Journal of Islamic Manuscripts 10 (2019) 292–326

brill.com/jim

A New Document in the Early History of the Qurʾān

Codex Mashhad, an ʿUthmānic Text of the Qurʾān in Ibn Masʿūd’s
Arrangement of Sūras

Morteza Karimi-Nia
Encyclopaedia Islamica Foundation, Tehran
[email protected]

Abstract

Codex Mashhad in the Āstān-i Quds Library (Mashhad), comprised of Manuscripts 18

and 4116, is possibly one of the most important documents for our understanding of
developments in the early history of Qurʾānic text. The combination of all features of
this codex as a whole is found in few copies of early Qurʾāns written in ḥijāzī style.
Considering the text, the spelling rules, the variant readings, the orthographic pecu-
liarities, and the arrangement of sūras, it may be concluded that the main part of this
codex was transcribed in a very early period, probably in the first/seventh century. Yet,
unlike other early Qurʾānic manuscripts, the whole Qurʾān in the initial status of Codex
Mashhad has been transcribed based on the official ʿUthmānic version but according
to Ibn Masʿūd’s arrangement of sūras.

Keywords

Codex Mashhad – Āstān-i Quds Library – ḥijāzī Qurʾāns – history of the Qurʾan – Ibn
Masʿūd’s Qurʾān – arrangement of sūras

1 Introduction1

By reviewing the evidence and documents from early Islam, academic Qurʾānic
studies have in recent decades achieved new findings on the text of the Qurʾān.
This may not be pleasant to the revisionists but can draw their attention to

1 Submitted in an earlier form on 16 November 2017. Accepted for publication on 10 May 2019.
a new document in the early history of the qurʾān 293

the importance of these documents and evidence. The discovery and analy-
sis of new inscriptions in old Semitic languages in the Arabian Peninsula and
Mesopotamia, the discovery and analysis of some non-ʿUthmānic fragments of
codices from Yemen, the results of radiocarbon dating of some early Qurʾānic
manuscripts, and, finally, the publication of some of the ḥijāzī-style codices
of the Qurʾān are among the achievements that have attracted both Muslim
and Western scholars. While some of these documents confirm our past knowl-
edge, which is mainly based on Islamic literary sources, others help us rebuild
or correct our information. More detailed information on the occurrence and
prevalence of rare variant readings (shawādhdh), the differences between the
regional scripts of the Qurʾān, the evolution of diacritical marks (al-naqṭ and
al-iʿrāb) in the text of the Qurʾān, and the use of illuminations while copying
the scripture will all be revealed through studying the most ancient fragments
and manuscripts of the Qurʾān.
Codex Mashhad is one of the most important remaining documents from
the first Islamic centuries that can give us a new and more inspiring picture of
the history of Qurʾānic text. In this article, the term Codex Mashhad refers to an
old codex of the Qurʾān, now mostly preserved in two manuscripts, MSS 18 and
4116, in the Āstān-i Quds Library. The first manuscript in 122 folios and the sec-
ond in 129 folios together constitute more than 90% of the text of the Qurʾān,
and it is also likely that other fragments will be found in Mashhad or elsewhere
in the world.2
In its current form, Codex Mashhad contains an ʿUthmānic text of the
Qurʾān in the ʿUthmānic arrangement of the sūras. Regardless of some variant
readings and spelling differences, the current form and content of the Codex
corresponds largely to the ʿUthmānic text, the current and standard text of the
Qurʾān among Muslims at all times in history, known by and attributed to the
third Caliph, ʿUthmān b. ʿAffān. However, this cannot be true of the initial form
of the Codex. There is some evidence in both volumes, MSS 18 and 4116, that
allows us to argue that the initial order of the sūras in this codex was not in
accordance with the current ʿUthmānic order. As we will see later in this article,
its initial order was in accordance with the Qurʾān associated with Ibn Masʿūd
(d. 32/653), a companion of the Prophet.
The four sections of this article deal with:

2 At the end of MS 18, there are two unbound, worn-out folios of the Codex, which are not num-
bered. The first contains the final verses of Sūra 19 (Maryam) and the beginning of Sūra 20
(Ṭāhā); the latter contains Q. 20: 57–67 and Q. 20: 72–82. I enumerate them as A121 and A122
of MS 18, respectively.

Journal of Islamic Manuscripts 10 (2019) 292–326

294 karimi-nia

1) the general features of the Codex;

2) the textual features, including orthography, verse division, variant read-
ings, all of which indicate its attribution to the first/seventh century;
3) the arrangement of sūras in both the initial and current forms of the
Codex, and its relationship to Ibn Masʿūd’s arrangement; and,
4) the historical explanation and dating of the Codex.

2 The General Features of the Codex

With the discovery of its 251 leaves, Codex Mashhad now covers more than 90
per cent of the Qurʾān. The current Codex is in two separate volumes, MSS 18
and 4116.3 The former contains the first half of the Qurʾān, from the beginning
to the end of the 18th sūra, al-Kahf, while the latter comprises the second half,
from the middle of the 20th sūra, Ṭāhā, to the end of the Qurʾān. We do not
know for sure why these two parts are bound separately. In the process of restor-
ing and binding the volumes in recent centuries, there have been misorderings
of the leaves, apparently due to a lack of familiarity or knowledge on the part of
the restorers or binders.4 For the sake of ease and consistency in our discussion,
a special system is required to refer to the various parts of this unique codex. In
referring to MS s 18 and 4116, I will use letters A and B, respectively. Thus, A12a
means folio 12 recto of MS 18 and B12b will be folio 12 verso of MS 4116.
We do not know for sure when, why, and how this Codex emerged from the
Ḥijāz or Iraq5 and by what process it was brought to Khurasan; however, the
deed of waqf at the beginning of MS 18 and the editor’s name at the end of

3 To my knowledge, MS 18 with its deed of waqf at both the beginning and end of the volume
has been in the Shrine of Imām al-Riḍa, and later in the Āstān-i Quds Library in Mashhad,
but was subjected to restoration and rebinding in the Qajar Period (1789–1925). The second
volume, MS 4116, was found during the Shrine discoveries in 1348 solar/1969 and added to
the Library following restoration and rebinding. The cataloguers in the library have not yet
acknowledged that the two manuscripts comprise the same codex. The parchments in the
first volume, MS 18, are in good condition while the folios in the second volume, MS 4116, are
often worn-out and many parts of their margins are missing. The situation is similar to that
of Codex Ṣanʿāʾ 1, the worn-out part of which is preserved in Dār al-Makhṭūṭāt (Rev. 01–27.1)
while the better-preserved part is kept in al-Maktaba al-Sharqiyya, both in Ṣanʿāʾ.
4 For example, leaf A18 of MS 18, which contains Q. 2:282 to Q. 3:7, is an annexed transcript
in a later Kufic script. This folio must actually be placed after f. A19. Similarly, f. A98 must
be placed between f. A93 and f. A94. Also, the sides of leaves A57, A69, A75, and A111 are in
reverse order in the current binding and, accordingly, the front and back of the folios must
be transposed. Such misplacements are uncommon in MS 4116.
5 As we will see, the Codex was probably produced in the Ḥijāz and kept in Iraq for a while.

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 295

the first chapter, Sūrat al-Fātiḥa, indicate that this work has been present in
Khurasan probably from the early fifth/twelfth century onwards, and may have
been donated to the Shrine of Imām al-Riḍā in Mashhad probably at the end
of that century. Since Khurasan was at the centre of the political and religious
developments of the Islamic empire from the mid-second/eighth century, it is
easy to assume that such codices reached the region during the events of the
second/eighth to fourth/tenth centuries. Accordingly, it can be estimated that
this codex was transferred from Ḥijāz or Iraq to Khurasan during the reign of
the Buyids (320–447/932–1055), or even at an earlier time, when the Abbasid
Caliphs Hārūn al-Rashīd and al-Maʾmūn were ruling there.
Like many other old Qurʾānic manuscripts in Iran, this codex is also ascribed
to Imām ʿAlī (d. 40/661), which is obviously a historical mistake. The ascription
is mentioned by the owner-donor on the first leaf of MS 18 (f. A1a) in New Style
Kufic, though it can also be found on numerous folios of both manuscripts in
the later naskhī script.6 Along with other copies related or ascribed to the Shi’a
Imāms, both parts of the Codex are now kept in the Āstān-i Quds Library. The
format is vertical and the size is relatively large. Based on my measurements,
the folio size of MS 18 is 46.2×34.5 cm, and that of MS 4116 is approximately
40×30cm. Because of the marginal damage to the parchments, the surviv-
ing written surfaces are unequal in the two manuscripts and vary between
40×29 and 39×27cm. Each folio usually has 22 or 23 lines, but sometimes the
number of lines decreases to 20 or increases to 25. The number of lines, the
vertical format, and the size of the Codex are unique among old manuscripts
in the Āstān-i Quds Library, while these features are distinguishable in com-
parison with all other known ḥijāzī or māʾil manuscripts outside Iran. Like
MS ŞE 71 (41.2×36 cm) in the Turkish and Islamic Arts Museum (Istanbul), MS
Is 1404 (47 ×38cm) in the Chester Beatty Library (Dublin), MS DAM 01–29.1
(42×30) in DAM (Ṣanʿāʾ), and MS Marcel 3 (41 ×37 cm) in the National Library
of Russia (St. Petersburg), our Codex can therefore be recognized as a folio vol-
ume.
In their present form, both parts of Codex Mashhad have been repaired,
partially completed with pieces from later Kufic Qurʾāns and sometimes in
a present-day nashkī hand. Ignoring the leaves added during the restoration
process, this study focuses particularly on the main body, which is in an old
ḥijāzī-style script, classified by François Déroche as BIa.7

6 For example, on f. A29b, at the end of Sūra 3 (Āl ʿImrān) and on f. A83a, at the beginning of
Sūra 10 (Yūnus).
7 This type of handwriting is very close to the Ḥijāzī handwriting used in the Tübingen Manu-
script (Ma VI 165). Only in two other Qurʾāns in the Āstān-i Quds Library, MSS 26 and 3540,

Journal of Islamic Manuscripts 10 (2019) 292–326

296 karimi-nia

The first 119 leaves of MS 18 are parchment, but a paper leaf, f. A120a, is
found at the end, on which the final verse of Sūra 18 is written in a recent
hand [figure 3]. In the margins of this particular leaf, there are seven royal and
administrative stamps indicating the formal inspections of the manuscript in
the Qajar period, three of which bear the dates of Ramaḍān 1267 / July 1851,
Shaʿbān 1286 / November 1869, and 29 Ṣafar 1289 / 8 May 1872. Then, at the end
of this manuscript, there are two separate damaged and unrestored leaves, f.
A121 and f. A122, containing parts of sūras 19 and 20. In order to preserve it from
severe decay, nearly all margins of MS 18 have been restored with glued tapes
of paper on four sides, probably in the Qajar period. Accordingly, the first and
last lines of many folios (e.g., f. A118b) are partially or totally covered and so not
visible.8
The second volume, MS 4116, contains 129 leaves; its last seven leaves, from
the beginning of Sūrat al-Nāziʿāt (Q. 79) to the end of the Qurʾān, are not orig-
inal and have been added later in a different Kufic hand. Although the leaves
of MS 4116 suffer from more severe damage than those of MS 18, no glued tape
has been used in restoring the parchment; rather, the edges of all leaves were
repaired in the periphery with patches of paper in December 2011.
A deed of endowment with the signature of the owner of the manuscript,
ʿAlī b. Abī al-Qāsim al-Muqriʾ al-Sarawī, at the beginning of MS 18 (f. A1a) reads:

‫هذا المصحف | وهو بخط امير المومنين | على بن ابى طالب عليه السلم | وقف على مشهد السي ّد‬
‫الامام السعيد | الشهيد ابى الحسن على بن موسى الرضا رحمة الل ّٰه عليه | الموضوع بالطوس وقفه‬
‫مالـكه | علي بن ابى القسم المقرى السروى تقر با الى الل ّٰه عز وجل | وطلبا لمرضاته بلغ الل ّٰه اماله‬

This muṣḥaf, which is in the handwriting of Amīr al-Muʾminīn ʿAlī b. Abī

Ṭālib, peace be upon him, is donated to the Shrine of the Lord, the blissful
Imām, the Martyr Abū al-Ḥasan ʿAlī b. Mūsā al-Riḍā, may God’s mercy be
upon him, who is buried in al-Ṭūs. This endowment is made by its owner,
ʿAlī b. Abī al-Qāsim al-Muqriʾ al-Sarawī, seeking closeness to the Almighty
God and His satisfaction. May God fulfil his wishes [figure 1].

