In the central Middle Ages, Bari was the most important mercantile port on the southern Adriatic coast. The city, with its majority Lombard population, had briefly been an Arab emirate in the 9th century and became the capital of the Byzantine territories in the Italian peninsula in 985. Less than a century later, it was conquered by Norman forces and absorbed into what would later become the Kingdom of Sicily. There are significant remains of the medieval fabric, mostly from the Norman Period, complemented by tantalising Byzantine fragments. The city is fruitful territory for those historians of medieval culture who are interested in peripheries and border areas. Medieval Bari was a border zone between the Byzantine Empire and the Latin West, and at times a border between the Islamic and Christian cultures of the Mediterranean. As such, it is also a disciplinary border zone for art historians, fitting awkwardly into narratives of both Byzantine and Romanesque art. The aim of this article is to explore this cross-cultural dynamic through an examination of a single motif: pseudo-Arabic, a form of ornament derived from Arabic script but illegible, either because the letters or letter-like shapes do not form words, or because the words or letters have been stylised or decorated to such an extent that they lose their meaning as language and instead function as a pattern. Despite some very visible and prominent monumental examples, pseudo-Arabic in medieval Puglia is often absent from existing scholarly discourse, and when it is acknowledged, it is often marginalised to footnotes and short descriptions. This may be partly attributed to local historians’ reticence to acknowledge such bold Islamic influence on Christian space, but it is equally due to the challenges that come from engaging with such an enigmatic motif. Aniconic ornament is marginalised within the story of Western art, both because figurative images are seen as more important and because, within conventional methodologies, it is difficult to assign meaning to non-figurative images. In this paradigm, pseudo-Arabic is given little attention, along with other non-figurative motifs, such as vine scrolls and geometric designs. Puglia is a region rich in well-preserved medieval churches from the 11th and 12th centuries, and yet it has been largely neglected by art historians, in large part because of a fascinating preference for non-figurative and non-narrative imagery, particularly in sculpture. We should remember the aspects of visual culture that the modern discipline of art history tends to classify as marginal—floors; ceilings; non-figurative designs; certain media such as textiles and metalwork—were not at all marginal to a medieval audience. What a modern eye may dismiss as ‘merely decorative’, the medieval eye may have been drawn to as precious and meaningful. Pseudo-Arabic details are therefore worthy of being placed at the heart of art historical discourse.

The motif features in both Islamic and Christian contexts in the central Middle Ages. In Islamic art, pseudo-inscriptions developed from the 9th century onwards, at first as a stylisation of legible inscriptions and later as an independent decorative motif (Ettinghausen, 1976). Early examples are found on a variety of media, such as pottery and wall painting (Ettinghausen, 1976). Very quickly, the motif found its way into Christian art. The earliest known examples in Byzantine art are the architectural ceramics from the church of Constantine Lips in Constantinople, which date from the very early 10th century (Gerstel & Lauffenburger, 2001: 31). In the 10th and early 11th centuries, there was a veritable explosion of the motif in Greece. Most notably, it is found in abundance at the monastery of Hosios Loukas in Boeotia, in mosaic, stone relief sculpture and terracotta decoration on the façades (Walker, 2015). Unsurprisingly, it is also found in France, Spain and Italy, appearing in objects such as wooden doors, manuscripts, metalwork, and mastic inlay, to name just a few examples (Cahn, 1974: 11–34; Harris, 1995; Pedone & Cantone, 2013). In 20th century scholarship, these motifs are labelled interchangeably as kufesque and pseudo-kufic. In recent years, this has given way to the more general term pseudo-Arabic, which I will use here. The historiography of the motif in Christian contexts has recently been insightfully analysed by Silvia Pedone and Valentina Cantone, who have grouped the scholarship into three categories (Pedone & Cantone, 2013). These three categories can be summarised as the exotic, the archaic and the apotropaic. To begin with apotropaic interpretation, in the mid 20th century, it was argued that the motif developed from the stylisation of the word Allah (God), as a protective device, and was understood in both Islamic and Christian contexts as an apotropaic invocation of divine power (Erdmann, 1954). Later, others interpreted the appeal of the motif as the allure of the exotic and as an archaicising trend (Cutler, 1999, 2002; Ferber, 1975). Recent studies have, rightly, shifted away from such overarching theories, towards case studies that demonstrate how the motif was used in specific objects and buildings in response to local circumstances (Johns, 2015; Walker, 2015). This article will build on this local, contextual approach with an examination of the pseudo-Arabic apse mosaic in the Basilica of San Nicola, executed in the early decades of the 12th century. The article is divided into two parts. The first part focuses on the examples of pseudo-Arabic in and around Bari, which date from the Byzantine era (from the 10th century until the Normans took the city in 1070), in order to ensure that the San Nicola mosaic is situated within its local context. The main section of the article, which deals with the San Nicola mosaic, looks in detail at the events that took place immediately before the creation of the pavement: the victory of the southern Italian contingent of the First Crusade in the battle of Antioch in 1098; Bohemond I’s subsequent donation of a Muslim tent to the basilica; and the Council of Bari, which took place in the church in same year. In doing so, this article contributes to the picture historians are building up of how motifs circulated in the medieval Mediterranean, and how Islamic objects and designs were received by Christian audiences.

This section will discuss three known examples of pseudo-Arabic from Byzantine Bari and the surrounding area: a wall painting from Palazzo Simi; a tombstone excavated at the cathedral; and a set of relief panels from Bitonto Cathedral (Figures 1, 2, 3, 4).

