A Rediscovered Almoravid Qurʾān in the Bavarian State Library, Munich (Cod. arab. 4)


 This article examines and contextualizes a small Quranic manuscript, copied in al-Andalus in 533/1138–1139, whose importance has so far gone unrecognized. Among its many interesting features are: its early date; its lavish illumination; its colophon and the information contained therein; its system of notation and textual division; its use of different calligraphic styles, including Maghribī thuluth; and a series of didactic notes written at the beginning and end of the codex. Presented in the appendix is an updated list of the extant Qurʾāns in Maghribī scripts dated to before 600/1203–1204, aimed at encouraging the digitization, publication, and comparative study of this still largely uncharted material. The advancement of scholarship on the arts of the book, the transmission of the Qurʾān, and the consumption of Quranic manuscripts in the Islamic West depends upon the analysis of these and many other surviving codices and fragments, related to Cod. arab. 4 of the Bavarian State Library and its context of production.

the German humanist Johann Albrecht Widmannstetter (1506-1557).2 Despite its unassuming appearance, this manuscript represents an essential piece in the complex jigsaw of the arts of the book in the Western Islamic world.
The codex has a roughly square format (17.3× 16 cm) and comprises 130 thin parchment folios sewn into ternions and arranged according to Gregory's law, with the flesh side of each folio always facing the flesh side of the following one. 3 The left and right margins of each page are scored in dry point, delimiting a written area that measures 11×10.5cm and features 23 lines of miniature Maghribī calligraphy. The codex is profusely illuminated with verse dividers and marginal vignettes, and the text of the Qurʾān is preceded and followed by an impressive set of frontispieces and finispieces. These include four polychrome carpet pages (ff. 2b-3a, 128b-129a), a lavishly ornamented page spread containing the colophon (ff. 127b-128a), and a series of didactic notes written in gold at the beginning and end of the book (ff. 1b-2a, 129b).
The codex has been variously attributed to the 11th, 12th, 13th, and even 14th century, which is indicative of the still inadequate state of scholarship on the history of Quranic calligraphy and illumination in al-Andalus and the Maghrib.4 However, a careful reading of the colophon [ fig. 1] gives not only a precise year for the completion of this copy, but also the name of one of the artists involved in its making: After a series of religious formulae and pious invocations, we learn that the manuscript was copied in 533/1138-1139, and that it was "illuminated and bound (dhahhaba-hu wa-saffara-hu)" by a certain Zakariyā b. Muḥammad b. Zakariyā al-Qurashī. This name, not attested in any of the published sources, must have belonged to an exceptionally skilled artisan if we consider the inventiveness, chromatic richness, and meticulousness of his geometric and calligraphic compositions [ fig. 2]. Unfortunately, the codex is no longer bound in its original cover: its current blind-tooled leather binding can be attributed to the 15th century on the basis of similarities with other Islamic bindings from this period.5 While Zakariyā signed and dated his work in the rectangular frame of the colophon, the central panel was originally occupied by the name of the patron and first owner of the manuscript, who "took care of the realisation of this Qurʾān, for himself and for his son after him …", but this was later scraped off the page in a puzzling act of effacement. al-Makhzūmī, a member of the Valencian family of the Banū Bīṭash (or Bīṭish).8 A fourth important example of patronage comes from a slightly larger codex dated 599/1203, completed in Marrakesh through the joint efforts of the calligrapher ("nāsikh") Muḥammad b. Muḥammad al-Sharīshī and his son-in-law Yūsuf, the illuminator ("al-mudhahhib").9 This is a royal Qurʾān, which, according to its colophon, was produced to celebrate the birth of the Almohad prince Abū Yaʿqūb, the son and heir apparent of the caliph al-Nāṣir (r. 595/1199-610/1213).
