Arab. arch. epig. 2008: 19: 71–108 (2008) Printed in Singapore. All rights reserved Christianity in the Gulf during the first centuries of Islam This article re-examines the ceramics of SBY-9, a church and monastery site on the island of Sir Bani Yas, Abu Dhabi emirate, U.A.E. It then explores other archaeological evidence for Christianity in the Arabian Gulf and compares it to the textual data, resulting in a reconsideration of the history and activities of the Church of the East. The pottery of SBY-9 indicates that the monastery complex was occupied some time between the second half of the seventh century and the mid-eighth century AD and not, as previously believed, the sixth–seventh centuries AD. Other excavated churches and monasteries in the Gulf region should also be redated to the eighth and ninth centuries AD (al-Qusur, Kharg). These findings cast a new light on the history of Christianity in the Arabian Gulf after the Muslim conquest. Keywords: Christianity, Nestorians, Arabia, Sir Bani Yas, pottery Introduction The monastery and associated buildings at SBY-9, on the island of Sir Bani Yas, Abu Dhabi, U.A.E. (Fig. 1) are crucial to our understanding of the history of Christianity in eastern Arabia and its survival during the first centuries of Islam. According to the texts, Christianity flourished in the Gulf from at least the fourth century AD until the early centuries of the Islamic era. The Syriac sources of the Church of the East (also known as the Nestorian Church) state that north-eastern Arabia, the Bahrain1 archipelago and the Qatar peninsula were part of a region known as Bet Qatraye, which contained monasteries at least from the mid-fourth century, and several named episcopal seats from at least the early fifth (Potts 1990: 241–247; Bin Seray 1996). The dioceses of Bet Qatraye did not form an ecclesiastical province, except perhaps for a short period during the mid- to late seventh century, but were subject to a Metropolitan at Rev Ardashir (Rishahr, on the Bushehr peninsula), 1 Note that the term ‘al-Bahrain’ is used in historical sources to refer to north-east Arabia in general as well as the island and archipelago that today are denoted by the term. R.A. Carter Dept. of Archaeology, Durham University, South Road, Durham DH1 3LE, UK e-mail: racbahr@hotmail.com together with the single bishop of Bet Mazunaye (Oman). The Christian community was tolerated and remained active in Bet Qatraye after the conversion of the region’s leaders to Islam in around 629, though it appears that the Christians of Bet Mazunaye soon switched faiths. The bishops of Bet Qatraye ceased to attend synods after AD 676, but textual indications show that Christianity in the region persisted until at least the late ninth century. A full exploration of the textual evidence is given below. Against this historical backdrop can be set the archaeological discoveries of churches and monasteries at Kharg (Steve 2003), al-Qusur (Bernard, Callot & Salles 1991; Bernard & Salles 1991), Akkaz (Gachet 1998; Calvet 1998), Jubayl and Thaj (Langfeldt 1994) (Fig. 1). A further church was reported from the island of Marawah (Abu Dhabi) (Elders 2001) but upon further excavation it became clear that this was in fact a Neolithic site (Beech et al. 2005). The location of these sites does not match well with the supposed locations of Christian centres mentioned in the historical texts (cf. Figs 1 & 2), a discrepancy which is further discussed in the conclusions. 71 R. A. CARTER Fig. 1. Map of the Gulf showing the locations of Sir Bani Yas and other archaeological sites with Christian remains. Fig. 2. Map of the Gulf showing supposed locations of toponyms mentioned in the historical sources. Perhaps because of the unarguable textual evidence for monasteries and bishops in Bet Qatraye between the fourth century and AD 676, the excavators have universally dated the foundation of these sites to within this time bracket. This paper not only redates the Sir Bani Yas complex to the late seventh–mid-eighth century AD but also reconsiders 72 the dating of the church at al-Qusur, placing it within the same time bracket and slightly later, and repositions the monastic complex at Kharg to an even later time period (late eighth ⁄ ninth century AD). These adjustments demand a complete reconsideration of the activities of the Church of the East in the Gulf after the advent of Islam. CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Site location and description SBY-9 was discovered in 1992 by the Abu Dhabi Islands Archaeological Survey (ADIAS), and brief descriptions were published in preliminary reports (King et al. 1995: 69–72; King 1998). Excavation commenced at SBY-9 in 1993 and continued there and at the surrounding sites until 1996. There was one season of survey and surface collection (1992) followed by four seasons of excavation (1993–1996). A preliminary report was published in 1997 (King 1997a), giving a brief account of excavations up to 1995 at SBY-9, SBY-3, SBY-7 and SBY-2. In 2001 Elders published an account of the monastery and church at SBY-9, setting it within the context of other archaeological finds in the Gulf (Elders 2001), with a similar publication following (2003). Finally, a brief account of the faunal remains from SBY-9, SBY-4, SBY-2 and SBY-7 is given by Beech (2004: 109–121). The dating of the church and associated complex at SBY was thought to be between the sixth and seventh centuries AD (King 1997a: 231; Elders 2001: 52; 2003: 231). SBY-9: architecture and stratigraphy The monastery (SBY-9) was found among a cluster of sites stretching across an area of the coastal plain measuring c. 1.5 km east–west and 2 km north–south (Fig. 3) (King 1997a: 221). The other sites include six courtyard houses (SBY-2, SBY-3, SBY-4, SBY-5, SBY-6, SBY-7) and other mounds, structures and pottery scatters (SBY-8, SBY-15, SBY-16, SBY-31) (King 1998: 17–43). These sites are located opposite a creek or embayment on the eastern side of the island, Khawr al-Janubi, which forms a shallow natural harbour. SBY-9 was described as a ‘low flattened occupation mound approximately 220 m · 160 m. reaching 1.5m. – 2 m. in height’ (King 1998: 25). Pottery and moulded plaster was found on the surface. It was initially divided into three sub-sites, SBY-9.1 to SBY-9.3, according to surface indications. These distinctions were not evident in the pottery sample. Excavations took place between 1993 and 1996 and it rapidly emerged that the SBY-9 complex consisted of a Christian site, with stucco cross fragments being found in 1994, and a church 2 Note that King gives the original dimensions of the church as 14 m E–W and 4.5 m N–S. This refers only to the first stage of building, consisting of the nave and apse. Fig. 3. Map of Sir Bani Yas showing the location of SBY-9 and related sites. excavated in 1995, measuring 16 m east–west and 11 m north–south (King 1997a: 226–227; Elders 2003: fig. 5)2. The church was tripartite, with a classic basilica plan consisting of a central nave and a square apse or chancel (Fig. 4). The nave was flanked by northern and southern aisles, and the chancel by two square rooms3, with a narthex (entrance lobby) at one end and perhaps a belltower at the other (Elders 2001: 49–51 and fig. 2; 2003: 231)4. It was built in two stages, or three if one counts the preparation and levelling stage. It is not entirely clear whether the two stages represent two phases of a single but punctuated construction 3 4 These were perhaps the prothesis and diaconicon, and were sometimes referred to as transepts in the excavation records, though it would be overly interpretative to assign any of these terms with certainty. As Elders points out, wooden clappers are more likely to have been used in the area rather than bells, as attested to by the poet Labid, a contemporary of the Prophet Mohammed, when he descended from the inland Yamama region towards the lowlands of Hajar (the al-Hasa oasis) (Langfeldt 1994: 55). 73 R. A. CARTER Fig. 4. Plan of church at SBY-9 (after Elders 2003: fig. 5). process, or whether the second stage represents an enlargement following a period of use, perhaps in order to accommodate a growing congregation (King 1997a: 227). This work and subsequent excavation, test trenching and geophysical survey showed that the church was in a large walled courtyard measuring c. 90 · 70 m (Elders 2001: 48) (Fig. 5). Buildings surrounded the inside of the courtyard, consisting of plastered rooms and generally devoid of occupational debris except for the case of some open yards. These may have been monks’ living quarters. A Fig. 5. Simplified plan of the monastery at SBY-9, combining the results of excavation and survey (after Elders 2003: fig. 2). Numbers refer to ‘Room’ numbers mentioned in the pottery analysis. 74 2.4 m-wide entrance was found on the eastern side of the perimeter wall, with an adjacent burial, presumed to be that of the founder, an important abbot or a saint (2001: 52). The stratigraphy of SBY-9 was divided into four phases, the first being the levelling and preparation phase for the building of the church (Phase 1); the second being the half-built church with unfinished northern aisle and apse (Phase 2a); the third being the completed church, and presumably any related deposits (Phase 2b); and the fourth (Phase 3) being a squatter occupation in and above the collapsing church and monastic buildings (Elders 2001: 49–52). It was noted that finds were sparse for the main occupation, with the church, the outbuildings in the enclosure and the external courtyard houses all being kept clean of domestic rubbish. Thus, much of the material included in the pottery sample may relate to an occupation that continued after the ruin of the monastery and church, perhaps associated with the abandonment of Christianity. In most cases it was not possible to relate the recorded pottery to each phase, as the phasing of the contexts was not provided in the records available to the author. The assemblage was notably homogeneous, however, and can be treated as a single unit without losing chronological focus. Architectural parallels Architectural parallels and differences can be found with the other churches and monasteries of the Gulf. The church at al-Qusur was also finely plastered, though built of mud brick, and had a central nave and apse, with northern and southern aisles and chapels, and a narthex (Bernard, Callot & Salles 1991: fig. 1). It was twice the size, however, at 35 · 19 m (1991: 146)5 and had rooms attached to the south-eastern end of the southern aisle, as well as spaces for doors between the nave and aisles, and niches in the chapels at the ends of the aisles. These features are not visible on the Sir Bani Yas plan. The church at Akkaz was very badly preserved, but resembles that of Sir Bani Yas and al-Qusur in plan and proportions. It is too damaged to know its exact plan but it stood around 15 · 12 m (Gachet 1998: 73), thus comparing to the one on Sir Bani Yas in length ⁄ width ratio. It was made of stone and it too 5 This is proportionally slightly longer than the SBY-9 church, the ratio of length to width being 1.84 as opposed to 1.45. CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM was plastered with fine stucco work, including a Nestorian cross (Calvet 1998: fig. 12). The church on Kharg very closely resembled that of al-Qusur, having the same conventional plan of a central nave and apse, with aisles on each side joined by doorways and a narthex, and being of similar dimensions at 31 · 15 m (Steve 2003: plan 12). A large monastic complex was attached to the south-east corner of the church (cf. the arrangements on the same corner at al-Qusur), including the library, refectory, sacristies and other rooms. Like Sir Bani Yas there was a large enclosure around the church, of similar dimensions and lined with cells (2003: plan 11), though the cells appear to have been more regular in construction than those of SBY-9. It is not yet known whether the church at al-Qusur had a comparable enclosure with monks’ living quarters, as work there was stopped by the first Gulf War. The church at Jubayl had a different design, apparently having a single square room or open courtyard, rather than a long nave with aisles, leading into three rooms at the eastern end, i.e. an apse or chancel with apses or rooms either side (Langfeldt 1994: fig. 2). The significance of this apparently completely different design remains uncertain: the square room could conceivably conceal a long apse flanked by aisles, perhaps separated by walls or columns which were not visible when the sketch plan was made. The stucco most closely resembles that of Kharg, hinting at a slightly later date, but also that of al-Qusur and Sir Bani Yas. The plan of the church at Thaj remains unknown. Similar churches to those at Sir Bani Yas, al-Qusur and Kharg, with a rectangular chancel flanked by two rooms or chapels (or prothesis and diaconicon), a nave divided from two aisles with walls or columns and a narrow narthex, can be seen at Hira, Ain Sha’ia and Rahaliya in Iraq (Talbot Rice 1932: figs 1–2; Okada & Numoto 1989: fig. 4; Finster & Schmidt 1976: 41, Abb. 13). These similarities in the architecture of the churches of the Gulf and the Iraqi Western Desert are summarised by Simpson (in press) including, among other features: 1. square-ended domed chancels 2. separation of the nave from the aisles with dividing walls or columns 3. a shallow narthex-like room at the western end 4. a typically ‘bent-axis’ approach leading into each aisle through as many as five doorways 5. ancillary rooms often attached on the northern and south-east sides The church at Sir Bani Yas shares most of these characteristics, though it cannot be certain whether the chancel was domed or not, and there do not appear to have been rooms attached to the northern or south-eastern sides. Along with al-Qusur and Kharg, it nonetheless appears to represent a fairly standardised plan characteristic of the Church of the East. Also characteristic of these churches is the use of moulded stucco. The stucco work at Sir Bani Yas, which is of relevance to the relative and absolute dating of SBY-9, compares well to that of Kharg and Jubayl, according to Simpson (see below). Elaborate stucco work is also found at contemporary or nearcontemporary churches at al-Qusur, Tulul al-Ukhaidir, Hira and Ain Sha’iah, indicating that this was a key part of the religious architecture of the Church of the East. SBY-4 and others As well as SBY-9, several other sites in the vicinity were excavated. These include SBY-4, of which the pottery was included in the study sample, and SBY-3. These are two of six buildings in the vicinity described as courtyard houses, of which five are said to have been excavated (Elders 2003: 232). SBY-3 is the only one which has been partially published (2003: 232). It consisted of a courtyard measuring c. 20 · 20 m, containing a plastered water reservoir and a small plastered house measuring 7.42 · 7.14 m (King 1997a: 229), with four rooms (Elders 2003: fig. 4). It is thought that these houses may have been living quarters for senior or solitary monks (Elders 2001: 53). The pottery assemblage Overview The pottery was recorded in a relational database (Microsoft Access), and quantified in two ways, using sherd counts and EVEs (Estimated Vessel Equivalents) (Orton, Tyers & Vince 1993: 168–171)6. Qualitative data was also recorded for each entry, 6 In practice, the surviving percentage of a rim sherd (or base) is measured on a diameter chart which has been split into 5 % divisions, and that number can be used as a measure of quantity. 75 R. A. CARTER i.e. method of manufacture, surface treatment (e.g. presence of glaze, slip, paint, incised or appliqué decoration). 