You are on page 1of 30

Journal of

Near Eastern Studies

The Sanʿā
̣ ʾ Palimpsest: Materializing
the Codices
É LÉ ONORE CELLARD, Colleg̀ e de France*

Introduction The analysis conducted here demonstrates that the


fragments did indeed form a complete codex, and
Considered one of the most important documents in
one which when recontstructed reveals an early non-
the history of the transmission of the Qurʾā n, the S ̣anʿā ʾ
ʿUtmānic text, sharing some similarities with the so-called
palimpsest (Codex S ̣anʿā ʾ 1) is remarkable because of its
“Companions’ codices.”
unique lower text (i.e., erased text), which differs from
the canonical edition in both wording and the arrange-
ment of the sū ras (chapters). To this day, the text remains The discovery of the Codex S ̣anʿāʾ 1
the only evidence of a separate textual tradition from
In 1973, during restorations of the roof of the S ̣anʿāʾ
the hundreds of early Qurʾā n manuscripts in which the
mosque in Yemen,1 thousands of ancient documents,
text resembles the Vulgate. As a fragmentary document,
including hundreds of old parchment Qurʾā nic manu-
however, the S ̣anʿā ʾ palimpsest is difficult to study and
scripts, were found between the ceiling and the roof.
has been the focus of diverse hypotheses concerning its
This storage site may have functioned as a genizah, a
nature and function. The present study seeks to physi-
cally describe Codex S ̣anʿā ʾ 1, including the lower text,
1
The exact date of the discovery remains unclear. The most likely
and provide a snapshot of its initial appearance, as well
date is 1973, as noted by the archaeologist Paulo Costa, who super-
as new perspectives regarding its textual organization. vised the restoration work of the mosque. According to his article,
published in 1974, the manuscripts were found during these restora-
tion work in the course of 1973. See Costa, “Moschea” (1974): 487,
1. Sergio Noja Noseda also suggests the year 1973 in Noseda, “Mia
* I am grateful to Christian J. Robin and the CNRS UMR 8167 Visita” (2003). Publications by the German-Yemeni project staff re-
for providing me the photographs and ultraviolet images of the fer to the year 1972, however. Actually, the Yemeni Director of An-
DAM 01-27.1 folios. I also wish to thank Michael Marx, who made tiquities and Libraries, the Qādı̄ Ismāʿı̄l al-Aqwaʿ reports only that the
Razan Ghassan Ḥ amdū n’s dissertation available to me, as well as Ministry of Religious Affairs took the decision to undertake the re-
François Dé roche, Fred Donner, Guillaume Dye, Marijn van Putten, construction of the walls in the year 1392 A H / 1972 A D (“Ğāmiʿ”
Behnam Sadeghi, and Hythem Sidky for their helpful comments. [1985], 21), while Gerd-Rü diger Puin indicates 1971/1972 in “Meth-
[Ed.: Note also Nick Posegay’s article on a related topic in this issue ods” (1985), 9. According to Christian J. Robin (personal communica-
of JNES entitled “The Marking of Poetry: A Rare Vocalization System tion), the manuscripts were probably found sometime between Novem-
from an Early Qurʾā n Manuscript in Chicago, Paris, and Doha.”] ber 1972 and July 1973.

[JNES 80 no. 1 (2021)] © 2021 by The University of Chicago. All rights reserved. 0022-2968/2021/8001-0001$10.00. DOI: 10.1086/713473

1
2 ✦ Journal of Near Eastern Studies

repository for worn documents.2 It is not known how tural areas and scribal traditions that ranged from Latin
long the site had been in use,3 nor is the precise to Greek and Syriac. In most cases, several centuries
condition of the manuscripts at the time of discovery elapsed between the first writing (“lower” text or scriptio in-
well understood, although they clearly showed signs ferior) and the reuse of the material for a second script
of neglect, dust, and water damage. Nor, finally, is it (“upper” text or scriptio superior). It is therefore assumed
known whether the leaves of the fragmentary books/ that the books that were erased were perceived as “obso-
codices were found together or were scattered and/ lete” or had become illegible because they were in “dam-
or mixed with leaves from various volumes.4 aged or defective condition.”7 Interestingly, palimpsests
Between 1980 and 1989, the Yemeni Department appear to be uncommon in the early Qurʾā nic scribal
for Antiquities launched a project, financed by the Ger- tradition. Indeed, only three or four examples of early
man Foreign Ministry, to restore and curate these frag- Qurʾā nic palimpsests are known besides the Codex
ments. A team of specialists under A. Noch of Hamburg Sạ nʿā ʾ 1.8
University restored approximately 8,000 parchment The Codex S ̣anʿā ʾ 1 is remarkable for a number of rea-
sheets, which they then classified, collating folios that sons. First, it consists of an extensive set of leaves repre-
had originally belonged to the same manuscripts. The senting more than 40% of a complete Qurʾā n codex
collection was then microfilmed5 and placed into stor- (see below, “The Codex S ̣anʿā ʾ 1 and its Fragments”).
age in the Dār al-maḫ tụ̄ tạ̄ t (DAM) alongside the Great Second, the codex was first produced (i.e., its scriptio in-
Mosque of S ̣anʿāʾ. ferior) and then reused (for writing the scriptio superior)
A few palimpsests were identified among these during the early Umayyad period, or slightly earlier accord-
Qurʾā nic fragments, including one of particularly signif- ing to carbon dating data.9 Its character as a palimpsest
icant scholarly interest. Before discussing this any case
further, however, it is important to provide some back- 7
Declercq, “Introduction” (2007), 13.
ground regarding palimpsests more generally. A pa- 8
This corresponds to approximately 0.01% of the entire available
limpsest is a document whose text has been intention- Qurʾā nic documentation. Aside from the Mingana-Lewis palimpsest
at Cambridge University Library (Ms. Or. 1287), which contains orig-
ally erased in order to reuse the material, sometimes
inally Qurʾā nic folios that were erased and reused in a Christian context,
more than once. This ancient recycling practice has been others were produced in different contexts. The S ̣anʿā ʾ collection ap-
noted in texts as old as the 3rd century BC. It became pears to preserve two additional palimpsests. A first was inventoried in
more widespread with the advent of parchment, partic- the collection of S ̣anʿā ʾ (DAM 18- ?.a). Like the DAM 01-27.1, both
ularly after the 5th century AD,6 in geographical and cul- layers of this folio palimpsest are also entirely Qurʾā nic; its scriptio supe-
rior could date from the 3rd or 4th century AH. An additional four Qurʾā n
leaves belonging to another copy (DAM 01-18.10) from the 3rd or
2
Sadan, “Genizah” (1986): 42 and Dé roche, Le Coran (2019), 4th century AH are also probably palimpsests. For the present, this as-
174–76 and 218. Both authors identify this storage site as a genizah sumption is based solely on my observations of the UNESCO photo-
(from the Hebrew root GNZ ). graphs and needs to be confirmed. Another palimpsest was identified
3
Costa, “Moschea” (1974): 505–506. The Qādı̄ Ismāʿı̄l al-Aqwaʿ from a facsimile available in the unpublished manuscript Mss. simulata
(“Ğāmiʿ” [1985], 20) assumes that the storeroom was already in use orientalia 6 from the Berlin Staatsbibliothek, Photographien von
in the beginning of the 4th/10th century, when the Arabic handwrit- ausgewählten Fragmenten aus der Omayyaden-Moschee in Damaskus
ing evolved from Kū fic to Nasḫ ı̄ and a number of old manuscripts had in verschiedenen Sprachen (1908), pls. 3a and 4a. It is a folio from the
become useless and illegible. collection of the Great Mosque of Damascus, with a Greek undertext
4
Dreibholz (“Treatment” [2003]: 132) mentions that “in a that was erased in the 3rd century AH to transcribe a Qurʾā nic text.
number of cases, a substantial part of the text block is still preserved,” The last example is the Copto-Qurʾā n palimpsest: a Coptic biblical text
later describing “such a fragmentary and dispersed state.” In another reused in the middle of the 2nd century AH for writing the Qurʾā n (see
Islamic storage site, located in the ʿAmr mosque in Fustạ̄ t,̣ extensive Cellard and Louis, “New Palimpsest” [2021]).
9
groups of folios were found, still stitched together. See Dé roche, These dates indicate that further investigation of the material
Transmission (2009), 21–22, and Cellard, Codex (2018), 3. and textual features of both layers is needed. First, the ornamentation
5
By Hans-Caspar Graf von Bothmer in winter 1996–97, who and titles of sū ras between chapters present in the text, as well as spe-
kept a copy of these microfilms (Puin, “Observations” [1996]: 107– cific symbols for separating groups of one hundred verses, are be-
111). For additional detail, see Sadeghi and Goudarzi, “S ̣anʿāʾ 1” lieved to represent later features from around the end of the 1st cen-
(2012). tury AH. Second, several carbon dating studies suggest an earlier date.
6
This practice was well known in antiquity, as the chemical proper- The Stanford ’07 folio, which was carbon-dated at the University of
ties of inks and erasure techniques are mentioned in Roman literature, Arizona, has a 68% (1σ) probability of belonging to the period be-
and several literary and documentary Greek papyri from the 3rd century tween 614 CE to 656 CE and a 95% (2σ) probability of belonging to
th
BC to the 9 century AD are palimpsests. See Schmidt, “Palimpsestes” the period between 578 CE and 669 CE, see Sadeghi and Bergmann,
(2009). “Codex” (2010): 353. Three other samples, from DAM 01-27.1, were
The S ̣anʿā ʾ Palimpsest ✦ 3