The script is Iranian or New Style Kufic. Later, a certain al-Warrāq al-Ṭabarī
has duplicated the text between the lines of the original waqfīyya in a smaller

do we find the BIa style; these two codices, however, are produced in oblong format and can
be dated back to the early Abbasid period in the second Islamic century.
8 This method of restoring with glued tapes of paper can also be seen in all leaves of MSS 54
and 55 and in the opening and closing parchments of MS 26 in the Āstān-i Quds Library.

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 297

figure 1 Codex Mashhad. The deed of waqf at the beginning (f. A1a)

Journal of Islamic Manuscripts 10 (2019) 292–326

298 karimi-nia

figure 2 Codex Mashhad. The editor’s name, Wajīh b. Ṭāhir al-Shaḥḥamī, at the end of the
Sūrat al-Fātiḥa (ff. A1b–A2a)

nastaʿlīq script. We cannot find any deed of waqf or sign of the donor in the
counterpart volume, MS 4116.
On f. 120a of MS 18, the deed of waqf is repeated in an old naskhī hand:

‫وقف مو بد الى الل ّٰه عز وجل بمشهد السيد الامام على بن مـ … | وقفه وملـكه علي بن ابى القسم‬
| … ‫بن الحسين المقرى السرو‬

[This is] an ever-lasting endowment to Almighty God in the mashhad [=

place of martyrdom] of the Lord al-Imām ʿAlī b. M[ūsā] […]. ʿAlī b. Abī al-
Qāsim b. al-Ḥasan al-Muqriʾ al-Sara[wī] was the owner of the Codex and
made the waqf [figure 3].

The name al-Ḥasan in the name of the wāqif mentioned here is missing from
the primary deed of waqf at the beginning of MS 18 (f. A1a). The waqf note on f.
A120a is followed by a few lines of a prayer text in Persian, in old orthography in
an old Ṭabarī dialect. On a number of pages there are more complete or partial
waqf texts in hands that are more recent than the Qurʾānic text of the Codex.
Examples can be seen on f. A105a [figure 5], f. A59b [figure 6], f. A83a [figure 9],
f. A29b [figure 10], f. A30a [figure 10] and f. A98b [figure 11].

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 299

figure 3 Codex Mashhad. The end of the first part with a repetition of the waqf deed (ff.
A119b (parchment)-A120a (paper))

Although historical sources provide no precise information, I surmise that

the wāqif must be Abū al-Ḥasan ʿAlī b. Abī al-Qāsim b. al-Ḥusain al-Muqriʾ, a
Qurʾān reader and scholar from Ṭabaristān, who was probably living in Khura-
san. He is mentioned in Sunni traditions, which indicate that he was a teacher
(shaykh) of al-Samʿānī (d. 562/1161).9 Moreover, Abū Bakr Aḥmad b. al-Ḥusain
al-Sarawī al-Muqriʾ, whose name is mentioned by al-Ḥākim (d. 405/1014) in his
Tarīkh Nīshapūr10 and who lived about 150 years before Abū al-Ḥasan ʿAlī in
Nishapur, was possibly of the same family.

9 See, for example, Abū al-Futūḥ al-Ṭāʾī al-Hamadānī, Kitāb al-Arbaʿīn fī Irshād al-Sāʾirīn ilā
Manāzil al-Muttaqīn, aw al-Arbaʿīn al-Ṭāʾiyya, edited by ʿAbd al-Sattār Abū Ghida, Beirut:
Dār al-Bashāʾir al-Islamiyya, 1420/1999, vol. 2, p. 155. See also Taqī al-Dīn al-Subkī, Ṭabaqāt
al-shāfiʿiyya al-kubrā, edited by Maḥmūd M. al-Ṭanaḥi, Cairo: Hijr li al-Ṭibāʿa wa al-Nashr,
1383/1964, vol. 6, p. 73.
10 Al-Samʿānī, al-Ansāb, edited by ʿAbd al-Raḥmān al-Yamānī et al., Hyderabad: Dāʾirat al-
Maʿārif al-ʿUthmāniyya, 1962–1982, vol. 3, p. 274. See also al-Ḥākim al-Nisābūrī, Taʾrīkh
Nīshapūr (revised Persian translated by Muḥammad b. al-Ḥusain al-Khalīfa al-Nisābūrī),
edited by M.R. Shafīʿī Kadkanī, Tehran: Āgāh Publishers, 1375 Solar/1996, p. 149.

Journal of Islamic Manuscripts 10 (2019) 292–326

300 karimi-nia

We find in the old bibliographical sources and catalogues11 two other schol-
ars named Abū al-Ḥasan ʿAlī b. Abī al-Qāsim, neither of whom can be the writer
of this endowment:
– Abū al-Ḥasan ʿAlī b. Abī al-Qāsim al-Bayhaqī, or Zayd b. Muḥammad b.
Ḥusain, known as Ibn Funduq, author of the Tarīkh Bayhaq, a theologian,
historian, literary scholar, astronomer, jurisprudent, and poet of the sixth/
twelfth century (ca. 490–565/1097–1170). The problem with this Abū al-
Ḥasan is that he was not a qāriʾ or muqriʾ (Qurʾān scholar), nor did he have
any connection to the city of Sārī in Ṭabaristān;
– Abū al-Ḥasan ʿAlī b. Abī al-Qāsim b. ʿAbdallāh b. ʿAlī al-Muqriʾ, who was born
in Zaragoza, lived in Toledo, and narrated hadīths in eastern Islamic lands.
In Qurʾān and tafsīr, he was a pupil of al-Māwardī (d. 450/1057). Although
he spent most of his life in the East and returned to Cordoba only in old age,
this Maghribī Abu al-Ḥasan died in 472/1079, much earlier than our period
of concern here. Nor did he have any connection to the city of Sārī.
On the back of the folio containing the deed of waqf, Sūrat al-Fātiḥa is writ-
ten in a later Kufic hand (f. A1b) in order to complete the beginning part of the
Codex that at a later time, near the fifth/eleventh century, apparently had gone
missing. At the bottom of the folio, it reads:

| ‫جدد هذه الـكتبه وجيه بن طاهر الشحامى | غفر الل ّٰه ⟩له و⟨ والديه‬

This writing was renewed by Wajīh b. Ṭāhir al-Shaḥḥāmī, may God forgive
[him and] his parents.

This is followed by taṣliya formula in a smaller script [figure 2]. Abū Bakr Wajīh
b. Ṭāhir al-Shaḥḥāmī (455–541/ 1063–1146) was a well-known religious scholar
in Khurasan and Nishapur. According to al-Dhahabī,12 he was a pupil of Abū
al-Qāsim al-Qushayrī (d. 465/1074) and also a teacher of such scholars as al-
Samʿānī (d. 562/1161) and Ibn ʿAsākir (d. 571/1176). This may show that the Codex
Mashhad had been in Khurasan since the fifth/eleventh century.

11 Yāqūt al-Ḥamawī, Muʿjam al-Udabāʾ, edited by Iḥsān Abbās, Beirut: Dār al-Gharb al-
Islāmī, 1414/1993, pp. 1759–1768; Al-Dhahabī, Taʾrīkh al-Islām, edited by ʿUmar A.S. al-
Tadmurī, Beirut: Dār al-Kitāb al-ʿArabī, 1413/1993, vol. 32, p. 72. See also Muḥammad
Qazvīnī’s introduction to Bayhaqī’s Taʾrīkh Bayhaq, edited by Aḥmad Bahmanyār, Tehran:
n.p. 1317 Solar/1936, pp. x–xxxiii.
12 Al-Dhahabī, Sīyar aʿlām al-nubalāʾ, edited by Shuʿayb al-Arnaʾūṭ, Beirut: Muʾassasat al-
Risāla, 1405/1985, vol. 20, pp. 109–111.

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 301

From f. A1b it is clear that that f. A1 is a later addition to the volume. That
means that also the waqf deed on f. A1a is written on an additional leaf.

3 The Textual Features of Codex Mashhad

Codex Mashhad has almost all the elements and features of the oldest known
Qurʾānic codices. The dual volumes of the main body, written in ḥijāzī or
māʾil script, are the only ḥijāzī manuscripts in vertical format in Iran. Like all
ancient ḥijāzī codices, Codex Mashhad contains variant readings, regional dif-
ferences of Qurʾānic codices, orthographic peculiarities, and copyists’ errors,
partly corrected by later hands. The script and orthography of the Codex show
instances of archaic and not-yet-completely-recognized rules, manifested in
various spelling peculiarities. Illumination and ornamentation are not found
even in sūra-headbands; rather, some crude sūra dividers have been added later
and are found only on adjoining sections.
In both parts of the Codex, the text of the Qurʾān mostly contains diacritical
points, but it is clear that some of these points have been added later. As we
will see, this sometimes allows for two concurrent readings of the text. How-
ever, because there is no difference of colour between these points and the
rest of the text, it is often difficult to differentiate the additional points from
the original ones. There are also vowel signs in the form of red dots, known
as Abu al-Aswad al-Duʾalī’s system, which are clearly later additions. There is
some evidence that these colour vocalizations were added later than the black
diacritical dots. Thus, sometimes, the double dots of letter tāʾ are placed verti-
cally and then above them there is a red dot for the vowel fatḥa. An example
can be seen in the words ‫( ت َکُن‬f. A37a), ‫( تعَ م َلوُ ن‬f. A38a), ‫( ٺتبع‬f. A66a), and ‫ٺثقفنّهم‬
(f. A73a).
Just as in most ḥijāzī manuscripts of the Qurʾān, the verse division in Codex
Mashhad was made at the time of the text was copied. The Codex Mashhad fea-
tures two types of verse ending: the scribe usually uses five or six strokes in a tri-
angular arrangement;13 in a few cases, three oblique strokes are used instead.14
Both these systems are sometimes observed on one folio. Examples are: f. A33a,

13 The triangular six-stroke sign is seen in many other Qurʾānic manuscripts, for example,
MS Arabe 331 in the Bibliothèque nationale de France, MSS Marcel 3, and Marcel 18 in the
National Library of Russia and MSS ŞE 264 and ŞE 3702 in the Turkish and Islamic Arts
Museum (Istanbul).
14 This diagonal three-stroke sign is also found in other ḥijāzī Qurʾāns, including the MSS
ŞE 3591 and ŞE 80 in the Turkish and Islamic Arts Museum (Istanbul).

