The Max Schloessinger Memorial Foundation
Offprint from
JERUSALEM STUDIES IN
ARABIC AND ISLAM
49 (2020)
Petra M. Sijpesteijn
ARABIC SCRIPT AND LANGUAGE IN THE
EARLIEST PAPYRI: MIRRORS OF CHANGE
THE HEBREW UNIVERSITY OF JERUSALEM
THE FACULTY OF HUMANITIES
STUDIES IN HONOUR OF ELLA LANDAU-TASSERON
I
Y. Friedmann The scholarly oeuvre of Professor Ella
Landau-Tasseron
i
Ella Landau-Tasseron — List of
Publications
vii
Christian J. Robin Allāh avant Muḥammad
1
Haggai Ben-Shammai “The Qurʾān has been brought down in
seven modes of articulation”: on
possible parallels (or antecedents) to
an old Islamic tradition
147
Adam Silverstein Unmasking maskh: the transformation
of Jews into “apes, driven away”
(Qurʾān 7:166) in Near Eastern context
177
Miklos Muranyi Visited places on the Prophet’s track in
Mecca and Medina
217
David Cook Was Kaʿb al-Aḥbār a prophet in Syria?
231
Jonathan Brockopp Ar. 3001 and the apotheosis of Mālik b.
Anas
249
Andreas Görke Criteria for dating early tafsīr
traditions: the exegetical traditions
and variant readings of Abū Mijlaz
Lāḥiq b. Ḥumayd
275
Amikam Elad Al-Masjid al-Aqṣā during the Umawī
period: seven miḥrābs with seven
domes
339
Petra M. Sijpesteijn The Arabic script and language in the
earliest papyri: mirrors of change
REVIEWS
Yaacov Lev New studies on Aghlabī and Fāṭimī
history: a review article
495
Vera B. Moreen Roberta Casagrande-Kim, Samuel
Thrope and Raquel Ukeles, eds.
Romance and reason. Islamic
trans-formations of the classical past
433
ARABIC SCRIPT AND LANGUAGE IN THE EARLIEST PAPYRI:
MIRRORS OF CHANGE
Petra M. Sijpesteijn
Leiden University
Abstract With the arrival of the Arabs in Egypt in the mid-seventh century CE, Arabic became the third administrative language, joining Greek and Coptic, in which most Egyptians continued to operate. Arabic papyri produced by the Muslim administration use standardised administrative formulae, which were shared throughout the Islamic empire, but which were consistently different from local practices. This suggests that these formulae have a common origin and that the Arab conquerors introduced them. The script and orthography in the Arabic papyri, however, do not show a similar consistency, pointing to the introduction of various scribal practices representing local Arabian customs. As scripts continued to develop, one writing style appears to have become dominant, but this was not a linear process. Archaic and newer letter forms existed side by side, sometimes even in the same document. The role of the chancery, where interaction with Coptic and Greek practices took place as well, was crucial. The development of a uniform chancery writing style also led to a divergence between the way texts were penned in different social domains, depending on the legal, political and commercial/private context. One striking example is the development of the logogram that appears in Arabian inscriptions as well as early Arabic papyri and Qurʾān manuscripts with the meaning ‘son of’ and follows the writing of Nabatean bar, ‘son’. This article traces the developments in administrative Arabic writings from Egypt, connecting them to political and social changes taking place in the first hundred years of Muslim rule in the province.
Keywords Egypt, Arabia, Arabic, Aramaic, Nabatean, papyri, chancery, administration
This work was supported by the European Research Council under Grant number 683194. The
article is based on a paper delivered in April 2013 in Paris at the conference entitled ‘Le contexte de naissance de l’écriture arabe: écrit et écritures araméens et arabes au 1er millénnaire après J.-C.,’ organised by Laïla Nehmé, Françoise Briquel-Chatonnet and Muriel Débie. I would like to thank Abdullah AlHatlani, Robert Hoyland, Ahmad al-Jallad, Fokelien Kootstra, Michael Macdonald, Marijn van Putten and Khaled Younes for their very useful comments. Any mistakes remain, of course, my own. For the bibliographical abbreviations used throughout the article, see the list of abbreviations in the bibliography.
With the arrival of the Arabs in Egypt in 639 CE, a new official script and language was added to the linguistic landscape of this province: Arabic. Like the Greek and Coptic of Byzantine Egypt, Arabic was also immediately put to use by the new rulers for internal administration and for communication with the local population — although the vast majority of Egypt’s inhabitants were Coptic and Greek-speaking and remained so for the next couple of centuries.1 Our earliest
pieces of dated Arabic text in Arabic script are in fact two papyri from Egypt dated 22 AH/643 CE, only three years after the Arabs had entered the country.2
The use of Arabic in these writings sets the Arabs apart from the local population, sending a powerful message about the ascendency of a new order. But they also continued practices used by the Arabs before their arrival in Egypt. The subsequent development of the Arabic script and language in the papyri was informed by these same two elements: symbolism and pragmatism. This paper will present this development during the first hundred years of Muslim rule in Egypt, discussing the kinds of administrative and scribal routines and know-how that the Arab conquerors brought with them, how these were related to indigenous traditions, and how bureaucratic scribal practice developed as a result. While the paper will focus on Arabic scribal practice in early Islamic Egypt, interactions with the Greek and Coptic writing traditions as they continued in the Islamic administration, will also be addressed in the discussion. Interestingly, these Coptic and Greek scribal traditions underwent their own developments and changes of practice well into the Islamic period, indicating that these remained viable and living systems, not mere tokens.3 Although references to non-official
writings will occur, the emphasis will be on monumental and official texts produced by the Arabs’ court or chancery, connecting administrative politics with Arabic linguistic expression.
1 See my forthcoming publication ‘Did the Muslim empire have a language policy?’ in A. Vacca
and A. Borrut, Navigating language (Houten: Brepols).
2 The first text is a receipt written in Greek and Arabic for sixty-five sheep received by the Arabs
from the two pagarchs of the Heracleopolite/Ihnās. For an image of the papyrus, see Grohmann,
Arabische Chronologie, pl. II.1, here fig. 1. The Greek and Arabic texts appear with an English
translation in Grohmann, From the world of Arabic papyri, pp. 113–114. See also here, Appendix 1. For the Greek text, see also Sammelbuch 6, no. 9576. The second text is a demand note for the payment of taxes in dīnārs. For a depiction of this text, see Grohmann, Arabische Chronologie, pl. 2. For the Arabic edition, see Rāġib, “Un papyrus arabe.” For the interpretation of this text as a tax demand note, see Sijpesteijn, “Arab conquest,” p. 446.
3 For a discussion of the development of Greek scribal tradition, see Morelli, Griechische Texte XV,
Schooling and training
When examining the Arabic of the earliest papyri and trying to discover a pattern from its first occurrence through its early development, a pertinent question that arises is where writing in Arabic was learned. What evidence is there for a structured system of schooling and scribal training in the first century of Islam in Egypt? For the earliest period, the chronicles and other literary sources, unfortunately, are almost uniformly silent. The Arabic sources mention in detail how specific caliphs or governors established dīwāns, government offices, for correspondence, land tax, the administration of crown lands, the postal service, etc.4 There are also anecdotal references to the scribes who worked there. There
is, however, no discussion of how the people manning these offices were trained, about the regulation of a scribal practice, or the presence of schools. The kind of handbooks for scribes (adab al-kuttāb) that survive in large numbers for the later period, listing model texts, how scribes should behave at court, etc., are also not attested for this early period.5
In general, it seems, basic writing and reading were trained in different stages, leading to different levels of expertise and specialisation.6 Reading and
writing were separate, although obviously closely related skills, which could, and were in certain circumstances, mastered in isolation. Although Egypt was what we would broadly consider a literate society in the sense that writing was everywhere present, people clearly operated at different levels of writing and reading ability and documents were used also by those unable to produce them. The instruction of reading and writing Arabic could take place at a young age and was important for many more people in society than scribes working in the chancery. The mastering of the terminology, expressions, formulae and other practices common to the administrative documents was already a degree of specialisation limited to those being trained for the bureaucracy. Finally, sophisticated knowledge of the style, layout and wording of different kinds of
4 Duri and Gottschalk, “Dīwān,” pp. 323–337.
5 For the adab al-kuttāb works, see especially van Berkel, Accountants, and Heck, Construction of
knowledge. On scribal education in medieval Egypt, see Shahin, “Schreibübung.”
