of raw materials for glazes in the area being discussed may be divided into two groups: vegetable and mineral. All vegetable matter appears to contain the alkali and associated elements in some amounts, and these may be concentrated for use by burning the plants and using the ashes.In particular, halophytic plants, which tolerate saline conditions, contain appreciable amounts of the requisite elements, and these are found across a great deal of the Middle East. Available analyses of such plants are very uneven. Brill (1970) compiled most of the available analyses and added some more, but it must be recognized that these were not collected consistently with regard to plant species identification (see Table 8.1). Species analysed by Brill include Salicornia, Salsola, especially S. soda, Tamarix, and another plant described by what appears to be its Arabic name of chinan. Bezborodov (1975) published analyses of two species, Salicornia herbacera and Kalidium caspicum, including analysis of different parts of the same plant, but does not appear to have provided specific locations for the site of collection. For the correlation of glaze and glass chemistries to plants of specific species and locations, it might be worthwhile to undertake a more comprehensive study of plant sources with plant species being identified accurately. The available analyses certainly do point out that different parts of the same plant and also the same plant from different locations can produce widely differing results. Regarding plant sources, Abu ‘l-Qasim writes in c. 1300 (Allan 1973a): "shakhar, which they call qali . . . is made by burning pure, fully-grown ushnan [Salsola soda] plants, not mixed with shureh [Salsola tragus], which is like ushnan. The best shakhar is that which has a red-coloured centre when broken, with a strong smell." This shows that the potter could be very selective about not only the variety of closely related species that was required, but also particularly desirable plants. Regarding preparation, Abu ‘l-Qasim states: "They take 105 parts of shukar-i sang [quartz] which has been . . . ground, and sifted . . . , and 100 parts of shakhar in lumps the size of hazelnuts or almonds, and mix them and put them in a kiln, technically known as bariz. The pungency or weakness of the shakhar varies depending on the place. Hence they need for 1 man [a contemporary weight measure] of stone 1.5 mans of Tabriz shakhar or 1 man of Baghdad shakhar. This is cooked over a slow fire for six hours . . . until . . . it is become one, like molten glaze, and this is the material of glass vessels." The molten glass is then ladelled directly into vats of water. Apart from showing the method of preparation, it again shows the importance of selection of raw materials; here, the location where the plant grew is important. Recent practices in Iran were described by Hans Wulff (1966:160-61). Apparently the preparation of "potash" (qaliya) is often undertaken by specialists (qalla'). A number of them at that time worked on the edge of the northern desert at Qom and were renowned for the high quality of their product. Halophytic plants were collected from the desert over a number of weeks while they were not completely dry, the best being Salsola varieties, with a less preferred species being Quilandia bonducella. The plants were burnt with a slow smouldering fire in a pit about one metre wide and two metres deep. The ash was then calcined into blocks of ten pounds each. Then fifty-five pounds of ground quartz and sixty-five pounds of "potash" were mixed, together with half a pound of manganese oxide, and then heated for eight hours until a clear bubble-free glass was produced, and this was then ladelled into vats of water. Important mineral sources for alkalis include the sodium carbonates natron and trona, both minerals having cognate etymologies. Natron, the sodium carbonate decahydrate (Na2CO3.10H2O), is actually quite rare and most reported incidences of sodium carbonate are of trona (Na3H[CO3]2.2H2O). Analysis of sodium carbonate from the Wadi Natrun in Egypt by Brill (1988) showed that the material was actually trona. Both are found as efflorescences or as more concentrated deposits created by alkaline or saline lakes. Evaporitic deposits of these lakes may include the varieties of sodium carbonate together with other carbonates, sulphates, and chlorides, including those of calcium. Mineralogical analysis of material from Wadi Natrun by Brill (1988, table 9-9) includes small amounts of chlorides of sodium (1.3%) and potassium (0.4%), calcium sulphate (1.5%), and magnesium carbonate (0.6%); while those by Lucas (Lucas and Harris 1962:493-94) include sodium chloride (2.2 to 26.8%) and sodium sulphate (trace to 39.3%). Major mineral sources of sodium carbonates are to be found principally in Egypt, with only minor efflorescences in Iran. In Egypt the chief sources are the Wadi Natrun, Barnugi in Beheira province in Lower Egypt, and al-Kab in Upper Egypt (Lucas and Harris 1962:263). Wadi Natrun is forty miles northwest of Cairo and comprises a depression twenty- one miles long and over twenty metres below sea level, and containing several lakes. Thirty miles due north of Wadi al- Natrun at Barnugi in al-Buhayra province, fourteen miles west of Kawm Ju‘ayf (Naucratis), is a smaller