1682 sherds were studied from the 1994, 1995, and 1996 seasons. These could be broken down into 1335 from SBY-9, 140 from SBY-4 (1996 season), two from SBY1 ⁄ 2 and 205 from unspecified sites on Sir Bani Yas. The latter were almost all from a Ware Typology devised by Christina Tonghini, and also included twenty-seven Late Islamic sherds. The sample appears to be composite and biased, taken through a combination of random and purposive selection. Sampling processes appear to have included: 1. the selection of what were considered to be the most reliable stratified contexts, taken rapidly from the store on the island of Sir Bani Yas when it became apparent that access to the site and excavated material was to be restricted. This material was brought to the UK and provided the bulk of the studied assemblage, and included body sherds and what appear to be whole excavated contexts. This collection also included bags that only contained diagnostic sherds, however, indicating that selection had already occurred in some cases, as well as Tonghini’s Ware Typology. Some unstratified material was also present. 2. a selection of thirty-four diagnostic sherds from the 1993 season, which were recorded in detail and drawn by Derek Kennet in 2000. These sherds were not seen by the present author, but the material is entirely comparable to the assemblage studied by this author, and Kennet’s notes allowed easy identification according to this study’s typology of pottery classes. These sherds were therefore included in the database as if they had been recorded in person, though it was not possible to assign a stratigraphic context to them7. Additionally, a significant quantity of other pottery from SBY-9 and SBY-1 to SBY-6 became available following the writing of this report. It was too late to 7 The sherds had been marked with identifying numbers, which were recorded by Kennet, but it was not possible to relocate the database that contained the correspondence between sherd numbers and contexts. Special thanks are due to Dr Kennet for allowing his records to be used in this analysis. 76 incorporate this material, but it can be confirmed that the dating of the main assemblage should not be changed, though there is very limited evidence that SBY-1 and -2 are slightly earlier in date (see below). Because of these different selection processes, totals and proportions of the different types of pottery are unreliable when taken from the bulk sample. An attempt has been made to rectify this by making such calculations after the exclusion of material that had obviously been subjected to purposive selection (see below). Pottery classes at SBY-9 Fifteen classes were identified or defined at SBY-9, plus a sixteenth category (‘Undiagnosed’) devoted to undiagnosed or very rare categories that could not be identified with a recognised type. Some of the classes are already well known, others less so. Table 1 gives a summary description of each, and they are discussed individually below, in order of frequency. The typology is based on that developed by Kennet (2004) and refined by Priestman (2005). The exact names of Kennet’s and Priestman’s classes have not always been used, as the SBY types do not always match exactly, but it is mentioned where there is concordance. Regarding frequency, Table 2 and Figure 6 show the breakdown of pottery types using material from whole stratified contexts at SBY-9 only, which are less likely to have been subject to selection bias. Thus, the following have been excluded: surface material (Context 1); the Kennet material; the Tonghini typology; material from unknown sites. The size of this sample is 1194 sherds, which is still a reasonable number. Seven of the wares are relatively common, each accounting for more than 5 % of the assemblage whether quantified by sherd count or EVE. If one takes EVE measurements to be more accurate indicators of the original composition of the assemblage (they have the advantage of not being susceptible to bias through differential breakage rates caused by transformation processes or variable friability), then the commonest type is Buff Ware (26 %), followed by Turquoise Glazed Ware (23 %), together accounting for almost exactly half of the assemblage. The amount of Turquoise Glaze may be overestimated in the EVE count, however: Turquoise Glaze rims may have been selectively picked out Table 1. Pottery types (arranged by name, alphabetically). Common name(s) Surface Treatments Buff grey or reddish ext. slip Crude Grey Brittle Thickness Manufacture Hardness Colours Inclusions ⁄ Voids Other details 0.4–2 cm W medium to hard buff to greyish buff; sometimes green Nearly always shows distinct ribbing int. and ext. from wheel manuf. 0.3 H soft to medium grey, sometimes brownish freq medium rounded quartz sand; mod lime, up to 1.5mm, sometimes exploded mod or freq flat angular incls, 0.5–2 mm; occ or mod lime, 0.5–1.5 mm occ, mod or freq angular grits, 0.5–1 mm; occ, mod or freq lime, sts exploded freq. v. fine grey and white sand; occ. subrounded lime or white particle, up to 2mm mod to freq. angular grits, red or grey, 0.5–1.5 mm; mod. whitish or lime incls 0.5–1.5 mm, sometimes exploded; mod or freq linear and irregular voids freq small voids; freq small lime or rusty incl, mod to freq tiny dark incls mod quartz sand; occ lime or white particle, up to 1mm No Visible Inclusions mod shiny micaceous incls, up to 0.5mm; mod to freq small voids; occ. or mod tiny dark particles; Finer, fewer finer gits than normal Torp, with glittery mica visible occ. burnished? Parallel incised lines. 0.4 W hard grey, somtimes brownish or whitish surfaces Grey Sandy incised wavy lines 1.7 ? medium grey Hard Gritty Incised. Grey slip ext. sometimes up to 2.3cm ? hard grey; red brown; red brown with dark grey surfaces; yellow streaking Honey thumb-prints ext. 1.5 ? medium to hard greenish buff Indian reddish or grey slip. Burnished ext. orange ⁄ red brown slip ext. Burnished ext often bitumen int. 0.5 W soft to medium dark grey 0.4 W medium orange ⁄ red brown 1 W medium red brown, paler ext surface IRPW Micaceous Torp cf. SMAG.B. Hard-fired grey biscuity ware. Some variation in the fabric. Only noted at SBY1 ⁄ 2. Jagged break with abrasive feel. Variable. Includes Kennet’s LISV fabric description matches Kennet’s HONEY Indian cooking pot ware? Very fine, no inclusions or voids 77 CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Grey Brittle Crude hand formation clear on the inside. R. A. CARTER 78 Table 1. Continued. Common name(s) Surface Treatments Pebbly Torp often bitumen int. Red Fine Sandy Thickness Manufacture Hardness Colours Inclusions ⁄ Voids Other details 1.2cm WH medium reddish brown As Torp but very large grits are present as well 0.3–0.6 ? medium red brown reddish brown; sometimes pale brown or cream red brown mod huge subangular translucent and opaque incls 0.4–4 mm, plus normal Torp inclusions v. freq v. fine sand; occ. subangular grit 0.5–1 mm or variable type; occ. lime, 1mm. freq. sand and fine grits, ->1mm Torp often bitumen int. 0.7–1 cm WH medium to hard Thin Torp-like grey or whitish slip ext. 0.5–0.7 W soft to medium Turq greenish turquoise glaze 0.7–2.3 W soft to medium yellowish buff freq tiny black angular particles; freq quartz sand No Visible Inclusions Undiagnosed Veg 1.5–2.1 cm ? soft to medium brown or reddish brown, grey core White Incl 0.6–1.5 W, H medium buff, pinkish or pale brown 0.4–1.2 W soft to medium buff White NVI parallel incised lines on shoulder mod med or coarse veg; occ. quartz sand or larger quartz particle; sts other incls; sts mica mod to freq v. fine mod to freq particles; sts fine irregular voids No Visible Inclusions or v. occ. tiny black particle; sts small irregular voids body sherds could be confused with finer Torp. Associated with Torpedo Jars V. like Torp, but thinner walls, more obvious quartz sand impossible to discern category, or only one occurrence looks Wadi Suq or Iron II. Residual? Variable amounts of v. fine particles. Overlap with fine cream Torp. CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Table 2. Frequency and percentages of pottery types at SBY-9, using only whole contexts and excluding surface material. Buff Turq White NVI Thin Torp-like Grey Brittle Torp Hard Gritty Pebbly Torp Red Fine Sandy Indian White Incl Crude Grey Brittle Veg Undiagnosed Micaceous Torp EVE total % of assemblage by EVE Count % of assemblage by Count 154 136 66 65 53 52 35 17 15 4 0 0 0 0 0 26 23 11 11 9 9 6 3 3 1 0 0 0 0 0 201 75 69 102 85 163 359 1 6 8 69 36 7 7 6 17 6 6 9 7 14 30 0 1 1 6 3 1 1 1 during excavation or processing, and the attempt to remove bias may not have been entirely successful. This is suggested by the low quantity of body sherds (6 %). If Sherd Counts are used then the commonest types are Hard Gritty Ware (30 %), followed by Buff Ware (17 %) and Torpedo Jar Ware (14 %). In the case of the Hard Gritty Ware, and probably also the Torpedo Jar ware, this high representation is due to the presence of only a few large vessels that had been smashed into very many pieces. Note that some rare types were present in this sample only as body sherds, and do not have EVE totals. Buff Ware (‘Buff’) (Figs 7–9) This class makes up the most frequent category at SBY-9 (26 % by EVE, 17 % by count). Forms consist of large straight-sided bowls or vats with a rolled rim (Fig. 7) and a selection of finer bowl types and medium-sized jars with straight vertical necks and rims, and handles at the shoulder (Figs 8–9). It is generally buff, yellowish or greyish and fairly soft, containing a relatively low quantity of rounded quartz sand inclusions, but harder-fired greenish examples are found. Of the sherds of this class 43 % were slipped (28.5 % by EVE) with a dark brownish black slip, which sometimes tends to reddish; it is likely that a much higher proportion is slipped, perhaps all of the jars, but the colouring erodes easily off the soft surface. The slipped examples (Fig. 9) do not show different forms to the unslipped ones. The outside of the jars is ribbed, a perhaps deliberate result of the wheel manufacture, and this often leads to a differential application or survival of the slip resulting in a striped appearance (Fig. 9.6). The slip is only rarely seen on the large straightsided bowls (Fig. 9.1). The fabric of these bowls can be coarser, but it is essentially the same. 400 350 Count EVE 300 250 200 150 100 Micaceous Torp Undiagnosed Veg Crude Grey Brittle White Incl Indian Red Fine Sandy Pebbly Torp Hard Gritty Torp Grey Brittle Thin Torp-like White NVI Turq 0 Buff 50 Fig. 6. Chart of frequency and percentages of pottery types at SBY-9, using only whole contexts and excluding surface material. 79 R. A. CARTER Fig. 8. Buff Ware (‘Buff’): bowls, jars and other sherds. Fig. 7. Buff Ware (‘Buff’): large straight-sided bowls. In colour and inclusions this type bears similarities to the ‘Common Ware’ of Bilad Qadim (BQ), Bahrain, found from the late eighth ⁄ early ninth century onwards8. Formal parallels include the large straight-sided bowl or vat, classified as Type 31 at BQ in Periods 1 and 2 (late eighth and ninth century) (Carter 2005a: fig. 4.11, and p. 440, table 4.4). Type 31 was rare at Bilad Qadim, and more common in Period 1 than Period 2, indicating that it was going out of use around the start of the Samarra Horizon. Other common Buff Ware forms at Sir Bani Yas were 8 Bilad Qadim Common Ware appeared in unglazed form until Period 3 (eleventh c. AD), when a glazed variety appeared, and was often slipped. By Period 5 (twelfth ⁄ thirteenth c. AD) wasters indicate that the glazed variety was being manufactured at the site, and similarities in the fabric with Periods 1–4 imply that the earlier Common Ware was being manufactured at or near Bilad Qadim, using the same clay sources (see Carter 2005a). 80 not found at Bilad Qadim (e.g. the small vertical jar rims), unless one counts small bowls with simple rims (e.g. Fig. 8.1–4), comparable to slipped examples from BQ Period 1 (Carter 2005a: fig. 4.1.7–8). It is considered likely, though unproven, that the Buff Ware at Sir Bani Yas is an earlier manifestation of BQ Common Ware, and thus originated from Bahrain. The straight-sided bowls have further parallels at the al-Qusur church, where two vessels from Period III are published (Bernard & Salles 1991: fig. 7.18– 19). The fabric description of the upper of these is appropriate (‘fine greenish clay, well fired’), but not that of the lower (‘coarse grey ware, with many grits’). A bowl with the same form and a yellow fabric from Hulayla Area D may also be related (Sasaki & Sasaki 1996: fig. 45.95–186). The Buff Ware vertical jar rims, sometimes with handles coming from the rim (e.g. Fig. 8.10) have parallels at al-Qusur (‘medium-fine yellow clay, fine grits’) (Bernard & Salles 1991: fig. 7.24) and Hulayla Area D (‘yellow fabric’) (Sasaki & Sasaki 1996: fig. CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Fig. 10. Turquoise Glaze (‘Turq’): bowls. Fig. 9. Buff Ware (‘Buff’): slipped examples. 45.95–187). Similar vessels are also found at Tulul al-Ukhaidir (Finster & Schmidt 1976: Abb. 42b–h, 43a– f and 44d), where they are described as medium to fine ware, which is greenish, whitish-grey or pale brown in colour. The forms are somewhat generic, however, and it is not certain that these parallels are significant. The Tulul al-Ukhaidir examples may instead relate more closely to the White Wares described below. The parallels obtained from the Turquoise Glaze are probably more useful (see below). Turquoise Glaze (‘Turq’. Kennet code: TURQ; Priestman code: ALK) (Figs 10–11) Because of its abundance, wide distribution and repertoire of chronologically bounded forms, Turquoise Glaze is one of the most useful types when attempting to diagnose the date of the assemblage. It has an alkaline glaze, coloured with copper to make a bluish or greenish colour, and petrographic studies suggest that it was manufactured at more than one locality around Basra in southern Iraq (Priestman 2005: 107–8; Mason & Keall 1991). By EVE measurement it is the second commonest type of ceramic at SBY-9 (23 %). The total by sherd count, however, was much lower, at 6 %. This large discrepancy suggests that Turquoise Glaze rims Fig. 11. Turquoise Glaze (‘Turq’): jars, lids and bases. were selectively targeted during excavation or processing, and the sherd count is more likely to be an accurate indication of its frequency, though also inflated. At Kush it dropped in frequency from the Sasanian to the Early Islamic Period, comprising (by sherd count) 13.4 % of the Period I assemblage (fifth–sixth century), 5.5 % of the Period II assem- 81 R. A. CARTER blage (seventh century) and 3.5 % of the Period 3 assemblage (eighth century)9 (Kennet 2004: 31, table 17). At Bilad Qadim Period I (late eighth century) it comprised 9.3 % by EVE and 5.7 % by count (Carter 2005a: 451, chart 4.20). Forms include medium-sized bowls, particularly a carinated variety (Fig. 10.1–6), and a variety of less common jar forms (Fig. 11.1–5). Also present were bases, which were glazed below as well as above and bore stacking scars on both sides (Fig. 11.12), and several lids (Fig. 11.9–11). The glaze was slightly greenish, suggesting affinity with Kennet’s variety TURQ.4, which occurs at Kush in declining quantities from the Sasanian Period onwards (Kennet 2004: 29–30, table 15). The carinated bowls (Fig. 10.1–6) provide some of the best dating evidence for the site. This form has been studied comprehensively by Derek Kennet, and occurs at Kush as Type 72, exclusively in Kush Period III (Kennet 2004: 30, table 16). It is also found in comparable assemblages at Hulayla Area D and al-Qusur, which Kennet dates to the eighth century AD. According to other authors (Kervran 2004: 296), such carinated bowls have a long lifespan from the second ⁄ third century BC onwards, but Kennet notes that, although present in Hellenistic-Parthian contexts, they are absent from intervening assemblages, such as Mouton’s PIR-C and D, and Kush Periods I and II (Kennet 2007: 97). He also observes that the Hellenistic-Parthian variety is generally more pronounced and vertical than the eighth-century version. As stated by Kennet, parallels for this type are found at Hulayla Area D (Sasaki T 1996: figs 10.94811-94-817 and 11.94-825-94-829), al-Qusur (Bernard & Salles 1991: fig. 5.4–6) and Suhar, Levels II and III (Kervran 2004: figs 10.17–18 and 12.6–8). An example is also found at Tulul al-Ukhaidir (Finster & Schmidt 1976: Abb. 46m), with another possible one at Kharg (Hardy-Guilbert & Rougeulle 2003a: pl. 20.4). Note that Kennet rejects Kervran’s dating of these layers at Suhar as Sasanian, preferring to see them as eighth-century with some residual Parthian material (Kennet 2007: 99). None of the other Turquoise Glaze forms represented at Sir Bani Yas have such a well-defined date 9 The dating of Kush Periods II and III has been altered here from Kennet’s dating, to take account of the revised chronology resulting from this study. 