is also unusual for two reasons. First, very short interval What is Currently Known About
of time—between a decade and a half a century— the Codex S ̣anʿāʾ 1?
elapsed between the two layers of script, and the two
The Codex S ̣anʿāʾ 1 has been the focus of a number of
scripts are oriented in the same direction, independently
parallel scholarly investigations in the past fifteen years
of any efforts to economize parchment. As a conse-
which have resulted in a range of divergent opinions
quence, this codex challenges our understanding of
and hypotheses about the reconstruction of the scriptio
the raison d’et̂ re behind the creation of the palimpsest.
inferior, as well as the nature of the document. The
Second, the most striking feature of this text is that both
present essay considers this second set of questions.12
the upper and lower texts are Qurʾā nic. The scriptio in-
While the physical nature of the documents is essen-
ferior, which is readily visible, has now been partially resus-
tially a matter of codicology, their implications extend
citated, revealing substantial variants in consonants and se-
far beyond the objects themselves to include important
quences of the sū ras 10 by comparison to the canonical
questions regarding the nature and function of the text.
ʿUṯ mānic text of the scriptio superior, as well as nearly
Indeed, there are currently two divergent opinions
all other known Qurʾā nic materials.11
about the nature of the Codex S ̣anʿāʾ 1.
One of the most significant challenges facing scholars
According to one hypothesis, Elisabeth Puin, Beh-
is gaining access to such a document. In addition to the
nam Sadeghi, and François Dé roche assume that the
black-and-white microfilms created by the German team,
Codex S ̣anʿāʾ 1 was a complete codex that became dis-
a second set of images was made between 2005 and 2008
persed over time.13 In this view, S ̣anʿāʾ 1 was a codex
using both normal light and UV light as part of the digi-
whose text was intended as a whole. It is clearly possible
tization project “De l’Antiquité tardive à l’Islam,” with
that Qurʾā nic codices were used for various purposes de-
support from the Agence nationale de la recherche
pending on the size of the volumes. A large, heavy codex
(ANR), a French research agency. These images were of
such as S ̣anʿāʾ 1 may have been publicly displayed and
decisive significance to scholars, although much work re-
made accessible to the Muslim community.14 Because
mains to be done, as illustrated by folios belonging to this
it was in the form of a bound volume, studying it entails
palimpsest that have surfaced since the German project.
comparisons to other representative examples of the
Since 2014, access to the S ̣anʿāʾ mosque libraries that
written transmission of the Qurʾā n. It is indeed possible
house most of the materials has become impossible due
to the Yemeni civil war. The preservation status of the
leaves of Codex S ̣anʿāʾ 1 and hundreds of other Qurʾā nic
12
In his article “La Mia Visita,” Sergio Noja Noseda provided a
manuscripts in the collection is presently unknown.
description of DAM 01-27.1, as well as information about the auc-
tioned folios from Sam Fogg and Bonhams. In addition to extensive
descriptions by Sotheby’s, Oriental (1992), lot 551 and Sam Fogg, in
dated by different laboratories (as part of the project CORANICA). Fendall, Islamic Calligraphy (2003), Alba Fedeli studied both auc-
Here are the results published by Robin, “L’Arabie” (2015), 65. The tioned folios in “Early Evidences” (2005). Elisabeth Puin published
results obtained in the Lyon laboratory were assigned a 68% (1σ) prob- five articles between 2008 and 2014, primarily based on the micro-
ability: F.2: 543–643 CE; F.11: 433–599 CE and F.13: 388–535 CE. films of DAM 01-27.1. See E. Puin, “Koranpalimpsest (2008),”
These were submitted to dating by three additional laboratories. Ox- “Koranpalimpsest II” (2009), “Koranpalimpsest III” (2010), “Ko-
ford obtained 1423 ± 23 BP, while the Zurich result was 1437 ± 33 ranpalimpsest IV” (2011), and “Koranpalimpsest V” (2014). Mean-
BP, which corresponds after calibration to 595–650 CE (1σ) or 565– while, Behnam Sadeghi published a study of the auctioned folio Stan-
660 (2σ) and 595–658 (2σ) for Oxford. Kiel found a result of 1515 ± ford ’07 in 2010: see Sadeghi and Bergmann, “Codex” (2010). In
25 BP, which corresponds to 430–495 CE (1σ) or 530–610 CE (2σ). 2012, B. Sadeghi and M. Goudarzi published the first complete edi-
The implications of these various dating efforts will be the subject of a tion of the lower text of DAM 01-27.1 in addition to four auctioned
future publication. folios, see Sadeghi and Goudarzi, “S ̣anʿāʾ 1” (2012). Dé roche also
10
Already recorded in the catalogue of the exhibition in Kuwait; addressed several issues regarding the Codex S ̣anʿāʾ 1 in Qur’ans
see Masạ̄ h ̣if S ̣anʿā ʾ (1985). (2014), and more recently in Coran (2019). Last, Asma Hilali pub-
11
I chose to use “ʿUṯ mānic text” as the official appellation of the lished several articles: “Palimpseste” (2011), “S ̣anʿāʾ” (2015), “Writ-
standard edition, whose textual tradition “reportedly began with the ing” (2018), and a partial edition of DAM 01-27.1: Sanaa (2017).
13
codices ʿUṯ mān disseminated as the ancestors of all the manuscripts in In E. Puin, “Koranpalimpsest III” (2010), the author used the
the textual tradition” (Sadeghi and Bergmann, “Codex” [2010]: 347). terms “Primären Kodex” (p. 236) and “Sekundäre Kodex” (p. 238).
With few exceptions, most Qurʾā nic manuscripts are consistent with Sadeghi and Bergmann, “Codex” (2010): 357. Sadeghi and Goudarzi,
the “ʿUṯ mānic text” in terms of the sequencing of sū ras, as well as “S ̣anʿāʾ 1” (2012): 11 n. 21. For Dé roche’s opinion regarding the S ̣anʿāʾ
verses within the sū ras and individual words, independent of spelling palimpsest, see Qurʾans (2014), 48–56, and Coran (2019), 201–229.
14
variations. Dé roche, Coran (2019), 216.
4 ✦ Journal of Near Eastern Studies

that this artifact offers unique “direct documentary evi- entifically problematic. After careful examination of the
dence for the reality of the non-ʿUṯ mānic text types.”15 photographs and a review of observations by scholars,
According to a second hypothesis, Asma Hilali sug- conservators, and scientists, however, I concluded that
gests that neither the first nor second Qurʾā nic layer a preliminary study was possible. It is my assumption
was intended to be part of a codex. Instead, Hilali that my additions to this knowledge base will be reex-
contends that the lower text was a “collection of dispa- amined and supplemented under more favorable con-
rate leaves”16 from the outset rather than a Qurʾā n co- ditions in future research.
dex. In view of the variations in layout and amend-
ments and corrections of both layers, Hilali believes
The Codex Sanʿa
̣ ̄ ʾ 1 and Its Fragments
that the S ̣anʿāʾ palimpsest consists of Qurʾā n passages
“written in a fragmentary fashion consisting of multiple The Dispersal of Fragmentary Books
sessions of teaching or dictation circle. [. . .] The frag-
Ancient books inevitably reach us in a fragmentary
mentary nature of the Qurʾā n seems to define the com-
state. Some are well-preserved, but most lose their bind-
position of the lower text.”17 In this view, if the Codex
ings or stitching, ultimately breaking bound volumes into
S ̣anʿāʾ 1 constituted only temporary student notes,
isolated folios. Fragments of a single manuscript can thus
completed and revised over time in a teaching circle,
become widely dispersed, as occurred with the Qurʾā n
it parallels the written transmission of the Qurʾā nic
manuscripts collection of the ʿAmr Mosque in Fustạ̄ t.̣
book. Its text therefore remains ʿUṯ mānic, with variants
In the early nineteenth century, two French scholars—
that result from dictation and interpretation, as well as
Jean-Joseph Marcel and Jean-Louis Asselin de Cher-
the “mobility” of specific passages in teaching environ-
ville—gained unrestrained access to the collection which
ments.18 If this is the case, however, the artifact does
enabled them to compile two significant collections, the
not, in fact, shed new light on the early history of the
first of which is now in St. Petersburg and the second
Qurʾā n codex and its formative stages.
currently housed in Paris. Although access to the manu-
Dé roche has very recently presented several elements
scripts in Egypt later became more difficult, the manu-
refuting Hilali’s hypothesis and supporting the idea of a
script trade never truly stopped circulating isolated
codex structure.19 In addition to the physical features,
folios from the ʿAmr Mosque, which still occasionally sur-
he assumes that the relationship between the lower and
face on the antiquities market.22
upper text reveals the true nature of the original manu-
Because it was apparently unknown until recent de-
script as a codex that was written and perhaps used for
cades, however, the S ̣anʿāʾ collection presents a unique
public rituals.20 However, Dé roche does not continue
case. Launched soon after the fragments were found,
his reconstruction of the initial structure of the codex,
the restoration project provided some protection for
emphasizing the lack of information about the hair and
the DAM collection, although these fragments are only
flesh sides in Hilali’s book and in the other publication.21
a part of a larger group. Indeed, some fragments had
This essay pursues this fundamental question by us-
already been discovered (damaged by heavy rains) dur-
ing the tools of codicology: what was the original ap-
ing a roof inspection in 196523 and transferred to the
pearance of Codex S ̣anʿāʾ 1? The primary purpose of
Awqāf Library (or Eastern Library, al maktaba al-
the study is to reconstruct the original physical struc-
s ǎ rqiyya). According to the Yemeni Director of Antiq-
ture of the Codex S ̣anʿāʾ 1 in order to evaluate how
uities and Libraries, the Qādı̄ Ismāʿı̄l al-Aqwaʿ, several
the text was organized.
manuscript leaves were removed from this storage site
I initially believed that codicological analysis without
and circulated on the commercial market trade.24 The
direct physical observation would be pointless and sci-

15 22
Sadeghi and Goudarzi, “S ̣anʿāʾ 1” (2012): 19. Cellard, “Manuscrits” (2019): 667–69.
16 23
Hilali, Sanaa (2017), 19. According to local newspapers, other hiding places holding man-
17
Ibid., 67. uscript fragments were discovered behind the upper section of a wall in
18
Ibid. March/April 2007, (http://www.sabanews.net/ar/news128835
19
Dé roche, Coran (2019). His observations are based on DAM 01- .html, last accessed March 17, 2019, but no longer available by Decem-
27.1 and the four auctioned folios. However, he mentions the existence ber 2020) and, in 2013, in the roof (http://yemenat.net/2013/01
of another part of the palimpsest in the mosque’s library (see below). /98807, last accessed December 2020). As of this writing, no addi-
20
Ibid., 216. tional information is available concerning these fragments.
21 24
Ibid., 215. al-Aqwaʿ, “Ğāmiʿ” (1985), 20.
The S ̣anʿā ʾ Palimpsest ✦ 5