Journal of Islamic Manuscripts 10 (2019) 292–326

302 karimi-nia

after the word al-muʾminīn; f. A115b, after the words mastūran and madḥūran;
and f. B123a, in Sūrat al-Nabaʾ.15 A red circle has later been added around every
tenth verse ending (taʿshīr), but there is no sign to represent groups of stanzas
in five lines (takhmīs). The instances of basmala at the beginning of all sūras
are marked clearly as independent verses.
Like many ḥijāzī and Kufic codices from the first two centuries, there is
always a gap of one line between the sūras. During subsequent centuries, the
names of sūras and the number of verses were added in these blank lines in
a distinct red script. Crude illuminations have also been added in some cases,
at the beginnings of the sūras, e.g. in Sūrat Yūnus (f. A83a) and Sūrat al-Isrāʾ
(f. A114a). The nomenclature of sūras, which is clearly a later addition, is in
some cases inconsistent with the current names. Instances of this inconsis-
tency include Ḥā-mīm al-Sharīʿa instead of al-Jāthiya (Q. 45),16 al-Nabī instead
of al-Taḥrīm (Q. 66),17 al-ʿIshār instead of al-Takwīr (Q. 81),18 al-Ḥafaẓa instead
of al-Infiṭār (Q. 82),19 al-Nāqa instead of al-Shams (Q. 91),20 and al-Zabāniyya
instead of al-ʿAlaq (Q. 96).21 This could either be the personal preference of the
scribe of the sūra-headbands, who likely lived in the fourth-fifth / eleventh-
twelfth centuries, or it could indicate a regional system of naming, probably in
Nishapur or Greater Khurasan.
In the text of Codex Mashhad, the recitation pause points are marked using
a primitive system. To identify these points, small circles, sometimes solid and
sometimes hollow, are placed throughout the text, including the main body, the
added sections, and the corrected and restored parts. Similar markings can also

15 This may be the result of concurrent reviewing of the text by another person who, in some
cases, has marked the verse endings in some other way. These two types of verse-end mark-
ings are also found in fragments of other ḥijāzī Qurʾāns, such as MS Marcel 17 (ff. 9v, 12r),
Marcel 18 (f. 33v), DAM 01–25.1 in Dār al-Makhṭūṭat, Ṣanʿāʾ (ff. 3r, 5v), and ŞE 87 (f. 1v).
16 Like MS 1 in the Raza Library (Rampur), MSS 9 and 25 in the Malek Museum (Tehran),
MS 394 in the Majlis Library (Tehran), MS 575 in the Imām ʿAlī Shrine (Najaf), MS 1207 in
the Āstāne Museum (Qum), and MSS Arabe 420, 385, 5935 in the BnF (Paris).
17 Like MSS 28, 31, 59 in the Āstān-i Quds Library (Mashhad), MS 1207 in the Āstāne Museum
(Qum), MS 4243 in the National Museum of Iran (Tehran), MS 27 in the Nuruosmaniye
Library (Istanbul), and MSS Arabe 5122 and Smith-Lesouëf 219 in the BnF (Paris).
18 Like MS 55 in the Āstān-i Quds Library (Mashhad), MSS 328 and 575 in the Imām ʿAlī Shrine
(Najaf), MS 28301 in the National Library of Iran (Tehran), and MS Vat. Ar. 711 in the Bib-
liotheca Apostolica Vaticana (Vatican).
19 This name clearly derives from the word la-ḥāfiẓīn in Q. 82:10; thus far, however, I have not
been able to find such a unique nomenclature in any other Qurʾānic manuscript.
20 Like MS Arabe 343 in the BnF (Paris).
21 Like MS 575 in the Imām ʿAlī Shrine (Najaf), MS 28301 in the National Library of Iran
(Tehran), and MS Wetzstein II 1914 in the SBB (Berlin).

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 303

be found in other old manuscripts of the Qurʾān. Examples are MS Qurʾān 16 in

the Imām ʿAlī Shrine (Najaf), MS Qurʾān 4154 in the Āstān-i Quds Library (Mash-
had), and the Qurʾān MS 1 in the Raza Library (Rampur). The same sign is also
used in the famous Qurʾān of ʿAlī ibn Shādhān al-Rāḍī (MS A6778 in Istanbul
University22 and MS Is 1434 in the Chester Beatty Library) dated 361 AH/972AD,
apparently to mark a verse ending.23

3.1 Orthographic Peculiarities in Codex Mashhad

The script of the Codex naturally follows the custom of early Islamic centuries,
which may be traced to the Late Antique tradition of scriptio continua. In this
custom, one pays no “attention to the spaces between the words and within a
word when it contains letters which are not connected to each other. Actually,
the groups of letters are scattered on the page in a rather regular way.”24 More
important are the spelling and orthographic rules of the Arabic script in the
main body, which are clearly inclined towards scriptio defectiva and are indica-
tive of the old paleography of the Codex. An important feature in this regard is
the lack of a mandatory character to indicate the long vowel |ā|, of which there
are many occurrences throughout the manuscript. Accordingly, many words
written in later periods with alif are without alif in this Codex.
The list of orthographic peculiarities in Codex Mashhad could be very long.
The following are some important examples that are less common in the
known ḥijāzī manuscripts of the Qurʾān: The word ‫ شئ‬is always written with
an additional alif as ‫ ;شای‬the word ‫ ذو‬is always written with an additional alif
as ‫ ;ذوا‬and the word ‫ اولوا‬is always written without wāw as ‫اولا‬. Moreover, the
omission of alif as the sign of long vowel |ā| is ubiquitous. The words ‫ قرآن‬and
‫ عذاب‬are almost always written without alif as ‫ قرن‬and ‫عذب‬. Apart from many
occurrences of ‫قلوا‬/‫ قل‬instead of ‫قالوا‬/‫ قال‬and ‫کنوا‬/‫ کن‬instead of ‫کانوا‬/‫کان‬, we
come across ‫( مل‬f. A16a) for ‫( مال‬Q. 2:247), and ‫( صل‬f. B60b) for ‫( صال‬Q. 37:163).
Other examples are: ‫( ایما‬f. A20b) for ‫( ایاما‬Q. 3:24), ‫( امما‬f. A8b) for ‫( اماما‬Q. 2:124),
‫( الذن‬f. A31b) for ‫( الذان‬Q. 4:16), ‫( اطع‬f. A35a) for ‫ اطاع‬in (Q. 4:80), ‫( ارد‬f. A42b)
for ‫( اراد‬Q. 5:17), ‫( مئدة‬f. A49b) for ‫( مائدة‬Q. 5:114), ‫( ثنی‬f. A77b) for ‫( ثانی‬Q. 9:40),

22 The colophon is in the Istanbul manuscript.

23 These can be regarded as the oldest examples of inserting pause signs in Qurʾānic manu-
scripts, which apparently took place first in Iran. For instance, in the colophon of the
Qurʾān MS Y-752 (Topkapi Palace Museum, Istanbul), dated 394/1004–1005, copied by an
Iranian scribe, Abū Bakr ʿAbd al-Malik b. Zarʿa b. Muḥammad al-Rūdbārī, it has been
added that Abū ʿAlī al-Ḥusain b. Muḥammad b. al-Ḥusain al-Muqriʾ made the pause signs
(al-wuqūf ) in the text.
24 François Déroche, Qurʾans of the Umayyads: A First Overview, Leiden: Brill, 2014, p. 18.

Journal of Islamic Manuscripts 10 (2019) 292–326

304 karimi-nia

‫( ا ین‬f. B29b) for ‫( ایان‬Q. 27:65), ‫( اینا‬f. A85a) for ‫( ایانا‬Q. 10:28), ‫( اعنه‬f. B19a) for ‫اعانه‬
(Q. 25:4), and ‫( فودک‬f. B20b) for ‫( فؤادک‬Q. 25:32).
The addition of alif in some words seems to indicate the vowel kasra. Exam-
ples are: ‫( جایت َهم‬f. A49a) for ‫جئت َهم‬
ِ (Q. 5:110), ‫( جایت ُکم‬twice on f. A22a) for ‫جئت ُکم‬ ِ
(Q. 3:49 and 50), ‫ت‬
َ ‫( شائ‬f. A67a) for ‫ت‬
َ ‫شئ‬
ِ (Q. 7:155), ‫ک‬
َ ُ ‫( جایت‬f. B22b), for ‫ک‬
َ ُ ‫جئت‬
ِ
(Q. 26:26), ‫ک‬
َ ‫( جاین‬f. B20b), for ‫ک‬
َ ٰ ‫جئن‬
ِ (Q. 25:33), and ‫( قد جائت ُکم‬f. B80a) for ‫قد‬
‫جئت ُکم‬
ِ (Q. 43:63). Also, words such as ‫آبائهم‬, ‫آبائنا‬, and ‫ آبائکم‬are always written with-
out middle alif as ‫ابیهم‬, ‫ابینا‬, and ‫( ابیكم‬f. A100b, B45a) and such words as ‫ایاه‬, ‫ایانا‬,
and ‫ ایاکم‬are always written without the middle alif as ‫ایه‬, ‫اینا‬, and ‫ایکم‬, respec-
tively (f. A52a, B52a, A85a).
The omission of alif while connecting two words together, in such cases as
‫( بلحق‬f. A58b) for ‫( بالحق‬Q. 6:151), ‫( بلبینت‬f. A103a, B80a & B103a) for ‫( بالبینت‬Q. 14:9,
43:63, & 57:25), and ‫( بلغیب‬f. B56a) for ‫( بالغیب‬Q. 36:11), is indicative of the oral
transmission of the Qurʾān in the first/seventh century. To this, one can add the
omission of yāʾ in such cases as ‫( اوف الکیل‬A96a) instead of ‫( اوفی الکیل‬Q. 12:59).
The omission of alif as a bearer of the ḥamza is common in the Codex.
Examples include: ‫( اطمننتم‬Q. 4:103) on f. A36b; ‫( وامرتن‬Q. 2:282) on f. A19b; ‫أطفها‬
(Q. 5:64) on f. A46a; ‫( اطمنوا‬Q. 10:7) on f. A83b; ‫( لاملن‬Q. 11:11, and 32:13) on f.
A98a & B44a, ‫( امتلت‬Q. 50:30), on f. B92b; ‫( اطمن‬Q. 22:11) on f. B7a; ‫( انشنا‬Q. 23:19,
and 28:45) on f. B13a & B23b; ‫( اشمزت‬Q. 39: 45) on f. B67b; ‫( نبرها‬Q. 57:22) on f.
B103a; ‫( یستخرون‬Q. 23:43) on f. B11a, ‫( تبرنا‬Q. 28:63) on f. B33b. On the other hand,
contrary to current usage, the addition of alif as a symbol of Arabic ḥamza
is common in the Codex; examples include: ‫( نباونی‬Q. 6:143) on f. A57a; ‫راوس‬
(Q. 2:279) on f. A19b; ‫( راوسهم‬Q. 17:51 and 22:19) on f. A116a & B7b; ‫( افادة‬Q. 14:37
and 16:78) on ff. A102a and A104b; ‫( الافادة‬Q. 67:23) on f. B114b; ‫( سیاه‬Q. 2:81)
on f. A5b; ‫( لا یاوده‬Q. 2:255) on f. A17a; ‫( تجارون‬Q. 16:53) on f. A109b; ‫لا تجاروا‬
(Q. 23:64) on f. B11b; ‫( سواة‬Q. 5:31) on f. A43a; ‫( فینباهم‬Q. 6:108) on f. A55b; ‫مبراون‬
(Q. 24:26) on f. B16a; ‫( سواتکم‬Q. 7:26) on f. A60b; ‫( السیا‬Q. 48:43) on f. B55b. More-
over, the word ‫ سوء‬ends almost everywhere with an additional alif as ‫سوا‬, for
example in Q. 6:157 (f. A59a), Q. 7:141 (f. A66a), Q. 7:167 (f. A68a), and Q. 16:59 (f.
A110a).
The omission of nūn while linking to another letter such as lam is notewor-
thy, for example in ‫( إلم یکن‬ff. A31a and A40b) for ‫( إن لم یکن‬Q. 4:12 & 176), ‫إلم تفعل‬
(f. A46a) for ‫( إن لم تفعل‬Q. 5:67), ‫( إلم ینتهوا‬f. A46b) for ‫( إن لم ینتهوا‬Q. 5:73), and ‫ألا‬
‫( تعبدوا‬f. A90a) for ‫( أن لا تعبدوا‬Q. 11: 26). The opposite is observed in ‫أن لو استقموا‬
(f. B118a) for ‫( ألوّ استقموا‬Q. 72:16).
The omission of wāw in words like ‫( الا‬f. A25b) instead of ‫( اولاء‬Q. 3:119), ‫ابنا الل ّٰه‬
(f. A42b) instead of ‫( ابنوا الل ّٰه‬Q. 5:18), ‫( نبا‬f. B62a & B111a) instead of ‫( نبوا‬Q. 38:21 and

