skip to content

T-S Ar.38.64: A Maġribī-Script Qurʾān Fragment[1]

Magdalen Connolly

While trawling through the Taylor-Schechter Arabic (henceforth T-S Ar.) collection, describing the Genizah’s Arabic-script fragments found therein, I happened upon a formerly mis-identified Qurʾān fragment: T-S Ar.38.64.[2] Although torn, stained and strewn with holes, this parchment fragment caught my eye.

The presence of Arabic-script Qurʾān fragments in the Genizah[3] is noteworthy on two levels: their placement in the Genizah, a Jewish “storeroom”[4] for old, but sacred texts, raises certain questions; were these Qurʾān transcriptions used by Jews, and if so, for what purpose? Beyond the realm of Genizah research, these Qurʾān fragments are also invaluable to the study of Arabic codicology, palaeography and vocalisation systems.

As these fragments contain neither colophons nor any other indication of ownership, attempts to address the former may appear speculative. However, the presence in the Genizah of Judaeo-Arabic Qurʾān fragments (cf. Paudice 2009) attests to Jewish engagement with the Qurʾān’s teachings. The purpose of these interactions – whether polemical or theological – remains open to investigation and interpretation. Yet, the simple act of depositing these Qurʾān fragments in the Genizah indicates that they were treated with the respect more commonly afforded Hebrew-script texts. This respect may have been engendered by these texts’ contents. Connolly and Posegay (forthcoming) point out that all the Qurʾān fragments found in the T-S Ar. collection contain references to prophets or non-Muslim figures (either in the fragments themselves or in the verses that are no longer extant). For example, T-S Ar.38.64 refers to the Hebrew Bible’s Moses and Pharaoh (ibid.; Qurʾān 79:15–26).

The following discussion concentrates on the latter, noting T-S Ar.38.64’s key (i) codicological and (ii) palaeographical features, (iii) briefly describing its vocalisation system, and (iv) offering a summary of its most notable orthographical features.


i. Codicology

As with the other Arabic-script Qurʾān fragments found in the Genizah, the size and style of T-S Ar.38.64 suggest that it was produced for personal use. The fragment comprises two leaves (bifolium), but four small holes, evenly spaced along the spine, indicate that it once formed part of a larger volume. The parchment fragment is arranged in a horizontal format. In its extant form it measures 12.8cm x 29.9cm, with the written area measuring 9.3cm x 14.6cm – 15.3cm. Each leaf contains seven lines of text on recto and verso.

The consonantal text is written in brown ink. The same brown ink is used to outline the gold sūra-headings found on Folio 1, recto line 3 and verso line 4. Rubricated vocalisation signs adorn the graphemes, while the end of an ʾāya (‘verse’) is signalled by three small dots in triangular formation.


ii. Palaeography

Written in maġribī (mabsūṭ)[5] script, in a well-formed hand, T-S Ar.38.64 contains gold sūra-headings composed in an Early Abbasid script-style,[6] finely outlined in brown ink. The script-style is also reminiscent of New Style I (NS I) and New Style III (NS III)[7] (see Déroche 1992).

The text is written on a straight writing-line, with ascenders that lean slightly leftwards, and descenders that curve downwards in deep, broad arcs. The relative lack of shading suggests that the scribe used a calamus with a rounded nib – favoured among maġribī scribes (Abbott 1939:42; Gacek 2012:148) – in the text’s execution. Left-side head-serifs occur frequently. Counters are generally closed.

The stylistic lengthening of letters in the basmala occurs in T-S Ar.38.64 between the ḥāʾ and mīm of al-raḥmān (fig.1) rather than between the sīn and mīm of bism. The former is a feature favoured by maġribī scribes, while the latter occurs predominantly in the eastern tradition (Blair 2006:226).


T-S Ar.38.64, F1, v.5

Figure 1. T-S Ar.38.64, F1, v.5



In isolation, ʾalif has a vertical shaft (red), which curves a little leftwards and sports a left-side head serif. In some instances, ʾalif has a barely perceptible left-turned foot. In final form, ʾalif has a left-turned hook at the top, and its vertical shaft tends to extend below the line (green).