There is more than one way in which these manuscripts, including the Munich Qurʾān, refer to their dedicatees and the role they played in their realization. The Qurʾān copied for the prince Abū Yaʿqūb features the simple expression "mimmā ʿamala-hu […] li-[…]", translatable as "made by […] for […]", and a similar turn of phrase appears in the Qurʾān dedicated to Ibn Bīṭash al-Makhzūmī ("kāna kamālu-hu li-[…]", meaning "completed for […]"). In both these colophons, the calligrapher's name appears before that of the patron, and the patron is not the subject of any verb; in other words, his agency is somewhat downplayed. On the contrary, in the two Qurʾāns copied by ʿAbd Allāh Ibn Ghaṭṭūs, the patron's name is mentioned before that of the copyist and is introduced as the grammatical subject of the sentence: the vizier Abū Muḥammad al-Madhḥijī "commanded the realization of this Qurʾān (amara bi-iqāmat hādhā al-muṣḥaf )", a verb particularly fitting to someone in a position of power, while Yāsīn b. Lub "realized it for himself (aqāma-hu li-nafsi-hi)". This ambiguous expression could be taken to indicate a hands-on involvement, were it not for the immediately following statement that Ibn Ghaṭṭūs "wrote and illuminated it (kataba-hu wa-dhahhaba-hu)". As we have seen, in the Munich Qurʾān the patron's (missing) name is preceded by the curious circumlocution "tahammama bi-iqāmat hādhā al-muṣḥaf ", which I have translated as "he took care of the realization of this Qurʾān". This is because the fifth form of the verb hamma conveys a sense of attention and solicitude perfectly reflected in the manuscript's extraordinary features, discussed below.10 The appearance of the verb aqāma (with its verbal noun iqāma) in the colophons of these three last Qurʾāns suggests that the term was used in connection with patrons and dedicatees, at least in the context of 12th-century Andalusī manuscripts. Supporting evidence comes from the copy of an Andalusī work of Ashʿarī theology completed in 540/1146-1147, which ends with a colophon stating: "fa-raḥima Allāh man aqāma-hu li-nafsi-hi wa-man kataba-hu wa-man qaraʾa fī-hi", translatable as: "may God have mercy upon [the person] who realized it for himself, and [upon the person] who wrote it, and [upon the person] who read it".11 Here again, whoever "realized" the manuscript did not actually copy it, but had it copied by someone else: the terminology may appear ambiguous to us, but it was likely not so for a 12th-century Andalusī reader. We are thus left with a tantalizing question: if Zakariyā al-Qurashī simply illuminated and bound the Munich Qurʾān, and if the patron did not take an active part in its creation, then who was responsible for its calligraphy? The absence of any reference to the act of writing or to the copyist in a colophon that gives the names of both the patron and the illuminator is a unique occurrence in this period, and indeed difficult to rationalize. It is tempting to imagine that the special care with which the patron tended to the creation of this codex, as mentioned in the colophon, meant that he personally copied and vocalized the text of the Qurʾān, and then entrusted Zakariyā al-Qurashī with its illumination and binding.
While this cannot be proven, it seems reasonable to suggest that Zakariyā did not copy the text of the Munich Qurʾān: had he done so, he would likely have mentioned it, as was customary. We are thus confronted with a manuscript that, although small, likely resulted from a collaboration between (at least) two individuals, a calligrapher and an illuminator, just like the Qurʾān dedicated to the prince Abū Yaʿqūb. This adds to the extraordinary features of the codex, especially since we know that, in 12th-century al-Andalus, Quranic copyists were also skilled illuminators, perfectly able to see to all aspects of book production.