6 Macdonald, “Literacy in an oral environment,” p. 48 n. 6 and pp. 61-66. Evidence also exists
from different historical contexts. For the Jewish Cairene milieu, cf. Olszowy-Schlanger, “Learning to read and write.” For the high Middle Ages in the eastern part of the empire, see Makdisi, Rise of colleges, p. 19.
documents, documentary practice and the etiquette of the scribal class were facets of a technical idiolect that would have been acquired only by the highest echelons of scribes. Additional educational hurdles existed for non-native or bilingual Arabic speakers who, for whatever reason, learned to write in Arabic. The latter might have been already accomplished writers in another language, or have learned the writing of Arabic according to the schedule described above, albeit possibly at a different stage in their lives.7
It is possible to examine the different stages of schooling in the papyrological material, especially when a group of documents illustrates the career of one particular individual. The scribe Aristophanes operated as a secretary to the Muslim authorities and in the private sphere in the first half of the eighth century in Jeme, near modern-day Luxor, ancient Thebes. There are 42 Coptic documents preserved on papyrus and potsherds which Aristophanes wrote, signed or figures in. Together with other documents produced by his father, brother, son, and contemporaries, they allow us to reconstruct Aristophanes’ development in writing and more sophisticated scribal skills. As argued in a recent study that examines these documents in all their details, it seems that Aristophanes took his first steps in literacy at home trained by his father, then moved to be schooled in specialized secretarial skills in the district capital, after which he returned to his hometown to work in the administration and private sphere until his writing skills deteriorated due to age or illness.8
Unfortunately there is not yet such a dossier of Arabic texts known, but the papyri nevertheless record all these stages in the education. There are examples of texts written in unpractised hands who have mastered just basic writing skills. These belong, generally, to the private sphere consisting of private letters.9 From the ninth century onwards witness statements written by
witnesses themselves are found on Arabic legal documents. It is clear that witnesses could not always write (well).10 Commercial letters and other
7 See below, n. 108. 8 Cromwell, Village life.
9 See for example the third/ninth-century literary text written in a carefully executed, but
unpracticed hand (Grohmann, Arabische Chronologie, pl. VIII.2). See also the second– third/eighth–ninth-century letter of introduction displaying archaic and more cursive forms of
dāl as well as a frequent use of diacritics, both pointing at a less practised hand (P.Berl.Arab. II,
no. 53).
10 See for example the different levels of accomplishment amongst the witnesses at the bottom
of the documents dated 288/901, some of whom had the document scribe sign for them (Sijpesteijn, “Profit following responsibility”). But see Macdonald’s discussion of what
documents show a higher level of fluency and expertise. Documents originating in the chancery are composed according to an even more developed and, from the eighth century onwards, clearly recognisable, scribal practice, eventually displaying the formulae and scripts prescribed in secretarial handbooks.11 But
papyri also show the secretaries trying out the administrative formulae, entries and expressions used in fiscal documents, as well as address formulae. They appear written repeatedly or in a sequence not usually applied in single administrative texts, most often on re-used pieces of papyrus.12
The documents that would be most helpful, however — school texts, which have been used to reconstruct with such success the system of education in Greek and Coptic in Egypt — have not been identified with certainty in Arabic.13 This is not to be explained by the absence of schools and schooling for
signatures might in fact say or not say about levels of literacy (“Literacy in Roman Judea,” pp. 833-834; idem, “Literacy in an oral environment,” p. 49).
11 For expressions in chancery scripts, see the examples given below. See also below, figs. 4, 5
and 6 for recognisable and standardised chancery scripts (also discussed in Abbott, “New papyrus”). For the development of proportional and geometric Kūfic styles in monumental and other public writings as well as Qurʾānic manuscripts in the early eighth century, see George,
Calligraphy, pp. 31-34, 60, and before, Imbert, “Réflexions.” For the use of the ‘thick pen’ and ‘thin
pen’ in the decrees written in reply to petitions in the Fāṭimī period, see Khan, Arabic legal and
administrative documents, no. 85.
12 For example, in Karabacek, Ausstellung, no. 594 (P.Vindob. G 39752), dating to 701-730,
provenance Madīnat al-Fayyūm, an address formula is repeated after the basmala following the primary Greek text. The basmala is written several times after some epistolary opening formulae on a third-fourth/ninth-tenth-century papyrus (Shahin, “Schreibübung,” no. 2). On the back of a third/ninth-century account mentioning administrative districts and Islamic month names some epistolary formulae are repeated and another contemporary Arabic papyrus attests the word sijill, register, repeated three times (P.Khalili II, nos. 104; 137). The back of a second/eighth-century administrative papyrus is re-used to practice opening formulae of letters (Shahin, “Schreibübung,” no. 3). Greek papyri of the eighth century show similar chancery exercises (Berkes, “Schreibübungen,” nn. 8 and 9).
13 For Greek and Coptic schooling systems, see Cribiore, “Education” and Bucking, Practice makes
perfect. Arabic literary fragments, especially of Qurʾānic and ḥadīth texts have been interpreted
as school exercises, but these do not show the kind of elementary and repeated practice that Greek and Coptic wordlists and copied texts show. See for example, a fragment with Qurʾān Sūra 109:1-3 and 6 (Karabacek, Ausstellung, no. 628, eighth century, provenance unknown) written in a crude hand on the back of a papyrus used for a Greek fiscal list. Another papyrus containing Qurʾānic Sūras written in a laboured hand with multiple orthographic and graphic peculiarities stops half way the verso of the papyrus and has been interpreted as a pupil’s text (Noseda, “Third Koranic fragment”). An eighth/early ninth-century ḥadīth fragment similarly written in a crude hand with lots of diacritical dots has been identified as a private copy produced by a not very
Arabic scribes. The explanation lies rather in the conditions of preservation. Greek and Coptic papyri have tended to come from excavations, however inadequate the archaeological methods of the time might be considered today and hence valued for their historical significance. Arabic papyri on the other hand have typically been bought from dealers and survival has meant meeting the (largely aesthetic) standards of crude sellability. Scribal exercises, alas, do not make especially beautiful documents. Even when the interest value of the contents was one of the criteria, Arabic scribal exercises did not score highly amongst the dealers who could read, to a certain degree, the Arabic as opposed to the Greek and Coptic texts. In more recent excavations, however, where the Islamic layers have also been studied, these documents do appear in greater numbers and should lead to an improvement in our knowledge. It has to be said, however, that editors of Arabic texts are also to blame, as they have generally not been interested in publishing these kinds of texts.14
An Arabic scribal tradition
The outlines of scribal practice can, however, be inferred from the documents that scribes produced, even if in the earliest period diverse styles and practices were used. The earliest Arabic papyri consist of demand-notes and receipts for goods needed by the Muslim armies, as well as legal documents and official letters or decrees. Only a handful of Arabic ‘private’ letters dating to the seventh century CE have been identified. These documents reveal an administrative and documentary practice that is not just consistent but consistently different from local practice and, as mentioned above, systematically expressed in Arabic. This is clearest in bilingual Greek-Arabic and Coptic-Arabic documents, in which the
accomplished writer (Sijpesteijn, “Ḥadīth fragment”) and a practice text produced by a student of ḥadīth (El Shamsy, “Debates on prayer”). The writing exercises generally show an already practiced hand training fluidity and appearance by the repetitious writing of phrases, combinations of or single words (e.g., Shahin, “Schreibübung,” no. 1, second/eighth century).