depression with saline lakes. Al-Kab in Upper Egypt has deposits described as being in fields, and presumably do not include lakes. The fourteenth-century geographer al-Kalkashandi describes two further deposits, one at Tarabiya near Behnesah in Upper Egypt, which he says had been worked since the time of ibn Tulun (AD 835-884), and the second in the Fakus district of the Eastern Delta. Neither of these places is presently known as a source of sodium carbonate. Ancient Egyptian texts refer only to Wadi al-Natrun and al-Kab, although classical texts refer to the Barnugi deposits (Lucas and Harris 1962:265-66). Various authors have used relative abundance of elements to propose either a mineral or a vegetable source. Sayre and Smith (1961) explained that the low levels of magnesia frequently found in soda-lime glasses result from the use of a mineral alkali, while high levels of magnesia result from the use of wood or plant ash. Pollard suggested that a group of glazes of high magnesia content, particularly in regard to lime, are also suggestive of particular materials, possibly plants (Pollard and Hojlund 1983). Henderson (1988) suggested that high levels of phosphorus are indicative of some plant ashes (none of the glazes discussed here had significant contents of phosphorous). Although some authors have attempted to attribute specific chemistries to particular plants, plant ashes appear to vary considerably within a species depending on where the plant grew (Brill 1970, table 2), and even vary for different parts of the same plant (Bezborodov 1975, table 5). A ternary plot of soda, potash, and lime, and another of soda, potash, and magnesia (Fig. 8.2) show the relative distribution of pre-Islamic Iraqi glazes, Early Islamic Iraqi glazes, and Chinese glazes, together with plots of the available plant ashes (see Table 8.1; data from Brill 1970, table 2; Bezborodov 1975, table 5). The pre-Islamic data include results from comparable analyses published by McCarthy et al (1995); other available data (Hedges and Moorey 1975; Hedges 1976; Pollard and Moorey 1982; Pollard and Hojlund 1983) obtained using other techniques were rejected as the ternary plot heightened the discrepancy due to loss of flux elements in analysis. The Chinese data comprise analyses of glazes of Tang to Yuan date published by Guo (1987). Here may be noted a close relationship between the pre- Islamic Iraqi glazes and the early Islamic Iraqi glazes, which may support the previously suggested hypothesis that the early Islamic industry continued the traditions of the pre-Islamic (Mason and Tite 1997). Similarly, a marked contrast may be noted between the Iraqi glazes and those of China, indicating a radically different glazing tradition. The Iraqi glazes plot quite closely among the plots of plant ash analyses, although the one sample that might have correlated closely, a sample of chinan from the suq at Baghdad (Table 8.1, no. 10; Brill 1970, sample 1326) is too sodic. The plot of Egyptian wares made between c. 975 and 1175 (Fig. 8.2) shows a division into four main groups, which are also divided by glaze chemistry and body type. The types technically most akin to some of the early Islamic Opaque- glazed wares of Iraq are the tin-glazed wares with clay bodies, a relationship underlined by a plot of tin oxide v. lead oxide (see Fig. 8.1). The ternary plots (Fig. 8.2) show a radical difference, with higher potash content and significantly lower magnesia. Glazes of Egyptian tin-glazed wares with stonepaste bodies differ from glazes of contemporary clay-bodied tin- glazed wares rather significantly in that they contain lower proportions of potash, and higher soda and magnesia. Presumably this divergence is due to some strategy aimed at counteracting any dichotomy of firing behaviour between the clay and stonepaste bodies, particularly as the lead content is comparable to the clay-bodied tin-glazed wares. Egyptian lead-alkali glazes are also found on both clay- bodied wares (with 10-15% lead-oxide against 8-9% soda and 3-4% potash) and stonepaste-bodied wares (with about 30% lead-oxide against 6-10% soda and about 1% potash). Once more there is a distinction between the glazes on the two bodies, with that on the stonepaste wares containing proportionally more soda, less lime, and less magnesia. The very low magnesia content of tin-opacified glazes on clay-bodied wares makes this group very distinct and isolates them from all the available plant ash analyses. This may, then, be suggestive of mineral sources as proposed for Egyptian glasses by Sayre and Smith (1961), which would be appropriate given the availability of sodium carbonates in Egypt. However, the analyses of sodium carbonate deposits do not provide a direct correlation for the presence of potash, as this is generally lacking from these analyses. This may be explained by use of a mineral potassium source, such as nitre or alum, which are both available in Egypt; by the presence of feldspars in the sand used to provide the silica (as noted in thin-section for the stonepaste bodies—see petrofabric descriptions in Appendix B); or by potassium from the body, which might explain its proportionally greater amounts in the clay bodies as opposed to the stonepaste bodies. The rest of the Egyptian wares appear to nestle comfortably