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 305

64:5), ‫( دعا‬f. B70b) instead of ‫( دعوا‬Q. 40:50), ‫( الی‬f. B53a) instead of ‫( اولی‬Q. 35:1),
and ‫( البلا‬f. B60a) instead of ‫( البلوا‬Q. 37:106) is also notable.
Using a denticle similar to |‫ |یـ‬to show the long vowel |ā| was common in the
early ḥijāzī Qurʾāns,25 a preference that has gradually disappeared over time.
There are ample instances of such usage in Codex Mashhad: ‫( إليه‬f. A46b) for ‫إله‬
(Q. 5:73), ‫( دیرهم‬f. A101a) for ‫( دارهم‬Q. 13:31), ‫( قرطيس‬f. A50b) for ‫( قرطاس‬Q. 6:7),
‫( سييتنا‬f. A29b) for ‫( سیاتنا‬Q. 3:191), ‫( کلیهما‬f. A115a) for the ‫( کلاهما‬Q. 17:23); ‫( ابنینا‬f.
A16a) for ‫( ابنانا‬Q. 2:246), ‫( بری‬f. B79a) for ‫( برا‬Q. 43:26), and ‫( جنتین‬f. B51a) for ‫جنتان‬
(Q. 32:15).
In sum, it should be noted that, unlike many other known ḥijāzī codices,
including MSS Paris BnF Arabe 328, London BL Or. 2156, Tübingen Ma VI 165,
Codex Mashhad does not show the regional features of Syria; rather, it contains
elements of belonging to the Ḥijāz (Medina and Mecca). Thus, the word ‫ا برهیم‬
never occurs as ‫ ا برهم‬in this volume while in the other ḥijāzī codices from Syria
the word is frequently written without yāʾ, which suggests an influence of the
dialect specific to Syria. This Syrian dialect is probably manifest in Ibn ʿĀmir’s
reading of Ibrāhām /‫ا برهام‬. Moreover, the word ‫ داود‬in this Codex is always writ-
ten in its present-day form, unlike other ḥijāzī codices of Syria, in which it is
sporadically written as ‫دواد‬. And, as we will see below, considering the regional
differences in early Qurʾānic codices, Codex Mashhad is closer to Medina.

3.2 Regional Differences in Qurʾānic Codices

Any early codex of the Qurʾān can be attributed to one of the five cities Basra,
Kufa, Medina, Mecca, and Damascus. The differences between these regional
Qurʾāns have been recorded in the early Islamic sources.26 The differences
shown in the following table suggest that Codex Mashhad is far from Damas-
cus, Kufa, and Basra and is very close to Medina and Mecca. The comparisons
between the variants in the table show that Codex Mashhad generally corre-
sponds to Medina reading; however, in two cases (Q. 9:100 and 18:95), we find
Meccan readings and in once case a Syrian reading (Q. 39:64), where the Codex
does not conform to Medina.

25 See François Déroche, Qurʾans of the Umayyads, p. 24.

26 Ibn Abī Dāwūd al-Sijistānī, Kitāb al-Maṣāḥif, edited by Muḥammad b. ʿAbduh, Cairo: Al-
Fārūq al-Ḥadītha, 1423/2002, pp. 144–155. For a list of the regional differences in early
Qurʾānic codices, based on Abū ʿUbayd al-Qāsim b. Sallām’s Faḍāʾil al-Qurʾān, Abū ʿAmr
al-Dānī’s al-Muqniʿ fī Maʿrifat Rasm Maṣāḥif al-Amṣār, and al-Mabānī li-Naẓm al-Maʿānī,
see Nöldeke-Schwally, Geschichte des Qorāns, Vol. 3, Leipzig: Dieterich’sche Verlagsbuch-
handlung, 1938, pp. 11–14. For an analysis of these variants, see Michael Cook, “The Stemma
of the Regional Codices of the Koran”, Graeco-Arabica 9–10 (2004) pp. 89–104.

Journal of Islamic Manuscripts 10 (2019) 292–326

 ‫‪306‬‬ ‫‪karimi-nia‬‬

‫‪table 1‬‬ ‫‪Regional differences in Codex Mashhad‬‬

‫‪folio‬‬ ‫‪Verse‬‬ ‫‪Medina‬‬ ‫‪Mecca‬‬ ‫‪Kufa‬‬ ‫‪Basra‬‬ ‫‪Damascus‬‬ ‫‪Mashhad‬‬

‫‪1‬‬ ‫‪A8a‬‬ ‫‪Q. 2:116‬‬ ‫وقالوا اتخذ‬ ‫وقالوا اتخذ‬ ‫وقالوا اتخذ‬ ‫وقالوا اتخذ‬ ‫قالوا اتخذ‬ ‫وقالوا اتخذ‬
‫‪2‬‬ ‫‪A8b‬‬ ‫‪Q. 2:132‬‬ ‫واوصی‬ ‫ووصی‬ ‫ووصی‬ ‫ووصی‬ ‫واوصی‬ ‫‪Not seen‬‬
‫‪3‬‬ ‫‪A26a Q. 3:133‬‬ ‫سرعوا‬ ‫وسرعوا‬ ‫وسرعوا‬ ‫وسرعوا‬ ‫سرعوا‬ ‫سرعوا‬
‫‪4‬‬ ‫‪A28b Q. 3:184‬‬ ‫والز بر‬ ‫والز بر‬ ‫والز بر‬ ‫والز بر‬ ‫و بالز بر‬ ‫والز بر‬
‫‪5‬‬ ‫‪A34b Q. 4:66‬‬ ‫الا قلیل‬ ‫الا قلیل‬ ‫الا قلیل‬ ‫الا قلیل‬ ‫الا قلیلا‬ ‫الا قلیل‬
‫‪6‬‬ ‫‪A45a Q. 5:53‬‬ ‫یقول‬ ‫یقول‬ ‫و یقول‬ ‫و یقول‬ ‫یقول‬ ‫یقول‬
‫‪7‬‬ ‫‪A45a Q. 5:54‬‬ ‫من یرتدد‬ ‫من یرتّد‬ ‫من یرتّد‬ ‫من یرتّد‬ ‫من یرتدد‬ ‫من یرتدد‬
‫‪8‬‬ ‫‪A51b‬‬ ‫‪Q. 6:32‬‬ ‫وللدار‬ ‫وللدار‬ ‫وللدار‬ ‫وللدار‬ ‫ولدار‬ ‫وللدار‬
‫‪9‬‬ ‫‪A53a Q. 6:63‬‬ ‫انجیتنا‬ ‫انجیتنا‬ ‫انجینا‬ ‫انجیتنا‬ ‫انجیتنا‬ ‫انجیتنا‬
‫‪10 A60a Q. 7:3‬‬ ‫تذکرون‬ ‫تذکرون‬ ‫تذکرون‬ ‫تذکرون‬ ‫یتذکرون‬ ‫تذکرون‬
‫‪11‬‬ ‫‪A62a Q. 7:43‬‬ ‫وما کنا‬ ‫وما کنا‬ ‫وما کنا‬ ‫وما کنا‬ ‫ما کنا‬ ‫وما کنا‬
‫‪12 A63b Q. 7:75‬‬ ‫قال الملأ‬ ‫قال الملأ‬ ‫قال الملأ‬ ‫قال الملأ‬ ‫وقال الملأ‬ ‫قال الملأ‬
‫‪13 A66a Q. 7: 141‬‬ ‫انجینکم‬ ‫انجینکم‬ ‫انجینکم‬ ‫انجینکم‬ ‫انجاکم‬ ‫انجینکم‬
‫‪14 A81a‬‬ ‫تجری تحتها ‪Q. 9:100‬‬ ‫تجری تحتها تجری من تحتها‬ ‫تجری تحتها‬ ‫تجری تحتها‬ ‫تجری من تحتها‬
‫‪15 A81b‬‬ ‫‪Q. 9:107‬‬ ‫الذین اتخذوا‬ ‫والذین اتخذوا والذین اتخذوا‬ ‫الذین اتخذوا والذین اتخذوا‬ ‫الذین اتخذوا‬
‫‪16 A84a Q. 10:22‬‬ ‫یسیرکم‬ ‫یسیرکم‬ ‫یسیرکم‬ ‫یسیرکم‬ ‫ینشرکم‬ ‫یسیرکم؟‬
‫‪17 A117a Q. 18:36‬‬ ‫خیرا منهما‬ ‫خیرا منهما‬ ‫خیرا منها‬ ‫خیرا منها‬ ‫خیرا منهما‬ ‫خیرا منهما‬
‫‪18 A119b Q. 18:95‬‬ ‫ما مکنی‬ ‫ما مکننی‬ ‫ما مکنی‬ ‫ما مکنی‬ ‫ما مکنی‬ ‫ما مکننی‬
‫‪19 B12a‬‬ ‫‪Q. 23:87‬‬ ‫سیقولون لل ّٰه‬ ‫سیقولون لل ّٰه‬ ‫سیقولون لل ّٰه‬ ‫سیقولون الل ّٰه‬ ‫سیقولون لل ّٰه‬ ‫سیقولون لل ّٰه‬
‫‪20 B12a‬‬ ‫سیقولون لل ّٰه ‪Q. 23:89‬‬ ‫سیقولون لل ّٰه‬ ‫سیقولون لل ّٰه‬ ‫سیقولون الل ّٰه‬ ‫سیقولون لل ّٰه‬ ‫سیقولون لل ّٰه‬
‫‪21 B 20a Q. 25:25‬‬ ‫ونزل‬ ‫وننزل‬ ‫ونزل‬ ‫ونزل‬ ‫ونزل‬ ‫ونزل‬
‫فتوکل ‪22 B 26a Q. 26:217‬‬ ‫وتوکل‬ ‫وتوکل‬ ‫وتوکل‬ ‫فتوکل‬ ‫‪Not seen‬‬
‫‪23 B25a‬‬ ‫‪Q. 27:21‬‬ ‫لیاتینی‬ ‫لیاتیننی‬ ‫لیاتینی‬ ‫لیاتینی‬ ‫لیاتینی‬ ‫لیاتینی‬
‫وقال موسی ‪24 B32b Q. 28:37‬‬ ‫قال موسی‬ ‫وقال موسی‬ ‫وقال موسی‬ ‫وقال موسی‬ ‫وقال موسی‬
‫ماعملته ایدیهم ماعملته ایدیهم ماعملته ایدیهم ماعملت ایدیهم ماعملته ایدیهم ماعملته ایدیهم ‪25 B56b Q. 36:35‬‬
‫تامرونی ‪26 B67b Q. 39:64‬‬ ‫تامرونی‬ ‫تامرونی‬ ‫تامرونی‬ ‫تامروننی‬ ‫تامروننی‬
‫‪27 B69a Q. 40:21‬‬ ‫اشد منهم‬ ‫اشد منهم‬ ‫اشد منهم‬ ‫اشد منهم‬ ‫اشد منکم‬ ‫اشد منهم‬
‫وان یظهر ‪28 B69b Q. 40:26‬‬ ‫وان یظهر‬ ‫او ان یظهر‬ ‫وان یظهر‬ ‫وان یظهر‬ ‫وان یظهر‬
‫بما کسبت ‪29 B76b Q. 42:30‬‬ ‫فبما کسبت‬ ‫فبما کسبت‬ ‫فبما کسبت‬ ‫بما کسبت‬ ‫فبما کسبت‬