The Notation System
If, as it seems, the copyist and the illuminator of the Munich Qurʾān were not the same person, it is anything but straightforward to determine which of the two did what exactly. The Andalusī biographical dictionaries claim that Quranic calligraphers were first and foremost religious scholars, trained in Quranic recitation (tajwīd) and variant readings (qirāʾāt), and praised on account of their accuracy in vocalizing (ḍabṭ, tanqīṭ) the text of the Revelation.14 In the Andalusī-Maghribī tradition, however, the practice of Quranic vocalization also had considerable aesthetic connotations, given the rich polychromy that characterizes this system, which differs significantly from that of coeval manuscripts from the eastern Islamic world. The distinctive Andalusī way of vocalizing the Qurʾān, based on the practice of early Medinan scholars, seems to have developed in the recitation circles established by Mālik b. Anas In his work, al-Dānī refers to Quranic manuscripts vocalized through dots, but during the 11th century, Andalusī Qurʾāns saw the introduction of a new vocalization system featuring the symbols for ḍamma, fatḥa, and kasra still in use today. In Andalusī Qurʾāns, however, these symbols continued to be penned in red ink. Moreover, the use of coloured dots was maintained in two cases: yellow (or orange) dots marking the hamza above or below alif, and green dots signalling alif al-waṣl. This is precisely what we see in the Munich Qurʾān [ fig. 3], and in virtually all the other Quranic manuscripts produced in al-Andalus and the Maghrib in the 12th century. As mentioned by al-Dānī, several other symbols continued to be traced "in a thin red pen": horizontal strokes for ṣila and madda, vertical ones for dagger alif, and small circles for zawāʾid and mukhaffafa letters. In the Munich Qurʾān, a significant deviation from al-Dānī's description is that the symbols of sukūn and tashdīd are traced in blue ink, and not in red. This practice, not mentioned in any of al-Dānī's treatises, is first referred to in the work of the Sevillian scholar Ibn Wathīq (d. 654/1256) as an alternative to the use of red.17 From the surviving manuscripts, it is clear that this was by far the 16 Abū inks (red, blue, yellow, and green) in addition to the black ink they used for transcribing the consonantal text of the Qurʾān (rasm), which definitely blurs the line between a copyist's job and that of an illuminator. For this reason, Quranic scribes and vocalizers were also artists and perceived as such, and their work was appreciated for its accuracy and reliability as well as for its beauty. This is evident from medieval accounts such as that of Abū This description could very well be applied to the Munich Qurʾān, although produced some thirty years before Muḥammad Ibn Ghaṭṭūs began his career in the 1160s. In this codex, moreover, the anonymous calligrapher went as far as to mark every single recitation pause (waqf ) with superscript letters traced in the same blue ink used for sukūn and tashdīd: a miniature tāʾ indicates the "perfect pause" (al-waqf al-tāmm), a miniature kāf signals the "sufficient pause" (al-waqf al-kāfī), and a miniature ḥāʾ marks the "good pause" (al-waqf al-ḥasan). This unique feature is explained in detail in a prefatory note written in gold ink and enclosed in two illuminated frames on folios 1b-2a [ fig. 4], which reads: with an explanation of its meanings and elucidation of its difficulties. He did so by drawing upon the opinions of the exegetes and the works of the reciters, grammarians, and religious scholars from among his masters, since the Prophet-peace be upon him-established the use [of these pauses] and encouraged their study. [Al-Dānī] divided them into categories, the explanation of which would be too long and, from these categories, he selected the three that we shall mention, namely the perfect pause, the sufficient pause, and the good pause. I have marked all of them in this manuscript, using symbols that I have traced in blue above each pause: a tāʾ above the perfect pause, a kāf above the sufficient pause, and a ḥāʾ above the good pause. Let the reciter be aware of all this, and may God help him, and may he stick to it, God willing.