amongst the plant ash data, but it is perhaps unlikely that the potters would have used one raw material for one type of pottery, and another raw material for an essentially identical type. It is perhaps more likely that the potters manipulated the raw materials to suit the required properties of the glaze. This is perhaps a warning from interpreting the source too liberally. Ternary plots of the alkali elements of the glazes of pottery made during the high point of production of wares with stonepaste bodies and lustre paints (Fig. 8.2 — note change of scale from previous figures) appear to show some continuation from earlier practice and some confinement of variability, presumably to fit the stonepaste body. The entire range of stonepaste wares fits within that defined by the two Egyptian stonepaste-bodied wares. The earliest wares in this plot are the Syrian soda-lead glazes of Syrian Stonepaste-bodied Group One including the "Tell Minis" Lustre-ware style. These plot between and slightly overlap the two groups of stonepaste-bodied Egyptian wares. By the period of production of Syrian Group Three (c. 1125-1150) the Syrian potters had switched to an alkali glaze, high in soda and lime. Early Syrian alkali glazes were not exactly like the alkali glazes of early-Islamic Iraq, which contained roughly equal parts of soda, potash, and lime. However,they are closer to pre-Islamic alkali glazes and so may be ultimately derived from them. Alkali glazes were certainly not used by Islamic potters in Egypt, and so presumably the potters had acquired the technology in Syria. The Syrian alkali glazes appear to show a consistent range of variation, although they represent pottery from a number of centres. The range includes most of the area covered by the "Tell Minis" wares and extends to slightly overlap that covered by contemporary Iranian wares. Importation of mineral alkalis from Egypt would have been possible but it is perhaps most likely that plant ash sources were used, although none of the Syrian ash data plot anywhere near the glaze data. The Iranian potters may have spent some time in Syria on their way from Egypt to Iran, and could also have picked up the alkali glaze technology there. The Iranian potters probably also used plant ash judging by Abu ‘l-Qasim and Wulff, and the Iranian data do plot nicely around one plant ash point in both diagrams, although this ash was from Iraq (no. 12). Earlier in this section I suggested that it might be fruitful to undertake a major analytical study of plant ashes, including a comprehensive collection, identification, and chemical analysis of plants from numerous sites and conditions. The results of the considerations made here may be taken to suggest that this may not be so fruitful. Although further work, including trace elements and possibly multivariate statistical treatment of the data, may enable some correlation between raw material sources and sites of ceramic production, the data presented here would seem to indicate that this would probably not be productive. Rather, it may be argued that the potters had complete control over the chemical proportions of the glazes on their pottery, through choice of different plants, or of plants from different conditions, or from different parts of the same plant, or from processing of the ash. Hence, it appears that the proportions of the various elements actually represent deliberate intentions of the potters to cope with technical problems, within the context of their continuing tradition. High-lead glazes The other glaze technology inherited by Islamic potters — the Romano-Byzantine high-lead variety—also had a long history under Islam. Although commonly associated with the Mediterranean, high-lead glazes were known in pre-Islamic times in Soghdian central Asia and T'ang China. Knowledge of the technology may have formed a continuum along the Silk Road. Hence it is not surprising that it quickly became the most widespread glaze technology in the Islamic world. High- lead glazes have been analysed in this study from centres as disparate as Fustat and Samarqand, and from all the lands between. Judging by the numerous and geographically widespread centres of lead-glazed ware production it seems likely that lead-glazing was in some way easier or cheaper to produce than alkali, lead-alkali, or tin-opacified lead-alkali glazes. It is a common assumption that high-lead glazes require lower temperatures for maturation than do other glazes. Based on experimental data, Bansal and Doremus (1986;33, 241) give values of 856 C for a high-lead glaze, 980 C for a lead-alkali glaze, and 990 C for an alkali glaze (see Table 8.2). Calculations based on the required degree of plasticity of the glaze (Tite et al. 1998) give 820-1080 C for a high-lead glaze, 900-1000 C for a lead-alkali glaze, and 950-990 C for an alkali glaze. Rye and Evans (1976) determined the firing temperature of a high-lead glaze at 950-1050 C in an ethnographic study of potters from Pakistan. These differences do not seem particulaly significant in regard to either simplicity or economy of production, although there may be other factors at play in the different firing procedures. A lower temperature for high-lead glaze has possibly been assumed on the basis of its apparent greater fluidity during the firing process, but