82 range, but parallels can be found for the triangular jar rims (Fig. 11.2, 4) at al-Ukhaidir, Hulayla and Kharg (Finster & Schmidt 1976: Abb. 46b, f, 47e; Sasaki T 1996: fig. 12.94-848; Sasaki & Sasaki 1996: fig. 44.95–64; Hardy-Guilbert & Rougeulle 2003a: pl. 17.1). The carinated jar rim (Fig. 11.3) has plentiful contemporary parallels in the Gulf, for example at Suhar and Hulayla Area D (Kervran 2004: figs 9.4–5 and 12.1–3; Sasaki & Sasaki 1996: fig. 44.95–104 and 95–107), as well as at Tulul al-Ukhaidir in Mesopotamia (Finster & Schmidt 1976: Abb. 46h, l, o). The small jar with a pronounced shoulder (Fig. 11.5) is closely comparable to an example from Tulul al-Ukhaidir (Finster & Schmidt 1976: fig. 47a), while lids comparable to Figure 11.9–11 are found at Hulayla Area D (Sasaki T 1996: fig. 9.94–807). Bases with scars from the firing stack on both sides (Fig. 11.12) are found at Hulayla (Sasaki & Sasaki 1996: fig. 43.95–61, 62). The fact that Turquoise Glaze represents the only glazed material in the assemblage is significant, as so-called Samarra Horizon glazed wares were introduced in the early ninth century and rapidly became common on numerous sites in the region (Northedge & Kennet 1994; Kennet 2004: 31, table 18). The absence of glazed ‘barbotine’ decoration suggests an even earlier terminus ante quem for Sir Bani Yas. This type corresponds to Priestman’s class ALK.3 (2005: 234, 238). It used to be described as ‘SasanianIslamic’ but it now appears that it is post-Sasanian. In its most typical form, that is to say with flattened appliqué handles, pellets and cordons on large rounded glazed jars (Whitehouse 1979: pl. 3a; Rosen-Ayalon 1974: fig. 381), it occurs during the Samarra Horizon but also immediately before it, for example in Period 1 at Bilad Qadim, implying that this variety was in use by the late eighth or early ninth century10. It is common, and its absence from Sir Bani Yas and Hulayla Area D dates these sites to some time before that phase. Certain absences from the Turquoise Glaze assemblage have chronological significance, providing a terminus post quem. These especially include Kush Type 94, a distinctive bowl with a notch on the inside 10 Note that an earlier variety of barbotine decoration is found at Susa A III, on smaller glazed vessels, with greener glaze and less standardised decoration, which does not apparently include the flattened appliqué handles (Rosen-Ayalon 1974: figs 367, 373 and 383). This type does not seem to be as widely or frequently distributed as the later variety. CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM White Wares (Fig. 12) Two varieties of fine cream-coloured pottery were defined, both soft and sometimes having a powdery surface. These were named White Incl (i.e. White Ware with Inclusions, Fig. 12.1–6) and White Ware NVI (i.e. with No Visible Inclusions, Fig. 12.7–15). The fabric of the former contained low to moderate amounts of very fine sand or grit, and at its sandier end could be confused with the paler version of Torpedo Jar Ware (see below). The latter was generally very fine, but sometimes contained small voids. Forms seemed to consist almost exclusively of small jars with ribbed shoulders or necks (Fig. 12.7– 13), while both varieties sometimes showed larger vessels with handles (Fig. 12.3, 15). Both these varieties would have been included under the category White Ware at Kush, which appears early in the sequence (in Period I) but only becomes abundant in Period IV (ninth–eleventh centuries) (Kennet 2004: 57). Only the small White NVI jars at Sir Bani Yas (Fig. 12.7, 8, 10) have parallels at Kush (2004: fig. 28, types 109–110), but similarities can be seen with ‘Common Wares’ from the church at al-Qusur. These include broad parallels with jars with handles (Bernard & Salles 1991: fig. 6.15), also found at Kharg and Hulayla Area D (Hardy-Guilbert & Rougeulle 2003a: pl. 14.5; Sasaki & Sasaki 1998: fig. 13), though in all cases these are also incised. The one from Hulayla, in ‘yellow fabric’, showed a line of impressed or rouletted decoration above the handle, similar to that of Figure 12.15. Interestingly, it also bore a round, stamped impression (see below). Several of the smaller jar rims (particularly Fig. 12.1, 9, 11, 12) can be compared to rims from Tulul alUkhaidir (Finster & Schmidt 1976: Abb. 43k–n), which fall into the category ‘medium to fine unglazed ware’, though it should be noted that it is hard to distinguish these sherds from the Buff Ware, which formed a distinctive category at Sir Bani Yas. A stamped impression was found which bears an indistinct round or horseshoe-shaped design (Fig. 12.4), comparable to examples from Hulayla Area D, the al-Qusur church, the al-Qusur settlement, Tulul al-Ukhaidir, Kharg and a well at Arad Fort, Muharraq, Bahrain (Sasaki & Sasaki 1998: figs 13, 15 – JHU98-32; 1996: fig. 46.95–232; Bernard & Salles 1991: fig. 6.16; Patitucci & Uggeri 1984: tav. 56–58; Finster & Schmidt 1976: Abb. 49; Hardy-Guilbert & Rougeulle 2003a: pl. 17: 6–7; Kervran, Hiebert & Rougeulle 2005: pl. 75). At Kharg and Arad they were interpreted as Sasanian (Hardy-Guilbert & Rougeulle 2003a: 131). According to Simpson, such stamps should instead be regarded as typical of the eighth–ninth centuries, on account of their occurrence at al-Qusur, Tulul alUkhaidir, Ain Sha’iah, Wasit and levels IV and V of Tell Abu Sharifa (Simpson, in press). As well as the stamped sherd, a single fine moulded piece was found with indistinct decoration (Fig. 12.14). Notable by its absence from Sir Bani Yas is Eggshell Ware, a particularly fine and thin variety of White Ware, less than 3 mm thick, often with incised and chattered decoration. This has chronological significance: according to Kervran it does not appear at Susa until the mid-eighth century, while at Kush it appears only in the latter part of Period III13. Coarser incised, punched or chattered versions of 11 13 of the rim which occurs in the Sasanian Periods I and II at Kush (Kennet 2004: 30, table 16 and fig. 5)11. Kush Type 64 (a thick-walled jar with a squared or notched rim), Type 62 (a bowl with a slightly thickened everted rim) and Type 93 (a bowl with a thickened notched rim) are also absent from Sir Bani Yas but associated with Periods I and II at Kush (Kennet 2002: 155–156, table 2, fig. 2; 2004: 30, table 16 and fig. 5)12. The Turquoise Glaze ceramics, particularly the carinated bowls, therefore indicate contemporaneity with Kush III, as well as Hulayla Area D, al-Qusur, Suhar I–IV and probably Tulul al-Ukhaidir. This supports the more tentative parallels obtained from the Buff Ware bowls and medium-sized jar rims and implies a date some time in the eighth or late seventh century (see below for a more detailed and precise assessment of the dating evidence). 12 There is also a single occurrence of Type 94 in Period III at Kush, but this is likely to be residual, given its previous abundance. Note that Types 93 and 94 in Kennet 2002 appear to have been amalgamated into Type 94 in Kennet 2004, though the numbers in the tables are not entirely consistent. Period III at Kush has two phases, E-04 and E-05 (see Kennet 2004: 100, fig. 3 for their stratigraphic relationship). Eggshell appears in the later one, E-05, which also contains YBTIN, the earliest class of Samara Horizon Pottery, thus the early ninth century. Note that there is also a stray occurrence of Eggshell in Kush Period I, which Kennet believes may be intrusive. 83 R. A. CARTER Fig. 12. White Ware with Inclusions (‘White Incl’, nos 1–6); and White Ware with No Visible Inclusions (‘White NVI’, nos 7–15). White Wares were also absent from Sir Bani Yas, as were turban handles and their smaller knobbed predecessors. This too may have chronological significance, and indicate that SBY-9 was slightly earlier than Ukhaidir, Hira and Kharg, where all these elements are present. Note, however, that al-Qusur, which is certainly close in date to Sir Bani Yas, has Eggshell and incised White Ware, as does Tulul al-Ukhaidir. The absence may therefore be because Sir Bani Yas was for some reason outside the circuit of such wares. Highly decorated White Ware vessels are known from Susa A III, thought to be seventh–eighth century (RosenAyalon 1974: figs 39–40, 62). They are rare at Kush 84 until late Period III, lending support to the argument that, at least until the early ninth century, Eggshell and other decorated White Wares became rarer or absent the further one goes to the south and east. Alternatively they should be regarded as a luxury element absent from the essentially basic and functional Sir Bani Yas assemblage. In sum, the White Wares of Sir Bani Yas tentatively suggest an earlier date than the closely related assemblages of al-Qusur and Tulul al-Ukhaidir, though other reasons may be sought for the differences. They also imply contemporaneity with the earlier part of Kush Period III rather than its later phase. CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Torpedo Jars (Kennet and Priestman code: TORP) (Fig. 13) Torpedo Jar fragments are among the more frequent and immediately recognisable sherds at Sir Bani Yas, the main type and its variants accounting for 15 % of the whole context assemblage by sherd count, and 12 % by EVE. This type of vessel is well known from numerous sites in Mesopotamia, western Iran and the Gulf, and is distributed as far as India, Sri Lanka and East Africa (Stern et al., in press; Tomber 2007; Kennet 2004: 63; Priestman 2005: 111). It has a distinctive amphora shape with a simple hole-mouth or collar rim, a pointed base (Fig. 13.8), and usually an abrasive sandy fabric (see Simpson 2007: fig. 140 for illustrations of the complete shape). Sometimes it has a ribbed upper part or shoulder (Fig. 13.2). It is doubtful that handles in the same fabric (Fig. 13.7) are typical of the classic torpedo jar; they may represent a different kind of vessel. Colours range from buff or pale brown to cream14. Sometimes the body was brown but the surface was cream in colour. Almost invariably Torpedo Jars were lined with bitumen, and any absence of bitumen is likely to be due to erosion of the sherd. Sometimes the bitumen was splashed over the rim and onto the outside of the vessels (Fig. 13.1). Apart from the gradations of colour, other distinct differences in Torpedo Jar fabrics could be identified, and three varieties were identified. The commonest is the one with the abrasive sandy fabric (‘Torp’, Fig. 13.1–8), but there was also a version with less sand and notable quantities of mica (‘Micaceous Torp’, Fig. 13.9), and a third with very large sub-angular translucent inclusions (‘Pebbly Torp’, Fig. 13.10). These may indicate different source areas or kilns. Consistent differences in the shapes between these types were not observed. The three Torpedo Jar fabric categories do not appear to correspond to the four identified by Priestman (2005: 208), except that the most abundant one at SBY (‘Torp’) probably includes both his TORP.1 (yellow fabric, cf. the cream examples at SBY) and TORP.3 (coarse sandy orange-brown fabric). Indeed, Priestman specifically equates a sample sherd from Sir Bani Yas with his class TORP.3 (2005: 111). It is possible that Priestman’s TORP.4 is equivalent to Micaceous Torp, which is finer than the other varieties at SBY, but he does not mention mica. Torpedo jars are not particularly useful for dating, having a long chronological range from the Parthian Period through to at least the ninth century (Kennet 2004: 63; Simpson 2007: 154). They are found at most of the sites which have assemblages similar to that of Sir Bani Yas, for example at Kush Period III, at the al-Qusur church and settlement, at Suhar II and V, and Tulul al-Ukhaidir (Kennet 2004: fig. 36; Bernard & Salles 1991: fig. 7.26; Kervran 2004: figs 10.15, 24.2; Finster & Schmidt 1976: Abb. 38g and 39). Torpedo Jars were clearly designed to carry liquid goods, and their shape is suitable for stacking in the hold of ships. The bitumen was used to waterproof the jars and it is likely that they carried wine (Simpson 2003: 353–355; 2007: 154). Wine would 14 Fig. 13. Torpedo Jars: common variety with sandy fabric (‘Torp’, nos 1–8); ‘Micaceous Torp’ (no. 9); ‘Pebbly Torp’ (no. 10). It appears that the brownish variety of Torpedo Jar Ware is more common at Sir Bani Yas (and at the Early Islamic sites at As-Sabiyah, Kuwait) than at Kush, where the paler creamcoloured variety predominates (pers. comm. Derek Kennet). 85 R. A. CARTER Fig. 14. Hard Gritty Ware (LISV). have been required for use in communion at Sir Bani Yas15. Hard Gritty Ware (Kennet and Priestman code: LISV) (Figs 14–15) Sherds from several large vessels were recorded, consisting of large bowls (Fig. 14.1), large vats (Fig. 14.4–5) and storage jars (Fig 14.3, 6, & Fig. 15). This fabric is here termed Hard Gritty Ware. It is hard-fired and wheel-made. The colour varies from red to red-brown or grey. Sometimes it is so highly fired that the colour turns to greenish, often with cream streaking caused by the explo15 According to Priestman, if we assume a connection between Torpedo Jars and wine, there may be scope in examining the distribution of Torpedo Jars with regard to the distribution of non-Muslim communities (pers. comm. S. Priestman), though the historical popularity of wine among the Muslim communities of the region must be acknowledged. 86 sion of lime grits. It usually has a greyish slip and is frequently incised, in which case it is comparable to Kennet’s class LISV (Large Incised Vessels). Inclusions consist of variable amounts of lime, with variable amounts of angular or sub-angular grits. The very high proportion of this type by Fig. 15. Hard Gritty Ware (LISV). CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM sherd count results from the recovery of a few large vessels broken into many pieces. According to Kennet, LISV is found from the fifth or sixth century into the Abbasid period (Kennet 2004: 58). The fabric aligns Hard Gritty Ware with Priestman’s LISV.A, and Priestman notes that LISV is concentrated in the seventh–ninth centuries at Kush (Priestman 2005: 178–179). Specific parallels for the Sir Bani Yas vessels can be found at Hulayla and Suhar, suggesting that this kind of vessel is more associated with the Lower and Central Gulf than Mesopotamia and the Upper Gulf. The large incised vats (Fig. 14.4–5) have an excellent parallel at Hulayla Area D (Sasaki & Sasaki 1996: fig. 49.95–149), as does the large incised storage jar (Fig. 15) (1996: fig. 50.95–1). Similar vessels were also found on Kennet’s survey of Hulayla, where they were included in Period 1 and then dated to third– seventh centuries (Kennet 1994: fig. 10.10, 14), though subsequent work suggests they extend later in date. The large open bowl with an extended rim (Fig. 14.1) has numerous parallels with Suhar Periods I–IV, which Kennet dates to the eighth century (Kervran 2004: figs 8.8, 10.4 and 11.23–26). Grey Brittle Ware (Kennet and Priestman code: SMAG) (Fig. 16.1–5) Grey Brittle Ware is fairly common (9 % of the assemblage by EVE), wheel-made, thin-walled, brittle and generally uniformly grey, with a dry biscuity appearance. The fabric includes lime particles and variable quantities of small angular dark grits. It sometimes has simple incised decoration of parallel lines on the shoulder (Fig. 16.5). It has affinities with Kennet’s SMAG (Kennet 2004: 63), which are described as ‘Small Grey Vessels’ and usually occur Fig. 16. Various Wares: Grey Brittle Ware (nos 1–5); Red Fine Sandy Ware (nos 6–8); Indian (no. 9); Thin Torp-like (no. 10); Undiagnosed (nos 11– 12); Honeycomb (no. 13). 