14 The publications remain scarce: CPR XVI, no. 35, third/ninth century; Shahin,
“Schreibübung,” nos. 1-3, second-fourth/eighth-tenth centuries; Berkes and Younes, “Trilingual scribe,” dating to 789-790; Vanthieghem, “Exercice,” dating to 811-812; P.Berl.Arab. II, p. 164,
P.Heid.Arab. II, no. 58, both third/ninth century, P.Hamb.Arab. II, nos. 1c, dating to 870-900, 11b,
42c-d, all dating 917-977; P.Vind.Arab. III, no. 17, dating 919-920, while unpublished pieces are numerous (Shahin, “Schreibübung,” p. 98). For examples of unpublished scribal exerchices, cf. Karabacek, Ausstellung, nos. 153, 594, 786 and P.Khalili II, nos. 21, 39, 63, 104, 137, 152, 173, 200, 203.
Arabic does not merely translate the Greek or Coptic but uses its own formulae to serve its own (legal) conventions and needs. The two versions of the text were written by different scribes, as can be deduced from the names that appear at the bottom of the documents.15 These scribes produced the text according to the
customs of their own linguistic and cultural milieu, even using different writing instruments in the process: the Arabic in these texts is invariably written with a much wider cut reed pen (Gr. calamus; Ar. qalam) than the Greek.16
In contents and expression, the Arabic documents differ from the Greek in several aspects (see fig. 1 and appendix 1). In the Arabic parts of the bilingual documents, titles are systematically omitted, while they are listed in the Greek.17
The Arabic part, on the other hand, describes individuals with more extensive names, including patronymics.18 Typically the Arabic documents are written in
15 Greek-Arabic bilingual texts mentioning a different scribe for the Greek and Arabic part:
Sammelbuch 6, no. 9576 (fig. 1 and appendix 1 here), dated 22/643, provenance
Ihnās/Heracleopolite, Arabic written by Ibn Ḥudayd and Greek written by Yohannes;
Sammelbuch 18, no. 13771 (Greek) and Stoetzer and Worp, “Zwei Steuerquittungen,” no. 2 (Greek
and Arabic), dated 72/692 or 88/707, provenance Ihnās/Heracleopolite, Arabic written by Juhaym and Greek written by Daniel; Sammelbuch 16, no. 13018 (Greek) and Stoetzer and Worp, “Arabisch-griechische Steuerquittung” (Greek and Arabic), dated 95/714 (for the date, see Gonis, “Reconsidering,” pp. 226-227), provenance Ashmūn/Hermopolite, Arabic written by Saʿīd and Greek written by Athanasios; Sammelbuch 18, no. 13218 (Greek) and Becker “Papyrusstudien.” (Arabic. With corrections in Diem, “Philologisches,” p. 256.), dated 95/714, provenance Ishqūh/Aphroditō, Arabic written by al-Ṣalt, Greek written by Amba Koum; P.Ness. III, no. 60, dated 54/674, Arabic written by Abū Saʿīd and Greek written by Alexander son of Ammonius;
P.Ness. III, no. 61, dated 55/675, Arabic written by Khālid and Greek written by Stephan of
Maïumas; P.Ness. III, no. 62, dated 55/675, Arabic written by Khālid and Greek written by Theodore son of John; P.Ness. III, no. 63, dated 55/675, Arabic written by Ḥamīd and Greek written by Procopius; P.Ness. III, no. 64, dated 56/676, name of Arabic scribe lost, Greek written by George of Maïumas; P.Ness. III, no. 66, dated 57/677, Arabic written by Abū Saʿīd and Greek written by Stephan of Maïumas; P.Ness. III, no. 67, dated 70/689, name of Arabic scribe lost, Greek written by Eulaeus; P.Ness. III, no. 56, dated 67/687, name of Arabic scribe lost, Greek written by Father George son of Victor, provenance of all is Naṣtān/Nessana.
16 For the Arabic qalam, see Grohmann, Arabische Paläographie I, pp. 117–123.
17 E.g., the titles symboulos, amiras, pagarchos and notarios used in the Greek lack in the Arabic part
(Sammelbuch 6, no. 9576 and Grohmann, From the world of Arabic papyri, pp. 113–114, fig. 1 and appendix 1 here, dated 22/643, provenance Ihnās/Heracleopolite; Sammelbuch 16, no. 13018 and Stoetzer and Worp, “Arabisch-griechische Steuerquittung,” dated 93/714, provenance Ashmūn/Hermopolite; Sammelbuch 18, no. 13218 and Becker, “Papyrusstudien,” dated 95/714;
P.Heid.Arab. I, nos. 22, a-l, provenance of all is Ishqūh/Aphroditō).
18 In Sammelbuch 6, no. 9576 and Grohmann, From the world of Arabic papyri, pp. 113–114, fig. 1 and
the third person as opposed to the second person of the Greek ones.19 Finally, it
is clear that in the two linguistic spheres different elements of the agreements are emphasised, appearing in the one text but not in the other, while similarly some formulae seemingly expressing administrative or legal conditions are confined to one or the other language.20 In the Greek parts, monetary amounts
Theodorakios as قرذث and Christoforos as رفطصا. For the latest edition, see
http://papyri.info/ddbdp/sb;6;9576 (accessed 20/7/2020).
19 See in the following tax receipts: hādhā mā akhadha fulān (this is what X took) in the Arabic
versus di’emou X (through me, so-and-so) in the Greek (Sammelbuch 6, no. 9576 and Grohmann,
From the world of Arabic papyri, pp. 113–114, fig. 1 and appendix here, dated 22/643, provenance
Ihnās/Heracleopolite) or barāʾa min fulān (release from so-and-so) in the Arabic versus X hymin (from so-and-so to you) in the Greek (Sammelbuch 16, no. 13018 and Stoetzer and Worp, “Arabisch-griechische Steuerquittung,” dated 93/714, provenance Ashmūn/Hermopolite;
Sammelbuch 18, no. 13771 and Stoetzer and Worp, “Zwei Steuerquittungen,” no. 2, dated 72/692
or 88/707, provenance Ihnās/Heracleopolite). Entagia, or tax demand notes, describe the addressee in the third person in the Arabic as opposed the second person in the Greek part of the text, but address the taxpayers in both the Arabic and Greek in the imperative (P.Ness. III, nos. 60–67; P.Heid.Arab. I, nos. a-l; Sammelbuch 18, no. 13218 and Becker, “Papyrusstudien”; Delattre, Pintaudi, and Vanthieghem, “Entagion bilingue”; Gonis and Morelli, “Two entagia,” p. 18 and nos. 1 and 2.). The Arabic and Greek parts of a bilingual release of labour contract are both constructed in the third person (P.Ness. III, no. 56, dated 67/687, provenance Naṣtān/Nessana. After consulting the original in the Piermont Morgan Library, I could conclude that the Arabic starts with hādhā mā, the equivalent of tauta ta in the Greek). A re-edition of this papyrus is being prepared by Robert Hoyland.