‫‪Journal of Islamic Manuscripts 10 (2019) 292–326‬‬

 a new document in the early history of the qurʾān 307

Table 1 Regional differences in Codex Mashhad (cont.)

folio Verse Medina Mecca Kufa Basra Damascus Mashhad

30 B80a Q. 43:68 ‫یعبادی‬ ‫یعباد‬ ‫یعباد‬ ‫یعباد‬ ‫یعبادی‬ ‫یعبادی‬

31 B80a Q. 43:71 ‫تشتهیه‬ ‫تشتهی‬ ‫تشتهی‬ ‫تشتهی‬ ‫تشتهیه‬ ‫تشتهیه‬
32 B85b Q. 46:15 ‫حسنا‬ ‫حسنا‬ ‫احسانا‬ ‫حسنا‬ ‫حسنا‬ Not seen
33 B88a Q. 47:18 ‫ان تاتیهم‬ ‫ان تاتهم‬ ‫ان تاتهم‬ ‫ان تاتیهم‬ ‫ان تاتیهم‬ ‫ان تاتیهم‬
34 B101a Q. 57:10 ‫وکلا وعد‬ ‫وکلا وعد‬ ‫وکلا وعد‬ ‫وکلا وعد‬ ‫وکل وعد‬ ‫وکلا وعد‬
35 B102a Q. 57:24 ‫الغنی‬ ‫هو الغنی‬ ‫هو الغنی‬ ‫هو الغنی‬ ‫الغنی‬ ‫الغنی‬

3.3 Variant Readings

Irrespective of the above-mentioned regional differences, any early Qurʾānic
codex simultaneously contains variant readings. In other words, no codex con-
tains only a single reading. However, it must be noted that the seven variant
readings attributed to the Seven Readers, which have been prevalent since the
fourth/tenth century, are only rarely evident in the Qurʾānic manuscripts of the
first two Islamic centuries. In these manuscripts, instead, one can find either
the above-mentioned regional differences (as between Mecca, Medina, Kufa,
Basra, or Damascus) or differences in lettering and dotting, which do not nec-
essarily reflect the canonical variants of the Seven Readers but can be traced
back to the readings of one of the Prophet’s Companions or Followers.
Codex Mashhad additionally contains two other, later types of markings
of differences. First, there are textual differences marked by coloured vowel-
dots indicating a canonical variant reading, and which were added later than
the copying of the original text. Second, there are textual differences that are
marked by dotting that can be concurrent with or later than the copying of the
original text. It should be noted that some of these added dots allow two simul-
taneous variant readings while some do not conform to any known canonical
or non-canonical reading. Thus, the words ‫( یاتیها رزقهُ ا‬Q. 16:112) are also dotted as
‫( تاتیها رزقهُ ا‬f. A107b) and the word ‫( ترکنا‬Q. 37:78, and 108) is dotted as both ‫ ترکنا‬and
‫( برکنا‬ff. B59b and B60a). Table 2 presents some instances of the various types of
readings in Codex Mashhad.

Journal of Islamic Manuscripts 10 (2019) 292–326

 308 karimi-nia

table 2 Variant readings in Codex Mashhad

Verse Leaf Ḥafṣ from ʿĀṣim Codex Mashhad Correspond to:

Q. 2:77 A5b ‫ تعلمون … تسرون … تعلنون یعلمون … یسرون … یعلنون‬Ibn Muḥayṣin

Q. 2:83 A5b ‫لا تعبدون‬ ‫لا یعبدون‬ Ibn Kathīr, Ḥamza, al-Kisāʾī, and
al-Ḥasan
Q. 2:96 A6b ‫یعملون‬ ‫تعملون‬ al-Ḥasan and Yaʿqūb
Q. 2:219 A14b ‫کبیر‬ ‫کثیر‬ Ḥamza, al-Kisāʾī, Ibn Masʿūd, al-
Aʿmash
Q. 2:259 A17a ‫لم یتسنه‬ ‫لم یتسّن‬ Ḥamza, al-Kisāʾī, Khalaf, and Yaʿqūb
when stopping
Q. 2:271 A19a ‫و یکّفر‬ ‫ونکّفر‬ Nāfiʿ, Ḥamza, al-Kisāʾī, Abū Jaʿfar,
Khalaf
Q. 3:13 A20a ‫یرونهم‬ ‫ترونهم‬ Nāfiʿ, Abu Jaʿfar, ʿAbān, Yaʿqūb, al-
Ḥasan
Q. 3:48 A22a ‫یعل ّمه‬ ‫نعل ّمه‬ Other than ʿĀṣim and Nāfiʿ + Ibn
Masʿūd
Q. 3:83 A23b ‫یبغون‬ ‫ یبغون‬/ ‫تبغون‬ Both readings

Q. 3:83 A23b ‫یرجعون‬ ‫ترُ جعون‬ Other than ʿĀṣim

Q. 3:90 A24a ‫تقبل‬ ‫ تقبل‬/ ‫یقبل‬ Both readings

Q. 3:115 A25a ‫یفعلوا … یکفروه‬ ‫تفعلوا … ت ُکفروه‬ Nāfiʿ, Ibn ʿĀmir, Ibn Kathīr, Abū
ʿAmr
Q. 3:120 A25b ‫یعملون‬ ‫تعملون‬ Al-Ḥasan, Abu Ḥātim, al-Muṭwaʿī

Q. 3:154 A27a ‫ل‬

ُ ‫الق َت‬ ‫ل‬
ُ ٰ ِ‫القت‬ Ḥamza, al-Ḥasan, al-Zuhrī, al-Azraq
Q. 3:163 A27b ‫یعملون‬ ‫ یعملون‬/ ‫تعملون‬ Both readings
Q. 4:34 A34b ‫تکن‬ ‫ تکن‬/ ‫یکن‬ Both readings
Q. 4:94 A36a ‫ فتبینوا‬/ ‫فتبینوا‬ ‫ فتثبتّ وا‬/ ‫فتثبتّ وا‬ Ḥamza, al-Kisāʾī, Khalaf, al-Ḥasan,
Ibn Masʿūd
Q. 4:104 A36b ‫تکونوا‬ ‫ تکونوا‬/ ‫یکونوا‬ Both readings
Q. 4:108 A37a ‫یعملون‬ ‫ یعملون‬/ ‫تعملون‬ Both readings
Q. 5: 50 A45a ‫یبغون‬ ‫تبغون‬ Ibn ʿĀmir
Q. 6:32 A51b ‫تعقلون‬ ‫ی ع ق لو ن‬ Ibn Kathīr, Abū ʿAmr, Ḥamza, al-
Kisāʾī
Q. 6:139 A57a ‫یکن‬ ‫ یکن‬/ ‫تکن‬ Both readings

Journal of Islamic Manuscripts 10 (2019) 292–326

 a new document in the early history of the qurʾān 309

Table 2 Variant readings in Codex Mashhad (cont.)

Verse Leaf Ḥafṣ from ʿĀṣim Codex Mashhad Correspond to:

Q. 6:132 A57b ‫یعملون‬ ‫ یعملون‬/ ‫تعملون‬ Both readings

Q. 6:135 A57b ‫تکون‬ ‫یکون‬ Ḥamza, al-Kisāʾī, Khalaf, al-Aʿmash
Q. 7:131 A65b ‫یطیروا‬ ‫ یطیروا‬/ ‫تطیروا‬ Both readings
Q. 7:147 A66b ‫یجزون‬ ‫ یجزون‬/ ‫تجزون‬ Both readings
Q. 7:186 A69b ‫یذرهم‬ ‫ یذرهم‬/ ‫نذرهم‬ Both readings
Q. 8:37 A72a ‫یعودوا‬ ‫ یعودوا‬/ ‫تعودوا‬ Both readings
Q. 8:39 A72a ‫یعملون‬ ‫ یعملون‬/ ‫تعملون‬ Both readings
Q. 9:31 A76b ‫یشرکون‬ ‫ یشرکون‬/ ‫تشرکون‬ Both readings
Q. 9:104 A81a ‫یعلموا‬ ‫ یعلموا‬/ ‫تعلموا‬ Both readings
Q. 9:117 A82a ‫کاد یز یغ‬ ‫کاد تز یغ‬ Other than ʿĀṣim
Q. 10:26 A84b ‫یرهق‬ ‫ ترهق‬/ ‫یرهق‬ Both readings
Q. 10:45 A85b ‫یحشرهم‬ ‫نحشرهم‬ Other than ʿĀṣim
Q. 12:12 A94a ‫یرتع و یلعب‬ ‫نرتع ونلعب‬ Ibn ʿĀmir, Ibn Kathīr, Abū ʿAmr, al-
Bizzī, Al-Yazīdī
Q. 12:43 A95b ‫رءیی‬ ‫ر یایی‬ Possibly a copyist error
Q. 12:109 A99a ‫یعقلون‬ ‫ت ع ق لو ن‬ Ibn Kathīr, Abū ʿAmr, Ḥamza, al-
Kisāʾī, Khalaf
Q. 16:86 A102a ‫ندعوا‬ ‫ ندعوا‬/ ‫یدعوا‬ Both readings
Q. 15:5 A105a ‫یستاخرون‬ ‫ یستاخرون‬/ ‫تستاخرون‬ Both readings

Q. 15:8 A105a ‫ما ن ُنزَ ِّل‬ ‫ما ت َن َزَ ّل‬ Ibn ʿĀmir, Ibn Kathīr, Abū ʿAmr,
Nāfiʿ, Yaʿqūb, Abū Jaʿfar
Q. 16:1 A108a ‫یشرکون‬ ‫تشرکون‬ Ḥamza, al-Kisāʾī, Khalaf, al-Aʿmash
Q. 16:3 A108a ‫یشرکون‬ ‫تشرکون‬ Ḥamza, al-Kisāʾī, Khalaf, al-Aʿmash
Q. 16:20 A108b ‫یدعون‬ ‫تدعون‬ Ibn ʿĀmir, Ibn Kathīr, Abū ʿAmr,
Nāfiʿ, Ḥamza, al-Kisāʾī
Q. 16:23 A108b ‫یسرون‬ ‫ تسرون‬/ ‫یسرون‬ Both readings
Q. 16:23 A108b ‫یعلنون‬ ‫ تعلنون‬/ ‫یعلنون‬ Both readings
Q. 16:28 A108b ‫ٺتوفیهم‬ ‫یتوفیهم‬ Ḥamza, al-Aʿmash,
Q. 16:48 A109b ‫اولم یروا‬ ‫اولم تروا‬ Ḥamza, al-Kisāʾī, Khalaf, al-Aʿmash,
al-Ḥasan
Q. 16:72 A110b ‫یومنون‬ ‫تومنون‬ Al-Sulamī, Qatāda

Journal of Islamic Manuscripts 10 (2019) 292–326

 310 karimi-nia

Table 2 Variant readings in Codex Mashhad (cont.)