Pseudo-Arabic is found on a painted textile, in the ruined church, usually referred to as the church in Strada Lamberti (Figure 1). The church was originally a small, cross-in-square structure built and decorated in the 10th century. Today, only the lower portion of the sanctuary wall survives decorated with wall paintings of church fathers, in pairs at the sides of the main apse (Depalo & Cioce, 2008). The ruins are preserved in the basement of the archaeological museum Palazzo Simi. This area of medieval Bari was home to the Armenian community, and the church was probably dedicated to the Armenian saint, Gregory the Illuminator (Lavermicocca, 2008: 62). One of the four figures wears an epitrachelion decorated with pseudo-Arabic. An epitrachelion is the Byzantine equivalent of the Western stole, a part of the vestments that symbolise the wearer’s ordination to the priesthood. It takes the form of a long, thin scarf that is sewn or buttoned together almost to the top, leaving a large enough opening for the priest to put it over his head, leaving the ends hanging straight down his front. It was often decorated with crosses, but could also be embroidered with figurative and other designs, as modelled by St Ignatius of Antioch in the mosaics of the Katholikon at Hosios Loukas (Parani, 2008: 413–6). In the church in Strada Lamberti, only the figure’s lower legs and feet have survived, so we can see only the ends of the epitrachelion, sewn together, hanging just above his feet, each end decorated with a single unit of pseudo-Arabic. Each unit is composed of two vertical, floriated ‘letters’ (evocative of the Arabic letter lam), joined together by an arch. The wall painting can be dated fairly securely to the Byzantine Period due to its stylistic similarity to a similar painting of a bishop in Bari’s pre-Norman cathedral (the painting and the early cathedral survived as a crypt beneath the current cathedral) (Bertelli, 2004: 91–3).

The story of pseudo-Arabic on textiles begins in the Islamic world with tiraz silks. The word tiraz refers to silks manufactured within the royal palace of the caliph and inscribed with his name (tiraz can be both the inscription itself or the workshop within the palace). Such silks were worn at court and distributed by the caliph as diplomatic gifts (Blair, 1998: 164–8). Over the course of the 10th to the 12th centuries, there was an increasing tendency for such inscriptions to be accompanied by bands of decorative motifs and to become highly stylised, often descending into pseudo-inscriptions (Blair, 1997b: 105, 1998: 164–7; Contadini, 1998: 62–9). Also during this period, such textiles became detached from the royal court and increasingly emerged as a commodity in international trade (Bierman, 1997). By the 12th century, if not earlier, inscribed tiraz fabrics could be made to order according to personal specifications. An example is a transaction from 12th century Egypt, in which an Indian merchant was able to order a turban cloth inscribed with his son’s name (Stillman, 1997: 49). Such Islamic silks were popular in Byzantium, thanks to both trade and diplomacy (Walker, 2012: 22–31, 34), and Byzantine weavers soon began copying tiraz (Muthesius, 1995: 89–93; Parani, 2003: 54), for example the shroud of St Potentianus, with its pseudo-Arabic medallions (Gonosova, 1997: 505–6).

Depictions of pseudo-Arabic on clothing and other textiles are found across the Byzantine Empire. Cappadocia is particularly rich with examples: on a depiction of the Mandylion in the apse of Chapel 21 in Göreme; on the vestments of a bishop on the north side of the apse in the monastery of Eski Gumus, and again in the same apse; on a textile draped over Christ’s throne and on the cloth covering the hands of an angel; in the Last Supper scene at the Karanlik Kilise; and, charmingly, on the socks of the Magi in the same church (Jolivet Lévy, 1991; Lidov, 2007: 203). On the north-western fringes of the Byzantine Empire, there are examples on a fictive wall hanging at the Cella Trichora at Pécs in modern-day Hungary and on the mandylion in Ohrid.

The depiction in Strada Lamberti was probably inspired by real textiles decorated with pseudo-Arabic, perhaps it is even an accurate depiction of real vestments used in this particular church. The pseudo-Arabic motif could have been rendered in appliqué (as in the 11th century ‘Starmantle’ of Emperor Henry II) (Ganz, 2014), woven into the fabric (as in the St Potentianus silk) or embroidered onto it (as in the Sicilian mantle of Roger II) (Bauer, 2006; Tronzo 2001). Textiles with woven designs tend to feature motifs repeated over a large area, so the single unit in Strada Lamberti is probably embroidery or appliqué. Weaving requires equipment and a highly-skilled workforce and therefore usually took place in large workshops in major cities where the craft was well-established, such as Cairo and Constantinople (Woodfin, 2012: xxv–xxvi). In contrast, embroidery and appliqué are easier to achieve in small workshops (Woodfin, 2012: xxv–xxvi). The pseudo-Arabic motif could have been added to a garment which was made in Bari, or it could have arrived ready-made, as part of Barese merchants’ trade with other areas of the Mediterranean. Even if the vestment was made in Bari, it is most likely that the motif arrived in Bari on a portable object, like a textile. It shows us how objects like textiles, both valuable and portable, were key to transmitting pseudo-Arabic across the Mediterranean. The wall painting in Strada Lamberti also demonstrates that, once in Bari, the designs on those portable objects were transmitted to other, more monumental media.

The second example is on a fragment of relief sculpture now on display in the south wing of Bari Castle (Figure 2). This object, which began life as a late antique tombstone, has a somewhat complex biography. It was excavated from beneath the building known as the trulla, the circular sacristy adjacent to the cathedral, in the 1990s (Ciminale, 2008). The piece was discovered underneath the mid 20th century floor, along with other fragments of sculpture with dates ranging from the Byzantine to the early modern period (Pellegrino & Belli D’Elia, 1996: 210–7).1 The slab is thought to be proconnesian marble, meaning it was quarried on the eponymous island in the Sea of Marmara (Zorzi, 2007: 40, note 16). The relief decoration is divided into three parts: at the bottom is a pattern of interlaced circles, in the centre is a Greek inscription and at the top is a band of pseudo-Arabic. Niccolò Zorzi (2007: 41–3) has shown that the inscription comprises verses from the books of Job and Genesis, which have funerary connotations, leading to the conclusion that the fragment was part of a tombstone.2 The circular pattern is found in late antiquity and can be compared to a similar motif on the mosaic pavements at Ravenna and at Basilica A at Nicopolis on the west coast of Greece. Likewise, the closest parallels for the script are from the same period. Consequently, both Zorzi (2007: 55) and Gioia Bertelli (2010c) have concluded that the tombstone was made in the 5th or 6th century.3 Since pseudo-Arabic does not appear until at least the early 10th century, it must have been added at a later date. This dual dating is supported by material evidence: there is a distinct difference in the two phases of carving. The earlier Greek inscription and interlace ornament is incised into the marble (with the letters and lines forming grooves in the stone), whereas the pseudo-Arabic is sculpted in relief (the Arabic ‘letters’ are raised from the excised background).