T-S Ar.38.64, F2, v.1–2

Figure 2. T-S Ar.38.64, F2, v.1–2



Final bāʾ/tāʾ/ṯāʾ constitutes a single, curved stroke which turns upwards at either end (fig.3). In isolation, bāʾ/tāʾ/ṯāʾ has a more angular form, comprising two strokes; a straight, horizontal baseline, with a short, vertical stroke on the right-hand side (fig.4). These homographs are consistently marked by a single sublinear dot (bāʾ), two supralinear dots (tāʾ) and three supralinear dots (ṯāʾ), respectively.


T-S Ar.38.64, F1, r.3

Figure 3. T-S Ar.38.64, F1, r.3


T-S Ar.38.64, F2, v.1

Figure 4. T-S Ar.38.64, F2, v.1



The graphemes ḏāl (red) and dāl are composed of two approximately equal strokes which meet at a 70-degree angle. The top stroke curves very slightly in the middle, while the end of the horizontal stroke is written with a slight serif, which juts below the textline. Ḏāl is distinguished from the homograph dāl by a single supralinear dot.


 T-S Ar.38.64, F2, r.1

Figure 5. T-S Ar.38.64, F2, r.1



The grapheme šīn is often adorned with three dots arranged in a horizontal line (see fig.4; see also Khan 1992:43, 100).[8]


The homographs ṣād (blue) ḍād (red), distinguished from one another by a diacritical dot above ḍād, are written in a distinctly curvilinear manner. The baseline of the grapheme’s body runs along the textline horizontally, a curved, oval-shaped loop rising up and along in parallel to the baseline, before coming back to rest on the textline. In final form, the grapheme’s tail extends well below the textline in a generous arc, tapering towards the end. In neither initial nor final form is there any evidence of the ‘tooth’, which usually projects upwards to the left-hand side of the grapheme’s body. The latter feature is characteristic of maġribī script-styles (Abbott 1939:42; Gacek 2012:149).


T-S Ar.38.64, F2, r.4–5

Figure 6. T-S Ar.38.64, F2, r.4–5



The homographs ṭāʾ and ẓāʾ are written with a rectangular body, resting on a long, horizontal baseline. The grapheme’s straight shaft extends diagonally towards the right.


T-S Ar.38.64, F2, r.4

Figure 7. T-S Ar.38.64, F2, r.4



In initial form, ʿayn/ġayn has a large, curved head, which rests on a straight, horizontal baseline. The baseline does not project far beyond the curved arm of the grapheme. In medial form, the grapheme is considerably smaller than in initial form. It has a closed counter and resembles a ‘triangle resting on its tip.’ (Déroche 1992: 136). The homographs ʿayn and ġayn are consistently differentiated from one another by a single dot placed above ġayn.


T-S Ar.38.64, F2 v.5–6

Figure 8. T-S Ar.38.64, F2 v.5–6



A later iteration of the early eastern practice of distinguishing qāf from fāʾ by placing a single dot above or below qāf is preserved here;[9] fāʾ is marked with a sublinear dot, while qāf is pointed with a supralinear dot. Although originally an eastern trend, this practice has become closely associated with maġribī script-styles.


The grapheme kāf has two distinctive forms; one for initial and medial forms; the other for word-final position. In initial (red) and medial forms, kāf is ‘s’-shaped (i.e. kāf mabsūṭa), with a horizontal baseline. In final form (green), kāf is written using two strokes, one horizontal and one vertical, which meet at a 90-degree angle. The horizontal baseline is shorter than the vertical stroke. The latter sometimes curves to the left in the middle. In final form, the grapheme is unmarked.[10]


T-S Ar.38.64, F1 v.6–7

Figure 9. T-S Ar.38.64, F1 v.6–7



In final form, lām has a long, curved tail that sweeps below the textline, often encompassing the following graphemes in its reach. The grapheme’s shaft has a slight curve in the centre, bends leftwards at the top, and is often adorned with a left-side head serif.