Verse Leaf Ḥafṣ from ʿĀṣim Codex Mashhad Correspond to:

Q. 16:76 A110b ‫یوجهه‬ ‫ یوجهه‬/ ‫توجهه‬ Both readings

Q. 20:96 B1a ‫بما لم یبصروا‬ ‫بما لم تبصروا‬ Ḥamza, al-Kisāʾī
Q. 16:96 A107a ‫ولنجز ین‬ ‫ولیجز ین‬ Ibn ʿĀmir, Abū ʿAmr, Nāfiʿ, Ḥamza,
al-Kisāʾī, Yaʿqūb
Q. 16:112 A107b ‫یاتیها رزقها‬ ‫تاتیها رزقها‬ Possibly a copyist error
Q. 20:109 B1b ‫تنف ع‬ ‫ینفع‬ Unknown reading
Q. 20:128 B2a ‫افلم یهد لهم‬ ‫افلم نهد لهم‬ Yaʿqūb, Ibn ʿAbbās, Qatāda, Warsh
from Nāfiʿ
Q. 21:45 B4a ‫یسمع‬ ‫ تسمع‬/ ‫یسمع‬ Both readings

Q. 24:35 B16b ‫یوُ قد‬ ‫توَ قَ َ ّد‬ Abū ʿAmr, Ibn Kathīr, Abū Jaʿfar,
Yaʿqūb, al-Ḥasan, Mujāhid
Q. 26:72 B 23b ‫یسمعونکم‬ ‫تشفعونکم‬ Unknown reading
Q. 27:59 B29b ‫یشرکون‬ ‫تشرکون‬ Other than ʿĀṣim and Abū ʿAmr
Q. 29:42 B37b ‫یدعون‬ ‫تدعون‬ Ibn ʿĀmir, Nāfiʿ, Ibn Kathīr, Ḥamza,
al-Kisāʾī, Abū Jaʿfar
Q. 29:61 B38b ‫یوفکون‬ ‫توفکون‬ Unknown reading

Q. 34:17 B 51a ‫نجَٰ زِی‬

ُ ‫یجزی‬ Qatāda, Ibrāhīm al-Nakhaʿī
Q. 34:40 B52a ‫یحشرهم … یقول‬ ‫نحشرهم … نقول‬ Ibn ʿĀmir, Abū ʿAmr, Nāfiʿ, Ibn
Kathīr, Ḥamza, al-Kisāʾī
Q. 37:78 B59b ‫ترکنا‬ ‫ برکنا‬/ ‫ترکنا‬ Both readings
Q. 37:108 B60a ‫ترکنا‬ ‫ برکنا‬/ ‫ترکنا‬ Both readings
Q. 39:36 B66b ‫عبده‬ ‫عباده‬ Ḥamza, al-Kisāʾī, Abū Jaʿfar,
Mujāhid, Khalaf, Aʿmash
Q. 40:51 B70b ‫یقوم‬ ‫تقوم‬ Ibn Hurmuz, Ismāʿīl al-Minqarī
Q. 52:43 B95b ‫یشرکون‬ ‫تشرکون‬ Unknown reading
Q. 58:17 B104a ‫تغنی‬ ‫یغنی‬ Unknown reading
Q. 63:8 B109b ‫لیخرجن‬ ‫لنخرجّن‬ al-Ḥasan, Ibn Abī Isḥāq, al-
Musayyibī
Q. 72:17 B118a ‫یسلـکه‬ ‫نسلـکه‬ Ibn ʿĀmir, Ibn Kathīr, Abū ʿAmr,
Nāfiʿ, Abū Jaʿfar, al-Ḥasan
Q. 78:4–5 B 122a ‫سيعلمون‬ ‫ سیعلمون‬/‫ستعلمون‬ Both readings

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 311

3.4 Verse Counting

As we have already observed, the process of dividing the verses in Codex
Mashhad was concurrent with the process of transcribing the text and has
been performed by the scribe. Nevertheless, the system of verse division in
the Codex does not follow any of the seven well-known numbering systems,
i.e. Kufa, Basra, Homs, Damascus, Mecca, Medina 1, and Medina 2.27 Unlike
these seven traditions, there is no sign of verse endings in many cases; at the
same time, Codex Mashhad indicates verse endings in places that are not rec-
ognized as such by any of the seven canonical systems. This suggests either
that the scribe was unaware of these systems or that they were not yet fully
developed. After yaʿlamūn (Q. 2:77), al-Qurʾān (Q. 2:185), mā yashāʾ (Q. 3:47),
mustaqīman (Q. 6:153), sabīlan (Q. 7:146), khuwār (Q. 7:148), mafʿūlan (Q. 8:44),
muʾminīn (Q. 9:26), baʿīr (Q. 12:72), al-muttaqūn (Q. 13: 34), yakrahūn (16:62),
kafīlan (Q. 16:91), ḥīn (Q. 14:25), al-ẓālimīn (Q. 14:27) al-muʾminīn (Q. 17:9), and
muʾminīn (Q. 18:80), for example, there are the signs of verse endings, a char-
acteristic of the Codex that does not conform with any of the above systems.
Contrary to what is found in all canonical systems, the words nahran (Q. 18:33)
and yuṭāʿ (Q. 40:18) on folios A117a and B69a, respectively, are not recognized
as verse-ending words in this Codex. In sum, the differences in verse endings
indicate that Codex Mashhad is closest in this regard to the systems of Basra
and Medina.

3.5 Mistakes by the Scribe(s)

As with nearly all ḥijāzī Qurʾāns, the scribe of this Codex sometimes made
mistakes in transcribing the text. Some of these mistakes have been corrected
by subsequent readers while others have been ignored, but none of them
can be considered a variant reading or a peculiarity in Arabic orthography.
Moreover, as well as correcting the errors of the scribe(s), the readers have, in
many cases, tried to change the Arabic orthography in accordance with what
they considered correct. Thus, after transcribing the word ‫( یبصط‬Q. 2:245) on
f. A16a, a subsequent reader has changed it into ‫یبسط‬. Also, the words ‫السیا‬
(Q. 35:43) and ‫( یذراکم‬Q. 42:11) have been modified into ‫( السیئ‬f. B55b) and ‫یذروکم‬
(f. B75b), respectively. Table 3 shows only the most outstanding mistakes by the
scribe(s).

27 For a fairly complete list of the early Islamic traditions in this regard, see Anton Spitaler,
Die Verszählung des Koran nach islamischer Überlieferung. Munich: Verlag der Bayerischen
Akademie der Wissenschaften, 1935, pp. 4–11.

Journal of Islamic Manuscripts 10 (2019) 292–326

 312 karimi-nia

table 3 Mistakes of the Scribe in Codex Mashhad

Qurʾān Folio The standard Qurʾān Codex Mashhad Modification flag

Q. 2:241 A16a ‫وللمطلقت‬ ‫والمطلقت‬ Modified by a subsequent reader

Q. 2:282 A19b ‫یکونا رجلین‬ ‫یکونوا رجلین‬ Modified by a subsequent reader

Q. 2:280 A19b ‫و ان کان‬ ‫فان کان‬ Not modified

Q. 3:86 A23b ‫یهدی الل ّٰه قوما‬ ‫یهدی الل ّٰه‬ qawman has been added later

Q. 3:180 A28b ‫خیرا‬ ‫خیر‬ Modified by a subsequent reader

Q. 4:150 A39a ‫ورسله‬ ‫ورسوله‬ Modified by a subsequent reader

Q. 4:163 A39b ‫النبیین‬ ‫النبیون‬ Modified by a subsequent reader

Q. 10:39 A85b ‫لم یحیطوا‬ ‫لما یحیطوا‬ Modified by a subsequent reader

Q. 10:67 A86b ‫لای ٰت‬ ‫لایة‬ Modified by a subsequent reader

Q. 13:33 A101b ‫ام تنبونه بما‬ ‫ ام تنبونه بما ام تنبونه بما‬Not modified: the repetition is
not corrected

Q. 14:36 A104b ‫رب انهن‬ ‫ر بنا انهن‬ Not modified

Q. 15:6 A105a ‫نزل علیه‬ ‫ا نزل علیه‬ Modified by a subsequent reader

Q. 15:77 A106b ‫لایة للمومنین‬ ‫لای ٰت للمومنین‬ Modified by a subsequent reader

Q. 18:49 A117b ‫یو یلتنا‬ ‫یو یلیتنا‬ Not modified

Q. 18:58 A118a ‫لهم موعد‬ ‫لهم موعدا‬ Not modified

Q. 21:17 B3a ‫ان نتخذ لهوا‬ ‫ان نتخذ ا لهوا‬ Not modified

Q. 21:74 B4b ‫کانوا قوم سو‬ ‫کانوا قوما سو‬ Not modified

Q. 26:116 B24a ‫المرجومین‬ ‫المخرجین‬ Modified by a subsequent reader

Q. 26:167 B28a ‫المخرجین‬ ‫المرجومین‬ Modified by a subsequent reader

Q. 27:80 B30a ‫الصم الدعا‬ ‫الصم الدعا الصم الدعا‬ Addition has been erased

Q. 27:82 B30a ‫القول علیهم‬ ‫علیهم القول‬ Reversal of the words

Q. 27:19 B27b ‫وقال‬ ‫قال‬ Modified: wāw has been added

Q. 29:23 B36b ‫لقائه‬ ‫لقاه‬ Modified by a subsequent reader

Journal of Islamic Manuscripts 10 (2019) 292–326

 a new document in the early history of the qurʾān 313

Table 3 Mistakes of the Scribe in Codex Mashhad (cont.)