This leads us to pose two questions: first, where was the 6th century tombstone made and used? And second, in what circumstances was the pseudo-Arabic added? The Greek inscription allows us to discount the possibility that the tombstone was made locally. Greek was not used in late antique Puglia. The city of Canosa di Puglia, which is close to Bari and is replete with 6th century archaeology and plenty of Latin inscriptions, has no Greek epigraphy.4 In Bari itself, we find no Greek epigraphy until the beginning of Byzantine rule in the 10th century. The presence of Greek, together with marble from the Sea of Marmara, suggests that the tomb was made in or near Constantinople and brought to Bari, where the pseudo-Arabic was added, during the period of Byzantine rule in the 10th or 11th century. The tombstone is not the only piece of sculpture to make such a journey. The 11th century monuments of Bari are peppered with reused late antique sculpture, much of which may have been brought from the Eastern Mediterranean (Todisco, 1994). For example, the bases of the columns that support the south porch of the cathedral have been reused from a 6th century building (Belli D’Elia, 1974: 107–9; Bertelli, Fornaro & Iorio, 1994).5 In Byzantine Bari, the residence of the Byzantine governors was to the north of the cathedral. This complex of buildings, which included churches, workshops, a military barracks and administrative offices, is known as the Court of the Catapan. After the end of Byzantine rule, the Basilica of San Nicola was built on the site. A number of pieces of sculpture from the court have survived, were reused in the basilica and are now either in situ in the church, or housed in the adjacent museum. They include some of the capitals in the crypt, which were carved in the Eastern Mediterranean, possibly Constantinople, in the 6th century and brought to Bari for the construction of the Court of the Catapan (Bertelli, 1991). It is most plausible that the tombstone arrived from Constantinople along with these capitals (or at roughly the same time). Since the tombstone was broken and used as rubble to level the floor of the trulla, we do not know the site of the grave, but we can assume that it was either at the cathedral or, more likely, in one of the churches within the Court of the Catapan. The Greek inscription would perhaps have been out of place in the Latin-rite cathedral. In the San Nicola Museum there are three pieces of Greek epigraphy, which came from the Court of the Catapan. The first is an inscription commemorating the reconstruction of the court complex by the Catapan, Basil Mesardonites, in 1011 (Guillou, 1970). The second is the tombstone of Basil of Mersin, a Byzantine official who died in 1075 and whose sarcophagus, with its Greek inscription, was later inserted into the external wall of the basilica (Bertelli, 2010b). The third is a small capital, originally from a templon screen bearing the monogram of a patrikios called Leone (Bertelli, 2010a). The tomb fragment adorned with pseudo-Arabic, which is now in the castle, slots perfectly into this milieu of the ruling elite in Byzantine Bari. The deceased was probably a Byzantine official who was buried in one of the churches in the Court of the Catapan. The pseudo-Arabic has been added to the tombstone, in a band along the top of the inscription and might have originally encircled the entire slab, as on the roughly contemporary tombstone of St Luke of Stiris at Hosios Loukas (Bertelli, 2010c; Connor, 1991: 71).

The mid 12th century cathedral of Bitonto lies 10 kilometres inland from our other examples in Bari. Excavations beneath the cathedral uncovered parts of its early Christian predecessor, paved with a 10th century mosaic. Neatly stacked in a chamber in the south-west corner, archaeologists discovered 52 limestone relief blocks, which could be assembled to form three small doorways (Belli D’Elia, 2006; Depalo, 1999; Spagnoletta, 1999). They are decorated with animal and vegetal designs, carved in the style of the sculptor Acceptus, who was trained in the local Lombard tradition and was active in northern Puglia in the first half of the 11th century. Among them are two pseudo-Arabic panels (Figures 3 and 4). The pseudo-Arabic ornament is arranged in two horizontal rows, at right angles to the curve of the arch, with three pseudo-letters in each row. In the other animal and vegetal panels, the sculptor has expertly fitted the designs into the space available with skilful and confident carving. In the pseudo-Arabic panels, however, the carving is cruder and the spacing of the pseudo-letters is uneven and awkward. This is brought into sharper focus when compared to the tombstone in Bari Castle, where the pseudo-letters form a delicate and uniform ribbon of ornament. The use of pseudo-Arabic at Bitonto, in short awkward horizontal rows, does not follow the convention of how the motif is usually used in Byzantine architectural sculpture. In Greece, where pseudo-Arabic is most commonly found in sculptural form in liminal zones of churches, the pseudo-inscriptions form continuous bands that wrap around entire buildings, doorways, templon screens and tombs. This can be seen in Figure 5, a fragment of an enclosure panel from a templon at Hosios Loukas. In contrast, the sculptor at Bitonto has made use of the concept of horizontal bands but without the harmonious repetition. The artist has interrupted the flow of decoration along the curve of the doorway in a way that seems to fundamentally misunderstand the way the ornament is usually used. This may be a Lombard artist’s experimentation with a new motif that he has not employed before. He had probably been trained, either in the north of Puglia where Acceptus worked, or inland, perhaps in Benevento, the capital of the Lombard principality. Perhaps he had recently come into contact with a Byzantine pattern book or luxury objects from Bari, such as textiles.6