87 R. A. CARTER as small jars with complex rims. The fabric is less dense than the version described by Kennet, and although it occurs as small jars the rims are invariably simple (Fig. 16.1–3). The fabric compares better to Priestman’s subdivision SMAG.B, described as having a ‘hard ‘‘dry’’ grey fabric’ (Priestman 2005: 175). Formal parallels may be drawn with the simplest rim shape found at Kush (Kennet 2004: fig. 37, type 58, upper). There is no tight dating reference for this class, though it is distinctive in appearance, and SMAG extends throughout the Kush sequence. Priestman attributes most of the Kush SMAG to the seventh–ninth centuries, and speculates that SMAG.B is slightly earlier, but this is not borne out by its presence at Sir Bani Yas. As well as the wheel-made SMAG, a cruder handmade variety existed with a very similar fabric and perhaps similar shapes, though no diagnostic sherds were found. It was perhaps softer and less hard-fired. This was called Crude Grey Brittle Ware (not illustrated). stump of a spout from a vessel resembling a pilgrim flask (Fig. 16.11), may also be Indian, judging from its fabric, but was included in the class ‘Undiagnosed’. Red Fine Sandy (Fig. 16.6–8) This wheel-made type had a fine red or red-brown fabric, with abundant very fine sand. It is not a particularly cohesive class, and forms seemed to relate to small or medium-sized jars with handles. One distinctive ribbed rim was present (Fig. 16.6) but no good parallels were found. In one case (Fig. 16.7) it may represent a particularly fine and reddish example of Torpedo Jar ware, drawn at the wrong angle. Vegetal Tempered Ware (‘Veg’) A few sherds of eroded vegetal tempered ware were found, from medium-sized or large vessels (not illustrated). No diagnostic elements were noted. This material resembles pottery from the early and midsecond millennium BC (Wadi Suq and Late Bronze Age ⁄ Late Wadi Suq). It is probably residual from an earlier occupation. Another site on Sir Bani Yas (SBY-37, near the airport), yielded Barbar pottery, characteristic of the Early Dilmun Period on Bahrain (Carter 2003: table 1). This material has now been examined by the author and its identification can be confirmed. It was not possible to pin it down to the City I or City II Period of Bahrain, but it provides supporting evidence of occupation on the island during the early second millennium BC. Indian (Fig. 16.9) A few sherds of probable Indian pottery were found, including an everted rim (Fig. 16.9). The fabric is light and grey or reddish, with medium or fine vegetal temper and small grits, with crudely burnished surfaces and a reddish or grey slip. The shape is not particularly distinctive and is no help in dating the site, as such Indian forms are found abundantly in the region throughout the first millennium AD, for example at Suhar (Kervran 2004: figs 8.10–11, 9.9–10 and 10.20). According to Kennet’s typology the pieces at Sir Bani Yas would belong to his classes SBBW (Black Burnished Ware), which is found at Kush from Period II onwards (Kennet 2004: 66). Note that another Indian rim was found at SBY-4 (Fig. 19.5), as well as a sherd of Indian Red Polished Ware (see below). A further sherd, apparently the 88 Thin Torp-like (Fig. 16.10) This variety was wheel-made and had an abrasive, very sandy fabric, similar to the sandier end of the Torpedo jar fabric. It was red-brown in colour, sometimes with dark grey or black interior or core, and appeared to have a whitish slip, though this could have been due to the action of salts. It was used for medium-sized jars with handles (Fig. 16.10), and is over-represented statistically due to the presence of an almost complete vessel which had most of the rim present and which was broken into many pieces. Similar vessels are found at Suhar Levels 0 and IV in a ‘medium coarse pinkish fabric’ (Kervran 2004: figs 8.6 and 13.10) and at Hulayla Area D in a black fabric (Sasaki & Sasaki 1996: fig. 51.95–174). Without examining these comparanda at first hand it is hard to assess their validity. Honeycomb Ware (Fig. 16.13) This distinctive finger-impressed ware is found in small quantities in eighth-century contexts in the Gulf, at Hulayla Area D and Kush Period III (Sasaki & Sasaki 1996: fig. 46.95–34; Kennet 2004: 59), with just one sherd found at the al-Qusur settlement (Patitucci & Uggeri 1984: tav. 60b). It is also found at Tulul al-Ukhaidir (Finster & Schmidt 1976: fig. 48d, Taf. 48a). Like Torpedo Jars, Honeycomb Ware is thought to have been used for the transportation of CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM bulk commodities (Priestman 2005: 111–112). There are some questions as to its exact date range. Originally it was thought to be Sasanian, but Simpson and Kennet believe it belongs to the Early Islamic Period, continuing into the ninth century at Samarra (Simpson 1992: 296; Kennet 2004: 59). It certainly appears to have been present, albeit uncommon, in eighth-century contexts in the Gulf. It is very rare at Sir Bani Yas, consisting of just one sherd from the studied SBY-9 assemblage (from the surface, and thus not appearing in Fig. 6), with one other in the sample which arrived too late to include in this analysis. Summary of ceramic dating evidence The comparative dating of the SBY-9 assemblage is fundamentally reliant on Kennet’s work at Kush and on other Sasanian and Early Islamic sites and assemblages in the region, which has allowed clear distinctions to be made between Sasanian and Islamic pottery assemblages for the first time (see especially Kennet 2004 and 2007). The dating of SBY9 corresponds to a fairly constrained ceramic horizon which is well marked by the presence of carinated glazed bowls (Kush Type 72), and which also occurs in the Gulf at al-Qusur, Kush Period III, Hulayla Area D and Suhar Periods II and III and, more rarely, at Tulul al-Ukhaidir in Mesopotamia. These correspondences are supported by other Turquoise Glaze forms, as well as parallels with Hard Gritty Ware forms in the Lower Gulf, particularly the incised vats and storage jars (LISV at Hulayla, Kush and Suhar). If the absence of elaborately decorated White Wares and Eggshell Ware is significant, and not the product of regional variation in distribution, it may indicate that SBY-9 is earlier than al-Qusur and Tulul al-Ukhaidir, probably aligning it with the early part of Period III at Kush (Phase E-04) A comparative terminus ante quem is indicated by the absence of any kind of glazed ware other than Turquoise Glaze. Thus, SBY-9 predates the Samarra Horizon, which began some time between AD 803 and 835 (Kennet 2004: 31). In fact, the absence of appliqué ‘barbotine’ decoration at SBY-9, which appears immediately before the Samarra Horizon, suggests a date prior to the end of the eighth century AD. Also significant are the parallels with Tulul al-Ukhaidir, which was abandoned by AD 762 (Finster & Schmidt 2005: 347). Given that evidence from the White Wares suggests that SBY-9 is earlier than Tulul al-Ukhaidir, albeit tentatively, it seems likely that SBY-9 dates to the mid-eighth century or earlier. The terminus post quem is slightly harder to ascertain, given the difficulty of distinguishing sixth-, seventh- and early eighth-century assemblages. Kennet’s work on this topic is important in this regard. According to the Kush sequence there are certain markers of the Sasanian Period. These include Fine Orange Painted Ware (FOPW), sometimes referred to as ‘Namord Ware’. This is concentrated in the first phase of Period I at Kush (fourth ⁄ fifth century) but is also found in Period II, suggesting that it continues beyond the range suggested by Potts in his study of the type (Potts 1998a). It is not found in Kush Period III. Another key marker is a kind of Turquoise Glaze bowl with a notch on the inner side of the rim (Kush Type 94). This is abundant in Kush Periods I and II, and only has one occurrence in Kush Period III, which is possibly residual. Period II at Kush is dated by C14 on charcoal (twigs) to 1340 ± 35 BP (Kennet 2004: 14, table 2; 2001: 44), which calibrates to AD 630–780 at 2r (94.5 % probability), or AD 640–690 at 1r (61.3 % probability). Thus, Period III at Kush could in theory begin as early as the mid-seventh century, and almost certainly not before AD 630. An early seventh-century date for SBY-9 can therefore be ruled out if the site is contemporary with Kush Period III. The significance of this is that, even at the earlier end of SBY-9’s potential date range, it puts the available evidence for occupation at the monastery to after the advent of Islamic rule in Eastern Arabia. The radiocarbon dates are further discussed below. Other absolute dating evidence comes from coins found at the sites with comparable assemblages. One from Hulayla Area D is said to date to the seventh or eighth century (Sasaki & Sasaki 1996: 69 and fig. 54.95–3). In a later comment at a workshop held at the British Museum, Sasaki was more specific, stating that the coin dated to the mid-eighth century, or perhaps the eleventh (see http://www.thebritishmuseum.ac.uk/ane/pottery_workshop.pdf). Three coins were also found at Tulul al-Ukhaidir, dating from the late seventh to the early eighth century (Finster & Schmidt 1976: 48). The precise 89 R. A. CARTER date ranges are: AD 685–705, AD 726–747 and AD 719. The sequence there was divided into four (Finster & Schmidt 2005: 347), but unfortunately it is difficult to know from which level the coins came, or how closely they were associated with the pottery. At the very least they suggest that the occupation deposits associated with the coins started at any time after the late seventh century and certainly continued after the early eighth. To summarise, SBY-9 dates to some time between the mid-seventh and the mid-eighth century AD. Examination of the C14 dates from SBY-9 confirms this range but does not narrow it significantly. the data in Oxcal 3.10, and incorporating sequential constraints and a terminus ante quem into the radiocarbon calibrations. The modelling was done using the following assumptions:  the date for Kush Period II is sequentially earlier than the one for SBY-9, because it originates from a phase which is definitely earlier than SBY-9 using the ceramic chronology. For the purposes of the calibration this is effectively the same as being stratigraphically earlier.  a terminus ante quem is provided by the absence of the Samarra Horizon, which is here taken to be AD 803, that being the earlier limit of Northedge and Kennet’s dating of the earliest Samarra types. In actual fact the true terminus ante quem is probably earlier still, given the absence of the barbotine jar horizon at Sir Bani Yas.  the earlier of the SBY dates is assumed to be from old wood and excluded from the model17. The results are shown on Figure 17. This shows that:  Kush Period II dates, at least in part, to between AD 630 and 730 (at 91.9 %) and probably the mid- ⁄ late seventh century (AD 645–685 at 68.2 %). The radiocarbon sample of twigs was taken from Phase E-03, the last part of the Period II sequence, suggesting that this is towards the end of its range.  SBY-9 dates to AD 660–780 (at 93.9 %), i.e. somewhere in the second half of the seventh century and the first three quarters of the eighth. The 68.2 % range does not narrow this significantly. Radiocarbon dating evidence at SBY-9 Two radiocarbon evaluations were taken using charcoal from a fireplace sealed beneath a collapsed wall (Elders 2001: 56)16. 1. 1460 ± 70 BP (GU-9185). This calibrates to AD 420–670 at 2r (94.5 % probability). The later part of this range is compatible with the ceramic evidence. 2. 1305 ± 50 BP (AA-40740 ⁄ GU-9185A). This calibrates to AD 640–830 at 2r (93 % probability). The earlier part of this range is compatible with the ceramic evidence. If the overlap were taken as the indicator of date it would suggest that SBY-9 was probably occupied between AD 640 and 670, i.e. the mid-seventh century. It is wise to treat the earlier of the dates with caution given that the samples were taken from the same context, and the possibility that old wood was burned. If only the later date is accepted, then occupation could have occurred between the midseventh and the early ninth centuries AD. Once the ceramic evidence is accounted for, the later part of this date range should be reduced to the mid-eighth century. Thus the combined evidence from the C14 and ceramics is the same as that from the ceramics alone, namely mid-seventh to mid-eighth century AD. It is possible to narrow slightly the potential ranges for Kush Period II and SBY-9 by modelling Other dating evidence and synthesis Glass According to St John Simpson, who briefly examined the glass from Sir Bani Yas, there are similarities with the Umayyad assemblage, particularly in the presence of appliqué ribbons in wavy lines on the shoulder of miniature bottles (pers. 16 17 It is not stated whether this was wood charcoal, though the d13C values are within the range expected from wood charcoal (see http://www.adias-uae.com/radiocarbon. html#sirbaniyas). 90 There is in fact sufficient overlap between the calibrated ranges for the model to be successfully made including the earlier SBY date, but there was poor agreement between it and the others in the sequence (around 38 %). CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Fig. 17. Modelled radiocarbon dates from SBY-9 and Kush Period II. comm. St J. Simpson). This would indicate a date during or after the second half of the seventh century AD. Plaster The stucco work of Sir Bani Yas has been compared to that of Kharg and the church at al-Jubayl 91 R. A. CARTER (Simpson, in press), particularly in the presence of a ‘horizontal arrangement of roundels enclosing alternating and heavily stylised rosette, clustered grape and acanthus patterns below a row of stepped merlons’ (Elders 2003: fig. 3 and unnumbered plate on p. 235; Langfeldt 1994: fig. 8) (Hardy-Guilbert & Rougeulle 2003b: pl. 12). According to Simpson’s detailed study of the architecture, stucco and pottery of Christian sites in the Gulf and Mesopotamia, an eighth- and ninth-century date should be assigned to these sites. Synthesis and ramifications of dating evidence According to both ceramic parallels and radiocarbon dates, the longest range for SBY-9 is mid-seventh to mid- ⁄ late eighth century AD. If a mid-eighth century date is accepted for the Hulayla coin, with close contemporaneity assumed between this site and SBY-9, and Kush Period II is dated to the mid- ⁄ late seventh century, then the later part of the date range is to be preferred. An early to mid-eighth-century date is therefore most likely, which is compatible with the likelihood that SBY is earlier than Tulul al-Ukhaidir, demolished in AD 762. However, unless further data comes to light it is not possible to entirely rule out a late seventh-century date. This has consequences for the dating of other sites in the Gulf with carinated glazed bowls:  Hulayla Area D has a very similar assemblage to Sir Bani Yas and can therefore be pinned down to the same range, i.e. early to mideighth century.  al-Qusur, which has the carinated bowls but also has more highly decorated white wares — and reports of a possible Samarra Horizon sherd from the settlement (Kennet 2007: 7, 115, n. 113) — may overlap in date but probably runs later, perhaps having a range of mid- to late eighth century and just extending into the first years of the ninth. This is the same as Kennet’s suggested range (2007: 89) and within Bernard and Salles’ date range, who place the pottery assemblage towards the end of a broader range of midseventh to mid-ninth century (Bernard & Salles 1991: 12).  Kush Period III dates from the late seventh century to the early ninth century (cf. Ken- 92 net’s dating of late eighth ⁄ ninth century). It seems sensible to split Kush Period III into its two phases, the first, E-04, being contemporary with Sir Bani Yas, and the second, which contains early Samarra Horizon sherds, being late eighth–ninth century.  