20 Only in the Arabic part of the entagia for the inhabitants of Naṣtān/Nessana is it stated that
the requisitioned oil and wheat are intended for the Arabs’ rizq (P.Ness. III, nos. 60–67). In two receipts, the Arabic notes that the payment was made to representatives of the pagarchs mentioned in the Greek (Sammelbuch 6, no. 9576 and Grohmann, From the world of Arabic papyri, pp. 113–114, fig. 1 and appendix 1 here, l. 6, dated 22/643, provenance Ihnās/Heracleopolite;
Sammelbuch 18, no. 13771 and Stoetzer and Worp, “Zwei Steuerquittungen,” no. 2 l. 3, dated
72/692 or 88/707, provenance Ihnās/Heracleopolite, with a corrected reading for lines 3–4:
khalīfa ṣāḥibay Habūyās, perhaps for the Greek Apiōn). While the Greek of a receipt merely states
that there were sixty-five sheep, the Arabic writes that there were fifty sheep for slaughtering (min al-juzur) and fifteen other (ukhrā) sheep (Sammelbuch 6, no. 9576 and Grohmann, From the
world of Arabic papyri, pp. 113–114, fig. 1 and appendix 1 here, ll. 2–3 vs 6–7, dated 22/643,
provenance Ihnās/Heracleopolite). The Arabic of this same receipt also specifies that the sheep are intended for “crew, squadrons and heavy armed soldiers” (aṣḥāb sufunihi wa katāʾibihi wa
thuqalāʾihi) where the Greek simply defines the recipients as “Saracens”. In an entagion
demanding the salary and maintenance of two and a half sailors to bring back other sailors having worked on the caliph’s fleet in North Africa (Ifriqiyā), the Greek gives a more extensive description of the community on whom the contribution is imposed (ahl Babūla min madīnat
are, for example, often written twice, in letters and numerals, adding other words such as “only” for surety.21 One striking example is the expression that
appears in the Greek part of the receipt for sixty-five sheep dated 22/643 which mentions that the document is written up as evidence of the transaction described therein, a phrase which is lacking in the Arabic, perhaps because of the Islamic legal condition that only witness statements, not documents, can be used as evidence.22 Interestingly, this phrase appears, on the other hand, relatively
frequently in other official Greek and Coptic papyri from the Islamic period, as opposed to the pre-Islamic period.23 Finally, the Arabic and Greek parts use
different calendars, the Greek using Coptic month names and indictions, the Arabic using Arabic month names and lunar years in a continuous calendar starting in 622 CE.24
The standardisation of Arabic formulae in the chancery at such an early period suggests broad usage and wide application. Moreover, some of the formulae and vocabulary of the early papyrus letters from Egypt also appear in contemporary documents found in the eastern Islamic empire, as well as in textual sources from al-Andalus, pointing to a common pre-Islamic origin (rather than originating with pre-Islamic Egyptian practice).25
the two kinds of boats used in the caliphal fleet (karaboi and akatēnaria. For the latter, see Morelli, “Legname,” p. 167, n. 4.). The Arabic mentions simply ‘boats’ (sufun) (Sammelbuch 18, no. 13218 and Becker, “Papyrusstudien,” dated 95/714, provenance Ishqūh/Aphroditō). The Greek and Arabic parts of the release of a labour contract are obviously written to fit different legal conditions, using different legal formulae. The Arabic is followed by the names of three male Muslim witnesses (following Muslim/Arabic practice?), while the Greek has one Christian male witness (P.Ness. III, no. 56, dated 67/687, provenance Naṣtān/Nessana).
21 Sammelbuch 6, no. 9576 and Grohmann, From the world of Arabic papyri, pp. 113–114, fig. 1 and
appendix 1 here, l. 3, dated 22/643, provenance Ihnās/Heracleopolite; Sammelbuch 16, no. 13018 and Stoetzer and Worp, “Arabisch-griechische Steuerquittung,” dated 93/714, provenance Ashmūn/Hermopolite. See also the Arabic “seventy mudd of wheat and the same of oil” vs the Greek “70 modii of wheat, 70 sextarii of oil” in P.Ness. III, no. 60 l. 7 vs 15, dated 54/674, provenance Naṣtān/Nessana. Cf. the description of the taxes paid in solidi in two ways, seemingly calculated and weighed in the Greek part of a receipt, while the Arabic seems to give only the total amount (Sammelbuch 18, no. 13771 and Stoetzer and Worp, “Zwei Steuerquittungen,” no. 2, dated 72/692 or 88/707, provenance Ihnās/Heracleopolite).
22 Schacht, Introduction, p. 193. 23 Morelli, Griechische Texte XV, p. 30.
24 See also the discussion of the ways that the Arabs’ calendar was described in the different
linguistic realms in the Muslim Empire, in Grohmann, Arabische Chronologie and Rāġib, “Ère inconnue.” For the indictions, see Bagnall and Worp, Chronological systems.
Multiple Arabic writing styles
The early Arabic documents thus show a certain standardisation in the formulae used albeit including evident variations and which are obviously different from local existing practices. The script and orthography on the other hand lack such likeness.26 The script of these earliest documents, while having been
characterised as “rather clumsy”, has nevertheless systematically applied letter shapes and a well-developed and regular form, although executed with a great variety of styles (see figs. 1-6).27 It is clear that different Arabic scripts or writing
styles were in use in Islamic Egypt, as attested in the papyri, which can be connected to Arabian inscriptions dating back to the pre-Islamic period. The same applies to the language or orthography attested in the papyri (see below). As scripts continued to develop, one writing style appears to have become dominant, eventually pushing alternatives out of use. The role of the chancery seems to have been crucial in this process. This was, however, not a linear development and archaic letter forms appear next to later shapes in earlier texts and continued to be used even when the latter had become dominant. Such parallel existence of multiple writing and script styles has been observed in other linguistic domains and historical periods as well.28 It shows that time was only
one factor in the evolution of the Arabic script which will become clear as we discuss the changes that occurred in the Arabic of the papyri below.
Most letter shapes typical for papyri from the seventh and eighth century are common across the written corpus of the first two Muslim centuries. This applies, for example, to the highly vertically extended lām and alif; the initial
ʿayn/ghayn with a horizontally extended bottom line, an open circle-formed loop
in mīm, which in its final form has a very short tail; horizontally extended
ṣād/ḍād/ṭāʾ/ẓāʾ; the writing of the fāʾ with one dot under and qāf with one dot over
26 See also George, Calligraphy, pp. 32-34.
27 For descriptions of the early script, see especially Khan, Selected Arabic papyri and Gruendler,
Development. Cf. Rāġib, “L’écriture”; idem, “Anciens papyrus”; Abbott, Ḳurrah papyri; idem, Rise of the North Arabic script and Sijpesteijn, “Arabic palaeography.”
28 For sixth and seventh-century Arabic inscriptions from Arabia (al-Shdaifat et al., “Arabic
graffito,” p. 315) and for much earlier Dadanitic inscriptions from Arabia (Macdonald, “Ancient North Arabian alphabets,” p. 5 and Kootstra, Writing culture). Older and later letter shapes can appear in one and the same legal document from eighth-century Khurāsān (Khan, Early Islamic
Khurasan, p. 67). The dome of the rock mosaic inscription shows three writing styles that reflect
different stages in applying standardised proportional letter shapes (George, Calligraphy, pp. 60-68).
the letter; the horizontally aligned dots above the letter shīn; and an oval-shaped medial, initial and freestanding hāʾ with a vertical extension on the right upper side (figs. 1-6).29 The extension of connecting strokes between letters and the
splitting of words over two lines occurs in Arabic papyri from the first two Muslim centuries.30 Some forms are only found in the seventh-century papyri.
These are the open medial ʿayn/ghayn (figs. 1 and 2);31 the medial jīm/ḥāʾ/khāʾ,
consisting of an oblique stroke through the baseline (figs. 1 and 3); a final mīm with the short tail extending in the direction of writing (figs. 1 and 2);32 and
medial and initial hāʾ written with a long vertical extension on the right side (see figs. 1, 2 and 3). The writing in the earliest papyri has a horizontal orientation; that is to say, all letters follow each other on the base-line and there is no stacking of letters. These letter shapes and writing conventions remind us of the pre-Islamic Arabic inscriptions and also appear in early pre-Islamic Arabic inscriptions.33
Some, such as the tail of final mīm extending into the direction of writing, as well as the absence of stacked letter combinations, can also be found in early Qurʾānic manuscripts.34
As discussed above, archaic and more developed letter shapes exist side by side in seventh- and eighth-century texts. The dāl/dhāl is an example of this.