These three examples from Byzantine Bari and its hinterland establish that pseudo-Arabic was part of Pugliese culture before the arrival of the Normans. Its use in Bari was facilitated both by international trade in luxury objects and by Byzantine rule, which brought trends from Constantinople and other areas of the empire. The question of how artists and viewers perceived the motif remains a challenge to answer. The merchant community in Bari were well-travelled, they relied on trade with Muslim merchants for a portion of their livelihood and were probably sufficiently literate to associate the motif with Arabic script (Han, 2017: 266; Kreutz, 1996: 83; Martin, 2002; Mondschein, 2016: 52; Safran, 2014: 54). Likewise, the Greek-speaking elite who were sent from Constantinople would have been educated in a heterogeneous courtly environment and would have recognised Arabic script when they saw it (Anderson, 2009; Reinert, 1998; Walker, 2012). But other sections of the population—the Armenian priests who may have worn vestments adorned with pseudo-Arabic and the inland population of Bitonto—are less likely to have associated the decorative motif with Arabic. The objects decorated with pseudo-Arabic that they encountered were as likely (probably more likely) to have come from merchants, cities and workshops within the Byzantine Empire as they were from the Islamic areas of the Mediterranean. Undoubtedly the motif would have been associated with the pan-Mediterranean trade networks that the local economy relied upon.

The second part of this article will focus on the most prominent example of pseudo-Arabic in Bari, this time from the Norman Period. In the Basilica of San Nicola, nestled in the curve of the apse, is a band of pseudo-Arabic ornament, rendered in a mosaic pavement (Figure 6). Acting as a border between the sanctuary and the apse wall, it is made up of elegant, white, foliated kufic letters, outlined in black, on a maroon background. On closer inspection, it is composed of a repeated motif consisting of two verticals joined together by an arch (as seen in previous examples). The apse itself has been altered by the addition of the early modern funerary monument to Bona Sforza, but we can safely assume that, when the basilica was first built, the mosaic abutted a synthronon, a bench or stepped benches where the clergy sat during the liturgy (the crypt of San Nicola retains its synthronon directly below the main apse). The mosaic is part of an impressive altar platform, the rest of which is paved in opus sectile. As described above, the Basilica of San Nicola was built, during the early Norman Period, on the site of the Court of the Catapan (Corsi, 1999: 62; Lavermicocca, 1995). The site is therefore highly symbolic of Byzantine rule, which ended in 1071. The site may have been vacant for a decade or so until, in 1087, a group of Barese merchants stole the relics of St Nicholas from Myra in Lycia and triumphantly transported them home with them (Geary, 1990: 94–102). Following the arrival of the relics on the 7th of May, the issue of housing them provoked violent civil disturbance. In the end, rather than entrusting the relics to the unpopular pro-Norman Archbishop Ursus, the citizens of the city chose a Benedictine abbot, Elia, to be the custodian of the relics and to oversee the construction of a new church (Vernon, 2015: 91–111). Work began immediately, with the destruction of the former Byzantine court and the clearance of the site. Within two years, the relics had been installed in the new crypt, consecrated by Pope Urban II in 1089. Work then began on the basilica above, which must have been more or less finished by the 1120s. The pseudo-Arabic mosaic can be dated more precisely with the help of an inscription. Three steps lead up to the sanctuary platform. The top riser is inscribed: quod pater Eustasius sic decorando regit. This refers to Abbot Eustasius, who took over from Abbot Elia as both custodian of the relics and manager of the construction project. Therefore, the construction and decoration of the sanctuary platform was completed in the period of Eustasius’ oversight, from 1105 to 1123. The platform, including the pseudo-Arabic mosaic, could have been designed and begun towards the end of Abbot Elia’s tenure, but was finished after 1105. The pseudo-Arabic mosaic should therefore be roughly dated to the first two decades of the 12th century.

This dating places the construction of the sanctuary platform in the years following two important events in the 1090s: the Council of Bari (to which we shall return later), and the First Crusade, which had a large contingent from southern Italy. The Norman Lord of Bari at that time was Bohemond I, son of Robert Guiscard, the leader of the conquest and first Norman duke of Apulia. Bohemond decided to join the crusade in 1096, and he attended mass at San Nicola before embarking for the East from the port of Brindisi (Panarelli, 1999). Bohemond’s biggest contribution to the crusade was his victory at the Battle of Antioch in the summer of 1098, after which he established himself as Prince of Antioch. As part of the looting that followed the battle, Bohemond acquired the tent of the vanquished Emir, Kerbogha of Mosul, and immediately sent it to Bari as a donation to San Nicola (D’Angelo, 2009: 89, chap. 13.57).7 The tent is described by Albert of Aachen (2007: IV.56) and William of Tyre (1986: bk. 6, 22.15–21), who both emphasise its size and complexity, describing it as a tent city, with turrets and an impressive central pavilion. Both sources mention that parts of the walls were made of precious multicoloured silks.8 The tent must have been a major acquisition for the new, half-finished basilica.