Suhar 0-IV has a complex assemblage, apparently with residual Parthian material and also strong Indian traits, but is broadly contemporary with Sir Bani Yas (contra Kervran’s dating to the Sasanian Period). Note that there is almost no evidence for contemporary occupation at the Kharg monastery, where all identifiable pottery is later than Sir Bani Yas, except for one possibly comparable carinated bowl (HardyGuilbert & Rougeulle 2003a: pl. 20.4), and belongs to the end of the eighth century or the Samarra Horizon. This has important ramifications for the history and chronology of monastic life in the Gulf, and is discussed in detail below. Distribution and functional aspects Given that some kind of selection had clearly occurred prior to the quantification of the pottery, it is unwise to draw detailed and concrete conclusions regarding the possible functions of the assemblage, or make judgements regarding the wealth or openness of the community. Some general comments can nonetheless be made. The SBY-9 assemblage contains some of the elements of what might be considered a standard domestic assemblage, having a mixture of large storage jars and processing vessels (the large vessels in Hard Gritty Ware, the larger Turquoise Glaze jar rims with lids); medium-sized open vessels for serving or processing (the large straight-sided bowls in Buff Ware); smaller jars for storage or serving liquids (the Buff Ware jars, the White Ware jars); and vessels for serving and consumption (the Buff Ware bowls and glazed bowls). Also present were the torpedo jars. It seems likely that these were brought in primarily for their contents (wine), though they could have been reused for storing other liquids. Not all the expected elements are present, however: cooking vessels were not identified with certainty, except for the Indian vessels, which may have been used for this purpose. An example was found in Context CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM 53 (Fig. 16.9), described as a monastic kitchen floor. It is feasible that the Buff Ware straight-sided bowls may have been used over a fire, but associated burning was not noted. Highly decorated pottery was not present, though this may be largely for chronological reasons. As noted above, the proportion of Turquoise Glaze is slightly higher (by count) than that found at contemporary or near-contemporary sites (Kush Periods II and III; Bilad Qadim Period I), but this is likely to be the result of selection bias. The glassware, which included fine bottles and goblets, may have compensated for the drabness of the pottery, and was probably used for significant occasions (e.g. communion). An examination of the distribution of the different pottery types reveals some spatial information. Table 3 shows sherd counts for each context included in the analysis, with brief descriptions of the contexts. The areas mentioned in the descriptions (termed ‘Rooms’, though some are courtyards) are marked on Figure 5. Leaving aside surface material, the largest registered assemblages, with more than fifty sherds, are mostly outside the church, in Contexts 122 and 179 (layers of debris in Room 20, an apparent courtyard in the north-eastern building complex within the compound); Context 124 (a hearth spread in Room 13, next to Room 20); and Contexts 125 and 128 (debris in collapsed walling of the church). The most significant church context, with fifty-one sherds, is Context 131 (sand filling the south aisle of the church). The largest collection, from Context 124, is dominated by many sherds of Hard Gritty Ware (Kennet’s LISV), nearly all from a single vessel (Fig. 15). It appears this room was used for cooking and storage. Although the church assemblage is small, it was considered worthwhile comparing it to other pottery from the site to see if any associations could be discerned (Fig. 18). Note that the total number of sherds from the church is not large (just 165 sherds). Moreover, no attempt has been made here to conduct detailed stratigraphic analysis in order to separate likely primary occupation deposits from putative earlier or later squatter occupations. For example, several contexts are described in the records as ‘make-up’ beneath the plaster of the floors of the church (the term preferred in Table 3 is ‘infill’), which may either consist of earlier settlement ceramics scattered around prior to and during the building of the church, or material from the usage of the church incorporated during replastering of the floors. The pottery in the sandy fill of the south aisle, Context 131, may in theory relate to occupation following partial abandonment, or at least the loss of the church’s sacred function. The following discussion is therefore tentative, but a clear pattern is apparent, and it is hoped that this broadly reflects the kind of pottery used in the church up to the point of its abandonment. It appears that certain types are more common in the church, particularly Buff Ware, the finer variety of White Ware, Turquoise Glaze and, to a lesser extent, Torp. Outside the church there is a greater quantity of coarser pottery, especially the types used for large storage vessels (Hard Gritty Ware) and smaller storage jars (Grey Brittle Ware, Crude Grey Brittle Ware, Thin Torp-like Ware). This pattern would fit the functional profile expected from a church context: torpedo jars containing communion wine could have been kept on site, and perhaps decanted into smaller Buff Ware vessels (seemingly small jars only), fine White Ware vessels (probably small jars, some decorated) and glazed bowls for ceremonial use. The bowls and jars may alternatively have been used for ablutions. A cruder and more domestic assemblage characterises the living and working areas in the compound surrounding the church. Imported pottery and regional interactions As might be expected from an island with apparently no clay resources, the pottery was all imported. Although it was not always possible to identify the source areas, material from southern Mesopotamia was present (Turquoise Glaze, probably Honeycomb, probably White Ware, possibly Torp), as well as pottery from India, and probably also Bahrain (the Buff Ware) and the perhaps the Lower Gulf (the Hard Gritty Ware). It is unclear whether Iranian material was present, unless the Hard Gritty Ware or Torpedo jars had an Iranian origin. Bitumen from Parthian or Sasanian Torpedo jars in Kuwait (Akkaz) turned out to originate from western Iran (Luristan) (Connan & Carter 2007), and but it is perhaps too 93 Context Context Description 1 9 16 44 53 55 122 124 125 126 127 131 150 157 161 162 163 170 173 176 No context number ascertainable Top layer of surface sand and debris, site-wide. Rubble in Room 8. Sand and rubble around northern compound wall. ‘‘Post-abandonment’’. Sand and rubble in ‘‘Room’’ 10, a yard. ‘‘Post-abandonment’’. Lower ashy fill of Room 6. ‘‘Monastic kitchen floor’’. Mudbrick and sand filling Room 5 (maybe a courtyard). ‘‘Monastic - abandonment’’. Rubble and sand in Room 20 (maybe a courtyard). ‘‘Post monastic’’. Fireplace in Room 13 with abundant broken pottery. Collapsed east wall of church with much decorated plaster. Sand and rubble in the chancel (choir). Collapsed north wall of church. Sand in south aisle of church. Partition wall between north aisle or transept and central nave of church. Infill layer beneath plaster floor of east room at end of north aisle (‘‘transept’’). Infill layer beneath plaster floor of south aisle. Plaster floor in central nave of church. Infill beneath plaster floor (162) in central nave of church. ‘‘Wash-out’’ beneath the collapsed north wall of the church. Infill beneath plaster floor in north aisle of church. Lower clay fill of east room at end of south aisle (aka ‘‘transept’’ or ‘‘apse’’). Crude Red Grey Grey Hard Micaceous Pebbly Fine Thin White White Buff Brittle Brittle Gritty Honey Indian Torp Torp Sandy Torp-like Torp Turq Undiagnosed Veg Incl NVI sum 10 20 1 5 20 1 1 1 4 1 1 3 28 12 20 2 5 3 1 2 5 1 3 1 2 1 38 5 6 1 6 1 1 1 26 1 2 81 14 11 303 14 8 10 1 1 2 8 5 1 3 1 23 1 4 2 6 51 15 36 15 277 23 1 5 2 368 33 13 2 4 1 1 3 78 11 1 23 1 3 1 7 2 6 1 1 1 3 13 2 3 2 1 1 23 5 1 2 10 2 1 54 17 9 39 102 1 128 51 38 2 6 7 1 11 1 2 7 4 1 6 17 15 6 4 8 2 1 2 3 1 1 1 1 17 8 47 14 2 7 R. A. CARTER 94 Table 3. Breakdown of pottery types by context, with context descriptions. Shaded contexts are those from within the church walls. 10 107 much to make a leap from that to positing an Iranian origin for Torpedo jars. Priestman believes that Torpedo jars are more likely to originate from southern Iraq, but acknowledges that conclusive petrographic work is yet to be carried out (Priestman 2005: 111). Notwithstanding the uncertainty surrounding the presence of Iranian pottery, the wide connections with other regions of the Gulf, southern Mesopotamia and the Indian Ocean indicate that the monastic community was well integrated into the region’s networks of trade and communication. This fits in with the highly active and expansionist nature of the Church of the East at the time, and with Elders’ observation that the position of the monastery facing directly onto the sea, and the elevated location of the church, invited contact with the outside world (Elders 2003: 233). This was not a closed eremitic community. 231 11 1 2 36 86 1 384 1 4 1 10 10 1 1 7 103 194 2 1 1 2 1 1 1 7 76 3 72 1335 4 13 2 1 14 1 81 1 77 3 sum 233 212 205 201 196 Burnt area above a clay surface in Room 20 (maybe a courtyard). Clay fill inside walls of north aisle. Sand outside eastern compound wall. Sand and rubble in east gateway. Lower infill layer beneath plaster floor of south aisle. Rubble, uncertain location (near eastern entrance of compound?). Context Context Description 179 Table 3. Continued. Red Crude White White Thin Micaceous Pebbly Fine Grey Grey Hard NVI sum Sandy Torp-like Torp Turq Undiagnosed Veg Incl Torp Buff Brittle Brittle Gritty Honey Indian Torp CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Other sites on Sir Bani Yas As well as the material that could be identified as having come from SBY-9, there were 140 sherds from SBY-4, a courtyard house near SBY-9; two from SBY-1 or SBY-2, two sites to the west of main cluster around the monastery; and 205 from unidentified sites on the island. The diagnostic sherds are illustrated on Figure 19. SBY-4 appears to be very similar in date to SBY-9, SBY1 ⁄ 2 may be earlier, and the others are compatible with the dating of SBY-9. SBY-4 (Fig. 19.1–7) Sherds included an unusually large straight-sided bowl in Buff Ware (Fig. 19.1). Turquoise glaze included a small simple bowl or lamp fragment (Fig. 19.3), a larger bowl (Fig. 19.2), and an unusual thin-sided bowl with glaze only coming down a short distance outside the rim (Fig. 19.4). There was also an Indian rim (Fig. 19.5) identical to many found at Suhar and elsewhere (see e.g. Kervran 2004: fig. 9.10), a small fragment of a Torpedo Jar rim (Fig. 19.6), and a large jar or amphora rim with a handle attachment in the variety of White Ware with small inclusions (Fig. 19.7). The latter was gritty enough to make it questionable whether to regard it as White Ware or a pale Torpedo Jar Ware. No new or different wares were observed at SBY4, and the assemblage is likely to be contemporary 95 R. A. CARTER 40 35 Church Not church 30 25 20 15 10 5 Thin Torp-like Crude Grey Brittle Red Fine Sandy Honey Grey Brittle Veg Indian White Incl Hard Gritty Turq Torp White NVI Buff 0 Fig. 18. The church assemblage (165 sherds) compared to material from elsewhere on SBY-9 (1022 sherds) expressed as percentages. The three varieties of torpedo jar fabric have been combined into one (Torp). Surface material and pottery of uncertain attribution has been excluded. 19.12–13). These would fall under LISV using Kennet’s classification, but had a very different fabric to the Hard Gritty Ware associated with the large incised vessels at SBY-9. The fabric, Grey Sandy Ware, was hard, grey and comparatively fine and sandy. In a quick review of the pottery which arrived too late to be included in this study, it was observed that SBY-1 and SBY-2 may be slightly earlier in date than SBY-9 and the surrounding courtyard houses, having higher quantities of ‘CLINKY’-type sherds (see below). Fig. 19. Pottery from SBY-4 (nos 1–7), SBY-1 ⁄ 2 (nos 12–13), and undetermined sites on Sir Bani Yas (nos 8–11). with SBY-9. There were, on the other hand, some slightly unusual features, including the partially glazed bowl and the unusually large Buff Ware bowl, so a slight difference in date should not be ruled out. SBY1 ⁄ 2 (Fig. 19.12–13) Two incised sherds from the same vessel, a large vat or storage jar, were included in the sample (Fig. 96 Site unknown (Fig. 19.8–11) No new wares were noted in this group, other than a single body sherd of Indian Red Polished Ware (IRPW) (not illustrated). It did not have the site marked on it, only the inventory number. No original database was provided so it was not possible to relate the inventory numbers to the sites. A second sherd of IRPW was noted but not recorded in the collection that arrived too late to study. IRPW has a fine red or orange fabric with a highly finished surface, covered with a reddish polished slip. It appears that it was first manufactured around the first century AD, and that it went out of use in the sixth ⁄ seventh or eighth century AD (Kennet 2004: CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM 65). The best dating evidence is probably provided at Kush, where it is relatively abundant in Period II and in the first phase of Period III, suggesting that it did indeed continue to circulate during the eighth century. The rest of the pottery was compatible with that found at SBY-9, and included a large Turquoise Glaze bowl (Fig. 19.8), some Torpedo Jar bases containing bitumen (Fig. 19.10–11), and the rim of a small jar in a fabric similar to the Hard Gritty Ware (Fig. 19.9), which was usually used for larger vessels. This, along with a small number of body sherds from SBY-9, would probably have been classified either as ‘CLINKY’ under the typology developed by Kennet and refined by Priestman, or SMAG.A, both of which have similar or identical fabrics to LISV.A (Priestman 2005: 174). The distinction between CLINKY and SMAG seems to be based on rim and vessel shape, and the chronological division is unclear, but SMAG predominates by Period III at Kush while CLINKY appears to be earlier. The best parallel for Figure 19.9 at Kush is CLINKY type 86 (Kennet 2004: fig. 35, K4856). It is feasible that this sherd came from SBY-1 or SBY-2, though it is not impossible for it have originated from SBY-9. Discussion: other archaeological evidence for Christianity in the Gulf There is good evidence for other churches, monasteries and Christian communities in the Gulf (Fig. 1), some of which has been mentioned in the preceding discussion of the architecture and pottery. Although these finds have been summarised before (Potts 1990; Langfeldt 1994; King 1997a; Bin Seray 1996) some brief comments are merited on their significance and dating. al-Qusur A church was excavated by a French team at alQusur on the island of Failaka, Kuwait (Bernard, Callot & Salles 1991; Bernard & Salles 1991). As noted above, its architecture and stucco work is comparable to that of Sir Bani Yas. Its architecture is also closely comparable to that of Kharg, while its stucco compares to that of Kharg and Jubayl (Simpson, in press). Very little pottery was associated with the inside of the church during its main occupation (Level I) though more ceramics were found in levels thought to be contemporary with its use just outside the church (Level II), and further sherds were found in the highest floors before the collapse (Level III). There was then a re-occupation of the smaller rooms (Level IV) (Bernard & Salles 1991: 11–12). Most of the published assemblage is from Level III, and is closely comparable to that of SBY-9, including glazed carinated bowls and circular stamps on White Wares. There is nothing from this or the other levels to indicate a date prior to that of SBY-9, and neither of the two Level I sherds picked out by the excavators as indicating ‘transitional period Sasanian-Early Islamic’ are diagnostic of the period prior to the eighth century. Level I lacks Eggshell but, as noted above, so does Sir Bani Yas and the first phase of Period III at Kush. Some sherds, for example chattered White Ware jars with carinated bodies (Bernard & Salles 1991: fig. 6.9, 14), may suggest an occupation stretching slightly later than SBY-9. The church is accompanied by a village settlement (Patitucci & Uggeri 1984), the pottery of which indicates a contemporary date, also probably extending