auctioned leaves of the palimpsest that emerged on the French-Italian project, although an illustration of its recto
antiquities market between 1992 and 2008 did not nec- is included in the catalogue of the exhibition in Kuwait.30
essarily originate in the DAM collection—which includes The leaves are damaged, clearly due to conditions
the fragments discovered in 1973—but may belong to an above the mosque ceiling.31 Nearly all of them consist
earlier find.25 of isolated folios, approximately half of which are dam-
aged only on their edges, while some are reduced to
one-tenth of their original size. The better-preserved
The Codex S ̣anʿā ʾ 1 and its Dispersed Folios
folios measure between 365 mm high × 280 mm long
Reuniting the leaves (a “leaf,” or folio, designates a sheet and 371 × 280 mm. These dimensions, let us remem-
of a book comprising two pages, the recto and the verso) ber, correspond to materials which have been restored
of a single manuscript demands close study of the pale- and processed using humidification and flattening.32
ography, codicology, and texts of leaves potentially be- There are three primary textual sequences of the
longing to the same manuscript. Paleographical and co- Qurʾā n in the scriptio superior, although they are inter-
dicological evidence have made it possible to unite eighty rupted by significant lacunae. The first two folios range
folios for the purposes of the present study. We can iden- from Kor 6, 49 to 7, 11 (with a lacuna); the third folio
tify three groups of folios. The first one, DAM 01-27.1, is isolated with Kor 9, 112 to 127. The final sequence
has thirty-five folios. The second one, preserved in the (ff.4 to 38) ranges from Kor 14, 32 to 60, 1, with many
Eastern Library (with no shelfmark until now), has forty short lacunae.
folios. The last group gathers five folios dispersed on the
antiquities market (see Table 1).
The Forty Folios in the Eastern Library (al maktaba
al-s ǎ rqiyya)33
The Thirty-five Folios of DAM 01-27.126
In 2012, an unexpected discovery revealed forty addi-
The DAM 01-27.1,27 discovered above the mosque ceil- tional folios that can be indeed connected with the same
ing, consists of thirty-eight leaves or fragmented leaves palimpsest. These leaves are stored in the Eastern Li-
and only thirty-five folios that were originally part of the brary, the awqā f library, which is located in the S ̣anʿāʾ
S ̣anʿāʾ palimpsest. Three small leaf fragments (F.8, 14 mosque. The text was identified thanks to a 2004 Mas-
and 2928), also numbered DAM 01-27.1, should be ex- ter’s thesis by Razan Ghassan Ḥ amdū n.34 In her study,
cluded from this set because they are not part of the pa- Ḥ amdū n provides images of nearly all of the leaves,35 as
limpsest but belong to another manuscript, DAM 01- well as the transcription and analysis of the upper text.
25.1.29 Folio 22 is missing from the photographs of the
30
Kuwait, Ibid., 59/44. The deciphering of the scriptio inferior
25
Witkam, “Review Masạ̄ h ̣if S ̣anʿā ʾ ”: 124. was performed by Puin, “Koranpalimpsest III” (2010).
26 31
My observations regarding DAM 01-27.1 are based on images The dry environment and unusually low humidity (10–20%)
from the DATI project, kindly provided to me by Ch. J. Robin. I primar- did not damage the parchment itself. However, curators noted a layer
ily used the set of unretouched pictures (folder “27.1 scatti”) and did not of dust as well as water and insect damage. See Dreibholz, “Preserv-
modify the folios’ numbering. The images are 4888 × 6510 pixels. ing” (1999): 21–25.
27 32
The signature represents the principal criteria of classification: Ibid.: 27–28.
33
“01” is related to an inconsistent number of lines per page; “27” is The existence of these folios is mentioned in scholarship pub-
the maximum length of the lines given in centimeters. The last number lished since 2012. See Dé roche, Qurʾans (2014), 52 n. 50, and Coran
“1” is the numbering of different volumes sharing these same criteria. (2019), 218–19; and Puin, “Koranpalimpsest V” (2014): 477–78.
28
My numbering follows that of the photographs from the Hilali’s edition does not take them into account, although she admits
French-Italian project. Folios numbering changes depending on the that “they might be a missing part of MS 01-27.1” but called for a
publications. “thorough codicological study [. . .] in order to confirm that there is
29
According to Hilali (Sanaa [2017], 33), the thirty-eight folios a link,” in Sanaa (2017), 15.
34
of DAM 01-27.1 belong to the palimpsest. However, E. Puin (“Ko- Ḥ amdū n, Maḫ tụ̄ tạ̄ t (2004). The information given by Ḥ amdū n
ranpalimpsest IV” [2011]: 312) and Sadeghi and Goudarzi limit the concerning the size of the leaves and the average number of lines per
total number to thirty-six. In addition to ff.14 and 29—already ex- page helped re-establish the link between these leaves and the S ̣anʿāʾ pa-
cluded in the publications of Puin and Sadeghi and Goudarzi—f.8 limpsest. See the Islamic Awareness website: https://www.islamic
should also be excluded from DAM 01-27.1. These three folios -awareness.org/quran/text/mss/soth.html (last accessed: December
should be reconnected to DAM 01-25.1: DAM 01-27.1, f.29 is part 2020). E. Puin also mentions the similarity between the script of this
of DAM 01-25.1, f.22; DAM 01-27.1, f.8 is part of DAM 01-25.1, set and DAM 01-27.1 in “Koranpalimpsest V” (2014): 477–78.
35
f.13; and DAM 01-27.1, f.14 is part of a lost folio of DAM 01-25.1. Seventy-two plates of a total of eighty pages.
6 ✦ Journal of Near Eastern Studies

Table 1—The Codex S ̣anʿāʾ 1 and its fragments


Textual Sequences of the Upper Text
Location Number of Folios (See Table 3 for detail concerning contents,)
S ̣anʿāʾ Great Mosque, Eastern Library (al maktaba al-s ǎ rqiyya) 40 Kor 2, 246 – 5, 111 (with lacunae)
Kor 7, 40 – 12, 49 (with lacunae)
Copenhagen, David 86/2003 1 Kor 2, 265 – 2, 277
Stanford ’07 1 Kor 2, 277 – 286
Bonhams 2000 1 Kor 4, 33 – 56
Christie’s 2008 1 Kor 4, 171 – 5, 9
Louvre Abu Dhabi 1 Kor 5, 9 – 32
S ̣anʿāʾ Great Mosque, DAM 35 Kor 6, 49 – 7,11 (with lacunae)
Kor 9, 112 – 127
Kor 14, 32 – 60, 1 (with lacunae)

Curiously, however, she appears to consider them as sep- David collection in Copenhagen (David 86/2003).40 Its
arate entities (“maḫ tụ̄ tạ̄ t”) and not as leaves from a single dimensions are 366 × 282 mm. The upper text begins
volume.36 Moreover, Ḥ amdū n does not establish a rela- from Kor 2, 265 and ends at 2, 277, with twenty-four
tionship between these folios and DAM 01-27.1. Be- lines per page on the recto and twenty lines on the verso.
cause of the war in Yemen, no further information about The lower text has Kor 2, 206–223.
these leaves is currently available. Stanford ’07:41 In October 1993, Sotheby’s sold an-
The leaves are in better condition than the folios of other folio to a private collector.42 Fortunately, in 2007
DAM 01-27.1. The outer edges of several folios are Sadeghi and Bergmann had the opportunity to study this
damaged, however, and have not been treated using hu- folio at Stanford University.43 The folio—the upper text
midification and flattening (like DAM 01-27.1), which runs from Kor 2, 277 to 2, 286—precisely follows David
explains the smaller dimensions—350 × 265 mm—pro- 86/2003 and measures 365 × 281 mm, with twenty-
vided by Ḥ amdū n.37 Furthermore, they remain gathered three lines per page on the recto and twenty-two lines
in eight quires (see below, “The S ̣anʿāʾ palimpsest: a on the verso. The lower text has Kor 2, 191–205.
multigathering codex”), with some missing folios. Bonhams 2000: Bonhams auctioned this folio in
These folios hold two primary and almost continuous 2000,44 which measures 368 × 276 mm. The upper
textual sequences: The first is from Kor 2, 246 to 5, 111 text—Kor 4, 33–56—is written with twenty-six lines
(ff.1-22), with several missing folios; and the second is per page on the recto and twenty-seven lines on the
from Kor 7, 40 to 12, 49 (ff.23-40), also with lacunae. verso. The lower text has Kor 5, 41–54.
Christie’s 2008: Christie’s sold this folio, now in a
private collection, in 2008.45 The folio measures 363 ×
280 mm. The upper text—from Kor 4, 171 to 5, 9—
The Five Folios From the S ̣anʿāʾ Mosque
has twenty-eight lines per page on the recto and thirty
In addition to these fragments, five isolated folios have
circulated on the antiquities market and are now in
public or private collections. All of them complete the and Manuscripts, 1 May 2001, lot 12) and came to Sam Fogg Art
group stored in the Eastern Library.38 These include Gallery. See Fraser, Ink and Gold (2006). It seems to have transited
through the Schøyen collection (MS 4584) before reaching the Da-
the following:
vid collection.
David 86/2003: This folio appeared for the first time 40
Images available on: https://www.davidmus.dk/en/collec
in a Sotheby’s auction in October 1992.39 It is now in the tions/islamic/materials/calligraphy/art/86-2003 (last accessed De-
cember 2020).
41
For this folio and the next three, I have retained the names as-
36
Ḥ amdū n, Ibid., 50. signed by Sadeghi and Bergmann, in “Codex” (2010).
37 42
However, she added that the height varies slightly from one fo- Sotheby’s, Oriental Manuscripts and Miniatures, 22 October
lio to the next: Ibid., 50. 1993, lot 31.
38 43
Previously referred to by E. Puin, “Koranpalimpsest V” (2014): Sadeghi and Bergmann, “Codex” (2010).
44
478. Fedeli, “Early Evidences” (2005). Bonhams, Islamic (2000),
39
Sotheby’s, Oriental Manuscripts and Miniatures, 23 October 1–14.
45
1992, lot 551. It was then sold by Christie’s in 2001 (Islamic Art Christie’s, Art (2008), 24–27.
The S ̣anʿā ʾ Palimpsest ✦ 7