this is because lead glazes have a significantly lower surface tension when molten than do alkali glazes (Bansal and Doremus 1986:103, 106). There are, however, other distinctions between the glaze types. Lead may be applied to the vessel in a number of ways (Tite et al. 1998), but for the wares we have covered here the evenness of the coat and its distinct boundaries indicate that it is probably added as a silicate-lead oxide mixture, possibly fritted, suspended in water. Alkali elements must always be fritted as they are soluble in the water used to apply the mixture. If applied to an unfired body the significantly lower surface tension of high-lead glazes would be more likely to allow gases to escape, and it is possible that alkali or lead- alkali glazes had to be biscuit fired. Lead glazes have lower thermal expansion coefficients than do the other types (Bansal and Doremus 1986:35, 131, 135), so it is easier to fit them to a clay body, while crazing is reduced because of their lower modulus of plasticity (i.e., they are more stretchable: Bansal and Doremus 1986:312, 315). Further, the above account of alkali sources makes it clear that alkalis require considerable manipulation to achieve the desired properties, while lead is always the same. With the introduction of stonepaste bodies after 1025, and their dominance after 1075-1100, high-lead-glazed wares begin to decline in importance, even for provincial and second-quality wares. Stonepaste probably had a firing temperature of 1000 C, which is above the optimum for the high-lead glazes. The lower surface tension of high-lead glazes would probably be a problem over the highly porous stonepaste bodies. Further, stonepaste wares presumably also had a high degree of thermal expansion, which would also make high-lead glazes unsuitable. Lead glazes would also be unsuitable for the Underglaze-painted wares, which were soon to become important, and which would dominate production after the end of the period covered in this study. However, Lead-glazedwares would continue to be important among the Crusader states, and in Byzantium. In Islamic lands they would see a revival under the Egyptian Mamluks after 1250, perhaps as a response to influence from the Christian areas. But on the whole, lead-glazed types did not continue long in competition with stonepaste-bodied wares. Decorative techniques Decoration on ceramic vessels may be executed by the creation of relief through moulding, carving, or incision, or by the application of some form of pigment. Moulding and other forms of relief decoration had a long pre-Islamic history, and although extensively used in the Islamic world, few new contributions were made in their execution. One exception to this concerns the development of slip-incised wares, or sgraffito. These are thought to have been made in imitation of imports of Chinese sancai wares, although at first they simply represented a clay-bodied ware, perhaps with a pale-coloured slip, and a pigment-splashed high-lead glaze. Sometime in the ninth century a potter began to make incisions through the pale-coloured slip into the red body, thereby creating decoration. This element of the decoration grew until it eventually overwhelmed the pigment- splashed element. A variant technique, often called champléve, involved excising or removing whole areas of slip. The slip-incised technique was the most extensively applied way of decorating pottery, and was the method used in the numerous smaller production centres across the Islamic world. The technique was also introduced to the Christian world, and became equally important in areas such as Italy and the Byzantine realm. Painting of vessels with highly fluid clay slips had been known in the Near East since the Chalcolithic. Such slips do not, however, seem to have been applied to pre-Islamic glazed wares. The presence of slip-paints on Islamic glazed wares date from the very earliest Islamic periods. The Semi-glazed wares of eighth-century Egypt are a continuation of practice represented by unglazed pre-Islamic Coptic Slip-painted wares (see Chapter 4). Similarly, in Iran and Syria slip-paints under lead glazes form a widespread early technology, although they are noticeably absent from Iraq and the Gulf coast of Iran. Slip-painted wares from the Iranian region are among the art- historically most significant wares. These often include slips of different colours, rendered in thick applications. Few pieces were available for analysis to investigate the cause of the different colours, and this would be a fruitful avenue of future research. In early Iraq the main types of paint were applied over the tin-opacified glazes of the region. The earliest of these was the cobalt-based blue pigment. This would appear to have been applied as a pigment-glaze mixture over the glaze, probably before firing (see Chapter 3). This technique appears to have been introduced c. 700 and has little obvious technological ancestry. However, certain pre-Islamic white opaque-glazed wares have areas of turquoise decoration, usually just bands on the rim. These may be the technological progenitors of the Blue-painted wares, and are worthy of further analysis. This general technique continued as a major type until c. 900, after which it is known from only a few pieces. However, the