The second important event took place a few months after the Battle of Antioch. In the autumn of 1098, Urban II held a council at Bari to discuss the relationship between the Eastern and Western Churches, including the filioque controversy. After centuries of debate, in 1014 the Latin Church had made an alteration to the Nicene Creed, adding that the Holy Spirit proceeds from both the Father and the Son (filioque), rather than just the Father. This was seen as a clarification of the theology of the Trinity among Latin theologians, but was unacceptable for Eastern theologians, not only because they disagreed with the addition, but also because they insisted that the pope was not entitled to make such a change to the creed (Chadwick, 2005; Gasper, 2004: xii; Siecienski, 2010: 3–5). Perhaps prompted by the increased contact between the two churches precipitated by the First Crusade, the primary purpose of the Council of Bari was to improve the relationship between the two branches of Christianity.9 The council was attended by 185 prominent bishops, from both East and West, including Rainier of Bleda, the future Pope Paschal II and St Anselm, who was at that time Archbishop of Canterbury, in exile (Gasper, 2004: 179; Somerville, 2011: 8–9). Urban II asked Anselm to argue the Latin position on the filioque issue, and his text, De Processione Spiritus Sancti, written a few years later, is probably based on the speech he gave at Bari (Fonseca, 1999: 39; Somerville, 2011: 126). Unfortunately, the proceedings of the council have been lost, so we do not have a record of the outcome of the filioque debate or any information about the Orthodox bishops who attended. We do have a brief account written by Eadmer, Anselm’s secretary, which he included in his Historia Novorum in Anglia. The fact that Anselm’s De Processione survives while the official records of the council have been lost means that historians’ understanding of the council is probably skewed. The filioque was probably one of many issues discussed, since other matters concerning the Byzantine east and the Holy Land were also on the pope’s mind. On the 11th of September, the crusaders had sent a letter to Urban II urging him to join them in the East. According to Heinrich Hagenmeyer, Urban II had received this letter before the council and therefore also knew that his representative on the crusade, Adhemar of Le Puy, had died at Antioch on the 1st of August and that he needed to appoint a new one (Hagenmeyer, 1973: 192–4). Additionally, Bohemond later claimed that, either at Bari or in Rome in 1099, the pope had promised to go to the Holy Land (Somerville, 2011: 126, note 58). In short, the victory at Antioch and Adhemar’s death must have been uppermost in the pope’s mind and those events must have been discussed at the council. There may even have been talk of the pope going to the East himself.

Kerbogha’s tent would have arrived by the time the council began. The Hystoria de via et recuperatione Antiochae atque Ierusolymarum states that the tent was sent to Bari by sea and left Antioch straight after the battle, on the 28th of June (D’Angelo, 2009: 89, chap. 13.57). The council opened at the beginning of October and three months would have been ample time for such a journey, especially given that the summer months were optimal for sailing (Eadmer, 1964: 108–14; Protospatarius, 1724: 197). John Pryor (1992: 117) has shown that commercial vessels were able to travel from the West to the Holy Land and back within a single sailing season (from March until late autumn), sometimes setting out as late as early August and still making it back to their home port before winter. Because of the prevailing winds in the Mediterranean, it was always slower to travel from East to West than vice versa. However, a journey from the Eastern Mediterranean to Italy could have been made comfortably within three months. In the 12th century, most voyages between the Latin Kingdom of Jerusalem and the ports of Italy took between four and eight weeks (Menache, 1996: 151, note 34). In the 9th century, Bernard the Monk travelled from Jaffa to Rome in 60 days (Wilkinson, 2002: 268), and Frederick II made a similar voyage in the summer of 1229, from Acre to Brindisi, in 40 days (Menache, 1996: 151). When the news of Bohemond’s great victory at Antioch reached Bari, together with the impressive tent that he had taken from his enemy, it must have caused excitement that would have been heightened by the arrival of Urban II, the instigator of the crusade. It is easy and logical to imagine that the tent would have been displayed in the church during and after the council. Eadmer tells us that the council took place ‘before the body of St Nicholas’, which might imply that it was held in the crypt, but that is impossible (Eadmer, 1964: 108–14). The crypt is nowhere near large enough to accommodate all the delegates, who must have numbered at least 200. Probably the discussions were held in the upper church, begun nine years earlier. Although the structure may have been in place, it is doubtful that the church was much more than a shell. All additional ornament would have been welcome. Although we have no evidence of how the tent was used in Bari, it is interesting to speculate. Perhaps the tent was erected in the space of the nave or outside the church, in order to provide a temporary shelter for the council. Another possibility is that it was cut up and used as carpeting or wall hangings within the half-finished building.

The symbolic value of Bohemond’s donation can be contextualised with other examples of tents being used as gifts. During the crusaders’ stay in Constantinople in June of 1097, before they travelled into the Holy Land, they were required to pay homage to the emperor. Bohemond’s nephew Tancred was more reluctant to comply than the others, but did so, begrudgingly. After he had sworn the oath of allegiance, Emperor Alexius offered him a gift of his choice, expecting that Tancred would ask for gold or something of monetary value. Instead, Tancred requested the emperor’s tent, despite the fact that it was cumbersome, requiring 20 camels to move, and would have been a hindrance to Tancred on the crusade. Alexius was very angry and refused. The tent was perceived to be akin to the emperor’s palace, and the request was seen as symbolic of Tancred’s ambition to usurp the emperor, who stated that the tent was part of his insignia and responded, ‘he desires nothing other than my palace, which is unique in the world. What more can he ask except to take the diadem off my head and place it on his own?’ (Ralph of Caen, 2005: chap. 18). A second illuminating parallel comes from a century later. During the Battle of Navas de Tolosa in 1212, King Alfonso VIII of Castile captured the tent of the Almohad caliph, al-Nasir. Alfonso donated part of the tent to the Abbey of Las Huelgas de Burgos and the other part to Pope Innocent III, along with the caliph’s lance, standard and a letter describing the battle (Ali-de-Unzaga, 2014). The pope ordered that the letter be read publicly in Rome, and the standard was hung in St Peter’s Basilica (O’Callaghan, 2003: 72). These two examples demonstrate the significance of the tent in medieval culture as highly symbolic of kingship. The Spanish example also shows the importance of donations to churches. It seems likely that the arrival of Kerbogha’s tent in Bari would have been similar to the arrival of al-Nasir’s tent in Rome: it would have been accompanied by a description of Bohemond’s victory that would have been read aloud during the council, in the presence of the pope, and the tent itself would have been displayed in the church.