lines on the verso. The lower text has Kor 63, 1–11; holes caused by weaknesses in the animal skin. The holes
62, 1–11 and 89, 1–90, 6.46 are smaller than in other manuscripts, such as Codex
Louvre Abu Dhabi: This last folio has recently been Parisino-Petropolitanus or BnF Arabe 330g. Another
identified, although only a photograph of its verso is common defect in early Qurʾā nic manuscripts is the irreg-
currently available.47 The upper text is consecutive to ular outer edges of the parchment, which correspond to
the Christie’s folio. The recto should have Kor 5, 9–18, the natural edges of the animal skin. These edges should
while the verso preserves Kor 5, 18–32. Following this have been eliminated by slight trimming. These natural
sequence, the text continues on f.19 the Eastern Library. edges can also be seen on some DAM folios, although
The lower text is partially decipherable: it has the end of they do not affect the layout or the script, except on
Kor 80, followed by the complete sū ra 83. f.9, where a slight misalignment is visible. Similar and
sometimes more obvious anomalies are frequent in other
early Qurʾā nic manuscripts, such as the Codex Parisino-
Codicology and Paleography of the Codex S ̣anʿā ʾ 1
Petropolitanus.
Eighty folios currently belong to this artifact. The upper
text retains approximately 40% of the ʿUṯ mānic text. The
The Lower Text
complete codex probably consisted of between 160 and
180 folios,48 slightly fewer than the reconstructed Codex Nearly all of the existing folios are palimpsests, except
Parisino-Petropolitanus.49 for the first of DAM 01-27.1, in which the scriptio in-
The manuscript is large, but in its original state, it was ferior cannot be discerned in the existing photograph.50
even larger. The dimensions of the largest folios currently These palimpsests all use folios from the same original
measure 365 mm high × 280 mm long and 371 × 280 mm. document with the following five features in common.
The codex margins appear to have been trimmed, as First, they were written by at least two scribes—A(SI)
shown by DAM 01-27.1, f.7, where the top line of the and B(SI)—who collaborated throughout the copy.51
scriptio inferior has been partly cut. The primary codex Their script styles were highly similar and sometimes in-
was probably over 370 mm in height. Moreover, parch- tersected because they do not have a fixed script typol-
ment retracts slightly over time; considering the treat- ogy. However, it is possible to discern different tenden-
ment using humidification, we could still reasonably add cies. Copyist A(SI) wrote with a thicker, stockier nib
10 mm, meaning the parchment’s original height would than B(SI) and made vertical strokes slightly angled to
have been 380 mm at the very least. A slight cropping the right. Regarding the letter shapes, four are highly sig-
and retraction would similarly be true of the horizontal nificant, as shown in Table 2.
axis or width of the folios. The original dimensions of Second, the copyists systematically alternated with
Codex S ̣anʿāʾ 1 could therefore have been at least 380 × each other between the recto and the verso of the same
290 mm. The volume would originally have been bigger folio.52 This collaboration is not surprising in early
than other early volumes, such as Codex Parisino-
Petropolitanus (330 × 240/48 mm) or Paris BnF Arabe
50
Darker traces between the lines of the upper script are present,
328c (330 × 245 mm).
however, suggesting that an earlier text was erased, possibly using a dif-
The Codex S ̣anʿāʾ 1 used a medium-quality parch- ferent process than the other folios. This needs to be confirmed based
ment that is consistent with the manufacture of Qurʾā nic on direct analysis or higher quality images. Regarding the folios in the
manuscripts in the early period. The parchment often Eastern Library, Fedeli visually inspected the leaves and confirmed that
has a high degree of opacity and reveals slight irregulari- they are palimpsests, see Hilali, Sanaa (2017), 29, n. 111. On f.19, the
ties. For example, some of the DAM folios have small lower script is no longer visible, but we should consider this observa-
tion with caution, as the image quality is poor.
51
The distinction of hands could not be performed on the other
46
This identification needs to be confirmed using higher quality leaves as they are not sufficiently clear for looking at such paleographic
images. details. Thus, all of the folios in the Eastern Library and the folio in
47
A newspaper published a photograph on Dec 29, 2017 (https:// Louvre AD have not been taken in account.
52
fridaymagazine.ae/life-culture/to-do/20-must-see-pieces-at-the Copyist A(SI) wrote the Stanford ’07 recto, Bonhams 2000,
-louvre-abu-dhabi-1.2146523, accessed December 2020; see no. 13). Christie’s 2008 verso, and DAM 01-27,1, ff.2, 3B, 4, 5A, 6B, 7,
48
This figure should be viewed with caution given variations in 10, 11, 16, 17A, 19, 20, 21, 23A, 24B, 32A, 33, 36B and the copyist
the number of lines of text from one page to the next, even within B(SI) wrote the others. Some copyists remain unidentified: DAM 01-
sections written by a single copyist. 27.1, ff.1 (presence of lower text?), 3B, 12, 15, 18, 22, 25, 24A, 34A,
49
Dé roche, Qurʾans (2014), 17. 35, 37, 38A.
8 ✦ Journal of Near Eastern Studies

Table 2—The Scribes of the Lower Text


Copyist A(SI) Copyist B(SI)

The letter hā ʾ has a rounded or pear shape,


The letter hā ʾ displays a heart-shape with the vertical bar bent to the left
(image: DAM 01-27.1, f.23A) (image: DAM 01-27.1, f.23B).

The final mı̄m has an almost nonexistent tail B(SI) prefers a clear upward tail
(image: Stanford ’07 recto). (image: Stanford ’07 verso).

The nū n has a semicircular shape and tends The nū n has a straighter, longer stroke,
to incorporate the dots used to mark the verses ending with a short return
(image: Stanford ’07 recto) (image: Stanford ’07 verso)

Wā w and fā ʾ/qā f have a triangular head Wā w and fā ʾ/qā f have a more rounded head
(image: Stanford ’07 recto). (image: Stanford ’07 verso).

Qurʾā nic scribal practices.53 It is worth noting that in scribal rule within the h ̣iğā zı̄ manuscripts55—requires
seven observed occurrences, the changing of hands does further study.
not necessarily coincide with the end of a verse (such as Third, the scribes employed a script style close to the
Stanford ’07 recto and DAM 01-27.1, f.23A), but often h ̣iğā zı̄ group,56 with slight variations in writing den-
occurs within a verse.54 Consequently, such a habit could sity.57 Fourth, both copyists also shared similar habits
imply copying from a written exemplar rather than via
dictation, in which case one would not expect copyists to 55
See for example Codex Parisino-Petropolitanus, S ̣anʿāʾ DAM 01-
relay to each other in the middle of a textual unit. In any 25.1 and 01-29.1.
event, this practice—which seems to be a well-known 56
For a definition of the h ̣iğā zı̄ script, see Dé roche, Catalogue
(1983), 35–36.
57
I.e., the number of lines per page. Both copyists have a close
writing density: A(SI) writes between twenty-two and thirty-one lines
53
For more detail, see Dé roche, Transmission (2009), 127–30. per page, and B(SI) writes between twenty-three and thirty-five lines
54
This phenomenon is clear in DAM 01-27.1, ff.5A-5B, 6A-6B per page. The latter corresponds to the verso of Christie’s 2008. This
and in Christie’s 2008 recto-verso. This could also be the case in the number does not take into consideration the unidentified folios. As a
fragmentary folios DAM 01-27.1, ff.32A-32B and f.36A-36B. result, it should be recalculated once all of the folios are identified.
The S ̣anʿā ʾ Palimpsest ✦ 9

Figure 1—DAM 01-27.1, f.20B, lower text: Ornament traced by


A(SI). (Courtesy CNRS-DATI).
Figure 3—Ḥ amdū n, MS 22A.

Figure 2—Christie’s 2008, recto, lower text: Ornament traced by


B(SI) in (Courtesy Christie’s).

to trace ornaments for separating the sū ras, as well as


the hundred-verse groups. Regarding the latter, they
drew medallions on Kor 2, 101 (DAM 01-27.1, f.2B)
and Kor 2, 203 (Stanford ’07 verso).58 When the copyists
arrived at the end of a sū ra, they almost always wrote a
subscript title formula (had̲ ihi ḫ atima sū ra . . .), and then
sometimes drew an elaborate ornament (see Figs. 1–3)
using the same black ink, with a single exception between
sū ra 8 and sū ra 9, where the title follows the basmala of
the following sū ra instead of preceding the ornament.59
Fifth and last (but not least), the basmala is not nec- Figure 4—Stanford ’07. Hand A of the scriptio superior (Courtesy of
B. Sadeghi and U. Bergmann).
essarily postponed at the beginning of a new line, but
was sometimes immediately written after the ornament,
tell us more about the background of the copyists or
when there was enough space. I have never encountered
possibly about their written exemplar.
such an economic presentation in the large Qurʾā nic
manuscripts like the Codex S ̣anʿāʾ 1, though this fre-
quently occurs in smaller early manuscripts.60 That could The Upper Text
Close examination of the upper script in the eighty folios
58
The medallion typology—close to 1.A.II in Dé roche, Cata- reveals that two copyists collaborated on the writing, as
logue (1983), 29—is similar to that of other early manuscripts, such in the lower text, although they organized their work
as Codex Amrensis 1. However, it is worth noting that all of the other differently. The first hand, A (see Figs. 4 and 5), may
manuscripts use these medallions to separate groups of ten verses, but have written the beginning of the codex. His contribu-
not for the hundred verses.
59
tion is now confirmed only in the first six existing fo-
DAM 01-27.1, f.5A. According to Puin’s reading in “Koran-
palimpsest III” (2010): 272, the title formula of sū ra 9 is written on lios (S ̣anʿāʾ, Eastern Library, Ḥ amdū n, MS ff.1-2, Stan-
lines 8 and 9. At the beginning of line 9, the end of the title “[a]l- ford ’07, David 86/2003 and Ḥ amdū n, MS ff.3-4).
anfal” is followed by a short basmala, used here for separating the con- His script is broad—the folios contain between nineteen
tents. My own reading supports this interpretation. Puin’s reading, as and twenty-five lines per page—and shows straight ver-
mine, does not include any reading instruction “lā taqul bi-smi llā hi ”
tical strokes, smooth angles—more in accordance with
on line 9, as it is assumed by Hilali in Sanaa (2017), 39–40, and
Dé roche, Coran (2019), 208, or, with a slight difference, by Sadeghi
documentary scripts—and a slight accentuation in the
and Gourdarzi in “S ̣anʿāʾ 1” (2012): 53, n. 157. According to the ed- treatment of the curves.61 However, the letter shapes
itors, the copyist “wanted to write lā taqul bi-smi llā hi” here but con-
fused the grapheme with the similar [a]l-anfal).
60 61
See, for example, Dé roche, Qurʾans (2014), Figure 16, Istan- Note the particular treatment of lā m-alif, with strokes curved
bul, TIEM, ŞE12827/1. towards each other and the final kā f shaped like a hairpin.
10 ✦ Journal of Near Eastern Studies

Figure 7—Ḥ amdū n, MS 31B. Hand B of the scriptio superior.

Figure 5—Ḥ amdū n, MS 1B. Hand A of the scriptio superior.


such as the final h ̣ā ’/ğı̄m/ḫ ā ’ or lā m-alif.62 The script
reflects a hybrid style, including certain tendencies of the
h ̣iğā zı̄ style, especially in the leading strokes and the
shapes of medial hā ’ and final/isolated tạ̄ , ended by a
short tail. On the other hand, it tends to reveal a more
elaborated angular script similar to C.I according to
Dé roche’s classification.63
In addition to the copyists, an illuminator (or possibly
several) was responsible for adding headbands between
the sū ras and medallions to mark groups of verses. The
ornaments are not systematic, however, and vary consid-
erably. Unlike the lower text, they were not traced during
the writing process—although the copyists left a larger
space for that purpose—but were added later.64 For mark-
ing the groups of ten and hundred verses, medallions
were inserted. They were sketched in brown ink (as in
Figure 6), and the majority were then painted in red
Figure 6—DAM 01-27.1, f.30A. Hand B of the scriptio superior and green, except on DAM 01-27.1, ff.30-38. This se-
(Courtesy CNRS-DATI). quence of folios also employed a different way to separate