T-S Ar.38.64, F2, v.4–5

Figure 10. T-S Ar.38.64, F2, v.4–5



Mīm resembles a triangle with softened, curved edges in initial form (red). The horizontal baseline of the soft-edged triangle rests on the textline, connecting to the following letter from the top left side of the triangle. In final form, mīm’s head, which is small with a closed counter, sits just below the textline, its long tail projecting from, and curving leftwards at the end.


T-S Ar.38.64, F2, r.6–7

Figure 11. T-S Ar.38.64, F2, r.6–7



Final nūn (red) begins on the textline, and has a small arm, which projects above the textline, veering diagonally leftwards. Nūn’s characteristic feature (evident in both final and isolated forms) is its large, sweeping bowl, which extends below the textline, before reaching back towards it in a tapered curve.


T-S Ar.38.64, F1, r.4

Figure 12. T-S Ar.38.64, F1, r.4



Initial hāʾ (red) comprises two small, closed-counter, oval-shaped loops: one sits below the line; the other above it. This rendition of hāʾ is referred to as ʾuḏn al-faras ‘the horse’s ear’. The same formation is attempted in medial form, but the loops here (green) are connecting by a single stroke, and do not meet neatly in the middle as with initial hāʾ.


 T-S Ar.38.64, F2, v.3–4

Figure 13. T-S Ar.38.64, F2, v.3–4



In isolation, yāʾ’s tail extends in a generous curve below the textline (fig.14). In final form, yāʾ mardūda (fig.15) is more common than yāʾ muḥaqqaqa. The double sublinear diacritical dots used to mark yāʾ are often written one on top of the other in medial form (see fig.10).


T-S Ar.38.64, F1, r.3

Figure 14. T-S Ar.38.64, F1, r.3


T-S Ar.38.64, F2, v.3

Figure 15. T-S Ar.38.64, F2, v.3


Lām-ʾalif ligature

Two styles alternate in this text. The first of these is lām-ʾalif warrāqiya (fig.16). The two vertical arms cross over in an elongated X-shape and are not joined at the bottom by a horizontal stroke, as is customary. In the second iteration of this ligature (fig.17), the two separate strokes are supplanted by a single curved stroke referred to as lām-ʾalif muḥaqqaqa.


T-S Ar.38.64, F1, r.4–5

Figure 16. T-S Ar.38.64, F1, r.4–5


T-S Ar.38.64, F2, v.3–4

Figure 17. T-S Ar.38.64, F2, v.3–4


iii. Vocalisation signs

The vocalisation signs used in T-S Ar.38.64 conform neither to the rubricated-dot system nor to the supralinear and sublinear oblique strokes used in the ‘modern’ or ‘standard’ Arabic vocalisation system prevalent today. The coloured dots (red, blue, green and yellow) often associated with earlier maġribī-script Qurʾāns (see George 2015a) are not encountered here. Instead, a combination of rubricated horizontal strokes, circles and semi-circles are used. Fatḥ (/a/) is represented with a red supralinear horizontal stroke, kasr (/i/) with a red sublinear horizontal stroke, and ḍamm (/u/) with a red miniature supralinear semi-circle or wāw. Šadda is either an upturned red supralinear semi-circle (fatḥ) or a downward-turned red sublinear semi-circle (kasr). Nunation (tanwīn) is marked in both red and black ink. The latter was probably added at a later date. In general, tanwīn is represented with two red supralinear horizontal strokes for tanwīn fatḥ, two red sublinear strokes for tanwīn kasr, and two red supralinear semi-circular shapes facing leftwards for tanwīn ḍamm. Sukūn is denoted with a small red supralinear open-counter circle.[11] Hamza is also drawn in red ink and resembles ʿayn/ġayn’s head in word-initial form.


iv. Orthography

T-S Ar.38.64 exhibits consistent plene spelling of long medial /ā/, e.g., ضحاها for ضُحَٮٰهَا (Folio 1 verso, line 4); مرْساها for مُرۡسَٮٰهَا (Folio 1 verso, line 2); ذكراها for ذِكۡرَٮٰهَآ (Folio 1 verso, line 2); الانسان for ٱلۡإِنسَـٰنُ  (Folio 2 recto 2, 4); and ولا نعامكم for وَلِأَنۡعَـٰمِكُمۡ  (Folio 2, recto 6). The plene spelling of long medial /ā/ is associated with personal-use Qurʾāns (Khan 1990–1991:56–57). In ‘model’ Qurʾāns, long medial /ā/ is generally written defectively.