The Calligraphy
The winding calligraphic style of both these notes, the Fātiḥa decision to abandon this misleading geographical designation. 26 Despite the opinion of some scholars, the fact that fāʾ, qāf and nūn in final position always bear a diacritical dot is not exclusively associated with this style.27 What sets this miniature script aside from all the other Maghribī hands is its angular aspect, obtained through the complete suppression of the oval bodies of ṣād, ḍād, ṭāʾ, and ẓāʾ, and of the rounded variant of initial and medial kāf, with its semi-circular stem. These elements are all replaced by parallel horizontal lines, often extremely elongated, joined together by short strokes, either vertical or oblique. The frequent elongation of letters (and of ligatures between letters) in the Munich Qurʾān harks back to the ancient Abbasid scripts used in early Qurʾāns, and must have also functioned as a visual encouragement to pronounce each word carefully, thus adding solemnity to the recitation [ fig. 5].28 The calligrapher's aim was arguably to give the script an archaizing aspect by evoking the angularity of Kufic, but also to impress the readers with his skills and mastery of even the most minute strokes. It should be noted that similar scripts are sometimes also found in non-Quranic manuscripts produced in the same period and region, mostly combined with illumination, as a means of visually conveying the prestige and authoritativeness of a particular recension of a work.29 The only attempt at a palaeographic analysis of this calligraphic style, made by the Algerian scholar Muḥammad Sharīfī in 1982, was based on the script employed by ʿAbd Allāh Ibn Ghaṭṭūs in a Qurʾān dated 557/1161-1162, but can perfectly be applied to the Munich Qurʾān, too [ fig. 6].30 In addition to the frequent elongation of certain letters and ligatures, Sharīfī remarks on the use of a double baseline in some words, through the rightward elongation of the body of medial and final jīm, ḥāʾ, and khāʾ, in the shape of a line stretching below 26 See Other calligraphic traits observable in the script of the Munich Qurʾān, common to most contemporary Quranic manuscripts from al-Andalus and the Maghrib, are: -The leftward inclination of the head serifs of alif, lām, and final kāf ; -Final bāʾ, tāʾ, thāʾ, and fāʾ terminate with a long horizontal stroke without the upward denticle at the end; -Long, oblique stems of ṭāʾ, and ẓāʾ; -Ample initial ʿayn traced with an oversized curl; -Final mīm always has a long, plunging tail curled leftwards; -Isolated lām-alif is always drawn as two separate strokes, both curved, intersecting near the baseline; -The baseline ligatures between the letters bāʾ, tāʾ, thāʾ, sīn, shīn, ʿayn, ghayn, fāʾ, qāf, lām, nūn and yāʾ present an accentuated saw-toothed profile. The same can be said about the lower part of medial ʿayn, ghayn, fāʾ, and qāf, often rendered as an open space in the shape of a triangle. The resemblance between the script of the Munich Qurʾān, that of ʿAbd Allāh Ibn Ghaṭṭūs 23 years later, and that of other 12th-century Andalusī calligraphers whose work has survived is indeed remarkable. This suggests that a well-codified and widely accepted mode of Quranic calligraphy-or rather, micrography-was among the hallmarks of Maghribī visual culture and religious praxis. But the characteristic script, format, and system of vocalization of 31 Sharīfī, Khuṭūṭ al-maṣāḥif, 277. these Qurʾāns find parallels in a number of equally distinctive textual features, first and foremost the adherence to the Medinan reading (qirāʾa) of Warsh from Nāfīʿ, which is also the one followed in the Munich Qurʾān.

Sūra Titles and Textual Divisions
A second interesting feature shared by Maghribī Quranic manuscripts is the preference for certain sūra titles different from those used in the contemporary Islamic East. For instance, in the Munich Qurʾān the title of sūra 26 is al-Ẓulla ("the Shadow"), and not al-Shuʿarāʾ ("the Poets") as in the eastern tra-dition; sūra 98 is called al-Bariyya ("the Creatures"), and not al-Bayyina ("the Clear Evidence"); sūra 104 is titled al-Ḥuṭama ("the Crusher"), not al-Humaza ("the Slanderer"), and so forth. As already remarked by Juan Pablo Arias Torres and François Déroche, these apparently trivial variations could help us date and identify regional trends or the work of specific scribes just as much as paleographic and codicological analyses, and are therefore very important to record and compare.32 Some of these specifically Maghribī titles are mentioned and employed by al-Dānī in his treatises, but others seem to be found exclusively in Quranic manuscripts. Also, the number of verses in each sūra (which is always indicated in sūra headings) presents conspicuous idiosyncrasies. The copyist of the Munich Qurʾān followed what al-Dānī calls "the