Frustratingly for art historians, Fulcher of Chartres and Albert of Aachen tell us only that the tent was an impressive, large, fairly complex structure and that at least part of it was made of multicoloured silk. Therefore, we must look at examples of other medieval tents to attempt a reconstruction. Most tents in the central Middle Ages seem to have been circular bell tents, with a pole through the centre. Seljuq tents like Kerbogha’s were usually domed pavilions of this type (Redford, 2012). This kind of tent can be seen in multiple images in the Madrid Skylitzes manuscript (Codex Matrit Bib. Nat. Vitr. 26.2), which is an illuminated copy of a Byzantine chronicle, full of military scenes, produced in the Sicilian royal chancery during the reign of King Roger II (Boeck, 2015; Cavallo, 1982: 35–6). The manuscript depicts 33 circular bell tents, all decorated with curved bands of ornament, either at the top, bottom or halfway up the sides (Figure 7). Both Byzantine and Arab tents are represented, and although the Arab tents are slightly more ornate, there is no difference between the two (Mullett, 2013: 277). The tents of prominent Islamic military leaders like Kerbogha were richly decorated with animals, ornamental designs and sometimes figurative and narrative scenes (Golombek, 1988: 31–2). Both Byzantine and Arab tents were sometimes embroidered with inscriptions, and sometimes with poetry or good wishes for the owner (Mullett, 2013: 277). Jeffrey Anderson and Michael Jeffreys (1994) have suggested that short poems were sometimes embroidered onto tents. We have two excellent examples of Arabic and pseudo-Arabic inscriptions on circular tents, both from 13th century manuscripts. Al Hariri’s Maqamat (St Petersburg Institute for Oriental Studies ms c-23, folio 43b), painted in Baghdad in the 1220s, depicts a pilgrimage scene with two inscribed tents (Figure 8). The tent on the right has an inscription running around the top of the tent, where the roof joins the side. The tent on the left has a pseudo-inscription in the same position. Another example can be seen in Alfonso X of Castile’s Book of Games, which features a scene in which two figures play chess within a tent inscribed with a band of Arabic. Therefore, it is entirely plausible that Kerbogha’s tent (or part of the complex structure that made up the ‘tent city’) was circular and that the hem of the fabric was decorated with a pseudo-Arabic motif, like the one we find on the semi-circular mosaic pavement at San Nicola.

For the Council of Bari, Urban II commissioned an episcopal throne for Abbot Elia, who was, by that time, not only custodian of the Basilica of San Nicola but also archbishop of the joint archdiocese of Bari and Canosa. The throne in question now sits in the sanctuary, at the centre of the opus sectile pavement, with the pseudo-Arabic mosaic fanning out behind it (Figure 9). The throne is inscribed ‘the illustrious and good patron, Elia Bishop of Bari and Canosa, sits on this seat’ (Dorin, 2008: 32).10 It is unclear whether or not the throne was finished and used at the council. A local chronicle narrates: ‘On the third of October, Pope Urban II arrived, bringing with him numerous archbishops and bishops, abbots and counts, they entered Bari and were greeted with great reverence and he [Urban] prepared a marvellous throne in the church of St Nicholas Confessor of Christ, for our archbishop Elia’ (Protospatarius, 1724: 197).11 The term ‘prepared’ does not indicate whether Urban II had commissioned the throne in advance and presented it to Elia on his arrival in Bari, or commissioned it at the council to be completed in the following months or years. Urban II had spent the summer at Salerno and September in Benevento (Hagenmeyer, 1973: 192), meaning that he was local enough to have perhaps arranged for the throne to be made in advance. The throne is carved from a single block of marble, almost certainly a Roman column shaft, which would have been a valuable and fairly rare material for a sculptor to work with. Perhaps the block of marble was presented to Elia as a gift from Urban II during the council and the carving began afterwards. The throne was later placed at the centre of the sanctuary platform and remembered as a gift from the pope. In one sense therefore, the sanctuary platform functioned as a visual commemoration of the council and the momentous events of 1098.

It would follow logically that the tent was also displayed in the sanctuary after the council. I want to suggest that the hem of the tent may have been decorated with a pseudo-Arabic band and that the design for the mosaic pavement may have been taken from the tent, although this is conjecture, because the tent is no longer extant. In the tent’s absence, this argument is necessarily ex silentio. My aim here is to present evidence for a plausible hypothesis and think through the possibilities for how the mosaic might relate to luxury portable objects circulating in the Mediterranean. Perhaps during or after the council, a circular part of the tent was used to carpet the sanctuary and the semi-circular apse, with the tent’s hem up against the synthronon, as the mosaic is now. And perhaps when the sanctuary platform came to be paved in the decade or so following the council, the design on the edge of this ‘carpet’ was rendered into mosaic in order to provide a more permanent depiction of pseudo-Arabic. In the Islamic world, designs on carpets were often transmitted into floor mosaics, hence some floor mosaics, both classical Roman and Ummayad, have fringe tassels, showing that they represent absent carpets (Golombek, 1988: 30; Strzygowski, 1930: 441). The appearance of the pseudo-Arabic mosaic is such that a textile model seems most likely. The mosaic is rendered in three colours: white letters outlined in black on a maroon background. This kind of polychromy is unlikely to come from other kinds of portable objects from the Islamic world: it does not seem to come from metalwork and it probably does not come from a ceramic (which are mostly either sgraffito or monochrome lusterware). The arrangement of three colours is similar to the tapestry-woven inscribed bands on the late 12th or early 13th century stockings which were part of the Sicilian royal regalia. The stockings feature gold letters outlined in red (Dolezalek, 2017: 20, Figure 13). If the pseudo-Arabic mosaic did indeed come from Kerbogha’s tent, then the effect on the original textile must have been similar to another part of the Sicilian royal regalia, the mantle, which has an Arabic inscription along its curved hem. However, on the mantle, in order for the inscription to be most legible it is arranged so that the base of each letter is closest to the hem. On the Bari mosaic however, the motif is the other way up, arranged so that the ‘letters’ can be ‘read’ from within the sanctuary (the base of each ‘letter’ is on the inner curve). This would mean that, if the mosaic were transposed from an Islamic textile, the pseudo-inscription has been flipped and that would indicate that the mosaicists had an awareness of, and a desire to engage with, the motif as script.