62
These allographs may have misled Hilali (Sanaa [2017], 73),
are not significantly different from those of the other who identified at least four scribes.
63
See Dé roche, Catalogue (1983). The shapes of several letters
copyist. He left his work to copyist B at the end of Ḥ am-
could indeed be consistent with C.I. First, the medial ǧım ̄ , does
dū n, MS f.4b, in the midst of a quire. not cross the baseline but is laid down, implying a shift in the base-
Copyist B (see Figs. 6 and 7), a professional copyist, line. Second, the final or isolated nū n has a crescent shape, with all
wrote the other seventy-four existing folios of the codex, three parts well-defined. Third, the final or isolated qā f has a long,
using a more variable writing density (from twenty-three oblique tail before a U-shaped return.
64
In many instances, medallions were not added in the intended
to thirty-seven lines per page). The general appearance
spaces, but elsewhere (See, for example, DAM 01-27.1, f.11B where
of the writing style tends to improve as the copy pro- the original space is left on Kor 20, 105; while the medallion appears
gresses. However, the letter shapes remain highly homo- on Kor 20, 102). These medallions were clearly added by a different
geneous, despite the use of allographs for certain letters, hand.
The S ̣anʿā ʾ Palimpsest ✦ 11

the sū ras, by adding colored headbands in the blank space The few known exceptions to these scribal practices,
(See ff.35-38).65 particularly in Qurʾā nic manuscripts, are exemplified by
Consequently, at first sight, the layout of the upper the S ̣anʿāʾ palimpsest (see Fig. 9).69 The direction of the
text appears formally heterogeneous, an impression writing in the lower and upper text is identical, except
that fades when we consider the original organization for seven instances in which the script is upside down
of the codex before it deteriorated. We shall return to (1807).70 This indicates that the format of the primary
this later. codex is maintained in the secondary codex, implying
the reuse of original bifolios and folds in the secondary
codex.71
The Sanʿā
̣ ʾ Palimpsest: A Multigathering Codex
The bifolios and their arrangement
Bifolios and Stitch-marks
By definition, a codex is composed of sheets made of
Hilali examined the DAM 01-27.1 manuscript in 2012,
papyrus, parchment, or paper which are folded to form
emphasizing the “scattered nature of the leaves”72 and
bifolios. The latter are then gathered into either one vol-
lack of evidence of existing bifolios. However, pictures
ume (a single gathering codex) or multiple quires (a
from the French-Italian project clearly show that ff. 23B
multigathering codex).66 As it is not always easy to find
and 24A remain joined together,73 as previously noted
sheets large enough to make bifolios, they are quite fre-
by Noseda, Puin, and Dé roche.74 The textual sequences
quently replaced by artificial bifolios, composed of two
in this bifolio are continuous from one folio to the next
smaller sheets joined together with a stub. When dealing
in both the scriptio inferior and scriptio superior (see
with fragments, detailed analysis—coupled with the col-
Table 3 below). Moreover, because this bifolio was placed
lation of the text—helps in the reconstruction of a docu-
in the same position in both primary and secondary codi-
ment’s original condition. In the specific case of palimp-
ces at the center of a quire, it is possible that part of the
sests, this analysis should go back gradually in time, first
arrangement of the leaves in the primary codex was re-
focusing on the presentation of the codex with the scrip-
tained when it was reused.
tio superior before considering the original codex and its
Previous studies also highlight other parallel se-
scriptio inferior. In the following demonstration, I use
quences in both codices. Puin, for example notes three
the term “primary codex” to refer to the original codex
similar sequences in thirteen folios of DAM 01-27.1.75
before it was disassembled, and “secondary codex” to des-
ignate the subsequent codex.67
Most currently-known palimpsests reveal a transfor- centuries, it is highly probable that the original bifolios were already
mation in format between the primary codex to the sec- disassembled. The craftsman thus had to work with isolated folios.
69
ondary one. Indeed, the direction of writing of the Among exceptions, the other Qurʾā n palimpsests found in
S ̣anʿāʾ appear to adhere to the same practice as the Codex S ̣anʿāʾ 1.
scriptio inferior is typically perpendicular to that of the
The Copto-Qurʾā n palimpsest constitutes also a unique case in which
scriptio superior. This implies the re-use of isolated folios the vertical Coptic folios were folded or cut, creating an oblong co-
that were rotated 907 to produce bifolios with a new dex with a fold strangely located at the bottom or at the top. See
fold, which is necessary for stitching quires together Cellard and Louis, “New Palimpsest” (2020). In addition to Qurʾā -
(see Fig. 8). A single folio of the primary codex thereby nic scribal practices, see Sokolov, “Christian” (1979).
70
produces two folios in a secondary codex. The primary For folios preserved in the Eastern Library, see ff.3, 8, 13, 14.
And DAM 01-27.1, ff. 17, 20 and 32.
motivation for this practice is clearly to economize valu- 71
Dé roche also discusses this issue in Coran (2019), 211–12.
able writing materials.68 72
Hilali, Sanaa (2017), 66.
73
The folios are joined together in their upper part for about
150 mm, including 80 mm glued using a sheet of paper (observation
65
Another header, sketched in pencil only, can be found in the based on unretouched pictures).
74
folios kept in the Eastern Library (Ḥ amdū n, MS f.11A). The orna- Noseda, “Mia Visita” (2003): 54; Puin, “Koranpalimpsest IV”
mental vocabulary used in these headers recalls that of several (2011): 316; Dé roche, Le Coran (2019), 212. Von Bothmer pro-
Qurʾā nic manuscripts in C.I script. See Cellard, Transmission (2015). vided an illustration of this bifolio in “Masterworks” (1987). The
66
Muzerelle, Vocabulaire (1985). same bifolio is also illustrated in Blair, Calligraphy (2008), 119.
67 75
These terms are used by E. Puin in “Koranpalimpsest III” See ff. 3, 4, and 5 (with a lacuna between f.3 and f.4); ff.21–24;
(2010): 236–38. ff.19–20. (E. Puin, “Koranpalimpsest III” (2010): 252). The num-
68
However, when the time interval between the lower text and bering of the folios provided by Puin has been converted using the
the reuse of the material for the upper text extends beyond several numbering of the CNRS-DATI images.
12 ✦ Journal of Near Eastern Studies

Figure 8—Qurʾānic palimpsest written over a Greek text, rotated 907 (Photographien von ausgewählten Fragmenten aus der Omayyaden-
Moschee in Damaskus in verschiedenen Sprachen).

Sadeghi also observes that Stanford ’07 and David 86/ less stitched together, revealing longer continuous se-
2003 are “adjacent in both the lower and upper lay- quences, at least for the upper script (ff. 3–9; 10–12;
ers.”76 These continuous sequences may therefore be 13–18; 19–22; 23–32; 33–40). It remains unclear
indicative of remnants of quires from both the primary whether the scriptio inferior is also continuous on these
and the secondary codex. leaves.
The folios in the Eastern Library are in better con- Regarding the stitching of the gatherings, unlike the
dition: several quires remain visible and remain more or set in the Eastern Library, DAM 01-27.1 has no thread
remnants. Wherever the inside edges of the folios are
preserved, however, they contain a number of holes that
76
Sadeghi notes that in Stanford ’07 and David 86/2003, “the
correspond to symmetrically arranged stitches (at least
2 folios are adjacent in both the lower and upper layers” (Sadeghi five, but several others can be observed at the top and bot-
and Bergmann, “Codex” [2010]: 357). tom edges). Some holes include small V-shaped nicks,
The S ̣anʿā ʾ Palimpsest ✦ 13

Figure 9—Codex S ̣anʿāʾ 1 (DAM 01-27.1, f.5B). The lower and upper texts are written in the same direction (Courtesy CNRS-DATI).

but in most cases, stitches are indicated by small, regular quire, the craftsman established a given number of
tears.77 bifolios, arranging the sides of the parchment—hair
and flesh—according to a specific pattern.78 For this
Identifying Hair and Flesh sides: Palimpsesting
Considerations
Since its emergence in the first centuries CE, the fabrica- 78
These rules varied from one cultural tradition to the next:
tion of the codex involved technical rules. For each Greek, Coptic, and Christian-Palestinian Aramaic traditions appear
to have promoted the quaternion (consisting of four bifolios, i.e.,
77
See DAM 01-27.1, ff. 1, 2, 5, 6, 7, 8 (here, one sewing station eight folios) with the parchment bifolios respecting Gregory’s rule:
is still visible, and there are maybe traces of another fold close to the flesh faces flesh, and hair faces hair; while the Syriac tradition adopted
text justification: evidence of a stub?), 10, 11, 16, 19, 21, 23, 24, 25, the quinion (composed of five bifolios, i.e., ten folios) alternating
30, 32, 33, and 38. such that flesh faces hair.
14 ✦ Journal of Near Eastern Studies