Dāl/dhāl written with an upward bend at the top is characteristic of papyri from
29 Alif and lām: Khan, Selected Arabic papyri, pp. 27-29; shīn: Khan, Selected Arabic papyri, p. 31;
ṣād/ḍād/ṭāʾ/ẓāʾ: Khan, Selected Arabic papyri, pp. 31, 41; ʿayn/ghayn: Khan, Selected Arabic papyri, pp.
32-33; mīm: Khan, Selected Arabic papyri, p. 42; hāʾ: Khan, Selected Arabic papyri, p. 42.
30 Grob, Documentary; Khan, Selected Arabic papyri, p. 39. Imbert, “Réflexions,” p. 120. 31 Khan, Selected Arabic papyri, p. 33.
32 See also, Grohmann, Arabische Chronologie, pl. II.3, dating to 645-650; Bruning, “Legal Sunna,”
dated 44/664-665. Imbert, “Réflexions,” p. 120.
33 For how these letter shapes of the early papyri compare to those from pre-Islamic inscriptions,
see Khan, Selected Arabic papyri, pp. 27–39, and Nehmé, “Nabatean script.” For early Islamic inscriptions showing the same letter shapes, see for example the graffito mentioning the death of the caliph ʿUmar from 24/645 (Ghabban, “Naqsh”); the inscription dated 58/678 commemorating the building of a dam near Ṭāʾif in Arabia under the caliphate of Muʿāwiya (George, Calligraphy, fig. 12); the graffito mentioning the murder of caliph ʿUthmān dated 36/656 (Imbert, “Califes, princes et poètes,” fig. 3); and the 7th-8th-century Arabic graffiti from Arabia
(al-Ghabbān, Kitābāt); Imbert, “Réflexions.” Gruendler compares the shapes of letters in pre-Islamic and early pre-Islamic inscriptions as well as papyri, but many new inscriptions have come to light since she completed her work (Development).
34 See, for example, in BNF Arabe 30 (Image available online at:
gallica.bnf.fr/ark:/12148/btv1b8415208w/f117.item.r=Arabe%20330.zoom, accessed August 15, 2018). I would like to thank Marijn van Putten for pointing this out to me.
the seventh and early eighth centuries. The form without the upward bend, which eventually becomes dominant in the eighth century, is already attested, however, in some seventh-century papyri.35 The hairpin-shaped medial and
initial kāf is attested in the seventh century and, besides the later dominant form with an oblique or straight stroke, continues to be used in the eighth.36 Similarly,
final qāf with a long dipping tail occurs occasionally in eighth-century papyri, even though the letter is much more frequently written then with a shorter tail.37
A straight final alif connected to the previous letter without lifting the pen from the papyrus appears together with a final alif written separately in a vertical or straight line, sometimes with the bottom part turning to the right in seventh and eighth-century papyri (table 3).38 Freestanding alif with an angular bend to the
right at the bottom, frequently occurring in early Islamic Arabic inscriptions, occurs side by side a straight vertical alif (table 3) which would eventually become the norm.39 Such archaic letter forms can occur in texts that otherwise
show more developed letter shapes, and differently written forms of the same letter can even be combined in one text. For example, in the basmala of late seventh-century papyri, raḥīm is sometimes written with a final mīm with a very short tail pointing in the writing direction, while in bi-sm the final mīm is written with a downwards pointing tail (see below, table 4). Even the short tail of final
mīm can sometimes point more downwards than horizontally (see below, table
4). Similarly, angular and curved forms of the dāl/dhāl can appear side by side (see below, table 2). In fig. 3 the name Amjad is written once with the mīm above the jīm, once with the mīm and jīm on the same base-line and jīm with an oblique stroke through the base-line. An alif with an angular bend to the right at the bottom occurs side by side straight ones in an administrative letter dated 65/684-685.40 Amongst the documents produced in the chancery of the governor Qurra
35 Khan, Selected Arabic papyri, pp. 30-31. Cf. P.Berl.Arab. II, no. 23, dating to the seventh century. 36 Khan, Selected Arabic papyri, p. 36.
37 Ibid., p. 34.
38 E.g., in the seventh-century papyrus P.Berl.Arab. II, no. 23. See below under eighth-century
chancery writings. Cf. Khan, Selected Arabic papyri, p. 29.
39 The papyri do not show the same pronounced angular rightwards bend as the inscriptions.
For the inscriptions, see, for example, Ḥaṣān, “Mukhtārāt,” nos. 4, p. 23 and 13, p. 27, al-Ghabbān, Kitābāt; Imbert “Califes, princes et poètes,” fig. 3; al-Ḥaṣān, “Mukhtārāt,” 27, no. 13; George, Calligraphy, fig. 12 as well as the milestones erected under ʿAbd al-Malik’s reign (George,
Calligraphy, figs. 44, 45).
b. Sharīk, the entagia attest the angular bent, freestanding alif as do some of the governor’s letters, but not all.41
While some generally recognisable letter shapes are restricted to papyri from the seventh and eighth century, subtle differences between them indicate that multiple writing traditions existed side by side with individual scribes sometimes combining them in one document. The differences are observable in the writing instrument (thinner versus thicker qalams), the direction of writing (orientation), indidividual letter shapes and the combination of letters, and the use of a fixed letter combination or logogram to represent ibn, son of, in names. The orthography, especially the use of wawation — the adding of wāw to Arabic anthroponyms –, as well as the use of tāʾ rather than hāʾ in feminine endings, the occurrence of medial long ā, the application of alif for yāʾ to represent final long
ā, and the writing of shayʾ with alif (ياش) also shows that multiple writing
standards co-existed.42
Although Arabic texts are generally written with a thicker qalam than contemporaneous Greek ones, a feature especially observable in bilingual texts as mentioned above, within the Arabic corpus both thicker and thinner pens were in use. The thicker ones are generally used in documents where the direction of writing is diagonal and the letter shapes (especially the dāl/dhāl, see below) are written in a more angular way without any stacked letters (figs. 2 and 8).43 Thick and thin qalams, diagonally and more vertically oriented, as well as
angular and more cursive writing existed, however, side by side in the seventh century and could even be used by different scribes in one document (fig. 3).
41 For the entagia, see P.Cair.Arab. III, nos. 160, 162, 163, for his letters, P.Qurra, no. 15, P.Cair.Arab.
III, no. 147. All are dated 91/709-710. It also occurs in some seventh-century papyri from Nessana (Hoyland, “Earliest attestation,” P.Ness. III, no. 60, dated 54/674) and in an Egyptian administrative papyrus dated 65/684-685 (Diem, “Gouverneur”). Other seventh-century Arabic papyri attest straight, vertical, freestanding alifs.
42 For the use of wawation in pre-Islamic Aramaic inscriptions and its disappearance in the
Islamic period, see al-Jallad, “One wāw to rule them all”; Hopkins, Grammar, pp. 55-56; Grohmann,
From the world of Arabic papyri, pp. 88-89. For the writing of feminine ending with tāʾ maftūḥa, see
Hopkins, Grammar, pp. 44-45, Grohmann, From the world of Arabic papyri, p. 88. The Arabic inscriptions on milestones from Syria erected under ʿAbd al-Malik also typically show this feature (e.g. Sharon, CIAP 2, pp. 5; 3, 104-105). The writing of medial and final long ā is dealt with in Hopkins, Grammar, pp. 11-14 and 14-15. The writing of shayʾ is discussed in Hopkins, Grammar, pp. 17-18, Grohmann, From the world of Arabic papyri, p. 88 and Diem, “Untersuchungen II,” pp. 103-104. These features are more elaborately discussed here below.