technology itself was also applied in Egypt for Opaque Polychrome-painted wares in the eleventh century, but not using cobalt blue. Although this may have been the earliest application of cobalt as a ceramic pigment, there is no continuity in the ceramics of the Islamic world with the later twelfth century blue Underglaze-painted wares. Another overglaze technique introduced in early Islamic Iraq was lustre-paint. This may have occurred occasionally on pottery before 800, but after this date it was used extensively in a series of major groups. Lustre-wares occupy the most prestigious niche in the fine ceramics market for the next five hundred years. No technological progenitor for this technology may be found in ceramics, and it may represent a direct transfer from the glass industry. The exact nature of the pigment and the methods of its application appear to be largely understood from the work of craft potters, and so were not considered a worthwhile line of analytical research in the current study. However, various other assumptions regarding the technology of Islamic ceramics have been disproven or questioned in this study, indicating that further analysis of the lustre-paints may be worthwhile. Such analysis would require application of techniques such as transmission electron microscopy to overcome the problems caused by the very thin pigment layer. A further overglaze-paint technique involves the application of low-temperature pigments after the main firing of the vessel. This technique was used in Iran from the latter part of the twelfth century, where it was used to make the Minai and Lajvardina types. As with lustre-paint there appear to be no obvious technological progenitors among ceramics, and it is possible that the origins of the technology must be sought in the glass industry. If true, this would be odd because the enamelled glass technology was in Syria, not in Iran at all. No attempt was made at analysis of these pigments, largely because of the difficulty of locating samples which would be available for analysis (although examples were available for sampling bodies and glazes). Such analysis would, however, be a useful avenue for future study. The original development of true underglaze pigment paints, comprising grains of oxide pigment applied to a vessel without a clay or other mounting medium, is another contribution of the Islamic potters. The earliest full use of this technology appears to be in Syria, in the Group Four wares of Damascus, dated to c. 1125-1150. Pigments used include chromium for black, cobalt for blue, copper for turquoise, and iron oxide for red, practically representing the full range of pigments available. The effect of all these colorants was known to potters and glassmakers prior to this date, but it remains to be seen how they developed the technology. It is possible that it developed from the Polychrome-relief (Laqabi) ware, made in Syria during Group One, c. 1075- 1100, and possibly slightly later. The carved relief decoration of these wares had a long history in Egypt with the incised- ware group which includes carved versions. Those wares with uncoloured glazes tended to be splashed with blue. If this blue was applied prior to the glaze, this may indeed be the ultimate origin. Whether Polychrome-relief was made in Egypt remains debatable—certainly some wares found in Egypt with an appearance that an art-historian would consider "reminiscent" of Egyptian Lustre-wares are known (Mathaf 1922, pl. 84). In Polychrome-relief wares the relief had the effect of separating the pigments. Unfortunately, no Polychrome-relief wares were available for analysis of their pigments (although bodies and glazes were sampled), and it remains to be seen in this case whether the pigment was applied under or over the glaze. If under, the underglaze-paint technology would simply be a matter of recognizing that the pigments formed effective designs without separation by the relief, once the potters had switched to an alkali glaze. This is because the alkali glaze would be less likely to make the pigments run than the soda- lead glazes of the Polychrome-relief wares. The technology of true underglaze-pigment paints would eventually be transferred to Iran. Here the potters had previously experimented with pigments that had a similar decorative effect to the true underglaze pigments, but in these cases the pigment was applied in a medium of stonepaste. Rather than being a major innovation, this was effectively a continuation of the highly developed coloured slip-paint technology of Iran. Underglaze-painted wares would eventually become more desirable than Lustre-wares in the market, although not until after the period covered in this study. Current evidence would suggest that this technology would also be transferred to China, where it formed the basis of the "blue and white" ceramic industry. Technological innovation Arnold (1985:202-28) relates technological change to a progressive movement towards greater production due to increased efficiency. This does not appear to have been a factor with the highest status Islamic wares. Innovations such as stonepaste, and decorative techniques such as lustre-paint that require secondary firings, clearly make the potter less productive. The potters who made the higher-status wares