slightly later than SBY-9, possibly into the ninth century AD (Kennet 2007: 89–92). As noted above, the stucco work also indicates an eighth–ninth-century date, rather than fifth–sixthcentury as suggested by the excavators (Simpson, in press). The dating of the al-Qusur church should therefore be revised. Akkaz Another church was identified on the small island of Akkaz, just off Kuwait City (Gachet 1998). As noted above, the architecture was poorly preserved but the plan appears to resemble those of al-Qusur, Sir Bani Yas and Kharg. As no in situ material was found (1998: 74), this church cannot be dated closely. Kharg An extensive monastic complex on the island of Kharg was excavated and dated by Ghirshman to the fifth–sixth century AD. This has recently been more fully published (Steve 2003) and is now said to have been founded around the end of the sixth or the start of the seventh century (Jullien & Jullien 2003: 155; Hardy-Guilbert & Rougeulle 2003b: 130). The date should be revised to the ninth century. Of the ninety-eight published sherds, not a single one indicates a date prior to the ninth century (Kennet 97 R. A. CARTER 2007: 92; Simpson, in press), except perhaps a small carinated glazed bowl (Hardy-Guilbert & Rougeulle 2003a: pl. 20.4) possibly comparable to the eighthcentury ones from Sir Bani Yas and elsewhere. Those that the analysts suggest are Sasanian, such as the ones impressed with circular stamps, are in fact better placed in the eighth–ninth century AD. The moulded stucco plaques and friezes of Kharg also indicate a post-Sasanian date (Simpson, in press). The late sixth ⁄ seventh century date given for its founding had been based on the stucco, which is considered to be slightly later in style than that of the church of al-Qusur (HardyGuilbert & Rougeulle 2003b: 130), which has now been redated (see above). There are also good parallels between the crosses and other stucco with Ain Sha’ia, which are said to indicate a date anywhere between the end of the sixth and the start of the eighth century (Steve 2003: 114). Note, however, that the earliest radiocarbon date for Ain Sha’iah suggests a date between the mid-seventh and the late tenth century AD18, and the excavator believes the foundation date to be late eighth or ninth century (Okada & Numoto 1989: 61; Okada 1992: 93, n. 9). This is compatible with the pottery of Ain Sha’ia19. There is therefore nothing to fix the date of the Kharg monastery to before the mid-seventh century, and if it is later than al-Qusur then it is likely to date to the mid-eighth century or later, as also indicated by the pottery of both sites. At the moment, a ninthcentury date is most likely. Thus it is slightly later than SBY-9 and al-Qusur, though perhaps overlapping in date with the latter. Jubayl A church in Jubayl was excavated by the Saudi Department of Antiquities in 1987, but the results have not yet been published. Unofficial reports 18 19 GaK-14261 (AS 103): 1260 ± 80, calibrating to AD 640–970 at 95.4 % probability, or AD 660–830 at 61.7 % (Okada & Numoto 1989: 61). Neither Okada nor Steve states the range of this calibration, which was done by the author using OxCal 3.1. Some of the pottery of Ain Sha’ia belongs to the Samarra Horizon, including splashed lead glazed wares (Okada & Numoto 1989: fig. 15.71, 73 and 75), indicating a date after the start of the ninth century, though the pottery of the lowest levels (1989: fig. 16) is somewhat too sparse to fix its date. 98 (Langfeldt 1994: 32–39) indicate the presence of a church with elaborate stucco, similar to that of the other churches in the Gulf, but of apparently different architecture (see above). The absence of published pottery makes it difficult to define a precise date, though the stucco indicates broad contemporaneity with al-Qusur, Sir Bani Yas and Kharg (Simpson, in press). If the difference in architecture between the Jubayl church and the others in the Gulf is genuine, it raises the possibility that it belongs to the rival Syrian Jacobite or Monophysite Church, which had adherents in the region. According to the Chronicle of Seert, the late sixth century saw a reorganisation whereby Mesopotamia and Bahrain were incorporated into a single bishopric (Potts 1990: 254). Thaj An unexcavated building at Thaj, the foundations of which had been incorporated into a later structure, was identified as a church according to the presence of stone slabs incised with crosses flanking the entrance (Langfeldt 1994: 44–47). There is no dating evidence. Although it seems likely, it cannot be proved that this is a church, though it certainly indicates a Christian presence at Thaj. Jabal Berri Potts reports the discovery of a bronze cross and a mother-of-pearl one from Jabal Berri, 9 km south of Jubayl (Potts 1994). The bronze example closely resembles those of the Church of the East, known from al-Qusur, Hira and Khargt. Settlement remains are reported nearby, so this may represent a separate Christian community close to that of Jubayl, despite its proximity. Hinnah Several stones engraved with crosses are reported from Hinnah, near Thaj. Langfeldt is likely to be correct in interpreting them as grave stones in a Christian cemetery (Langfeldt 1994: 49). Other reports Several stelae found in Bahrain are referred to as ‘Coptic, probably Christian’, but they do not bear Christian symbols such as the cross (Bin Seray 1996: 326–327). In Kuwait a Torpedo Jar sherd with a stamped cross motif was found on the north coast of CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Textual evidence for Christianity in the Gulf There is excellent evidence from various Syriac texts of the Church of the East, and other sources, for the presence of Christians in the Gulf before and during the early centuries of Islam. These sources are discussed elsewhere (Fiey 1979; Potts 1990: 150– 151, 241–247, 252–254, 256–257, 262; Bin Seray 1996; 1997) but are summarised below. Fig. 20. Torpedo Jar sherds from As-Sabiyah Kuwait. Upper: site SB4; incised with Arabic (?) lettering, perhaps part of the word ‘Allah’. Lower: site SB5; stamped with cross shape. Kuwait Bay at as-Sabiyah (Fig. 20.Lower). This was from one of a string of very small sites which were within sight of Failaka and bore pottery comparable to that of al-Qusur (Carter et al. 1999: 46–47). Interestingly, another sherd was found at a nearby site, which bore an eroded incised symbol resembling part of the name ‘Allah’ (Fig. 20.Upper). Although highly speculative, these sherds may suggest the presence of both Christian and Muslim communities in Kuwait at this time. A stamped sherd with a cross motif was also found near the probable church at Thaj (Langfeldt 1994: 47–48, fig. 20). This kind of motif is also seen on glass at Tulul al-Ukhaidir (Finster & Schmidt 1976: Abb. 68g), and stamped on Susa A II pottery (Rosen-Ayalon 1974: fig. 143). Origins The sixth-century Chronicle of Arbela, whose authenticity is disputed, attests to a bishopric at Bet Qatraye at around AD 225 (Potts 1990: 241; Bin Seray 1996: 318–319). Bet Qatraye refers to the area of the north-eastern shores of the Gulf, which had an uncertain border, somewhere east of Qatar, while Bet Mazunaye, encompassed the Oman peninsula (King 1997b: 82; 1997a: 233). Both were under the Metropolitan of Rev Ardashir (Bushehr, Iran), who answered directly to the Catholicos in SeleukiaCtesiphon (Potts 1990: 244). Some time between AD 343 and 346, according to the Life of Jonah, the monk Jonah built a monastery ‘on the borders of the black island’ (Potts 1990: 245; Bin Seray 1996: 319–320; King 1997a: 234). Soon after this, in the time of the Patriarch Tomarsa (AD 363–371), a monk named Abdisho (‘Awdish o) went to ‘an island of Yamama and Bahrain’ called Ramath, baptised the inhabitants and founded a monastery there (Potts 1990: 245; Bin Seray 1996: 320). It has been suggested that this would have been on the island of Abu ‘Ali just north of Jubayl (Potts 1990: 245, n. 275), and also that it may refer to the Christian establishment on al-Qusur, Kuwait (Bernard, Callot & Salles 1991: 145) (see below for discussion). By the time of the establishment of the hierarchy of the Nestorian Church of the East in AD 410 at a synod held at Seleukia-Ctesiphon by Patriarch Mar Isaac, several bishoprics existed in the Gulf. These included the important seat of Deirin on Tarut, and Meshmahig, identified with Muharraq, an island off the main island of Bahrain (Bin Seray 1996: 320, n. 34 and 36), as well as the unidentified diocese of Todoro. Potts speculates that Christianity would have spread to the region through two routes: firstly, via the movement of partly Christianised Arab tribes who were in contact with the Christian centre of al-Hira (central Iraq), including the majority of the 99 R. A. CARTER Abd al-Qays (the inhabitants of the region of al-Bahrain, i.e. northeast Arabia) and sections of the Tamim (Potts 1990: 242). The Christian settlements mentioned in the Syriac sources are largely in Abd al-Qays territory (Bin Seray 1996: 325). Secondly, Christianity could have spread through the missionary activities of the Nestorian church (the Church of the East), which split from the western Syrian Jacobite or Monophysite Church in the fifth century AD (Potts 1990: 242–243). Bin Seray adds that the persecution of Nestorians in Persia by Shapur II (310–379) caused migration into the Arab areas. The sixth and seventh centuries Following a gap until the end of the fifth century, perhaps caused by the Lakhmid’s loss of control of the region to the Kinda (Potts 1990: 247), bishops from Bet Qatraye continue to crop up in the records of the Nestorian Synods until AD 676. Bishops of Hajar (Hagr) and Hatta are mentioned, both settlements on the mainland opposite the Bahrain archipelago (see below for locations), along with bishops of Mazun ⁄ Bet Mazunaye (Bin Seray 1996: 320–321). By the mid-seventh century (c. AD 647) the bishops of Bet Qatraye felt independent enough to attempt to break away from the authority of the church, and the Catholicos of Seleukia-Ctesiphon, Isho’yahb III, unsuccessfully wrote letters and sent two bishops to try to resolve the dispute (Brock 1999: 86–87). The ringleader of the breakaway appears to have been a certain Abraham, bishop of the church at Meshmahig. The bishops of Bet Qatraye eventually went as far as to name their own Metropolitan (Fiey 1979: 210–211), but the schism was eventually resolved by Isho’yahb III’s successor, George I (Giwargis I), who went personally to the Gulf to hold a synod in AD 676 at Darein, Tarut, which was attended by the bishops of Darein, Mazun, Hagar and Hatta (Potts 1990: 261–262; Bin Seray 1996: 323; Jullien & Jullien 2002: 240). Despite these problems, it appears that Bet Qatraye was a centre of Christian scholarship during the seventh century. Brock observes a concentration of highly educated and widely read writers from the area at the time, including Isaac of Nineveh, various Gabriels, Ahob, Dadisho and others (Brock 1999: 88–95). He speculates that there was one or more church or monastery schools in Bet Qatraye ‘whose teaching was on the level of 100 higher education, and thus comparable to that of the famous School of Nisibis’. Christianity in Bet Qatraye during and after the late seventh century George I’s synod at Deirin was the last at which bishops from Bet Qatraye are mentioned. Indeed, the term vanishes from the sources apart from mentions in prayers of the twelfth century (Bin Seray 1996: 324). There is textual evidence, however, that Christianity persisted in Eastern Arabia and the Gulf, and that both the early Islamic rulers and the Abbasids tolerated Christianity in this region (see below). Although Bet Qatraye is not mentioned, it seems that Christian communities persisted in the Gulf during the ninth century. An anecdote concerning a fearsome fish recounted by Michael the Syrian indicates the existence of Christian pearl fishers around Bahrain at about AD 835 (Beaucamp & Robin 1983: 186–187), while in the last decade of the ninth century the patriarchal chronicle of Mari indicates that Christians were still present in the regions of Yamama and Bahrain and were well treated by a rebel leader (Fiey 1979: 211, n. 216a). After the ninth century the evidence for Christians in the Gulf becomes extremely patchy, but there is no doubt that many Christians chose to pay the jizya, the poll tax on non-Muslims, rather than convert. According to Isho’yahb III’s letter to Simeon of Rev Ardashir, as early as the mid-seventh century the Christian community of Mazun was dissolving through conversion, in order to avoid giving up their worldly possessions (Potts 1990: 346). There can be little question that financial as well as political and social incentives would eventually have eroded Christian numbers even in strongholds such as Bet Qatraye. Locations and place names in the texts It is worth comparing the known Christian archaeological sites with the Syriac place names associated with Christians (Fig. 2). Several authors have discussed the relevant nomenclature (Fiey 1979: 209– 219; Beaucamp & Robin 1983: 173–179; Potts 1990: 150, 244–246; Bin Seray 1996; 1997), and only a summary will be presented here. CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM Bet Qatraye This referred to the region of north-eastern Arabia and the islands (Bin Seray 1997: 211–212), but was neither a bishopric nor the seat of a Metropolitan. The various bishops of Bet Qatraye were subject to Metropolitan of Rev Ardashir (Bushehr, Iran), who answered directly to the Catholicos in Seleukia-Ctesiphon (Potts 1990: 244). In AD 676 there is mention of a Metropolitan Thomas of Bet Qatraye, (Fiey 1979: 210–211). This new Metropolitan province, probably based at Meshmahig (see below) was either self-declared following the schismatic ructions detailed in the correspondence of Isho’yahb III (Potts 1990: 261), or deliberately created by Isho ‘yahb III in response to the disobedience of the Christian leaders of the region (Young 1974: 98). It seems that the actions of George I were directed at ending this aberration and returning Bet Qatraye to the authority of the Metropolitan of Rev Ardashir. Mazun Mazun (Bet Mazunaye) was the seat of a bishopric, and referred either to the Oman peninsula or a specific town in Oman, probably Suhar, an identification supported by Mas’udi (Fiey 1979: 215; Potts 1990: 230; Kervran 2004: 341). Together with Bet Qatraye, Mazun was subject to the Metropolitan of Rev Ardashir. Bishops of Mazun attended synods in AD 424, 544, 576 and 676 (Bin Seray 1996: 320–321). Unlike the former, however, it appeared to have only one bishop, hinting at a smaller or geographically restricted Christian population. The border between Mazun and Bet Qatraye is unknown and must have lain somewhere between Suhar and Qatar (King 1997a: 233). Meshmahig This is mentioned as an episcopal seat in AD 410 and 576 (Bin Seray 1996: 320–321). In around AD 647 its bishop, Abraham, was leading the congregation into a split with the Patriarchy. Its bishop was not mentioned as being present in George I’s synod in 676, leading some authors to suspect that the seat and bishop had been elevated (without patriarchal approval) to the short-lived Metropolitanship of Bet Qatraye, under a certain Thomas (Beaucamp & Robin 1983: 184). Meshmahig is also referred to as a source of pearls. It is widely accepted that Meshmahig refers to Muharraq, an island off the north coast of the main island of Bahrain. The identification rests largely on the reasonably detailed descriptions of its location in the Arabic sources, where it takes the form Samahij, and the known existence of a locality on Muharraq still known as Samahij (Beaucamp & Robin 1983: 173–176). Moreover, a modern village on Muharraq is known as adDayr (Potts 1990: 254), meaning monastery. No Christian sites or finds have yet been reported from Muharraq, however. Deirin Deirin (Dayrin, Diren) is identified with the extant village of Darin on Tarut island, opposite Qatif. It was also sometimes written as Darai (Bin Seray 1996: 320 n. 34). Bishops of Darin attended synods in AD 410 and 676, the latter being held by Patriarch George I at Darin itself (1996: 320, 323). Another bishop of Darin, Jacob, addressed twenty questions to the Patriarch Isho’yahb I (582–592) (1996: 321). Ardai Bin Seray interprets this as being an alternative spelling of Darai and thus Deirin (1996: 320 n. 34). Beaucamp and Robin, however, see it as a separate toponym, perhaps related to ancient Arados, usually equated to Muharraq (Beaucamp & Robin 1983: 179), in which case it could be an alternative name for Meshmahig, or a second Christian centre on the island of Muharraq. Hagar Hagar (Hagr, Hajar) was sometimes used an alternative term for the region of Bahrain (north-eastern Arabia and the Bahrain archipelago). It appears also to have referred specifically to an inland town of the region of Bahrain (Fiey 1979: 218), probably Hofuf in the al-Hasa oasis, or perhaps the town of Thaj, c. 90 km inland from Jubayl (Bin Seray 1997: 214). A bishop Isaac of Hagar was mentioned in conjunction with the synod of Ezechiel in AD 576 (Bin Seray 1997: 213)20. A bishop of Hagar also attended the synod of George I in AD 676 at Darin (Bin Seray 1997: 213; 1996: 323). 