Qurʾān Folio The standard Qurʾān Codex Mashhad Modification flag

Q. 29:31 B37a ‫قالوا انا‬ ‫قال انا‬ Modified by a subsequent reader

Q. 29:53 B38a ‫لولا اجل‬ ‫لو اجل‬ Modified: lā has been added

Q. 30:30 B40a ‫لخلق الل ّٰه‬ ‫لخلق‬ Modified: Allāh has been added

Q. 30:41 B40b ‫بما کسبت ایدی‬ ‫بما کسب ایدی‬ Not modified

Q. 31:4 B42a ‫هم یوقنون‬ ‫یوقنون‬ Modified: hum has been added

Q. 33:4 B45a ‫تظهرون منهن‬ ‫تظهرون‬ Modified: minhunna has been

added

Q. 33:37 B47b ‫ادعیائهم‬ ‫ادعیهم‬ An additional dent is left

Q. 33:54 B48b ‫کان بکل شی‬ ‫بکل شی‬ kāna has been added later

Q. 35:33 B55a ً ‫لولوا‬ ‫لولو‬ may be a difference in orthogra-

phy or a variant reading

Q. 36:10 B56a ‫سواء‬ ‫سوی‬ Modified by a subsequent reader

Q. 37:71 B59b ‫لقد ضل قبلهم‬ ‫لقد اضل قبلهم‬ Modified: alif has been deleted

Q. 37:101 B60a ‫غلام حلیم‬ ‫غلام علیم‬ Modified by a subsequent reader

Q. 43:49 B79b ‫اننا لمهتدون‬ ‫انا لمهتدون‬ Modified: a dent has been added

Q. 46:11 B85a ‫واذ لم یهتدوا‬ ‫واذا لم یهتدوا‬ Not modified

Q. 48:29 B90b ‫فاستوی‬ ‫ــــ‬ Modified: the word has been

added in naskhī script

Q. 49:12 B91b ‫اجتنبوا کثیرا‬ ‫اجتنبوا کثیر‬ Modified: alif has been added

Q. 62:7 B109a ‫ایدیهم‬ ‫ایدیکم‬ Modified by a subsequent reader

Q. 65:4 B111a ‫والئی‬ ‫والتی‬ Not modified

Q. 67:2 B114a ‫الذی خلق‬ ‫الذی‬ Modified: khalaqa has been

added

Journal of Islamic Manuscripts 10 (2019) 292–326

314 karimi-nia

4 The Arrangement of Sūras in Codex Mashhad

Generally, emendations by original scribes or later readers are common in

Qurʾānic manuscripts. Moreover, it is not surprising or unusual to find addi-
tional leaves in old Qurʾānic manuscripts. These leaves have usually been added
by those who have discovered that parts of the manuscript are missing or dam-
aged by, for example, water. Therefore, the opening or closing parts of many
Qurʾānic manuscripts have been rewritten and added by later hands on leaves
of equal size. Moreover, the additional leaves have occasionally been taken
from other manuscripts and attached to the codices.
A unique phenomenon in the emendation process of Codex Mashhad dis-
tinguishes it from all other known early Qurʾāns. Quite apart from the addition
and completion of the initial and final leaves, almost all the leaves contain-
ing the respective closing and opening parts of two adjacent sūras in Codex
Mashhad have been manipulated. At the end of the preceding sūra and/or at
the beginning of the following one, the parchment has been rewritten entirely
or partially at a later date in a different Kufic script. This suggests a possible
attempt, in the form of a cut-and-paste operation, to bind together the leaves
containing the sūra-headbands and to rearrange the sūras in accordance with
the standard ʿUthmānic sequence. A careful examination of the folios reveals
that the emendator has scraped off the ink from the parchment to use the
space. This process of scraping and rewriting parts of the parchment means
that Codex Mashhad is a partial palimpsest, in which the upper texts result
from rewriting in ʿUthmānic sequence the beginnings or ends of sūras after
erasing the lower texts containing a sequence probably attributable to a com-
panion of the Prophet.
In the leaves containing the opening and/or closing parts of the sūras, the
kind of manipulation varies from folio to folio. When a sūra ends, the begin-
ning of the next one usually continues on the same folio in a different and
late Kufic hand. The emending scribe has usually adjusted the size and exten-
sion of letters and words to accommodate the beginning or closing parts of
the sūras he is joining. Occasionally, two or more lines are erased from the
bottom of the leaf but nothing is added instead and, according to the stan-
dard ʿUthmānic sequence, the next sūra is to start at the beginning or in the
middle of the next leaf [figure 8]. In some cases, the closing part of a sūra at
the top of a left leaf is erased and, instead, the end of another sūra is tran-
scribed in a different hand in order to connect it to the former portion of this
sūra on the previous folio [figure 4 and figure 5]. Likewise, the beginning part
of a sūra at the bottom of a right leaf is sometimes erased and, instead, the
beginning of another sūra of the ʿUthmānic sequence is written in a differ-

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 315

ent hand to connect it to the latter portion of this particular sūra in the next
folio [figure 6]. In such cases, one can easily see that the emending scribe has
scraped off the ink or has washed it from the parchment to reuse the space.
Often, the beginning portion of a sūra is erased from the lower layer and, in the
upper layer of this palimpsest, the beginning of the next sūra of the ʿUthmānic
order is replaced in a different Kufic hand. Sometimes, the parchment is erased
after the conclusion of a sūra and is either left blank or occupied by a crude
illumination. In such cases, the next sūra begins at the top of the next folio
[figure 7 and figure 8]. Likewise, the original script is sometimes erased before
the beginning of a sūra and this upper half of the folio, which had apparently
contained the end of another sūra, is always occupied by a crude illumination.
Thus, Sūrat Yūnus (Q. 10) in the Codex (ff. A82b–A83a) begins in the middle
of the folio, the top of which is erased and then illuminated. Clearly, this por-
tion could not be occupied by the end of Sūrat al-Tawba (Q. 9), since the sūra
concludes on the previous folio. This therefore suggests that the lower layer
of the portion contained the end of another sūra in a sequence other than
ʿUthmānic [figure 9]. Again, in two distinct cases, the first between sūras 2
and 3 and the second between sūras 3 and 4, complete leaves are added in
each of which the end of the respective preceding sūra and the beginning of
the following one are rewritten so as to restore the connection between the
sūras [figure 10]. Finally, the emending scribe has completely rewritten the last
part of the Codex containing the short sūras 79 to 114 on seven separate leaves
(ff. B123a–B129b), since the cut-and-paste operation was almost impossible in
these chapters.
Based on the evidence presented above, we can see that the arrangement
of the sūras in Codex Mashhad was not initially in accordance with the stan-
dard ʿUthmānic order, although the text was basically ʿUthmānic. The current
arrangement of the Codex is the result of a later, broad cut-and-paste opera-
tion. In only seven places, however, are the closing and opening verses of two
consecutive sūras intact, suggesting that the Codex had an order in these places
that concurred with the standard ʿUthmānic sequence, and that the emending
scribe accordingly left this order unchanged. Now, before further examining
Codex Mashhad and attempting to explain its initial status, it is necessary to
find the individual Companion to whom this initial arrangement of sūras is
ascribable.
Considering the significance of intact sequences of sūras in the examina-
tion of codices, we see that among the various arrangements attributed to
Companions like Imām ʿAlī (d. 40/661), Ibn Masʿūd (d. 32/653), and Ubayy
b. Kaʿb (d. 29/649), the initial order of sūras in Codex Mashhad concurred
with that of Ibn Masʿūd. The ʿUthmānic order and Ibn Masʿūd’s are similar

Journal of Islamic Manuscripts 10 (2019) 292–326

316 karimi-nia

figure 4 Codex Mashhad. End of sūra 5, al-Māʾida, and beginning of sūra 6, al-Anʿām (ff.
A49b–A50a)

figure 5 Codex Mashhad. End of sūra 14, Ibrāhīm, and beginning of sūra 15, al-Ḥijr (ff.
A104b–A105a)

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 317

figure 6 Codex Mashhad. End of sūra 6, al-Anʿām, and beginning of sūra 7, al-Aʿrāf (ff.
A59b–A60a)

figure 7 Codex Mashhad. End of sūra 56, al-Wāqiʿa, and beginning of sūra 57, al-Ḥadīd (ff.
B100b–B101a)

Journal of Islamic Manuscripts 10 (2019) 292–326

318 karimi-nia

figure 8 Codex Mashhad. End of sūra 60, al-Mumtaḥina, and beginning of sūra 61, al-Ṣaff
(ff. B107b–B108a)

figure 9 Codex Mashhad. End of sūra 8, al-Tawba, and beginning of sūra 10, Yūnus (ff.
A82b–A83a)

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 319

figure 10 Codex Mashhad. End of sūra 3, Āl ʿImrān, and beginning of ṣūra 4, al-Nisāʾ (ff.
A29b–A31a)

in eight—or, according to Ibn al-Nadīm, in twelve—places.28 Thus, the se-

quence of the sūra-pairs Hūd/Yūsuf (Q. 11/12), al-ʿAnkabūt/al-Rūm (Q. 29/30),
Sabaʾ/Fāṭir (Q. 34/35), al-Zumar/Ghāfir (Q. 39/40), Fuṣṣilat/al-Shūrā (Q. 41/42),
al-Dhāriyāt/al-Ṭūr (Q. 51/52) and al-Mursalāt/al-Nabaʾ (Q. 77/78) is identical in
both ʿUthmānic and Ibn Masʿūd’s versions. It was exactly in such places that
the Codex did not need any rearrangement by the emending scribe [figures 11
and figure 12]. In the Codex, however, there is no evidence of agreement with
other sūra arrangements. Thus, Islamic literary sources cite sixteen places in

28 The arrangement of the sūras in Ibn Masʿūd’s Codex varies a little in the historical
accounts. For details, see Nöldeke-Schwally, Geschichte des Qorāns, Vol. 2. Die Sammlung
des Qorāns, Leipzig, 1919, pp. 39–40. According to Ibn Ashta’s Kitāb al-Maṣāḥif (see al-
Suyūṭī, al-Itqān, vol. 1, pp. 223–224), Ibn Masʿūd’s arrangement of sūras corresponds to
the standard ʿUthmānic version in the following eight positions: Hūd (11)-Yūsuf (12), al-
ʿAnkabūt (29)-al-Rūm (30), Sabaʾ (34)-Fāṭir (35), al-Zumar (39)-Ghāfir (40), Fuṣṣilat (41)-al-
Shūrā (42), al-Takwīr (81)-al-Infiṭār (82), al-Humaza (104)-al-Fīl (105)-Quraysh (106), and
al-Masad (111)-al-Ikhlāṣ (112). But an old tradition on the authority of Faḍl b. Shādhān
(see Ibn Nadīm, al-Fihrist, p. 29) adds four other positions: al-Dhāriyāt (51)-al-Ṭūr (52),
al-Mursalāt (77)-al-Nabaʾ (78), al-Inshiqāq (84)-al-Burūj (85), and al-Ḍuḥā (93)-al-Sharḥ
(94). Except for the five pairs of short sūras in the last part of the Qurʾān, which are com-
pletely rewritten in the current status of Codex Mashhad, the rest of the above-mentioned
twelve positions remain intact. Therefore, without any manipulation and alteration, Sūrat
Hūd (11) is followed by Sūrat Yūsuf (12), al-ʿAnkabūt (29) by al-Rūm (30), Sabaʾ (34) by Fāṭir
(35), al-Zumar (39) by Ghāfir (40), and Fuṣṣilat (41) by al-Shūrā (42). These are mentioned
in the lists of both Ibn Ashta and Ibn Shādhān. Besides, here we find Sūrat al-Dhāriyāt (51)
followed by Sūrat al-Ṭūr (52) and Sūrat al-Mursalāt (77) followed by Sūrat al-Nabaʾ (78),
both are mentioned only in Ibn Shādhān’s account.

Journal of Islamic Manuscripts 10 (2019) 292–326

320 karimi-nia

figure 11 Codex Mashhad. The sequence of sūra-pairs Hūd (11)/Yūsuf (12) without any
alteration (f. A98b)

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 321

figure 12 Codex Mashhad. The sequence of sūra-pairs al-Mursalāt (77)/al-Nabaʾ (78) with-
out any alteration (f. B122a)

which the ʿUthmānic arrangement of sūras match Ubayy’s;29 neither can be

found sequentially in Codex Mashhad.

5 The Historical Explanation and Dating of the Codex

The initial status of Codex Mashhad was quite different from that of all other
early Qurʾānic manuscripts. Why should someone transcribe the entire Qurʾān
based on the official ʿUthmānic version but according to Ibn Masʿūd’s arrange-
ment of the sūras? Before a radiocarbon dating was available, various hypothe-
ses were suggested about the origin of the Codex. Here, I present three plausible
hypotheses, the third of which, I aim to show, is the best historical explana-
tion.

29 See Al-Suyūṭī, al-Itqān, vol. 1, pp. 222–223; Ibn al-Nadīm, al-Fihrist, pp. 29–30. See also
Nöldeke-Schwally, Geschichte des Qorāns, Vol. 2. Die Sammlung des Qorāns, Leipzig, 1919,
pp. 30–33.