studied here were clearly well disposed towards technological innovation. Potters are often considered to be highly conservative in their approach to technology and other aspects of manufacture, including decoration. This is justifiably based on a number of ethnographic studies illustrating the point. In studies of other fields, there seems to be a strong case for varying levels of innovation at different levels of craftsmanship (Homans 1961, Silver 1981). This may be thought of as a threefold division. Craftsmen at the bottom of the hierarchy, who have just entered the craft and have no reputation, are inclined towards innovation through desperation. The overwhelming majority of established artisans, who have regular markets and steady trade, do not innovate. The very highest calibre of craftsman does innovate, in order to maintain the prestige accorded his rank. Many of the potters that made the Lustre-painted, Mina'i, Underglaze-painted, and other wares studied here clearly fit into this third category. In each period the potters with the highest status would innovate to attract the custom of the most status-conscious buyers. PETROGRAPHY The application of petrographic analysis to the provenance problems of Islamic pottery has been exceptionally successful. In no small part this relies on the fact that much of the Islamic pottery studied was made in a few production centres with well-defined and easily distinguished characteristics. The methodology developed for characterization of stonepaste ceramics has been particularly valuable in the separation of these wares. This is especially useful as for the most part the characteristics of these wares that could be defined by chemical analysis would most probably have been obscured by technological considerations. This methodology should also be applicable to other quartz-based ceramic types, such as ancient Near Eastern "faience" bodies and perhaps European soft-paste porcelain. The finding that the highest-status wares were made in a very few centres at any one time is one of the most significant general findings of this part of the study. This begins with the attribution of all the fine wares associated with early `Abbasid Iraq to Basra, including all Lustre-painted and Blue-painted wares. Strong evidence for Basra may have been suggested by the documentary evidence and lustre-painted glass inscribed with production at Basra. However, the only documentary reference to lustre-pottery production is that to a Lustre-ware potter from Baghdad making the tiles at Qairouan (see Chapter 3). Without petrographic analysis, it would always be possible to suggest that one or several other sites could have been making these wares. Production of Fatimid Lustre-painted wares has traditionally been attributed to Fustat, and no new information has been supplied here. However, petrographic analysis has revealed chronologically specific variations in the use of raw materials. These include the use of the modified Nile alluvium, proto-stonepaste, and stonepaste in the first century of production (c. 975-1075), and the use of a highly calcareous red-bodied clay in the last century of production (c. 1075-1175). In Iran a similar situation existed to that in Iraq, except that there was an abundance of evidence indicating a single centre—Kashan. However, the allocation of all Lustre-painted wares and a number of other types to production at this centre has remained questionable to a number of people. Effectively it seems to depend on whether one is sympathetic to the concept of a single centre, or instead prefers a number of centres. Without petrographic analysis it would never have been possible to resolve this question. Now it is clear that Kashan was indeed the overwhelmingly dominant production centre in Iran during the twelfth and thirteenth centuries. Lustre-painted wares appear to be entirely assigned to this centre, as do the Overglaze-painted Minai wares. Perhaps surprising is Kashan's dominance in coarser stonepaste wares. This pottery, predominantly of Monochrome Incised styles, dominates the archaeological assemblages of sites such as Gurgan and Ghubeyra, and indicates the significance of the industry at Kashan. Several other centres appear to have also begun production in Iran, including Rayy and Gurgan, as well as a number of unrecognized centres. These produced a number of types that are also commonly attributed to Kashan. At the moment it is unclear in some cases if there are any obvious stylistic differences between wares of the same type made in different centres, although in other cases it is clear that there are. Further research on these wares could be fruitful, combining further petrographic and typological study. In Syria there has traditionally been a tendency to attribute wares to Raqqa, the only serious exceptions being the western Syrian "Tell Minis" wares and the largely speculative attribution of polychrome Underglaze-painted wares to Rusafa. Lately, even these attributions have been questioned, leaving Syria apparently with no obvious contenders as centres of production. Petrographic study has made a serious beginning in the attribution of wares to specific production centres. Little contribution has been made to designating a production centre for the Syrian Group One "Tell Minis" type, although