Table 3—Upper and lower script contents of folios included in the present investigation and parchment sides (hair/flesh).
Hypothesis
Traces about the side
Quire Folio (+ Sadeghi’s numbering if different) Upper text Lower texta of erasures of the parchment
1 Stanford ’07, recto Kor 2, 265 – 271 Kor 2, 191 – 196 No Hair
Stanford ’07, verso Kor 2, 271 – 277 Kor 2, 197 – 205 Yes Flesh
David 86/2003, recto Kor 2, 277 – 282 Kor 2, 206 – 217 No Hair
David 86/2003, verso Kor 2, 282 – 286 Kor 2, 217 – 223 Yes Flesh
Ḥ amdū n, MS 3Ab (reversed folio) Kor 2, 286 – 3, 10 [Kor 7, ? – 37] No Hair
Ḥ amdū n, MS 3B Kor 3, 11 – 20 ? Yes Flesh
Ḥ amdū n, MS 4A Kor 3, 20 – 28 ? No Hair
Ḥ amdū n, MS 4B Kor 3, 28 – 37 [Kor 2, 243 – ?] Yes Flesh
Ḥ amdū n, MS 5A Kor 3, 37 – 47 [Kor 2, 249 – 255] No Hair
Ḥ amdū n, MS 5B Kor 3, 47 – 57 [Kor 2, 255 – 259] Yes Flesh
Ḥ amdū n, MS 6A Kor 3, 57 – 71 [Kor 2, 259 – 266?] Yes Flesh
Ḥ amdū n, MS 6B Kor 3, 71 – 81 [Kor 2, 266? – 275] No Hair
Ḥ amdū n, MS 7A Kor 3, 81 – 94 [Kor 2, 275 – 284] Yes Flesh
Ḥ amdū n, MS 7B Kor 3, 94 – 107 [Kor 2, ? – 7, ?]c No Hair
Ḥ amdū n, MS 8A (reversed folio) Kor 3, 107 – 119 [Kor 2, 230/1?] Yes Flesh
Ḥ amdū n, MS 8B (reversed folio) Kor 3, 119 – 133 [Kor 2, 223 – 230/1?] No Hair
Ḥ amdū n, MS 9A Kor 3, 133 – 145 [Kor 7, 38 – 53?] Yes Flesh
Ḥ amdū n, MS 9B Kor 3, 145 – 154 ? No Hair
2 Ḥ amdū n, MS 10A Kor 3, 179 – 187 ? No Hair
Ḥ amdū n, MS 10B Kor 3, 187 – 199 [Kor 5, 115 – 120] Yes Flesh
Ḥ amdū n, MS 11A Kor 3, 199 – 4, 7 [Kor, 5, 91 – 99?] No Hair
Ḥ amdū n, MS 11B Kor 4, 7 – 14 [Kor 5, 100 – 107] Yes Flesh
Ḥ amdū n, MS 12A (no picture) ?
Ḥ amdū n, MS 12B Kor 4:24 – 4:33 ? Yes Flesh
Bonhams 2000 Kor 4, 33 – 43 Kor 5, 41 – 48 Nod Hair
Bonhams 2000 Kor 4, 43 – 56 Kor 5, 48 – 54 Yes Flesh
Ḥ amdū n, MS 13A (reversed folio) Kor 4, 56 – 64 [Kor 10, 59 – 69] ? ?
Ḥ amdū n, MS 13B Kor 4, 64 – 76 [Kor 10, 44 – 58] ? ?
Ḥ amdū n, MS 14A (reversed folio) Kor 4, 77 – 84 ? Yes Flesh
Ḥ amdū n, MS 14B Kor 4, 84 – 92 [Kor 5, 72 – 81] No Hair
Ḥ amdū n, MS 15A Kor 4, 92 – 100 [Kor 10, 97 – 11, ?] Yes Flesh
Ḥ amdū n, MS 15B Kor 4, 100 – 108 ? No Hair
Ḥ amdū n, MS 16A Kor 4, 108 – 120 ? Yes Flesh
Ḥ amdū n, MS 16B (no picture) ?
Ḥ amdū n, MS 17A Kor 4, 131 – 140 [Kor 10, 22 – 31] Yes Flesh
Ḥ amdū n, MS 17B Kor 4, 140 – 151 ? No Hair
Ḥ amdū n, MS 18A Kor 4, 151 – 161 [Kor 10, 1 – 10] Yes Flesh
Ḥ amdū n, MS 18B Kor 4, 161 – 171 ? No Hair
3 Christie’s 2008, recto (reversed folio) Kor 4, 171 – 5, 3 Kor 62, 11 – 89, 1 – 90, 6 No Hair
Christie’s 2008, verso Kor 5, 3 – 9 Kor 63, 1 – 62, 11 Yes Flesh
Louvre AD, recto (reversed folio) (no picture) [Kor 83, ? – ?]
Louvre AD, verso Kor 5, 18 – 32 [Kor 80, 31 – 83, 36]
Ḥ amdū n, MS 19A Kor 5, 32 – 42 ?
Ḥ amdū n, MS 19B (no picture) ?
Ḥ amdū n, MS 20A Kor 5, 49 – 61 [Kor 78, 24 – 76, 13?] No Hair
Ḥ amdū n, MS 20B Kor 5, 61 – 71 [Kor 76, 14 – 75, ?] Yes Flesh
Ḥ amdū n, MS 21A Kor 5, 71 – 82 [Kor 75, 18 – 81, 28/9?] Yes Flesh
Ḥ amdū n, MS 21B Kor 5, 82 – 93 [Kor 79, 1? – 40] No Hair
Ḥ amdū n, MS 22A Kor 5, 93 – 104 [Kor 79, 40 – 64, 10?] Yes Flesh
Ḥ amdū n, MS 22B (no picture) ?
4 DAM 01-27.1, 4A Kor 14, 32 – 41 Kor 11, 105 –112 ? ?
4B Kor 14, 52 – 15, 16 Kor 11, 120 – 8, 3 ? ?
5A Kor 16, 73 – 89 Kor 8, 73 – 9, 7 Yes Flesh
5B Kor 16, 89 – 102 Kor 9, 7 – 16 No Hair
The S ̣anʿā ʾ Palimpsest ✦ 15

Table 3— (continued)
Hypothesis
Traces about the side
Quire Folio (+ Sadeghi’s numbering if different) Upper text Lower texta of erasures of the parchment

6A Kor 16, 102 – 118 Kor 9, 17 – 26 Yes Flesh


6B Kor 16, 118 – 17, 6 Kor 9, 26 – 34 No Hair
5 7A Kor 17, 40 – 58 Kor 22, 15 – 26 No Hair
7B Kor 17, 59 – 77 Kor 22, 27 – 39 Yes Flesh
9A Kor 19, 38 – 64 Kor 33, 51 – 57 Yes Flesh
9B Kor 19, 64 – 98 Kor 33, 57 – 72 No Hair
10A Kor 20, 1 – 43 Kor ? – 24, 13 Yes Flesh
10B Kor 20, 44 – 74 Kor 24, 13 – 23 No Hair
11A Kor 20, 74 – 98 Kor 24, 23 – 32 Yes Flesh
11B Kor 20, 98 – 130 Kor 24, 32 – 40 No Hair
12A Kor 21, 16 – 29 Kor ? Yes Flesh
12B Kor 21, 38 – 42 Kor ? No Hair
6 13A Kor 21, 42 – 72 Kor 16, 26 – 37 No Hair
13B Kor 21, 72 – 92 Kor 16, 37 – 59 Yes Flesh
15A (14A) Kor 21, 111 – 22, 1 Kor 16, 68 – 69 No Hair
15B (14B) Kor 22, 15 – 16 Kor 16, 78 – 79 Yes Flesh
16A (15A) Kor 25, 10 – 34 Kor 20, 23 – 61 Yes Flesh
16B (15B) Kor 25, 34 –59 Kor 20, 61 – 80 No Hair
7 17A (16A) (reversed folio) Kor 26, 155 – 176 Kor 28, 30 – 35 Yes Flesh
17B (16B) Kor 26, 198 – 221 Kor 28, 19 –24 No Hair
18A (17A) Kor 27, 25 – 29 ? ? ?
18B (17B) Kor 27, 46 – 49 ? ? ?
19A (18A) Kor 28, 58 – 74 Kor 15, 4 –33 Yes Flesh
19B (18B) Kor 28, 74 – 86 Kor 15, 33 – 74 No Hair
20A (19A) (reversed folio) Kor 29, 29 – 40 Kor 25, 14 –27 No Hair
20B (19B) Kor 29, 43 –54 Kor 15, 87 – 99; 25, 1 – 8 Yes Flesh
8 21A (20A) Kor 30, 26 – 40 Kor 9, 70 – 81 No Hair
21B (20B) Kor 30, 40 – 54 Kor 9, 81 – 90 Yes Flesh
22A (21A)e Kor 31, 24 – 32, 4 Kor 9, 106 – 113 No Hair
22B (21B) Kor 32, 4 - 20 Kor 9, 114 – 120 Yes (?) Flesh (?)
23A (22A) Kor 32, 20 – 33, 6 Kor 9, 121 – 19, 5 No Hair
23B bifolio (22B) Kor 33, 6 – 18 Kor 19, 6 – 29 Yes Flesh
24A, bifolio (23A) Kor 33, 18 – 29 Kor 19, 29 – 53 Yes Flesh
24B (23B) Kor 33, 30 – 37 Kor 19, 54 – 74 No Hair
9 25A (24A) Kor 34,52 – 35, 9 ? No Hair
25B (24B) Kor 35, 10 – 18 Kor 30, 38 – 50 Yes Flesh
26A (25A) Kor 37, 38 – 59 Kor 39, 25 – 36 Yes Flesh
26B (25B) Kor 37, 73 – 88 Kor 39, 42 – 47 No Hair
27A (26A) Kor 37, 102 – 134 Kor 39, 51 – 70 Yes Flesh
27B (26B) Kor 37, 134 – 172 Kor 39, 70 – 40, 8 No Hair
28A (27A) Kor 38, 73 – 75 ? ? ?
28B (27B) Kor 39, 6 ? ? ?
10 30A (28A) Kor 41, 17 – 27 Kor 37, 15 – 33 Yes Flesh
30B (28B) Kor 41, 33 – 43 Kor 37, 33 – 43 No Hair
31A (29A) Kor 41, 47 – 42, 5 Kor 37, 82 – 103 Yes Flesh
31B (29B) Kor 42, 10 – 16 Kor 37, 118 – 144 No Hair
32A (30A) (reversed folio) Kor 42, 21 – 29 Kor 21, 5 – 19 Yes Flesh
32B (30B) Kor 42, 38 – 48 Kor 20, 122 – 133 No Hair
11 33A (31A) Kor 43, 63 – 69 Kor 12, 17 – 20 No Hair
33B (31B) Kor 43, 89 – 44, 11 Kor 12, 27 – 31 Yes Flesh
34A (32A) Kor 47, 15 – 20 Kor 12, 111 – 18, 5 Yes Flesh
34B (32B) Kor 47, 32 – 48, 2 Kor 18, 15 – 18 No Hair
12 35A (33A) Kor 55, 16 – 56, 4 Kor 34, 13 – 23 No Hair
16 ✦ Journal of Near Eastern Studies

Table 3— (continued)
Hypothesis
Traces about the side
Quire Folio (+ Sadeghi’s numbering if different) Upper text Lower texta of erasures of the parchment

35B (33B) Kor 56, 5 – 69 Kor 34, 23 – 33 Yes Flesh


36A (34A) Kor 57, 1 – 10 Kor 34, 40 – 47 No Hair
36B (34B) Kor 57, 16 – 22 Kor 13, 1–5 Yes Flesh
37A (35A) Kor 57, 27 – 58, 6 Kor 13, 6 – 14 No Hair
37B (35B) Kor 58, 11 – 22 Kor 13, 16 – 21 Yes Flesh
38A (36A) Kor 59, 1 – 10 Kor 13, 25 – 31 No Hair
38B (36B) Kor 59, 14 – 60, 1 Kor 13, 33 – 40 Yes Flesh
a
The content of the auctioned and DAM 01-27.1 folios is based on Sadeghi’s edition (“S ̣anʿāʾ 1” [2012]: 37–39). The content of the folios of
the Eastern Library, based on Sidky’s partial decipherment (pers. comm.), is set in brackets, indicating that this should be verified and com-
pleted when better images become available.
b
In Ḥ amdū n’s dissertation (Maḫ tụ̄ tạ̄ t [2004]), the photographs of the leaves and the printed edition are numbered consecutively. For the
present purpose, I renumbered the folios according to manuscripts’ rules (ex. F.1A for recto, F. 1B for verso).
c
The lower text is very difficult to read on that folio.
d
This observation is based on a picture of the printed catalogue and needs to be confirmed.
e
The pictures of this folio (recto and verso) are missing in the French-Italian pictures. Only the recto is illustrated in the catalogue Masạ̄ h ̣if.