43 Cf. P.Ness. III, no. 60. But see also, the papyrus dated 22/643 written with a thick qalam in a
Another feature that seems to represent different writing traditions is the way in which (i)bn is written in names. In many papyri of the seventh century (i)bn is written with an idiosyncratic, fixed shape. Starting with a rounded hook for initial bāʾ, it is followed by an oblique or vertical stroke extending above and below the line for final nūn. The vertical line sometimes extends far below the baseline (see in table 1, below).44 Sometimes an alif is added to the word (see figs.
1, 2 and table 1, below). The letter shape of final nūn (?) in ibn differs profoundly from other final nūns in these papyri.
The letter combination brings to mind the Aramaeogram bar (son) used in pre-Islamic Arabic inscriptions from Arabia. This particular writing of bar in Nabatean Aramaic inscriptions with a hook for bāʾ and a short vertically extending line for rāʾ initially co-existed with other writings of bar although the former became increasingly popular over time.45 In Arabic inscriptions written
in Nabatean and Nabatean-Arabic script this particular form of bar continued to be used with the meaning “son of” while it was probably pronounced as Arabic
ibn. 46 It seems that the same logogram remained popular amongst some scribes
of Arabic papyri, early Islamic Arabic inscriptions and Qurʾānic manuscripts.47
The final letter in the logograms attested in the Arabic papyri and other early Islamic writings, moreover, compares well with the writing of the rāʾ in some early Arabic inscriptions, although the vertical extension in those cases
44 Additional examples can be found in P.Ness. III, no. 60, dated 54/674; P.Ryl.Arab. I XV, no. 39,
plate 30, edited in Diem, “Gouverneur,” dated 65/684-685; Hoyland, “Earliest attestation,” dated 60s/680s; Moritz, Arabic palaeography, pl. 101, dated 87/706; P.Khalili II, no. 97, dated 104/723;
P.Sorrow, no. 30, 8th c.
45 See the different examples in Nehmé, “Nabatean script,” p. 53 and fig. 4. For the tendency for
bar to be written increasingly with a short vertical rāʾ, see pp. 61-63. Interestingly, the
sixth-century pre-Islamic Arabic inscriptions from Zebed (south of Aleppo), Harran and Jabal Says (both south of Damascus) all show the use of the small curved r in bar as in other places where the r appears. I would like to thank Michael C. A. Macdonald and Robert Hoyland for pointing this out to me.
46 See the photographs of the sixth-century Arabic inscriptions located in the North of the
Arabian Peninsula published on the website: http://alsahra.org/?p=17938, accessed July 7, 2018). I would like to thank Ahmad al-Jallad and Abdullah AlHatlani for confirming for me the continued use of bar in Arabic inscriptions and the latter also for referring me to this website.
47 For early Islamic inscriptions and papyri, see below, table 1 al-Ḥaṣān, “Mukhtārāt,” nos. 3 and
4, p. 27, dating to the early and late Umayyad period respectively and al-Ṣanduq, “Ḥajar Ḥafnat al-Abyadh,” p. 214. For examples in early Qurʾānic manuscripts, see BNF Arabe 330, folio 55r and DAM 01-29.1, folio 8r (both Q 4:157 ميرمنباسيع). I would like to thank Marijn van Putten for pointing me to these examples.
extends equally far above and below the line (see below, in table 1: al-Shdaifat et
al., “Arabic graffito” and al-Ḥaṣān, “Mukhtārāt,” nos. 3 and 4, p. 27). On the other
hand, the rāʾ in the early Arabic papyri invariably has a small crescent shape (see below, table 1). In pre-Islamic and early Islamic Arabic inscriptions both a vertically written and crescent-shaped rāʾ are attested.48 In other words, while
the combined letter shape in the Arabic papyri is not exactly the same as the Aramaeogram bar and the logogram was clearly adjusted in Islamic writings, it continued to constitute an idiosyncratic symbol which was applied in a similarly fixed form as it occurs in the pre-Islamic Arabic inscriptions. This fixed letter combination was based on the Nabatean-Aramaic writing of bar, but no longer understood as such; it was pronounced as ibn and interpreted as “son of” by the Arabic scribes of the Islamic period. The letter combination had obtained a fixed shape, with the final letter r neither looking like final rāʾ nor like final nūn as they are attested in Arabic writings from the Islamic period. Eventually, as will become clear below, the logogram developed further and then went out of use entirely, being replaced by a conventionally and recognisably written ibn. Compare the following examples from inscriptions and papyri for the early shapes of these letters:
48 Al-Shdaifat et al., “Arabic graffito,” p. 317; al-Ghabban, Kitābāt. See also in the milestones
erected during ʿAbd al-Malik’s reign (George, Calligraphy, figs. 44-45; Sharon CIAP 2, pp. 5; 3, 95; 104-105).
49 Photograph credit of CNRS/SIFAP DAJ, Guillaume Charloux. 50 Al-Shdaifat et al., “Arabic graffito.” Accessible online:
onlinelibrary.wiley.com/doi/full/10.1111/aae.12105, accessed August 16, 2018.
51I would like to thank Michael C. A. Macdonald and the Epigraphy and landscape in the hinterland
of Taymāʾ project, which discovered the inscription, for permission to use it. For a full image of
the inscription, see Imbert, “Califes, princes et poètes,” fig. 3. I would like to thank Abdullah AlHatlani for pointing me to this example.
Table 1: logogram ibn versus shapes of final nūn and rāʾ in inscriptions and papyri
52 The cutouts are taken from P.Vindob. AP 519 © Österreichische Nationalbibliothek. 53The cutouts are taken from P.Ness. III, no. 60 (The Morgan Library & Museum, Colt Pap. 60.
Photography by Graham S. Haber). For an edition and discussion of this papyrus, including its date, see Hoyland, “Earliest attestation.”
54The images are taken from P.Gen.Inv., no. 609 to be published as P.Gen.V, no. 1. I would like to
The Arabic papyri show, on the other hand, that the writing and understanding of the logogram bar changed in the Islamic period. The addition of the alif to the logogram suggests that the letter combination was perceived (and pronounced) as ibn and no longer perceived of as the Aramaic bar. In addition, as is visible in the examples from fig. 4 in the table above, although the logogram maintains a distinctive visual character, the final letter mutated from the Aramaic rāʾ. It was already mentioned that the letter extends further below the line than the rāʾ in contemporary Arabic inscriptions and in some cases it in fact approaches the writing of final nūn suggesting it was already seen in some way as the combination tooth-final nūn. This is confirmed by two attestations where the writing of tooth-final nūn in another word appears to be written exactly as the logogram ibn, while final nūn (not connected to a preceding bāʾ, tāʾ, thāʾ or yāʾ) in other letter combinations appears with a much deeper and curved letter shape (see in table 1, Chrest.Khoury I, no. 48; Delattre and Vanthieghem, “Ensemble archivistique,” fig. 6, dated 694-695). However, this can be compared with fig. 1 in table 1, where the same words appear — here ibn is written with a vertically extended final nūn and ʿashrīn with a deep, curved final nūn. As well as in seventh- and eighth-century Arabic papyri from Egypt, this distinctive writing of ibn is also attested in Arabic papyri from Nessana dating to the second half of the seventh century (table 1: P.Ness. III, no. 60). It also appears, as was mentioned before, in two early Qurʾānic manuscripts and Arabic inscriptions from the Islamic period.