one of this group has been included in the waster group attributed to the Tell Minis region ("Ma‘arrat" Petrofabric). This may suggest that the rest of these wares may also be made somewhere in the area. For the bulk of the later wares, traditionally attributed to Raqqa, we find a number of centres. Among the most significant findings from the petrographic study of this material is the demonstrated importance of Damascus in this period. A significant proportion of all the sampled Underglaze-painted wares were attributed to Damascus, including the overwhelming majority of polychrome Underglaze-painted wares. Wares with dark blue glazes and lustre-painted decoration, frequently put with technologically similar but later Syrian wares and assigned by the typological study to the twelfth century, may also be attributed solely to this site. Raqqa continues to be an important production centre, particularly for Lustre-painted wares without coloured glazes. However, some Lustre-wares of "Raqqa" type may be attributed to another petrofabric, and probably, hence, to another, presently unknown centre. Contrasting with the few centres defined for these fine wares, the wares with clay bodies and high-lead glazes represent the products of a bewilderingly large number of centres. These wares, including slip-painted and slip-incised types, seem to represent the products of smaller local kilns, with limited outputs and distributions. Although some centres appear to have stylistically highly distinctive products, in other cases it seems presently impossible to distinguish the different centres on typological criteria. This would be a very fruitful field for future study, again incorporating extensive typological and petrographic analysis. Given the number of centres involved, such a project would be very large. The results of the petrographic study raise the question as to the mechanisms by which a centre of ceramic manufacture commences and then ceases production. Two main models may be posited. One is that masterpotters wandered about the Islamic world taking their skills and techniques with them. The second is that local potters responded to the stimulus of a local court by producing fine pottery. There is documentary evidence for at least one case of the former model, in the forced removal of Damascus potters to Samarqand by Timur in 1402 (see below). In the period covered in this study we also have the reference to the potters of Basra and Kufa who were brought to Samarra. Traditionally it has been hypothesized that Lustre-ware potters moved from Iraq to Egypt, and from there to Syria, Iran, and Europe. The various studies in typology, technology, and provenance made in this study would strongly support this hypothesis. The previously existing evidence included the nature of the lustre-pigment, which would probably have been a secret technique, not easily redeveloped by others. We now also have evidence of the continuing development of stonepaste, which may also be considered a technological secret. Glaze technology also forms a link from centre to centre. The typological study shows the precise timing of these movements, and also the continuing tradition of the forming methods and decorative styles. It would seem that the collective evidence for the movement of potters—highly skilled artisans with an extensive range of technological knowledge—is pretty formidable. This does not rule out local development, which might be suggested in some cases, for instance for Iranian slip- painted wares. However, in the case of the high-technology wares, it seems that a continuing, possibly even genealogical succession is implicit. ISLAMIC POTTERY IN CONTEXT This study has produced results of significance to a number of general theoretical questions regarding pottery generally, and Islamic pottery in particular. One observation may be made in regard to the often assumed correspondence between major dynasties and significant ceramic groups. This had been suggested from the early days of study of Islamic ceramics, with designations such as "Abbasid" or "Fatimid" Lustre-wares. This linkage may appear questionable to some, as it once did to the author. However, the new data regarding provenance combined with the new typological dating supplied by this study would strongly reinforce the concept that these wares may be linked in time and space to significant dynasties. There may be nothing in this, as economically powerful regions with a good market for pottery may also be good bases for powerful rulers; while those authorities which wisely invest in infrastructure, notably agricultural works and roads, may encourage the prosperity that creates markets for fine pottery. However, there may be a case for a more direct link between ruler and potter (see Chapter 7), such as the Il-Khan Ghazan's patronage of Abu ‘l-Qasim, the reference to potters from Basra and Kufa who were brought to Samarra, and the foundation of pottery manufactories at Samarqand by Timur. At other times it was the ruler's implicit role to ensure peace and prosperity for the population, and to provide the conditions under which they might safely enjoy their fine ceramics. When they failed in this, there seems to be evidence of it in the archaeological record of ceramics (see Chapter 7). Return to Table of Contents