first Medinan count" in his Bayān, a system transmitted by the early Medinan scholar Nāfīʿ that departs noticeably from other eastern counts, such as the Basran or the Kufan.33 What follows is a complete list of sūra titles and verse counts as given in the Munich Qurʾān; all differences from the 1924 Cairo edition-which follows the reading of Ḥafṣ from ʿĀsim and the Kufan verse count-are highlighted in bold: 105. Al-Fīl (5) 106. Quraysh (5) 107. Al-Dīn (6) 108. Al-Kawthar (3) 109. Al-Kāfirūn (6) 110. Al-Fatḥ (4) 111. Al-Masad (5) 112. Al-Ṣamad (4) 113. Al-Falaq (5) 114. Al-Nās (6) If similar lists could be drawn for all the dated Maghribī Qurʾāns known to us, they would provide an invaluable benchmark for attributing undated manuscripts to specific contexts and periods with more confidence. However, most of these Qurʾāns are kept in Middle Eastern and North African libraries that only grant very limited access to their collections, and the single survey published to date covers a limited sample that only includes one manuscript from the 12th century.34 A third and final idiosyncrasy of Maghribī Qurʾāns is found in their complex system of textual division. In the Munich Qurʾān, the standard verse separator is represented by a cluster of three gilded roundels arranged into a triangle. However, every fifth verse is marked with a gilded Kufic hāʾ (representing the digit 5 in the Arabic alphanumeric system or abjad), and every tenth verse ends with a gilded roundel surrounded by eight alternating blue and red dots. Each tenth-verse marker in the text block is always accompanied by a corresponding gilded medallion in the margin, consisting of a petalled circle enclosing a roundel in which the verse number is spelled out in Kufic chrysography (ʿashara for the tenth verse, ʿishrūn for the twentieth, and so forth). The text of the Qurʾān is then divided into sixty equal parts (aḥzāb, sing. ḥizb), which are signalled by larger illuminated medallions in the margins, enclosing a roundel with the word ḥizb in Kufic chrysography. While these are all rather standard features, the Munich Qurʾān is further divided into ten equal parts, as well as ninths, eighths, sevenths, sixths, fifths, fourths, thirds, and halves, all of which are accurately indicated in the margins in bold Kufic chrysography, dotted and vocalized in blue [ fig. 7]. Once again, it seems that the scribe placed all these division marks in accordance with the instructions provided by al-Dānī in his Bayān, with only very rare and minor departures.
In addition, the text is further divided into 27 equal sections called tajziʾāt Ramaḍān, rather than into the canonical thirty ajzāʾ (sing. juzʾ). These sections refer to the first 27 nights of the holy month of Ramaḍān, and allow to recite the entire Qurʾān during tarāwīḥ, or night prayer, ending on the 27th night of the month, known as Laylat al-Qadr (the "Night of the Decree"), when the prophet Muḥammad received his first revelation. The end of each tajziʾa is indicated by figure 7 Munich, bsb, Cod. arab. 4, f. 48b, detail. Part of Sūrat Yūsuf (12: [47][48][49][50][51][52][53][54][55][56]. In the margin is a ḥizb medallion (25th ḥizb, 12:53), a smaller medallion marking the 50th verse of the sūra, and two notes marking the completion of the second fifth and fourth tenth of the Qurʾān. © bayerische staatsbibliothek illuminated trilobed devices in the margins, containing the last word of the relevant passage in Kufic chrysography over a green background [ fig. 8]. This kind of textual division, unparalleled in the Islamic East, seems to be specific to a group of Qurʾānic manuscripts produced in al-Andalus and Northwest Africa that await further study.35 The Munich Qurʾān represents one of the earliest surviving witnesses of this practice, which lasted until the early 17th century among the Moriscos living in Habsburg Spain.36

Conclusions
Based on these elements, it is possible to draw some conclusions regarding the nature and purpose of this and other similar Qurʾānic manuscripts from the same period. As we have seen, the patron and/or copyist of the Munich Qurʾān must have been a religious scholar competent in Qurʾānic sciences, but also a man deeply concerned with producing an exemplar of the highest possible accuracy and visual appeal. The use of the atypical expression tahammama bi-iqāmat hādhā al-muṣḥaf ("he took care of the realization of this Qurʾān") in the colophon emphasizes the patron's commitment to the task, confirmed by his choice to employ a professional illuminator-Zakariyā b. Muḥammad b. Zakariyā al-Qurashī-to provide the manuscript with lavish frontispieces and finispieces. The Munich Qurʾān was not commissioned to be endowed to a mosque or religious institution, but for the owner's private use ("li-nafsi-hi") and for his son after him, and was arguably