reason, hair sides must be distinguished from flesh sides tial treatment was performed to correct the scriptio in-
in order to reconstruct the quires. ferior where parts of the text of the primary codex had
In many cases—and often with early Qurʾā nic manu- been scraped off 83 and re-written in black ink in some
scripts—the hair or grain side is characterized by traces instances. This correction process was never completed,
of hair follicles on the folio edge, in the form of residual however, as it was decided to completely erase the pri-
pigmentation and a yellowish tinge, while the flesh side mary codex by soaking the sheets and stretching them
is whiter. In the present case, however, these differences again on a frame. Puin’s assumption of such a process ac-
are not discernible in part due to the specific treatment tually implies that the original sheets were considerably
of the parchment during its original manufacture or sub- larger because their edges, fastened to the frame, would
sequent palimpsesting, which made the two sides highly have been cut after they had dried.84
homogeneous. During my examination of DAM 01-27.1, I discov-
This raises the difficult but critical question of the ered a significant feature of the technique used to erase
techniques used to delete the scriptio inferior. Scholars the scriptio inferior that enabled me to distinguish the
who examined the S ̣anʿāʾ palimpsest offered a number hair side from the flesh side. In effect, one side of each
of observations. Hilali reports “various type of erasures, folio of DAM 01-27.1 (except for f.1) is more damaged
which could be due to damage to the parchment or due than the other (see Figs. 10 and 11), because some ar-
to the palimpsesting technique.”79 Dé roche states that eas of the scriptio inferior were submitted to a deeper
only scraping could remove the metal-gall ink of the erasure technique, permanently destroying the script
lower text.80 Sadeghi occasionally observes such scrap- remnants.85 The same observation can be made con-
ing on the verso of Stanford ’07,81 but does not provide cerning the auctioned folios Stanford ’07, David 86/
further comment, noting only that the lower script was
“erased by scraping or washing.”82 Puin offers the most 83
E. Puin, “Koranpalimpsest III” (2010): 253–55.
affirmative hypothesis by identifying two stages in the 84
Ibid.: 253. This process is similar to the initial treatment of
deletion of the scriptio inferior. She assumes that an ini- skins. For more information, see Reed, Skins (1972). Stretching im-
plies that frames were specifically manufactured to fit the dimen-
sions of the bifolios.
79 85
Hilali, ibid., 32. Sadeghi notes examples of similar scraping on the verso of
80
Dé roche, Coran (2019), 211. Stanford ’07. E. Puin assumes that some instances of scraping were
81
Sadeghi and Bergmann, “Codex” (2010): 357. linked to a first attempt to correct the scriptio inferior before the com-
82
Sadeghi and Goudarzi, “S ̣anʿāʾ 1” (2012): 5. plete removal of the text, in Puin, “Koranpalimpsest IV” (2011): 314.
The S ̣anʿā ʾ Palimpsest ✦ 17

Figure 11—DAM 01-27.1, f.16A: Flesh side with traces of scraping


(Courtesy CNRS-DATI).

Figure 10—DAM 01-27.1, f.16B: hair side (Courtesy CNRS-DATI).


mersion in a lime bath to soften and release the hair and
2003, and Christie’s 2008, as well as the leaves in the epidermis and to treat the surface during drying.88 The
Eastern Library, as far as can be determined based on use of lime was not systematic in ancient times, how-
available photographs. ever, as illustrated by the Qurʾā n samples from Dublin
Given this background, the question is what tran- and Munich.89 In the present case, the use of lime could
spired with the S ̣anʿāʾ palimpsest. Chemical analyses of be related either to the original parchment manufactur-
the writing materials would help clarify this question. ing process or to later palimpsesting. Indeed, because
A non-destructive analysis performed in 2007 by Uwe the palimpsesting process implies chemical removal,
Bergmann on the Stanford folio revealed the presence of as in most of the cases, it is possible to resuscitate the
calcium in the parchment.86 Later analysis by Ira Rabin buried script.90 On the contrary, a mechanical abrasion
and Oliver Hahn (of Hamburg University) as part of would permanently destroy the material. Several medi-
the CORANICA project confirmed this finding. Analy- eval Latin recipes record the use of lime in various treat-
ses based on X-Ray fluorescence (XRF) and Raman ments,91 followed by polishing with pumice stone and
spectroscopy were performed on several samples from chalk. Mechanical abrasion was perhaps used to com-
S ̣anʿāʾ manuscripts and two Egyptian Qurʾā n fragments plement and finalize the work, as prescribed by certain
preserved in Munich and Dublin.87 The results high- Latin formulas.
lighted the singularity of the DAM 01-27.1 parchment, It is not currently possible to precisely identify the era-
which contains substantial traces of calcite, unlike other sure technique applied to the S ̣anʿāʾ palimpsest.92 My
Egyptian samples that were prepared using starch or
alum. It is therefore possible that the S ̣anʿāʾ palimpsest 88
Reed, Skins (1972), 132–51.
was treated with lime. 89
Liming (immersion in a lime bath) may have been replaced by
It is worth recalling a few aspects of the fabrication enzyme treatments. For a discussion of parchment production tech-
of parchment in ancient times. Lime was used at differ- niques, see Rabin, “Building” (2017).
90
ent stages of the preparation of parchment, both via im- According to numerous studies of palimpsests, most were washed
off. For example, see Irigoin, “Palimpseste” (1951): 447. See also
Tchernetska, “Cambridge” (2002): 318, and Easton and Knox, “Dig-
86
Bergmann, “Chemical” (2012). ital” (2004): 3.
87 91
The six samples from S ̣anʿāʾ were obtained as part of the DATI As a mixture of unslaked lime, vitriol, and alum. See Declercq,
project. Additional fragments came from Dublin CBL (Is.1615 I) “Introduction” (2007), 8.
92
and Munich, Bayerische Staatsbibliothek (Cod. Arab. 2569). I. Ra- This should be compared to another palimpsesting technique,
bin presented her conclusions in the Symposium Autour de Fustạ̄ t,̣ which occurs in DAM 01-29.1, f.33A. See Cellard, “Nouveau té mo-
Collège de France, 25 June 2015. ignage” (2020).
18 ✦ Journal of Near Eastern Studies

Figure 12—Stanford ’07, verso in natural light, non-scraped area Figure 13—Stanford ’07, verso under XRF analysis, copper on non-
(Courtesy of B. Sadeghi and U. Bergmann). These images were scraped area (Courtesy of B. Sadeghi and U. Bergmann).
uploaded to the web by B. Sadeghi, U. Bergmann, and M. Goudarzi
as companions to the papers “The Codex” and “S ̣anʿāʾ 1.”

My observations clearly demonstrate that the S ̣anʿāʾ


palimpsest consisted of bifolios, one of which has been
observations indicate two steps, the first one soft—per- preserved intact, evidence that the separated leaves
haps involving washing with a sponge—and a second originally formed an earlier codex that was subsequently
mechanical abrasion of only one side. The need to recycled into another one. The question remains, how-
scrape the parchment may have been due to the metallic ever, whether it is possible to reconstruct the original ap-
compounds in the ink used for the scriptio inferior. Indeed, pearance of these codices. Was each codex fabricated by
the XRF analysis of Stanford ’07 has revealed a high level simply assembling sheets of parchment, or was it a multi-
of copper, as well as iron and zinc, in the ink used for the gathering codex that resulted from a more complex
scriptio inferior.93 Metals, particularly copper, are highly structure consisting of several quires? What technical
corrosive of parchment. When applied to the flesh side, rules were observed in each codex? Reconstructing the
which is more permeable than the hair side, metallic com- unity of the codex could help us understand the organi-
pounds are able to penetrate into the collagen fibers be- zation of scribal work.
neath the surface. As a consequence, the script is more
difficult to delete on the flesh side (note the difference be-
tween the hair side in Figs. 12–13, and the flesh side in The Remnants of the Quires: A Hypothetical
Figs. 14–15)94 and its traces remain visible—although per- reconstruction
haps not as clearly as today95—after washing. This could
explain frequent signs of scraping on the flesh side, al- The following investigation focuses first on the continu-
though dating this operation remains problematic.96 ous sequences of the second codex in order to physically
Traces of scraping likely characterize the flesh side, in recreate the quires (see Table 4). Second, I attempt to
any event, meaning that it is possible to reconstruct the assess the links between the structure of the second
organization of the consecutive folios that have been and primary codices based on both textual sequences.
preserved. I expect that this detailed investigation of the structure
of the initial book also provides new insights regarding
the organization of the initial text.
This reconstruction of the quires will include the fo-
lios DAM 01-27.1, ff.4-38 (the first three are excluded
93
Sadeghi and Bergmann, “Codex” (2010): 358 n. 13.
94
because they are isolated text sequences), the auctioned
Cohen, “Composition” (2015).
95
This is because the metals eventually tend to oxidize.
folios and, to a lesser extent, the ff. 3–22 preserved in
96
The correction of the rasm has been scraped as well, implying the Eastern Library, as these are connected to the pre-
that this correction was made prior to erasure. vious examples. The contents of the scriptio inferior in
The S ̣anʿā ʾ Palimpsest ✦ 19

Figure 14—Stanford ’07, verso in normal light, scraped area (Cour-


tesy of B. Sadeghi and U. Bergmann). Figure 15—Stanford ’07, verso under XRF analysis, scraped area
(Courtesy of B. Sadeghi and U. Bergmann).
these folios is partially identified thanks to the efforts of
Hythem Sidky (see Figs. 16 and 17).97
and 9 were composed of four bifolios, folded to obtain
My reconstruction of the quires of the secondary co-
eight folios (quaternion). Quire 7 was an exception be-
dex rests on several assumptions. First, I assume that the
cause it may have been composed of six bifolios (senion).
erasures identify the flesh side. Second, because the
All of the quires conformed to the same practice of col-
number of lines per page is slightly irregular, my compu-
lating leaves of parchment so that the hair side was
tation of the missing folios depends on the proportion of
systematically outside of the quire, while inside of the
text kept on the undamaged folios within the same
quire the flesh side faced the hair side.100 This practice
quire.98 Third, on many occasions, several patterns could
was not systematically observed, however, in the re-
be possible. For this reason, I chose the most coherent
versed folios—where the scriptio inferior turned to
option based on comparison with other quires.
1807, as in ff.17 and 20.101
In any event, this reconstitution of the quires enables
The Quires reexamination of the physical features of the secondary
codex and the supposed heterogeneity of the lay-out.
From a physical point of view, this reconstitution clearly
Indeed, during the realization of a codex (writing and
demonstrates that the folios currently preserved in DAM
illuminating), it is commonly agreed that craftsmen did
(except ff.1 to 3) represent a continuous sequence of
not work on previously-bound manuscripts, but on sep-
fragmentary quires. Like the folios in the Eastern Library
arate quires.102 How much time elapsed between the
(including the auctioned folios), the DAM folios be-
writing of two quires? Did the copyist again carve his
longed to a series of nine consecutive quires in the sec-
nib? A number of parameters could alter the conditions
ondary codex.99 A final quire that would have included
of copying and generate wide disparities from one quire
the final section of the Qurʾā nic text (from Kor 60, 1 up
to the next. The quire thus constitutes the only genuine
to Kor 114, 6) is presently missing.
unit of work for the copyists and illuminators.
This reconstitution reveals at least three types of
In fact, reshaping the quires provides valuable insight
quires. The majority typically consist of five bifolios that
into the working logic of Codex S ̣anʿāʾ 1, and allows us
were folded to yield ten folios (quinion). Quires 6, 8,
to evaluate its homogeneity. As previously stated, the
97
density of copyist B’s writing varied from one page to
My thanks to Hythem Sidky for this information (pers.
the next (density refers here to the number of lines per
comm.). Twenty-four pages have been partially deciphered (see Ta-
ble 3) based on Ḥ amdū n’s dissertation. On most pages, only a few page). However, reconsidering this question by taking
lines—sometimes only a few words—are legible. The identified text the reconstructed quires into account reveals that varia-
is set in brackets in Table 3. tions of writing density reflect a planned and organized
98
When all of the folios in a quire were damaged, I performed writing process. On the one hand, the variation is sig-
several calculations based on the previous quires with respectively
nificantly reduced within the quires. Thus, Quire 2 has
greater or fewer lines per page, which permitted me to select the most
coherent results for reconstructing a regular quire.
99 100
My hypothesis is that the folios in the Eastern Library, including Except, naturally, for the central bifolio, in which flesh side
the five sold at auction and DAM 01-27.1, ff.1-3, belonged to seven faces flesh side.
101
consecutive quires in the secondary codex. A first quire—correspond- And perhaps the folio in the Louvre Abu Dhabi Museum.
102
ing to the beginning of the Qurʾā n—appears to be completely missing. See Turner, Typology (1977), 74.
20 ✦ Journal of Near Eastern Studies