An apotropaic interpretation of pseudo-Arabic may be relevant here. As mentioned above, it has been argued that the origins of pseudo-Arabic lie in evoking divine protection. Most pseudo-Arabic, including all four of the examples described here from Bari, conform to what is known as the ‘tall-short-tall syndrome’. That is, each unit is composed of two verticals joined together by a shorter decorative element (in the case of the mosaic, two tendrils that twist together to form an arch). This is thought to represent the two central letters (lams) in the world Allah. This abbreviation was first found on amulets with the accusative prefix ‘ya’ (meaning, ‘oh god’, as a supplication for, for example, ‘oh god protect the wearer of this amulet’) but soon became an apotropaic symbol in its own right (Ettinghausen, 1976: 28–47). If there were an apotropaic pseudo-Arabic inscription on Kerbogha’s tent (or another textile that served as a model), that protective function may have been understood by the mosaicists and transferred to the mosaic. The apse of the basilica is just a few meters from Bari’s city wall. On the other side of the wall there is now a road on reclaimed land, but in the Middle Ages the sea came right up to the city walls. The apse therefore is almost a liminal zone, a border between the safety of the city, protected by St Nicholas, and the potential danger of the wider Adriatic world. The sense of the liminality of the apse is reinforced by some of the iconography on Abbot Elia’s episcopal throne. Crouching at the back of the throne are two lionesses, violently devouring their human prey. Pairs of lions are often to be found guarding doorways in southern Italian churches (including at San Nicola) so their protection of liminal zones is well-established. The pseudo-Arabic mosaic needs to be read in conjunction with the lionesses, as part of an apotropaic programme of imagery. What is interesting about the apotropaic function of pseudo-Arabic in this context is that it demonstrates that Christian patrons, mosaicists and audiences possessed an understanding of how the motif was used in the Islamic world and were keen to harness its apotropaic power.

Bohemond was not the only crusader to send textiles home from the Holy Land and to have them used and displayed. Fellow crusader, Gouffier of Lastours, acquired textiles that he used to decorate a tower at his home in Limousin and that remained on display for at least a century (Lester, 2017; Paul, 2013: 90–4). The 12th century English crusader, William Marshal, left instructions that textiles acquired on crusade were to be his burial shroud (Tyerman, 1988: 214). A further example is the Veil of St Anne, a tiraz silk with Arabic inscription, donated by crusaders to the cathedral of Apt in Provence (Snyder, 2004: 154). Furthermore, it is well-documented that designs from textiles were often transferred to other media, and there are some particularly strong examples of this from 11th century Italy. Textiles were often depicted in wall paintings in sanctuaries in the 11th century, particularly in Rome and Campania. This can be seen, in particular, in the apses of Santa Maria in Foroclaudio in Ventaroli, and San Sebastiano al Palatino in Rome (Edwards, 2016: 108–15). Jill Caskey has argued very convincingly that the stucco panels at Santa Maria dei Terreti and Caltagirone in Calabria both borrow their design from tiraz silks and both feature pseudo-Arabic (Caskey, 2011: 110). The stucco panels probably framed the sanctuary as chancel screens. Textiles donated to medieval churches were most commonly used to demarcate the most sacred space, the sanctuary. The use of real and fictive veils in the sanctuary is representative, at least in part, of biblical descriptions of the Holy of Holies in the desert, when the Tabernacle took the form of a tent, hung with purple silks, and the most sacred area inside was demarcated with linens (Exodus, 26: 1–2). This tent structure was well-known to medieval theologians. For example, William of Durandus, writing in the 13th century, explains that the curtains and veils in the sanctuaries of contemporary churches are symbolic of the biblical Tabernacle (Durand, 1893: 20; Edwards, 2016: 82–4). Thus, the use of real textiles, and designs taken from textiles, to frame, veil and delineate the sanctuary was common practice. When Archbishop Elia received Bohemond’s gift of Kerbogha’s tent, it would have been instinctive for him to display it in his new church, and it is extremely plausible that mosaicists took inspiration from the tent when they created the pseudo-Arabic pavement under the direction of Abbot Eustasius a few years later. It was common for designs taken from Islamic textiles to be monumentalised in different media in Christian contexts, particularly in church sanctuaries.