Journal of Islamic Manuscripts 10 (2019) 292–326

322 karimi-nia

5.1 The First Hypothesis

One can speculate that the Codex was produced in Kufa by supporters of Ibn
Masʿūd who were still familiar with Ibn Masʿūd’s arrangement of the sūras
and who were uncomfortable with ʿUthmān’s version. Following the canoniza-
tion and spread of the ʿUthmānic version, the copyist was seemingly unable
to write the Qurʾān differently, so he utilized the ʿUthmānic text, but with Ibn
Masʿūd’s arrangement. When, in later centuries, the ʿUthmānic arrangement
became firmly established everywhere, including in Kufa, this manuscript was
rearranged according to the ʿUthmānic order. Evidence from early literary
sources provides some support for this hypothesis. Thus, Sulaymān b. Mihrān,
known as al-Aʿmash (d. 148/765), reportedly said in the second/eighth century
that the qirāʾa of Zayd, i.e. the ʿUthmānic version, was unknown in Kufa in
comparison with that of Ibn Masʿūd.30 Also, in his Maʿānī al-Qurʾān, al-Farrāʾ
(d. 207/823), a literary exegete from Kufa, repeatedly cited Ibn Masʿūd’s ver-
sion,31 suggesting the currency of Ibn Masʿūd’s codex and reading in his time.
Moreover, the fatwā32 issued by Mālik b. Anas (d. 179/795) against Ibn Masʿūd’s
reading suggests that the reading was prevalent then. According to such histori-
ans as Ibn al-Jawzī (d. 597/1201) and al-Subkī (d. 756/1355),33 codices attributed
to Ibn Masʿūd were known in Kufa and Baghdad even during the fourth/tenth
century and were in the possession of some Shīʿa Muslims.
This hypothesis still fails to answer the various dimensions of an important
question. Why would an Ibn Masʿūdist who is still loyal to Ibn Masʿūd’s order of
sūras, want to adopt the ʿUthmānic text, particularly the ʿUthmānic text in the
Medinan regional version? If this Qurʾān was written in Kufa by Ibn Masʿūd’s
proponents, why are none of the characteristics of his readings and his codex
found in Codex Mashhad? More importantly, why does our examination of the

30 Ibn Mujāhid, Kitāb al-sabʿa fī al-qirāʾāt, edited by Shawqī Ḍayf, Cairo: Dār al-maʿārif,
1400/1980, p. 67.
31 Al-Farrāʾ, Maʿānī al-Qurʾān, edited by Aḥmad Yūsuf al-Najātī et al., Cairo: Dār al-Miṣriyya,
1374/1955, vol. 1, pp. 12, 16, 26, 28, 95, 145, 155, 192, 249, 300, 315, 318, 393, and 437. For a list of
Ibn Masʿūd’s variant readings see: Ibn Abī Dāwūd al-Sijistānī, Kitāb al-Maṣāḥif, pp. 166–
186.
32 See Abū Shāma al-Maqdisī, Al-Murshid al-wajīz fī ʿulūm tataʿallaqu bi al-kitāb al-ʿazīz,
edited by Tayyar Altikulac, Beirut: Dār Ṣādir, 1395/1975, vol. 1, p. 182. Abū Zakariyā al-
Rahūnī, Tuḥfat al-masʾūl fī sharḥ mukhtaṣar muntaha al-sūl, edited by al-Hādī b. al-Ḥusain
Shabīlī, Dubai: Dār al-Buḥūth, 1422/2002, vo. 2, p. 162.
33 Ibn al-Jawzī, Al-Muntaẓam fī tārīkh al-mulūk wa al-umam, edited by Muḥammad A.Q. ʿAṭāʾ
and Muṣṭafā A.Q. ʿAṭāʾ, Beirut: Dār al-Kutub al-ʿIlmiyya, 1412/1992, vol. 15, p. 59; Taqī al-Dīn
al-Subkī, Ṭabaqāt al-shāfiʿiyya al-kubrā, edited by Maḥmūd M. al-Ṭanaḥi, Cairo: Hijr li al-
Ṭibāʿa wa al-Nashr, 1383/1964, vol. 4, p. 65.

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 323

regional differences in maṣāḥif, i.e. the early codices, show us that Codex Mash-
had has almost all of the specific characteristics of the Medinan text and differs
drastically from the Kufan text?34 Examples of these regional differences (ikhti-
lāf maṣāḥif al-amṣār) are mentioned in Table 1 above.

5.2 The Second Hypothesis

A second hypothesis would be that the Codex was not transcribed in the first
century, but rather at some point in the following centuries when the offi-
cial ʿUthmānic version was prevalent throughout the Muslim world, includ-
ing even Kufa, and that the scribe(s) relied on the early Islamic historical
reports to produce a codex in accordance with Ibn Masʿūd’s arrangement of
the sūras. Thus, the Codex would date from a period significantly later than
the first/seventh century. The only significant supporting evidence that can
be found in favour of this hypothesis is that the scribe(s) has written some of
the Qurʾānic words based on apparently late Islamic reports. The word ‫یت َسَن ّه‬
(Q. 2:259) was written without hāʾ as ‫( ی َت َسََّن‬f. A17a) and, in Q. 26:116 and 176,
the words “‫ ”المخرجين‬and “‫ ”المرجومين‬are swapped (ff. B24a and B28a), though the
passages are corrected afterwards in accordance with the ʿUthmānic version.
These examples evoke a famous report from Sijistānī’s Kitāb al-Maṣāḥif stat-
ing that al-Ḥajjāj b. Yūsuf made eleven changes, including these three, to the
ʿUthmānic Qurʾān.35 Someone in later centuries has apparently tried to write a
Qurʾān according to Ibn Masʿūd’s arrangement of the sūras, reversing al-Ḥajjāj’s
changes!
The main weakness of this hypothesis is that it fails to account for the ortho-
graphic peculiarities found in Codex Mashhad. The strangeness of the orthog-
raphy of the Codex is so pronounced that it cannot be attributed to the rules of
Qurʾānic spelling as they were from the second/eighth century onwards.

5.3 The Third Hypothesis

By modifying our first hypothesis we may perhaps find the best explanation.
The initial status of Codex Mashhad, comprising an official ʿUthmānic text
with regional Medinan readings and Ibn Masʿūd’s arrangement of the sūras,

34 Given that the text of Codex Mashhad does not resemble the Kufic tradition and reading
of Ibn Masʿūd, are Ubayy b. Kaʿb and the Medinan tradition not better alternatives? I think
not. As we have seen before, the Islamic literary sources cite sixteen places in which the
ʿUthmānic arrangement of sūras match Ubayy’s one (al-Itqān, vol. 1, pp. 222–223; al-Fihrist,
pp. 29–30); none can be found sequentially in Codex Mashhad. See also Nöldeke-Schwally,
Geschichte des Qorāns, vol. 2. Die Sammlung des Qorāns, Leipzig, 1919, pp. 30–33.
35 Ibn Abī Dāwūd al-Sijistānī, Kitāb al-Maṣāḥif, p. 157 and pp. 272–273.

Journal of Islamic Manuscripts 10 (2019) 292–326

324 karimi-nia

figure 13
MS Ṣanʿāʾ, DAM 01–32.1, Sūrat al-Shuʿarāʾ (Q. 26) is fol-
lowed by Sūrat al-Ṣāffāt (Q. 37)
UNESCO image No. 060042B

was not the work of the followers of Ibn Masʿūd, but rather that of supporters
of ʿUthmān, or better yet, the supporters of government. Accordingly, since the
dominant tradition in Kufa was Ibn Masʿūd’s arrangement and reading, a scribe
from either inside or outside of Kufa has written the governmental Qurʾān in
the form and arrangement of Ibn Masʿūd. Therefore, the textual characteris-
tics of the Codex match neither those of Ibn Masʿūd’s Qurʾān, nor the regional
ʿUthmānic codices of Kufa.
Supported by historical reports, the availability of different versions of Ibn
Masʿūd’s codex during the first Islamic centuries is a well-known fact. Thus, in
the fourth/tenth century, Ibn Nadīm states, “I have seen manuscripts claimed
by the copyists to be Ibn Masʿūd’s codex; however, no two such manuscripts
are similar. I have even seen a manuscript in which Sūrat al-Fātiḥa is also
included.”36 Again, al-Farrāʾ frequently refers to baʿḍ maṣāḥif ʿabdillāh, to some
of Ibn Masʿūd’s codices, and not to a single certain codex of his.37 These all
illustrate that there were different codices attributed to Ibn Masʿūd and that
some of these Qurʾānic manuscripts may not have been much different from
the official ʿUthmānic version in terms of the text.
Broadly speaking, the same combination, i.e. an official ʿUthmānic text in
Ibn Masʿūd’s arrangement of the sūras, may be claimed for one of the Ṣanʿāʾ
manuscripts, i.e. DAM 01–32.1. Based on the few images published by UNESCO,38

36 Ibn al-Nadīm, Muḥammad b. Isḥāq, Al-Fihrist, ed. Riḍā Tajaddud, Tehran: (the author),
1971, p. 29.
37 See al-Farrāʾ, Maʿānī al-Qurʾān, vol. 1, pp. 202, 220, 289; vol. 2, p. 350; vol. 3, pp. 21, 30. 38,
102, 132, 160 and 274.
38 UNESCO ‘Memory of the World’ Program—Sanʿa Manuscripts CD. For a three-part review
of this work see Keith E. Small and Elisabeth Puin, “UNESCO CD of Ṣanʿāʾ MSS. Part 1”,
Manuscripta Orientalia 12ii (2006) pp. 65–72; “UNESCO CD of Ṣanʿāʾ MSS. Part 2: Qurʾan

Journal of Islamic Manuscripts 10 (2019) 292–326

a new document in the early history of the qurʾān 325

one can clearly observe that Sūrat al-Ṣāffāt (Q. 37) begins on one of the folios
right after the conclusion of Sūrat al-Shuʿarāʾ (Q. 26) [figure 13]. This arrange-
ment is only reported in the versions attributed to Ibn Masʿūd.39 It is worth
noting that, like Codex Mashhad, the text of this Ṣanʿāʾ manuscript is totally
ʿUthmānic.

6 Conclusion

Based on the above evidence, I arrive at the conclusion that Codex Mashhad
is an early version combining the official text of the Qurʾān with Ibn Masʿūd’s
arrangement of the sūras from the first/seventh century, later subjected to a
broad cut-and-paste operation to conform to the prevailing ʿUthmānic arrange-
ment. Although it remains impossible to accurately determine when and how
the Codex was produced, based on intertextual evidence, the Codex should
be considered one of the oldest, even one of the first-century compilations
probably in Medina/Ḥijāz, since the extent of utilization of incomplete orthog-
raphy (scriptio defectiva) as well as the amount of spelling peculiarities, variant
readings, and copyist’s errors are far more impressive and divergent than what
is found in the known ḥijāzī and Kufic manuscripts of the second/eighth to
fourth/tenth centuries. More accurate answers as to the origin of this Codex
will hopefully be available upon radiocarbon dating of its two manuscript vol-
umes after the agreement of the authorities from the Āstān-i Quds Library in
Mashhad.

Acknowledgments

The first version of this article was presented at two conferences: “Aspects
of Quranic Scholarship: Philology meets Theology”, Berlin, Freie Universität
Berlin, 23–25 September 2016 and “Paleo-Qurʾānic Manuscripts Conference:
State of the Field”, Budapest, Central European University, 4–6 May 2017. I
would like to thank the authorities in the Office of the Manuscripts of the
Āstān-i Quds Library for assisting my access to the manuscripts of the Codex;
Behnam Sadeghi, Ala Vahidnia, Nicolai Sinai, and François Déroche for their

MSS. Contents in Sūra Order”, Manuscripta Orientalia 13i (2007) pp. 62–72; “UNESCO CD
of Ṣanʿāʾ Mss. Part 3: Qurʾān Palimpsests, and unique Qurʾān Illustrations”, Manuscripta
Orientalia 13ii (2007), pp. 59–71.
39 Ibn al-Nadīm, Al-Fihrist, p. 29.

Journal of Islamic Manuscripts 10 (2019) 292–326

326 karimi-nia

constructive comments during the last five years; Michael Cook for his fruitful
comments and suggestions and his excellent advice; my son, Mojtaba Karimi-
Nia for his help in re-transcribing the text of the Codex; Mojtaba Ebrahimzadeh
Ghias for his help in editing the English text; and Sayyid Jawād Shahrestānī, the
head of the Āl al-Bait Institute in Qum, who agreed to publish the Codex in a
facsimile edition. An expanded version of this article will be my introduction
to that edition.

Journal of Islamic Manuscripts 10 (2019) 292–326