Final /ā/ is more frequently represented with ʾalif ṭawīla than ʾalif maqṣūra (as is customary in the 1925 edition of the Qurʾān), e.g., وتولا for وَتَوَلَّىٰٓ (Folio 1, verso line 5); الاعما for ٱلۡأَعۡمَىٰ (Folio 1, verso line 5); and ونها النڢس عن الهوا for وَنَهَى ٱلنَّفۡسَ عَنِ ٱلۡهَوَىٰ (Folio 1 verso, line 1). Again, this substitution was probably made with the intention of avoiding confusion; final or isolated long /ī/ vowel, which is denoted with yāʾ and two sublinear dots (ي) may easily be confused with ʾalif maqṣūra (ى) /ā/ if the two sublinear dots are omitted.

Words written with supralinear dagger ʾalif in Classical Arabic, such as hāḏā هَـٰذَا are written defectively in T-S Ar.38.64. However, a red vertical stroke is inserted above the grapheme. These amendments and additions likely resulted from a desire to ensure the correct pronunciation during private study.

This brief exploration of T-S Ar.38.64 merely touches on the invaluable contribution the personal-use Qurʾāns found in the Cambridge University Library’s Cairo Genizah collections make to the study of Arabic codicology, palaeography and orthography, and their importance for understanding Medieval Jewish engagement with the Qurʾān.





Abbott, Nadia, The Rise of the North Arabic Script (Chicago: University of Chicago: Oriental Institute, 1939).

Blair, Sheila, S., Islamic Calligraphy (Edinburgh: Edinburgh University Press, 2006).

——— “Kufic,” in Encyclopedia of Arabic Language and Linguistics, ed. Kees Versteegh et al (Leiden; Boston: Brill, 2007), II: 597–602.

Déroche, Francois, The Abbasid Tradition: Qur’ans of the 8th to the 10th centuries AD (London; Oxford: The Nour Foundation in association with Azimuth Editions Ltd. & Oxford University Press, 1992).

Gacek, Adam, “Maġribī,” in Encyclopedia of Arabic Language and Linguistics, ed. Kees Versteegh et al (Leiden; Boston: Brill, 2008a), III: 110–113.

——— Arabic Manuscripts: A Vademecum for Readers (Leiden; Boston: Brill, 2012).

George, Alain. “Coloured Dots and the Question of Regional Origins in Early Qur’ans (Part I).” Journal of Qur’anic Studies 17, no. 1 (February 2015): 1–44.

Gruendler, Beatrice, “Arabic Script,” in Encyclopedia of the Qurʾān, ed. J.D. McAuliffe (Leiden: Brill, 2001), I: 135–144.

Khan, Geoffrey, “Standardisation and Variation in the Orthography of Hebrew Bible and Arabic Qurʾān Manuscripts,” Manuscripts of the Middle East 5 (1990–1991): 53–58.

——— Arabic Papyri. Selected Material from the Khalili Collection (London; Oxford: The Nour Foundation in association with Azimuth Editions Ltd. & Oxford University Press,1992).

——— “The development of Early Arabic Documentary Script,” in Manuscrits hébreux et arabes. Mélanges en l’honneur de Colette Sirat, ed. Lange and J. Olszowy-Schlanger (Turnhout: Brepols, 2014), 279–293.

Paudice, Aleida, “On Three Extant Sources of the Qurʾān Transcribed in Hebrew,” European Journal of Jewish Studies, 2 (2) (2009): 213–257.



[1] This fragment and other Arabic-script Qurʾān fragments encountered in the T-S Ar. collection are discussed in greater detail in Connolly and Posegay (forthcoming).