20 Note that Bin Seray also refers to a bishop Ezechiel of Hagar at this synod, as well as Isaac, presumably erroneously (Bin Seray 1996: 321). 101 R. A. CARTER Hatta Hatta is generally accepted to refer to Qatif (Fiey 1979: 218; Beaucamp & Robin 1983: 171), the coastal town opposite Tarut which traditionally served as an outlet for the al-Hasa oasis. A single bishop responsible for Hagar and Hatta attended the synod of Ezechiel in AD 576, but by the synod of George I at Darin in AD 676 the two had separate bishops (Bin Seray 1996: 321; Fiey 1979: 218). Talun Talun (also Talon, Talwan) was referred to in a letter of Isho’yahb III during the problems of the mid-seventh century. It appears to have a Christian community with monks, though no episcopal seat is directly associated. It is tentatively thought to refer to the main island of Bahrain, the toponym having developed from Tilmun and Tylos (Bin Seray 1997: 219; Beaucamp & Robin 1983: 178). Todoro A bishop of Deirin and Todoro attended the synod of Isaac in AD 410 (Bin Seray 1997: 212). Todoro is not identified, though may equate to the island of Tarut (Fiey 1979: 214), in which case it would be logical for the bishop of Deirin to be described in this way21. Ramath According to the Chronicle of Seert this was an island off Yamama and Bahrain, and thus could be located in any part of coastal eastern Arabia between and including Kuwait and Qatar. It was visited by the monk Abdiso (‘Awdish o) around the mid-fourth century, who baptised the inhabitants and founded a monastery. It is said to have been 68 parsangs from Ubullah (near modern Basra). According to Bernard, Callot and Salles this would place it at al-Qusur, Failaka, Kuwait (1991: 173), but according to Potts it would be 300 km further down the Gulf at the island of Abu Ali, just north of Jubayl (1990a: 245 n. 275), which seems more likely to be correct22. Ruha In the records of George I’s 676 synod there is a reference to the island of ‘Talwan and Ruha’ (Fiey 1979: 214). Talwan is thought to equate to Talun, i.e. the main island of Bahrain (see above). Ruha may therefore have been a separate settlement on the main island. The Black Island and other attestations The ‘black island’, where the monk Jonah is supposed to have founded a monastery in the mid-fourth century, is thought to lie somewhere between Qatar and Oman (Bin Seray 1996: 319–320; Potts 1990: 245; King 1997b: 234). It has been speculated that this foundation may refer to Sir Bani Yas (Potts 1997: 67; 1998b: 64), but there is no evidence yet for occupation of that date on the island. A more fruitful place to search for Jonah’s monastery would be the islands of Ghagha and the Yasats, to the west of Sir Bani Yas and close to Qatar. These have yielded sites and ceramics of an appropriate date (King & Tonghini 1998). Others have speculated that the Black Island was actually one of the bishoprics known from later sources, for example, Ardai or Todoro (Jullien & Jullien 2002: 108). A certain Bishop Sergius of Trihan is recorded as having attended the synod of 676 at Darein (Potts 1990: 150; King 1997b: 234). This is the only attestation of this place name in relation to the Gulf, and it has been suggested that Talun ⁄ Talwan should be read instead. Arabic sources refer to a church in the Abd al-Qays town of Juwatha, in the al-Hasa oasis (Bin Seray 1996: 325). Finally, there have been rumours of at least one church site discovered on Qatar (Elders 22 21 There is some confusion in the secondary sources over whether the bishop of Todoro had a separate diocese or not: Bin Seray referes to dioceses of Ardai (Deirin) and Todoro at the 410 synod in his 1996 article (Bin Seray 1996: 320), but to the diocese of Ardai (Deirin) and Todoro in his 1997 article (1997: 212). 102 Although variable, one parsang equates to around 5.6 km, thus the distance is c. 381 km. Measured as the crow flies this would put Ramath at Potts’s location. Even if the distance is measured along the winding Shatt al-Arab and then the coast of Kuwait, the location of al-Qusur would be far short of 68 parsangs. If measurement were made by land, and a parsang is considered to be an hour’s travel by caravan, al-Qusur would still be too close. CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM 2001: 55), but these turned out to be false (pers. comm. Fran Gillespie). Summary and geographical distribution of toponyms There is a good level of agreement between historians for some of the toponyms, including Meshmahig (Samahij on Muharraq), Hatta (Qatif) and Deirin (on Tarut). There is less certainty or agreement concerning the others, but the equation of Hagar with Hofuf seems reasonable, as does that of Talun with Tilmun, suggesting the main island of Bahrain. As for Mazun, the name appears to have referred to the whole region of the Oman peninsula, and it is likely that its bishop was based at its main town, i.e. Suhar. There is an almost comical discrepancy between the identifications mentioned above (Fig. 2) and the archaeological evidence (Fig. 1). Note also that there are some with multiple identifications:  Deirin, Todoro and Ardai may all equate to or be on the island of Tarut;  Meshmahig and Ardai may equate to or be on the island of Muharraq;  Talun and Ruha may equate to or be on the main island of Bahrain. There is also the problem that the main island of Bahrain, the major population centre of the region throughout history, does not appear to have had its own episcopal seat. In the light of these problems, further data is required before these textual identifications can be made with certainty, including even Meshmahig ⁄ Samahij and Deirin ⁄ Darin. The duplication of place names is extremely common in the region, and philological work is required to establish the significance of similarities between the ancient and modern toponyms. There are no place names that can be assigned convincingly to the monastery at Sir Bani Yas. Leaving aside Hagar, Hatta, Meshmahig and Deirin, most of the remaining names occur too early (Ramath, Todoro, the Black Island) and should be discounted unless Christian remains of the fifth century AD are confirmed on the island. This leaves only Talun and Ruha, but as noted above these have previously been identified with the main island of the Bahrain archipelago. The names of the Christian sites at al-Qusur and Kharg also remain unknown. Conclusions Reconciling the evidence The redating of Sir Bani Yas, al-Qusur and Kharg to the first centuries of Islam leads to an intriguing divergence between the archaeological and the historical evidence. These discrepancies amply illustrate the shortcomings of both the archaeological evidence and the historical sources, but also indicate the significant geographical and chronological range of Christianity in the region. Firstly, the texts very clearly indicate the presence of monks, churches and Christian communities in the Gulf from the fifth century AD or earlier, but no traces of this have yet been confirmed in the archaeological record. Secondly, the archaeological evidence has revealed a period of active church building and apparent expansion in the late seventh ⁄ eighth and ninth centuries AD, but Bet Qatraye and its bishops disappear from the texts after AD 676. Regarding the first problem, it is likely that the early Christian sites simply remain undiscovered in Bahrain and the north-east of Saudi Arabia, the focus of the early missionary efforts. The churches at Jubail and Thaj have not been properly dated, while Tarut, Muharraq and the main island of Bahrain are now very densely populated and urbanised and therefore poorly investigated. Further down the Gulf, the possibility remains that earlier sites on Sir Bani Yas (perhaps SBY-1 and -223) and neighbouring islands (e.g. Ghagha, the Yasats) may harbour Christian remains. Moreover, the nature of the early church in the Gulf remains unknown: it is quite possible that worship took place in people’s homes or in lightly constructed ephemeral buildings during the earlier centuries. The late seventh- to ninth-century remains that have been identified do not reflect the introduction of Christianity but simply a change in the quantity or disposition of resources, evident as a burst of building activity. This burst in the Gulf after the late seventh century is mirrored elsewhere: according to Thomas of Marga, in the early ninth 23 The only faint evidence comes from the small assemblages of SBY-1 and SBY-2, dominated by CLINKY ware. The pottery from these sites was not sufficient to establish a firm dating, however, and CLINKY can date to the Early Islamic Period as well as the Sasanian Period. 103 R. A. CARTER century the Metropolitan Isho’yahb of the monastery of Beth Abhe (near Mosul) pulled down the old church, which had been built from clay (presumably mud brick or pisé) in the mid-seventh century, under the Patriarch Isho’yahb III (648–660) and replaced it with one built with stone and fired bricks bonded with lime (Wallis Budge 1893: 393, 397–400). Thus, an earlier church built of easily degraded material was replaced by one of more solid construction. This appears to have happened under the Patriarch George II (828–835) (1893: 404). Although this is distant from the Gulf, it testifies to building activities in Mesopotamia both in the midseventh century and the early ninth. It is possible, though unproven, that such initiatives originated from the patriarchal authorities, and may therefore have been reflected in rebuilding elsewhere in the territory of the Church of the East. A church has recently been uncovered at Urgut, near Samarkand, which was founded around the ninth century24, and was still flourishing in the tenth (Savchenko 1996; 2005). The second problem, i.e. the disappearance of Bet Qatraye from the texts after the 676 synod, requires further historical research. Either the bishops ceased to attend the synods, or there are large gaps in the historical record. If the former is true it could simply be that George I was not successful in his attempts to bring Bet Qatraye back into the fold at the 676 synod. The Christian communities of the Gulf retained or regained their independence after George’s departure from Darein, under their own self-created Metropolitan. Their absence from the records of the Church of the East may therefore be because they had ceased to be a part of that organisation. A contrasting argument could be made, however: if George I’s synod was wholly successful in laying down a framework for the regulation of the churches of Bet Qatraye, then there would be little need for further interference and therefore no further mention in the historical record. Regarding the gaps in the textual evidence, it is clear that the record is very incomplete, as demonstrated by the apparent absence of any 24 The dating of the Urgut church is still being researched. An account of its excavation was given by Dr Alexei Savchenko at the Christianity in Iraq IV seminar held at SOAS, London, 5th May 2007. 104 mention of the large and flourishing monastic complex on Kharg. It is to be expected that the written record is extremely patchy for this period: very few texts concerning the Church of the East survive, and most of these are primarily concerned with affairs in other areas, the texts having been preserved by communities where Syriac Christianity has survived until recent years, such as in northern Mesopotamia25. The social and economic context of the monastery on Sir Bani Yas Monasteries cannot exist without people, and as noted above, the architecture, location and ceramic assemblage of SBY-9 do not imply a closed community, though it may have been geographically isolated26. It is unlikely that there were sufficient agricultural resources on the island to sustain settlement without external trade, so some kind of revenue must have been generated. There is good evidence that the Christian communities in the Gulf were closely involved in mercantile activities, including maritime trade and the pearl industry. The latter had been an important part of the regional economy at least since the Hellenistic period (Carter 2005b: 143–145). In the sixth century, the Patriarch Isho’yahb I (582–592) delivered a ruling, in response to questions from Bishop Jacob of Deirin (Tarut), that pearlers may work on a Sunday (Bin Seray 1996: 321). There is also the well-known reference in the Chronicle of Se’ert to a certain Ezekiel, who undertook a pearling expedition on behalf of the Sasanian Emperor Khosrau I (531–579) and rose high in his service (Colless 1969-1970: 29). Meanwhile, the Babylonian Talmud (containing texts of the third– sixth centuries) refers to pearls being brought to the port of Meshmahig (Simon 1938: 99), while the 25 26 I have Sebastian Brock to thank for this observation. To give an example of the incompleteness of the record, we know that 200 letters of the Patriarch Timothy were preserved in his day in two volumes, but only fifty-nine have survived to this day, all from the first volume (Young 1974: 129). The only evidence so far for a contemporary site in the Abu Dhabi Islands region is found on the island of Marawah, where a lime-kiln (MR-6.1) was excavated, which yielded radiocarbon dates of 1239 ± 45 BP and 1317 ± 46 BP respectively. These calibrate to AD 630–890 and AD 630–810 respectively, at 95.4 % probability (see http://www.adiasuae.com/radiocarbon.html#marawah). CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM island of Kharg was also renowned for its pearls (Fiey 1979: 196). Significant Christian involvement in trade was not restricted to pearls: Colless refers to sixth- and seventh-century instances of Christian monks who had previously been active as merchants or sailors in the sea-trade between Mesopotamia and India (1969-1970: 32). Other authors stress the importance of mercantile activities and dispersals in the process of evangelisation (Jullien & Jullien 2002: 219). Just as Christian monasteries in inland Arabia had earlier functioned as caravanserais (Langfeldt 1994: 53), so might the monastery at Sir Bani Yas have acted as a maritime staging post, providing support services for merchant shipping. If we accept that the Christian communities of the Gulf were closely involved in pearling and ⁄ or mercantile shipping activities, then the foundation of a community on Sir Bani Yas makes perfect economic sense. The island is located right on the edge of the major pearling banks of the Gulf, and was used in recent centuries as a base for pearlers; it is just 30 km from the island of Dalma, also used as a base for pearling fleets, a seasonal pearl market and a centre for the administration of the pearl dive (Carter 2005b: 176–177). It is possible that the early Muslim rulers preferred not to interfere with a lucrative Christian-dominated pearl trade, being content to reap its rewards indirectly through taxation. This would have encouraged the survival of Christian communities in comparatively isolated areas of the Gulf, such as Sir Bani Yas. Un éclat du christianisme? The theory that there was a burst of Christian activity in the Gulf region (un éclat du christianisme) in the latter part of the seventh century was developed