After its donation to the basilica, Kerbogha’s tent disappears from written records. There is no trace of it in the basilica’s inventories and no other documentary reference to it. By the time the inventories began it must have either left the basilica, or the memory of what it was had been lost (the earliest surviving inventories are 16th century) (Massafra 1988: 155). There are two possible explanations for this. The first is that the tent may have been used as a floor covering from its arrival in 1098—that would mean for at least a decade during the ongoing construction and decoration of the church. If that were the case, it would have been worn out through years of use. Perhaps by the time the pavement was laid, the tent was already worn out and the pavement was seen as a more durable replacement. The second possibility (not mutually exclusive with the first) is that the tent was cut up and made into something else, such as vestments. This would explain why it does not appear in inventories (at least not as a tent). The Fermo chasuble provides a parallel here: the chasuble began its life as a tent, made in Spain, probably in Cordoba in the middle of the 11th century. It was originally octagonal, with a pole in the centre, and decorated with a kufic inscription (Rice, 1959; Shalem, 2014; Simon-Cahn, 1993).12 It is not known how the tent left Spain, but Rosamond Mack (2002: 27) has, logically, suggested that it was captured in battle. It may have changed hands during the instability that followed the fall of the Umayyad caliphate in 1031. Part of the tent was later made into a chasuble that has traditionally been associated with St Thomas Becket, and is now in the cathedral at Fermo in Le Marche. According to local tradition, it has been in Fermo since soon after Thomas’s martyrdom in the 12th century, but it is only documented as being there from the 17th century (Shalem, 2014: 257). The fabric is made up of medallions and stars containing figurative images and animals. It has a rectangular panel, upon which is embroidered a large kufic inscription that has been placed vertically, so that it runs down the back of the priest wearing it. This panel would, presumably, have originally been a horizontal band on the tent. There are further kufic inscriptions running along the hem of the garment. This example shows how prized textiles, including tents, could be cut up and put to new purposes. Something similar could have happened to Kerbogha’s tent in Bari. These tents were spoils of war, objects symbolic of the enemies’ power and military strength. To cut them up and re-form them as Christian vestments was to symbolically dismantle that power and strength, whilst also putting a precious and beautiful object to good use—an act that was both thrifty and triumphant.

The apse mosaic at San Nicola is not the only mosaic pavement in southern Italy to include pseudo-Arabic in a design that is likely to have come from a textile. There is a group of mosaic pavements from the mid 12th century, most of which feature pseudo-Arabic in medallions. The most famous and complete example of this group is the mosaic that paves the cathedral of Otranto in the Salento. The mosaic was commissioned by Archbishop Jordan and laid between 1163 and 1165 by a priest named Pantaleone (Castiñeiras, 2004; Ungruh, 2013). It is structured around the Tree of Life, which extends up the nave, accompanied by monsters, biblical and mythological scenes. The Labours of the Months and certain figures and animals are enclosed within rows of tightly packed medallions, in a manner highly evocative of Byzantine medallion silks. The medallions act as frames for Latin inscriptions, geometrical patterns and pseudo-Arabic ‘inscriptions’, again, in the manner of textiles. The pseudo-Arabic can be most clearly seen surrounding the figure of Adam. Other pavements within this group of mosaics also feature pseudo-Arabic, including pavements at Brindisi, Taranto and Santa Maria del Patir in Calabria (Antonucci, 1942; Coscarella, 1997; Rash-Fabbri, 1974: 435). These pavements feature pseudo-Arabic as a framing device, either in medallion frames or in linear borders, echoing the use of both script and pseudo-script on textiles. The circular frames, found at Otranto and Santa Maria del Patir, can be compared to those on Byzantine medallion silks, such as the 12th century St Potentianus shroud, which features pseudo-Arabic medallions (Blair, 1997a: 129–31; Gonosova, 1997: 505–7).

Pseudo-Arabic also appears on another building associated with Bohemond. After the successful conquest of Antioch, Bohemond refused to return the city to the Byzantine emperor and proclaimed himself Prince of Antioch. In 1104, Emperor Alexios, furious at Bohemond’s betrayal, launched a campaign to retake the city, and Bohemond, struggling to retain control, returned to the West in order to gain support and resources to enable him to retain his conquest (Albu, 2000). In pursuit of this, he embarked on a tour of France and married Constance, the daughter of King Philip I (Suger, 1992: 43–5). The money and support he acquired enabled him to launch an ultimately unsuccessful attack on Alexios in the Balkans. Bohemond conceded defeat and was forced to sign the Treaty of Devol, which relegated him to a vassal of Alexios (Savvidēs, 2007: 80–1; Suger, 1992: 43–5). He returned to Puglia and died there in 1111 (Flori, 2007: 287–9; Gadolin, 1982). His nephew Tancred, who was acting as his regent in Antioch, refused to honour the treaty and, later, so did his son and heir, Bohemond II (Flori, 2007: 286).

Bohemond was buried in a chapel attached to the south transept of the cathedral at Canosa di Puglia (which officially had equal status with the cathedral at Bari, within the joint archdiocese of Bari and Canosa). The funerary chapel is an opulent construction, clad in marble, enclosed by bronze doors and originally decorated inside with mosaics (Cilla, 1993). Inscriptions on the doors and around the drum make exaggerated references to Bohemond’s victories over the Byzantine emperor. The doors, made by Roger, a bellmaker from Melfi, probably the first to be cast in Italy since antiquity, have been dated to the decade following Bohemond’s death, which corresponds roughly to the period of Eustasius’ patronage at San Nicola (Vona, 2009). The design of the doors is strongly influenced by Islamic art. On the right-hand door are two medallions containing stars at the centre, surrounded by intricate interlace. On the left-hand door are two medallions with pseudo-Arabic frames (Figure 10). The doors are complex and interesting objects, which merit in-depth discussion in their own right, but here I will limit myself to focusing on how they might illuminate the use of pseudo-Arabic in Bari. The epigraphic programme on Bohemond’s funerary chapel shows that the visual programme is intended to glorify Bohemond’s role in the First Crusade and his acquisition of Antioch. This throws some light on the pseudo-Arabic pavement in Bari, which was laid roughly contemporaneously with the decoration of the funerary chapel in Canosa. It adds another piece of evidence (albeit circumstantial) that pseudo-Arabic, in the diocese of Bari-Canosa in the early 12th century, may have been viewed as a monumentalisation of the memory of Bohemond and his contribution to the crusade. Ultimately, we must acknowledge a tension in the way the motif is used at Bari and Canosa. On the one hand, Kerbogha’s tent, the mosaic pavement and the bronze doors allude to a military victory over a hostile Other and the appropriation of the Other’s visual culture as plunder. At the same time, objects and designs from that culture are acknowledged as precious and beautiful, and worthy of glorifying sacred Christian space.