meant to become a cherished family heirloom handed down from father to son. Codices such as this probably functioned as personal or travel copies, to be read from during acts of familial devotion, especially in the holy month of Ramaḍān, as suggested by the division into 27 tajziʾāt. However, they also had a strong pedagogic purpose, since the extra-textual notes and the profusion of vocalization and orthoepic notation were clearly intended to instruct the reader in the art of correct recitation and memorisation. The plethora of textual divisions would have allowed different types of learners to tackle the Quranic text at different paces: ḥizb by ḥizb, tenth by tenth, seventh by seventh, and so forth. While larger Qurʾāns in multiple volumes were endowed to Andalusī and Maghribī mosques by the ruling elites for public recitation, it was through small manuscripts such as this that students and scholars prayed and exercised their reading in a domestic environment. The Munich Qurʾān was produced at a turbulent time when the fortunes of the Almoravid dynasty were declining, and yet the cities of al-Andalus witnessed in this period a renewed cultural effervescence brought about by the rise of new local families of governors, judges, and jurists. The effacement of the owner's name may point to a moment of regime change, possibly coin-ciding with the downfall of the Almoravids, the second ṭāʾifa period, and the Almohad conquest of Iberia (1147-1172). The obliteration of names of patrons, owners, or calligraphers from medieval Maghribī colophons is an extremely rare phenomenon, and cannot simply be explained in terms of changes of ownership. A new owner would have added their name at the beginning of the book or after its colophon, without any need to erase someone else's name and titles from it. The only other instance of effacement in a 12th-century Andalusī manuscript (that I am aware of) was carried out at the expenses of an Almohad prince in a luxury copy of the Ṣaḥīḥ Muslim produced for his library in 573/1178, in Murcia.37 His name, titles, and lineage were entirely expunged from the colophon, without any attempt to replace them with those of a new owner, while the signature of the calligrapher and the indication of the date and place of production were left intact. It is therefore possible that the patron of the Munich Qurʾān was also a governor, appointed by the Almoravid emir, or at least an Almoravid supporter of some consequenceperhaps the imām or preacher at a major mosque-whose memory had to be eradicated.
Even if just speculatively, it is tempting to attribute the Munich Qurʾān to the scholarly and artistic milieu of Córdoba, the Almoravid capital of al-Andalus, based on the striking similarities in format, layout, and script with a codex copied and illuminated in Córdoba five years later, in 538/1143-1144 [ Fig. 9].38 This, too, is a lavishly illuminated manuscript penned by an anonymous calligrapher, who added after the colophon two handsome finispieces emblazoned with the words: "Abū ʿAbd Allāh Muḥammad b. ʿAlī b. Muḥammad b. Yaḥyā al-Ghāfiqī al-Shārrī was born in Murcia, in the month of Shawwāl of the year 537, may God make him happy and successful". The scion of the important Murcian family of the Banū Yaḥyā al-Ghāfiqī, Abū ʿAbd Allāh al-Shārrī became a renowned transmitter of works of ḥadīth and fiqh and a scholar of qirāʾāt.39 This manuscript was evidently commissioned to celebrate his birth, probably by his proud father. Similarly to the Munich Qurʾān, this is a fascinating example of a personal codex imbued with a sense of familial self-assertion and domestic intimacy. Interestingly, its copying and illumination were not entrusted to a Murcian master but to a Cordoban one, which might indicate the pre-eminence of this city in the arts of the Qurʾān during the last decades of Almoravid rule. For its historical and art historical significance, the Munich Qurʾān can be considered one of the most remarkable artefacts to have survived from 12thcentury al-Andalus, and should be given pride of place within the increasingly growing corpus of the earliest Maghribī Qurʾāns known to us. Appended to this article is a list of dated Qurʾāns in Maghribī script copied until the year 600/1203-1204, which updates the previous lists published by David James in 1992 and by François Déroche in 2001.40 A comparative study of these 27 manuscripts and their textual, codicological, and aesthetic features would be crucial to the advancement of scholarship on the arts of the book and the transmission of the Qurʾān in the medieval Islamic West, and should not be deferred any further. An important piece has been added to the jigsaw, but a lot more remains to be done as new material waits to be identified in libraries, museums, and private collections all over the world. 40 James