Also in the letter shapes of individual letters subtle differences can be observed revealing different writing traditions. The upward turned bend of the
dāl/dhāl, which is so characteristic of papyri dating to the first two Islamic
centuries, is written with different degrees of angularity sometimes occurring in the same text. The following examples from papyri dating to the seventh century can be compared; it is striking that the angular dāl/dhāl, cursive writing direction, and use of a broader qalam are frequently combined.55
Fig. 1 خا د ان ىدمج Fig. 2 هش د دهش دهش Fig. 3 هش د ديبع دجملا Table 2: dāl/dhāl
In other letter shapes too we find variations which appear in combination with other elements suggesting different writing traditions that originated regionally. The extending of connected alif below the line and bending to the right of the bottom tip differs from the one directly connected to the previous letter. The former shows clearly that it was written by taking the qalam from the papyrus and starting from the top, while the latter sometimes does not show any extension below the line as if the alif is written from the bottom up connected to the previous letter. Similar variation is encountered in freestanding alifs with straight ones and those with an angular bend to the right at the bottom, both occurring in the papyri. Straight stand-alone alifs occur exclusively in the two earliest Arabic papyri dated to 22/643 as well as some other ones from the seventh century, both from Egypt and Palestine (table 3). Both alifs — with and without a bend at the bottom towards the right — appear in seventh-century Arabic inscriptions.56 The extension in papyri is, however, never as pronounced
56 For straight alifs, see in table 3 and the tombstone from Egypt dated 31/652 (George,
Calligraphy, fig. 11); for alifs with a bend at the bottom towards the right, see the inscription from
Khashnain in the Ḥijāz, dated 52/672 (George, Calligraphy, fig. 35), an early Umayyad graffito from al-Mafraq or Maʿān in Jordan (al-Ḥaṣān, “Mukhtārāt,” no. 4, p. 23), and the one near the dam of Ṭāʾif dating to 58/678 (George, Calligraphy, fig. 12).
as it is in the inscriptions. In both cases, the alif shows the seventh-eighth-century characteristic of being extended high above the line.
Table 3: final and freestanding alif
The writing of medial jīm/ḥāʾ/khāʾ with an oblique stroke through the baseline is opposed to the writing that became more prominent later with a ‘simple’ triangle shape. Both forms can be found in pre-Islamic Arabic-Nabatean inscriptions.57
They also both occur in early Islamic Arabic inscriptions.58 Final yāʾ is attested
57 Nehmé, “Nabatean script,” p. 49.
with and without an extended backwards pointing tail.59 In Arabic-Nabatean
inscriptions it is written with a backwards pointing tail.60 This is also how it
occurs in early Qurʾānic manuscripts.61
As for orthographic variation in the early Arabic papyri related to an interface between Aramaic and Arabic which can be traced in pre-Islamic inscriptions and early Islamic writings, the case of wawation in names has been best studied. The addition of a final wāw to names disappears in Arabic, but still occurs in a papyrus dated 22/643 in the name Ḥudayd(ū).62 Another phenomenon
in this category is the writing of feminine endings with a tāʾ maftūḥa rather than a hāʾ or tāʾ marbūṭa.63 While the use of tāʾ is standard practice in the status
constructus of feminine words in Aramaic and Nabateo-Arabic inscriptions, hāʾ is attested more frequently in Arabic inscriptions from the immediate pre-Islamic and pre-Islamic period.64 In papyri the writing with hāʾ is also dominant.
There are nevertheless still some occurrences of tāʾ in papyri and inscriptions pointing at the continuation of the older practice. Early papyri, coins and inscriptions attest a handful of cases in which the feminine ending is written with a tāʾ maftūḥa, including one in a third/ninth-century list of taxpayers.65 Raḥmat
allāh and ibnat written with tāʾ maftūḥa appear relatively frequently in later
59 Gruendler, Development, pp. 114-115. 60 Nehmé, “Nabatean script,” p. 50. 61 George, Calligraphy, pp. 50-51.
62 For the use of wawation in pre-Islamic Aramaic inscriptions and its disappearance in the
Islamic period, see al-Jallad, “One wāw to rule them all”; Hopkins, Grammar, pp. 55-56; Grohmann,
From the world of Arabic papyri, pp. 88-89; Diem, “Untersuchungen III,” pp. 359-360.
63 Hopkins, Grammar, pp. 44-45, Grohmann, From the world of Arabic papyri, p. 88.
64 For Nabateo-Arabic inscriptions, see Nehmé, “Aramaic or Arabic?,” pp. 84-87 and Diem,
“Untersuchungen III,” pp. 355-357. For sixth-century Arabic inscriptions, see ibid., pp. 364-365 and for Qurʾānic manuscripts, see ibid., pp. 378-381. The use of hāʾ to indicate feminine endings is considered an innovative feature of Arabic (Nehmé, “Aramaic or Arabic?,” p. 84; Robin, “Réforme.”).
65 The following feminine words are written with tāʾ maftūḥa in status constructus: sanat on a
tombstone dated 31/652 found in Egypt (El-Hawary, “The most ancient Islamic monument”) and in two graffiti, one dated 70/869 located in modern-day Jordan (al-Ḥaṣān, “Mukhtārāt,” pp. 21-22), the other in Saudi Arabia dated 100/718 (al-Ghabbān, Kitābāt, pp. 187-188, no. 116); khalīfat on coins dated 75/694 (Hopkins, Grammar, p. 17 n. 1; 44); ḍarāʾibat on a papyrus dated 90/709 from Qurra b. Sharīk’s chancery(P.Heid.Arab. I, no. 9). For later examples, see below nn. 103-104.
centuries, but this seems to belong to a different category.66 Final long ā is
sometimes written with an alif where classical Arabic uses final yāʾ.67 Defective
writing of medial ā was the norm in Aramaic and remains dominant in early Arabic papyri and inscriptions.68 Shayʾ is written with an alif before the yāʾ in
several early papyri.69 This way of writing is also frequently attested in early
Qurʾānic manuscripts.70
At this early stage, even though a standardised practice can be observed, there seems to be no strong distinction in scribal convention according to the genre of the text. In other words, private and official letters, protocols and legal documents all seem to be written in the same kind of script, albeit with many variations.71 Wether it be in the lay- out, formulae or expressions used, it
66 For ibnat, see Hopkins, Grammar, p. 45. A late example of raḥmat allāh written with tāʾ maftūḥa
is for example a legal document dated 797/1395 from al-Ḥaram al-Sharīf (P.Haram II, no. 34). The expression is also attested in the Geniza material dating to the same period.
67 In Aramaic final ā is generally written with yāʾ, sometimes with alif (Diem, “Untersuchungen
I,” pp. 214-219). For the papyri, see Hopkins, Grammar, pp. 14-15. The following words occur written with an alif in inscriptions and papyri (as well as with the classical Arabic yāʾ): Jumādā in an inscription dated 150/774 found in Jordan (Al-Jbour, Inscriptions, pp. 137-138); fatā in an inscription found in Saudi Arabia dated 90/709 (al-Ghabbān, Kitābāt, no. 38, p. 123) and in papyri from the 2nd-3rd/8th-9th centuries (P.Sorrow, no. 26, p. 184; Rāġib, “Lettres,” no. 9, 2nd/8th century;
P.Marchands III, no. 35, 3rd/9th century); iḥdā in two contracts, one on papyrus dated 271/884
(P.Vindob. AP 1102), another on parchment dated 461/1068 (Chrest.Khoury I, no. 59); ḥattā is attested many times written with an alif in all periods (see the Arabic Papyrology Database). I would like to thank Abdullah AlHatlani for pointing me to some of these examples.
68 The writing of medial and final long ā in the papyri is dealt with in Hopkins, Grammar, pp.
11-14 and 11-14-15. For the development of the writing of medial long ā in inscriptions and Qurʾānic manuscripts, see also George, Calligraphy, pp. 30-31.