[2] T-S Ar.38.64 contains Qurʾān 77:27–50; 78:1–12; 79:40–46; 80:1–37.

[3] The Taylor-Schechter New Series (henceforth T-S NS) collection also comprises a number of Arabic-script Qurʾān fragments, see, e.g., T-S NS 183.79; T-S NS 192.11a, T-S NS 192.11b and T-S NS 19211c (these three folios belong with T-S Ar.41.93, my thanks to Estara Arrant for pointing this out); and T-S NS 306.145.

[4] A similar practice is known to have been performed by Muslims: the ʿAmr b. al-ʿĀṣ Mosque in Fusṭāṭ, situated very near the Ben ʿEzra Synagogue, had just such a storeroom (Déroche 1992:12–13; James 1992:12; cf. also Blair 2006:101). It has been suggested to me that these Arabic-script Qurʾān fragments may have been deposited in the Ben ʿEzra Synagogue’s Genizah by Muslims as they wandered past the building. However, given that the ʿAmr b. al-ʿĀṣ Mosque was located so near-by this seems improbable.

[5] The term maġribī, when used in relation to script-styles, refers to the curvilinear script-styles favoured in Spain, North Africa and sub-Saharan Africa from approximately the tenth century CE onwards (Abbott 1939:41; Blair 2006:223; Gacek 2008:110–111). These script-styles are thought to have developed from Early Abbasid Script-styles and their variants (see n.6). As Khan (2014) has demonstrated, the curvilinear script-styles that prevail in the eastern and western provinces after the ninth century CE were most probably borne of Pahlavi-influenced cursive script-styles that are attested in eighth century CE documents from Khurasan (283, 290–291). Mabsūṭ is a sub-category of maġribī script that is generally reserved for Qurʾān composition (Gacek 2012:149).

[6] The term ‘Early Abbasid script’ was introduced by Déroche to replace ‘kufic’, which he considers inadequate (1992:11). Blair has since defended the use of the term (2007:597–598), redefining it as a ‘general rubric of the angular style used in early Islamic times to inscribe monuments and transcribe the Qurʾān.’ (ibid.:598). Early Abbasid script-styles are arranged into six main categories by Déroche (1992).

[7] Also referred to as ‘broken cursive’ (cf. Blair 2006:144), NS I and NS III began to appear during the ninth century CE (Gruendler 2001:142), becoming popular in Qurʾān composition from the tenth century CE onwards (Gacek 2012:167). It is from NS I and NS III that the maġribī script-styles are thought to descend (Gruendler 2001:142).

[8] Khan records this horizontal pointing of šīn in a papyrus document dated 104 AH/723 CE (1992:99–100).

[9] In eighth-century CE texts only qāf receives a diacritical dot, which may be written either above or below the grapheme. This was later supplemented by an opposing dot, marking fāʾ (Gruendler 2001:140). While this system has been retained in the maġribī tradition, the eastern trend of pointing qāf with two supralinear dots and fāʾ with one supralinear dot, which become popular during the ninth century CE (ibid.), is more generally used today.

[10] Final kāf began to be marked with a miniature ‘s’-shaped kāf during the ninth century CE (Gruendler 2001:140). This later transformed into a hamza. This supplementary sign was intended to differentiate final kāf from final lām. The scribe may have deemed such an addition unnecessary here; final lām’s baseline stroke is markedly more curvilinear than final kāf in T-S Ar.38.64 (cf. fig.9 and fig.10).

[11] The circle sometimes looked as if it were closed, but this is probably due to the ink spreading or the effects of humidity/water damage (Connolly and Posegay, forthcoming).


Cite this article

Connolly, M. (2019). T-S Ar.38.64: A Maġribī-Script Qurʾān Fragment. [Genizah Research Unit, Fragment of the Month, June 2019].


If you enjoyed this Fragment of the Month, you can find others here.

Contact us:

The zoomable images are produced using Cloud Zoom, a jQueryimage zoom plugin:

Cloud Zoom, Copyright (c) 2010, R Cecco,

Licensed under the MIT License