by Beaucamp and Robin, and recently highlighted by Kennet (Beaucamp & Robin 1983: 186; Kennet 2007). It is based on the activities of George I in AD 676, who visited the Gulf to resolve the tensions that had led to the separation of Bet Qatraye from the authority of the Patriarch. Detailed canons resulted from his synod at Darein, including rules regarding the ordination of bishops and other clergy, and the construction of churches and monasteries (Beaucamp & Robin 1983: 184). These are clearly not the activities of a moribund Christian community. Further canons cover the behaviour of the faithful, while others are concerned with the maintenance of episcopal authority in legal matters between Christians, independently of the secular authorities, i.e. the ruling Muslims. The canons also reveal that some Christians held official positions under Muslim rule, for example as tax collectors: they are told to exempt bishops from the customary poll tax on non-Muslims (1983: 186). Moreover, as noted by Brock, Bet Qatraye was a centre of Christian scholarship during the seventh century and probably the home of at least one high-level church or monastery school (Brock 1999: 95). Again, this shows that the Gulf had a vibrant and thriving Christian community during the early years of Islam. There is further evidence for a high level of Christian activity in the region during the next century. According to Colless the apogee of the missionary activity of the Church of the East was during the later eighth and early ninth centuries AD, under the Patriarchate of Timothy I (780–823) (Colless 1969-1970: 31). In his Letter to the Maronites (792 ⁄ 3) he refers to his jurisdiction in ‘Babel, Pars and Athur’ as well as among the Indians, Chinese, Tibetans and Turks. Significantly, he refers to monks passing by sea to India and China, a route which would have taken them through the Gulf and thus potentially into contact with surviving Christian monasteries and secular communities. It is at this time, in AD 781, that the famous inscription was erected in northern China at Sigan-Fu (Xi’an, or Hsian-Fu), which commemorated the recognition of the religion by the Tang emperor in AD 635 (Mathews 2000: 932). The cross depicted on this stele is stylistically very similar to a stucco example from Kharg (Jullien & Jullien 2002: 109 n. 291; Steve 2003: pl. 10.1). It is not surprising that Christianity flourished in the region in the late seventh, eighth and ninth centuries, as there is evidence for periods of good relations with the local Muslim rulers. As far as Isho’yahb III was concerned in a letter of AD 647 to the separatist Metropolitan Simeon of Rev Ardashir, the threat to the Church came not so much from the Muslim conquerors but from separatist elements from within his own communion. This letter amply demonstrates his good relations with the Muslims: 105 R. A. CARTER These Arabs, to whom God for the time being has given the Empire of the world, are also, as you know, very close to us; and not just because they do not attack the Christian religion, but they praise our faith, honour the priests and the saints of the Lord and award benefits to the churches and monasteries (Potts 1990: 260 and n. 389). At the time of the AD 676 synod Christians still held administrative positions as tax collectors (Potts 1990: 262). Following this there was a downturn in the Church’s fortunes, with discriminatory measures being enforced upon Christians in the Islamic realm (Baumer 2006: 151). By the last quarter of the eighth century, however, relations between the Church of the East and the Abbasid court appear to have improved: Patriarch Timothy I (780–823) earned the respect and friendship of the Caliph al-Mahdi (775– 785) and was subsequently close to the court of Harun al-Rashid (786–809), having a circle of Christian friends with influence at the Abbasid court (Young 1974: 134–5, 138–139; Baumer 2006: 154; Steve 2003: 152). The Caliph Ma’mun (830–833) was also said to be favourable towards Christians (Baumer 2006: 154; Steve 2003: 153). The Abbasid rulers relied heavily on Christians, who were experienced in translating the Gospels from Greek into Aramaic, to translate the Greek philosophers and medical authors from Greek into Arabic, via Aramaic (Baumer 2006: 157). The redating of the SBY-9 assemblage to the late seventh or early eighth century, and a similar date for the foundation of the al-Qusur church, fits squarely with Beaucamp and Robin’s assertion, and demonstrates the diversity of the region and tolerance of its Muslim rulers during the first centuries of Islam. The later monastery at Kharg may relate to the highly active and expansionist policies of Timothy I, effectively amounting to a second éclat du christianisme. The eighth and ninth centuries are sometimes referred to as the Golden Age of the Church of the East (Baumer 2006: 156), and this is reflected in the archaeological record of the Gulf. Acknowledgements Firstly I would like to thank Dr Sami El-Masri and ADACH (Abu Dhabi Authority for Culture and Heritage), for allowing the material to be studied and putting forward the sponsorship for this analysis. Thanks are due to Peter Hellyer and Mark Beech for expediting this process, and to Joe Elders for providing stratigraphic information. I would also like to express my gratitude to Derek Kennet, St John Simpson and Seth Priestman for allowing me to see and cite their forthcoming works, and for their ever insightful comments on the SBY material and its dating. Thanks also to Sebastian Brock for his comments on Syriac sources and to Fran Gillespie for explaining the background to the false claims of a Christian church being found in Qatar. Extra thanks are due to Derek Kennet for allowing the use of the SBY material he recorded in Ras al-Khaimah. Thanks also to Lloyd Weeks for helping me with the radiocarbon dating model, and to Alexei Savchenko for his insights into the expansion of the Church of the East in the Gulf and in Central Asia. Pottery drawings are by Robert Carter and Derek Kennet, with digital inking by Robert Carter. References Baumer, C 2006. The Church of the East. An illustrated history of Assyrian Christianity. London & New York: I.B. Tauris. Beaucamp, J., Robin, C 1983. L’évêché nestorien de Mâšmahı̂g dans l’archipel d’al-Bahrayn. BBVO 2: 171–196. Beech, M. 2004. In the Land of the Ichthyophagi: Modelling fish exploitation in the Arabian Gulf and Gulf of Oman from the 5th millennium BC to the Late Islamic period (Oxford, BAR Int Ser S1217 [=Abu Dhabi Islands Archaeological Survey Monograph 1]). 106 Beech, M., Cuttler, R., Moscrop, D., Kallweit, H., Martin, J 2005. New evidence for the Neolithic settlement of Marawah Island, Abu Dhabi, United Arab Emirates. PSAS 35: 37–56. Bernard, V., Salles, J-F 1991. Discovery of a Christian church at al-Qusur, Failaka (Kuwait). PSAS 21: 7–21. Bernard, V., Callot, O., Salles, J-F 1991. L’église d’al-Qousour Failaka, État de Koweit. Rapport préliminaire sur une première campagne de fouilles, 1989. AAE 2: 145–181. Bin Seray, H.M. 1996. Christianity in East Arabia. Aram 8: 315–332. Bin Seray, H.M. 1997. The Arabian Gulf in Syriac sources. New Arabian Studies 4: 205–232. Brock, S 1999. Syriac Writers from Beth Qatraye. Aram 11: 85–96. Calvet, Y 1998 Monuments paléochrétiens à Koweit et dans la région du Golfe. Orientalia Christiana Analecta 256: 671–699. Carter, R.A. 2003. Tracing Bronze Age trade in the Gulf: Evidence for waystations of the merchants of Dilmun CHRISTIANITY IN THE GULF DURING THE FIRST CENTURIES OF ISLAM between Bahrain and the Northern Emirates. Pages 124–131 in D. Potts, H. Al-Naboodah & P. Hellyer (eds), Archaeology of the United Arab Emirates. Proceedings of the First International Conference on the Archaeology of the U.A.E. (London, Trident Press). Carter, R.A. 2005a. The Pottery. Pages 107– 192 & 397–451 in T. Insoll (ed.), The Land of Enki in the Islamic Era. Pearls, Palms and Religious Identity in Bahrain (London, Kegan Paul). Carter, R.A. 2005b. The history and prehistory of pearling in the Persian Gulf. JESHO 48: 139–209. Carter, R.A., Crawford, H.E.W., Mellalieu, S., Barrett, D 1999. The Kuwait ⁄ British Archaeological Expedition to as-Sabiyah: Report on the first season’s work. Iraq 61: 43–58. Colless, B.C. 1969-1970. The traders of the pearl. Abr Nahrain 9: 17–38. Connan, J., Carter, R.A. 2007. A geochemical study of bituminous mixtures from Failaka and Umm an-Namel (Kuwait), from the Early Dilmun to the Early Islamic period. AAE 18: 139–181. Elders, J. 2001. The lost churches of the Gulf: New discoveries on the islands of Sir Bani Yas and Marawah, Abu Dhabi Emirate. PSAS 31: 47–57. Elders, J 2003. The Nestorians in the Gulf: Just Passing Through? Recent Discoveries on the Island of Sir Bani Yas, Abu Dhabi Emirate, U.A.E. Pages 230–236 in D. Potts, H. Al-Naboodah, P. Hellyer (eds), Archaeology of the United Arab Emirates. Proceedings of the First International Conference on the Archaeology of the U.A.E. (London, Trident Press). Fiey, J.M. 1979. Communautés syriaques en Iran et Irak des origines à 1552. London: Variorum Reprints. Finster, B., Schmidt, J 1976. Sasanidische und frühislamische Ruinen im Iraq. BaM 8. Finster, B., Schmidt, J 2005. The origin of ‘desert castles’: Qasr Bani Muqatil, near Karbala, Iraq. Antiquity 79: 339–349. Gachet, J 1998. Akkaz (Kuwait), a site of the Partho-Sasanian period. A preliminary report on three seasons of excavation (1993–1996). PSAS 28: 69– 79. Hardy-Guilbert, C., Rougeulle, A 2003a. La céramique et les verres du monastère. Pages 131–132 in Steve 2003. Hardy-Guilbert, C., Rougeulle, A 2003b. Les fragments de stucs décoratifs. Pages 121–130 in Steve 2003. Jullien, C., Jullien, F 2002. Apôtres des Confins. Processus missionnaires chrétiens dans l’Empire iranien. Bures-sur-Yvette: Groupe pour l’Étude de la Civilisation du Moyen-Orient. Jullien, C., Jullien, F 2003. Le monachisme dans le Golfe Persique: six siècles d’histoire. Pages 155–183 in Steve 2003. Kennet, D. 1994. Jazirat al-Hulayla — early Julfar. JRAS 4 ⁄ 2: 163–205. Kennet, D. 2001. An Archaeological Study of the Sasanian and Islamic Periods in Ras alKhaimah (U.A.E.). Unpublished PhD dissertation, University of London. Kennet, D 2002. Sasanian pottery in southern Iran and eastern Arabia. Iran 40: 153–162. Kennet, D 2004 Sasanian and Islamic Pottery from Ras al-Khaimah. Classification, chronology and analysis of trade in the Western Indian Ocean (Oxford, BAR Int Ser 1248). Kennet, D 2007. The decline of eastern Arabia in the Sasanian period. AAE 18: 86–122. Kervran, M 2004. Archaeological research at Suhar 1980–1986. JOS 13: 263–381. Kervran, M., Hiebert, F.T., Rougeulle, A 2005. Qal’at al-Bahrain: A trading and military outpost: 3rd millennium B.C.– 17th century A.D. Turnhout: Brepols. King, G.R.D. 1997a. A Nestorian Monastic Settlement on the Island of Sir Bani Yas, Abu Dhabi: a Preliminary Report. BSOAS 60: 221–235. King, G.R.D. 1997b. The history of UAE: The eve of Islam and the Islamic period. Pages 74–94 in E. Ghareeb, I. Al-Abed (eds), Perspectives on the United Arab Emirates (London, Trident). King, G.R.D. 1998. The Abu Dhabi Islands Archaeological Survey. Season 1. London: Trident Press. King, G.R.D., Tonghini, C 1998. The western islands of Abu Dhabi Emirate. Notes on Ghagha. Pages 117–141 in C.S. Phillips, D.T. Potts, S. Searight (eds), Arabia and its neighbours: Essays on prehistorical and historical developments (Turnhout, Brepols). King, G.R.D., Dunlop, D., Elders, J., Garfi, S., Stephenson, A., Tonghini, C 1995. A report on the Abu Dhabi Islands Archaeological Survey (1993–4). PSAS 25: 63–74. Langfeldt, J.A. 1994. Recently discovered early Christian monuments in Northeastern Arabia. AAE 5: 32–60. Mason, R.B., Keall, E.J. 1991. The ‘Abbasid glazed wares of Siraf and the Basra connection: Petrographic analysis. Iran 29: 51–66. Mathews, E.G. 2000. Nestorian Monasticism. In W.M. Johnstone (ed.), Encyclopedia of Monasticism (Chicago & London, Fitzroy Dearborn Publishers). Northedge, A., Kennet, D 1994. The Samarra Horizon. Pages 21–35 in E. E. Grube (ed.), Cobalt and Lustre: The first centuries of Islamic pottery. The Nasser D. Khalili Collection of Islamic Art, Volume 9 (London, Nour Foundation in association with Azimuth Editions and Oxford University Press). Okada, Y 1992. Ain Sha’ia and the early Gulf churches: An architectural analogy. Al-Rafidan 13: 87–93. Okada, Y., Numoto, H 1989. Excavations at Ain Sha’ia ruins and Dukanin caves, Pt. III: Fortified building, Site F. Al-Rafidan 10: 35–61. Orton, C., Tyers, P., Vince, A 1993. Pottery in Archaeology. Cambridge: Cambridge University Press. Patitucci, S., Uggeri, G 1984. Failakah: insediamenti medievali islamici: ricerce e scavi nel Kuwait. Rome: L’Erma di Bretschneider. Potts, D.T. 1990. The Arabian Gulf in Antiquity, Volume II: From Alexander to the Coming of Islam. Oxford: Clarendon Press. Potts, D.T. 1994. Nestorian Crosses from Jabal Berri. AAE 5: 61–65. Potts, D.T. 1997. Before the Emirates: An archaeological and historical account of developments in the region c. 5000 BC to 676 AD. Pages 36–73 in E. Ghareeb, I. Al-Abed (eds), Perspectives on the United Arab Emirates (London, Trident Press). Potts, D.T. 1998a. Namord Ware in Southeastern Arabia. Pages 207–220 in C.S. Phillips, D.T. Potts, S. Searight (eds), Arabia and its neighbours: Essays on prehistorical and historical developments (Turnhout, Brepols). Potts, D.T. 1998b. Seas of Change. Pages 44–67 in P. Hellyer (ed.), Waves of Time. The Maritime History of the United Arab Emirates (London, Trident). Priestman, S 2005. Settlement and Ceramics in Southern Iran: An Analysis of the 107 R. A. CARTER Sasanian and Islamic Periods in the Williamson Collection. Unpublished PhD dissertation, University of Durham. Rosen-Ayalon, M 1974. La poterie islamique. Paris: MDP 50. Sasaki, T 1996. Umayyad and Abbasid finds from the 1994 excavations at Jazirat al-Hulayla. Bulletin of Archaeology, The University of Kanazawa 23: 179– 222. Sasaki, T., Sasaki, H 1996. 1995 excavations at Jazirat al-Hulayla. Bulletin of Archaeology, The University of Kanazawa 23: 37–177. Sasaki, T., Sasaki, H 1998. 1998 Excavations at Jazirat al-Hulayla. Bulletin of Archaeology, The University of Kanazawa 25: 118–169. Savchenko, A 1996. Urgut revisited. Aram 8: 333–354. Savchenko, A 2005. Excavations in Urgut August–October 2005. Progress Report. Kiev: The East Sogdian Archaeological Expedition. Simon, M 1938. Rosh Hashanah. London: The Soncino Press. 108 Simpson, S.J. 1992. Aspects of the archaeology of the Sasanian period in Mesopotamia. Unpublished PhD dissertation, Oxford University. Simpson, S.J. 2003. From Mesopotamia to Merv: reconstructing patterns of consumption in Sasanian households. Pages 347–375 in T. Potts, M. Roaf, D. Stein (eds), Culture through objects: Ancient Near Eastern studies in honour of P.R.S. Moorey (Oxford, Griffith Institute). Simpson, S.J. 2007. Bushire and beyond: some early archaeological discoveries in Iran. Pages 153–165 in E. Errington (ed.), From Persepolis to the Punjab. Exploring ancient Iran, Afghanistan and Pakistan (London, British Museum Press). Simpson, S.J. in press. Christians on Iraq’s desert frontier. in H. Baker, E. Robson, G. Zólyomi (eds), Your Praise is Sweet. A memorial volume presented to Jeremy Allen Black by colleagues, students and friends (Oxford, Griffith Institute). Stern, B., Connan, J., Blakelock, E., Jackman, R., Coningham, R., Heron, C in press. From Susa to Anuradhapura: Exchange in bitumen-coated ceramic vessels between Iran and Sri Lanka from the third to the ninth centuries AD. Archaeometry. Steve, M-J 2003. L’ı̂le de Kharg. Une page de l’histoire du Golfe persique et du monachisme oriental. Neuchâtel: Civilisations du Proche-Orient. Talbot Rice, D 1932. The Oxford Excavations at Hira, 1931. Antiquity: 276–291. Tomber, R 2007. Rome and Mesopotamia — importers into India in the first millennium AD. Antiquity 81: 972–988. Wallis Budge, E.A. 1893. The Book of Governors by Thomas, Bishop of Marga: Volume II: English Translation. London: Kegan Paul. Whitehouse, D 1979. Islamic glazed pottery in the Arabian Gulf: The ninth and tenth centuries. AIUON 39: 45–61. Young, W.G. 1974. Patriarch, Shah and Caliph. Rawalpindi: Christian Study Centre.