69 Hopkins, Grammar, pp. 17-18, Grohmann, From the world of Arabic papyri, p. 88 and Diem,
“Untersuchungen II,” pp. 103-104.
70 Diem, “Untersuchungen II,” pp. 103-104.
71 See also the comparisons between some letter shapes in the papyri and that of the so-called
Ḥijāzī Qurʾāns (Abbott, Rise of the North Arabic script). There are not many Arabic documents datable with certainty to this early period: a receipt (Sammelbuch 6, no. 9576 and Grohmann,
From the world of Arabic papyri, pp. 113-114, fig. 1, provenance Ihnās/Heracleopolite), tax demand
note (Rāġib, “Un papyrus arabe”) and an unpublished text fragment (P.Vindob. AP 87) are all dated 22/643; under a protocol dating to 642-650, the first two lines of a notice about a wheat delivery are written (P.Mich. 6714. The protocol text is published in Grohmann, “Papyrusprotokoll,” p. 2. The Arabic text is unpublished. For a description, see Grohmann, “Papyrusprotokoll,” p. 3. For an image, see Grohmann, Arabische Chronologie, pl. III.1); P.Vindob. AP 94 is an unpublished papyrus fragment of an official document in the Austrian National Library datable to 25-30/645-650 (described by Grohmann, From the world of Arabic papyri, script
is difficult to distinguish between these different kind of texts. Moreover, there does not seem to be a clear distinction between the writing material and the shape of the letters. That is to say, the specific features of these early texts recur in different media, such as coins, inscriptions, papyri, manuscripts and so on.72
First developments towards a chancellery script
Papyri from the end of the seventh century CE, although exhibiting the general characteristics of the script of the first two Muslim centuries, can be distinguished from the earliest Arabic writings (see figs. 1, 2 and 3 versus fig. 4 and P.Ryl.Arab. I XV, no. 39 = Diem, “Gouverneur” in table 4 below.). In spite of the great variations even within documents, these texts, with their more rounded, more pronounced and different letter shapes, can be grouped together. Some letter shapes acquire different forms, whether because of developing writing practices or because one form becomes dominant. Amongst the most notable changes are the closing of the medial ʿayn and the development in the shape of the final and medial hāʾ, which can still have a short vertical extension on the right side but is tilting to the left, often taking a horizontally or vertically
table and Grohmann, Arabische Chronologie, p. 91. For an image, see Grohmann, Arabische
Chronologie, pl. II.3); six debt acknowledgements recorded in one document (Rāġib, “Ère
inconnue,” no. 1) are dated 42/662-663; a poll tax debt acknowledgement from the Louvre (Bruning, “Legal Sunna”) is dated 44/664-665; demand notes for oil and wheat dating to between 54/674 and 70/689 were sent to the inhabitants of Naṣtān/Nessana (P.Ness. III, nos. 60–67); a debt acknowledgement is dated 57/677 (Rāġib, “Ère inconnue,” no. 2), while two bilingual tax demand notes, entagia, can be dated to 57/677 and 59/679 respectively (Sammelbuch 18, nos. 13771; 13771 and Stoetzer and Worp, “Zwei Steuerquittungen,” no. 2; SPP VIII, no. 1198, provenance of both is Ihnās/Heracleopolite. See below n. 76 for the argument to date these two texts later than the accepted dates); a bilingual release of a labour contract is dated 67/687 (P.Ness. III, no. 56, provenance Naṣtān/Nessana). A commercial letter (Hanafi, “Business letter”), a letter concerning agricultural matters (unpublished P.Vindob. AP 26), and two other letter fragments (unpublished P.Vindob. AP 444 and AP 466), a debt acknowledgement (Chrest.Khoury I, no. 48) and a legal document (unpublished P.Vindob. AP 11154. For an image, see Grohmann, “The problem of dating,” pl. IIIb) can be assigned to the first half of the first/seventh century on the basis of the script.
72 See the examples from non-papyrological sources mentioned in Khan, Selected Arabic papyri,
pp. 27-39, Gruendler, Development, pp. 154-156, Grohmann, Arabische Paläographie II, Hoyland,
Seeing Islam, pp. 687-695, and Moritz, Arabic palaeography. Cf. Grohmann, From the world of Arabic papyri, p. 69; George, Calligraphy, pp. 32-34.
orientated heart shape. In inscriptions and on coins, the open medial ʿayn and oval-shaped medial and initial hāʾ, with pronounced vertical extension, are preserved, indicating that the orthography of papyri and inscriptions did not exactly coincide anymore.73 In the shape of the final mīm a variation can be
observed. While the final mīm in every case shows the characteristic short final tail, in some cases the tail points in the writing direction, in others it points downwards. It is this latter shape that will become prevalent, developing eventually into the regular downward-pointing longer tail used in documents in the eighth century. In Qurʾānic manuscripts of the eighth century, as well as on coins, by contrast the short-tailed final mīm remains dominant.
73 This seems more likely than that a “monumental epigraphic script” had developed which was
used consciously (Gruendler, Development, p. 131) as there does not seem to have been an alternative for writing on these kinds of materials and both graffiti and monumental inscriptions show the same features. For the hāʾ, see the graffito dated 56/676 (tracing in Gruendler, Development, p. 154 E3). For ʿayn, see for example the inscription on a bronze cannister from Baṣra dated 69/688-689 (a tracing appears in Gruendler, Development, p. 154), the tombstone from Aswan dated 71/691 (image in Grohmann, Arabische Paläographie II, pl. X.2), the milestones from the time of caliph ʿAbd al-Malik (r. 65–86/685–705) (images in Grohmann,
Arabische Paläographie II, p. 83; Sharon, CIAP 3, p. 95), an inscription from Ḥafnat al-Abyaḍ near
Karbalāʾ in Iraq dated 64/684 or the inscriptions in the Dome of the Rock (images in George,
Calligraphy, figs. 36, 37). In general for these lettershapes in the graffiti, see the images in
al-Ghabbān, Kitābāt, all dating to the 1st-2nd/7th-8th centuries. George and Imbert point at a
movement towards a more uniform and systematically applied Kūfic script in the third quarter of the seventh century developing further in the early eighth century in monumental inscriptions and Qurʾānic manuscripts (Calligraphy, pp. 60, 74ff;Imbert, “Réflexions,” p. 120).
Table 4: final mīm in basmalas (ميحرلانمحرلااللهمسب) and other words in seventh-century papyri
Another development is the introduction of stacked combinations of letters, which appear first in papyri dating to the last quarter of the seventh century (fig. 3; P.Ryl.Arab. I XV, no. 39). Such stacking of letters is only introduced in Qurʾānic manuscripts several centuries later, again showing how writing traditions according to genre had already diverged.75
In this period, separate linguistic scribal traditions persisted, with bilingual Arabic-Greek documents continuing to be written by different scribes for the Arabic and Greek parts, producing different formulae and expressions and using a thicker pen for the Arabic writing.76 Even in a papyrus dating to between
74The cut-out is taken from P.Ryl.Arab. I XV, no. 39, plate 30, edited in Diem, “Gouverneur.” I
would like to thank the curator of the collection for permission to use this image.
75 I would like to thank Marijn van Putten for pointing this out to me.
76 The tax receipts SPP VIII, no. 1198 and Sammelbuch 18, no. 13771 were both produced in name
of the Arab official ʿAbd al-Raḥmān b. Abī ʿAwf for a community of tax-payers. The script points to a date after the first half of the first Muslim century, while the presence of a Muslim (as opposed to Christian) fiscal official issuing the receipt to a community of tax-payers (as opposed to an individual tax-payer) places the documents in the late first/seventh or early second/ eighth century. This reduces the possible dates of these documents to 72/692 and 74/694 or