Table 4—The Quires: partial reconstruction of the quires of the secondary codex (author drawings)

Quire 1 Quire 2 Quire 3


Stanford 07, David 83/2006,
al-maktaba al-šarqiyya (MS), ff.3–9 MS ff.10–18, Bonhams 2000 Christie’s 2008, Louvre AD, MS ff.19–22a

Quire 4 Quire 5 Quire 6


DAM 01-27.1, ff. 4–6 DAM 01-27.1, ff.7–12 DAM 01-27.1, ff.13–16

Quire 7b Quire 8 Quire 9


DAM 01-27.1, ff.17–20 DAM 01-27.1, ff. 21–24 DAM 01-27.1, ff.25–28

Quire 10c Quire 11d Quire 12


DAM 01-27.1, ff.30–32 DAM 01-27.1, ff.33–34 DAM 01-27.1, ff.35–38
a
The DAM 01-27.1, f.1 may have been part of either this quire or the next (with DAM 01-27.1, f.2). Traces of a potential lower script are rare,
however, and there are no traces of erasure. For these reasons, it is difficult to distinguish the hair side from the flesh side.
b
Because Quire 7 is irregular, it is difficult to firmly establish a link to the other quires. There are two possible explanations. The first hypothesis,
which is supported here, was to reconstruct a senion for Quire 7 (a quire with six bifolios), and a quaternion for Quire 8, as well as Quires 6
and 9. The other possible reconstruction was to identify Quire 7 and 8 respectively as a quinion and a quaternion, with one additional folio
with a stub in each one of them. My initial hunch was to identify f.17 and f.20 as a bifolio for two reasons: they share the same arrangement
in parchment, unlike the others, and the primary script is reversed in both relative to the secondary script. However, the calculation of the
missing folios does not support this assumption.
c
For this quire, it is assumed that the missing folios were written using twenty-three to twenty-five lines per page, as with the previous folio.
d
The calculation for the missing folios in Quires 11 and 12 is based on folios with twenty-eight to thirty-one lines per page.
The S ̣anʿā ʾ Palimpsest ✦ 21

Figure 16—Ḥ amdū n MS 15A. Scriptio superior: Kor 4, 92 – 100. (Image from R. Ḥ amdū n, Al-maḫ tụ̄ tạ̄ t al-qurʾā niyya fı̄ S ̣anʿā ʾ min al-qarn
al-awwal al-hiğrı̄, plate 75, p. 178).

between twenty-four and twenty-eight lines per page, has a high writing density (thirty-one to thirty-five lines
while Quire 3 has between twenty-six and thirty lines. for f.35), but it is no greater than previous quires,
On the other hand, there is no significant increase of den- particularly the last one from the Eastern Library (from
sity in the last pages of each quire, or apparently in the last approximately Kor 9, 112 to Kor 12, 49),104 which
quires of the codex. These observations are unfortunately
based on fragmentary evidence.103 We should note that
Quire 12, the penultimate one in the complete codex,
would need to take into account the ruling unit, the distance between
two lines of script.
103 104
All of the final folios of DAM 01-27.1 except f.35 are dam- According to my reconstruction, DAM 01-27.1, f.3 could be
aged and reduced to fragments. To overcome this difficulty, one the first folio of this quire.
22 ✦ Journal of Near Eastern Studies

Figure 17—Ḥ amdū n MS 15A. Scriptio inferior in red: [Kor 10, 97 – 11, 5]; H. Sidky tracing. (Image from R. Ḥ amdū n, Al-maḫ tụ̄ tạ̄ t
al-qurʾā niyya fı̄ S ̣anʿā ʾ min al-qarn al-awwal al-hiğrı̄, plate 75, p. 178).

exhibits the greatest writing density (thirty-one to thirty- its deterioration. Among possible scenarios, one could
seven lines). assume that the illuminator began by the last quires with
Similarly, the illuminator’s interventions, previously an overly ambitious project and reduced his work to
described as heterogeneous, turn out to be a more log- only the medallions. The disappearance of the colored
ical mode of operation in light of the structure of the headbands may be linked to the subsequent painting
quires. Indeed, the ornamental headers, added only of the medallions, which were initially only drawn with
on DAM 01-27.1, ff.35–38, actually belong to a single a lead pencil (Quires 10 to 12) and were later filled with
quire: the penultimate quire in the whole codex before colors in the other quires. In any event, and whatever
The S ̣anʿā ʾ Palimpsest ✦ 23

Figure 18—Arrangement of quires (author drawing).

the possible scenario, the structure of the quires con- the DAM folios are probably close to each other and even
tains critical cues to the process of writing and illustrat- consecutive, as is the case in the secondary codex. The
ing a manuscript on parchment. fragmentary condition of the quires—perhaps even at
the time of their re-use—offers evidence of only two
continuous sequences of text between the quires: the
first between Quires 4 and 8, and the other between
The Sequence of the Quires: From the Primary
Quires 6 and 10. This signifies that these two sequences
to the Secondary Codex
of quires were consecutive in the primary codex but
The tables reveal that several consecutive textual se- have been separated by three quires respectively in the
quences of the scriptio superior are also adjacent in the secondary codex (see Fig. 18). Following the same rule,
scriptio inferior. This would confirm Puin’s hypothesis Quires 5 and 9 may also have been consecutive. It is
that the quires of the primary codex remained more currently difficult to establish a clear connection, because
or less bound together before they were dismembered, the recto of the first folio of Quire 9 is illegible. It is im-
quire by quire. This ultimately explains why the leaves portant to explore these questions further by extending
from the same original quire were reused together within this investigation to the physical folios in the Eastern
the same quire in the secondary codex, sometimes in the Library.
same sequence (as ff.21–24; ff.35–38). The codicological analysis of the Codex S ̣anʿāʾ 1 thus
By contrast, the process did not necessarily respect the supports a codex structure for both secondary and pri-
sequence of the original quires. Indeed, Quires 1 and 2 mary folios. In addition to the existing bifolio (DAM
could be placed at the beginning of the primary codex, 01-27.1, ff.23-24) and traces of stitches, reconstituting
while Quire 3 (including Christie’s 2008 and Louvre the quires, most of which follow the same structure, re-
AD) could be the penultimate quire—or at least one of veals that both primary and secondary codices were re-
the last—in the primary codex. However, the quires of spectively a multigathering codex.
24 ✦ Journal of Near Eastern Studies

Table 5—From the secondary to the primary quire: The case of Quire 5 (author drawings)

1. The secondary quire 2. Reordering of the folios in the primary quire

The Quires of the Primary Codex fluctuations in the number of lines per page (between
and the Order of the sū ras twenty-two and thirty-five lines).107
According to this reasoning, it is possible to establish
Previous studies of the Codex S ̣anʿāʾ 1 have highlighted
connections between seemingly unconnected textual
distinctive features of the organization of the primary
sequences, provided that they are preserved within
text. While the sequence of the verses within a sū ra is
the same quire. For this reason, it can be assumed that
nearly identical to the standard ʿUṯ mānic sequence,105
sū ras 22, 24, and 33 were consecutive because they
the sequencing of the sū ras reveals a significant number
were bound within the same quire. Table 5 proposes
of discrepancies. Indeed, the existing fragments from
a reorganization of the likely position of the folios within
DAM reveal eleven transitions between two sū ras oc-
Quire 5 (ff.7–12) in the primary codex. This reorganiza-
curring in the same folio. Only four transitions follow
tion conforms to the rule of situating the hair side to the
the standard ʿUṯ mānic order (sū ras 8–9; 20–21; 39–
exterior (meaning that ff.9–11 necessarily belong to the
40 and 89–90106). Another transition is simply reversed
second part of the quire) and relies on the calculation of
(sū ras 63–62), while the six others follow a different or-
the missing folios.
der, sharing similarities with the codices of the Com-
Indeed, there are two missing sequences of text.
panions, according to traditional reports (see below).
The first missing sequence is between Kor 22, 39 (end
The present demonstration allows us to pursue the
of f.7B) and Kor 24, 1 (second text line of f.10A). The
reconstruction of the lower text and its arrangement
single missing folio exactly spans the verses of sū ra 22,
of the sū ras. Indeed, while the structure of the quires
from the verse 40 until the end. The second missing se-
was maintained from the primary to the secondary co-
quence is between Kor 24, 40 (end of f.11B) and Kor 33,
dex, this practice limits the extent to which the folios in
51 (beginning of f.9A). Only two folios are missing
each quire are out of order. As a consequence, because
from the reconstituted primary quire (including the il-
the sequence of the verses in the primary text is nearly
legible f.12). It is possible that they span Kor 24, 40
identical to the standard sequence, it supports calcula-
to verse 63 and Kor 33, 1 to verse 51, with some adjust-
tions regarding the missing folios, although this calcu-
ments.108 Sū ras 24 and 33 were therefore consecutive.
lation could slightly vary due to rasm variations and

107
My calculations are based on the textual reconstruction pro-
105
Except in two instances: one the omission of Kor 9, 85 that was posed by Sadeghi and Goudarzi, ibid., which is currently the most
probably due to copyist error, and the second a reversal of verses 31 complete edition of the S ̣anʿāʾ palimpsest.
108
and 32 in sū ra 20. See Sadeghi and Goudarzi, “S ̣anʿāʾ 1” (2012): 23. Calculations based on the Cairo edition regarding the missing
106
For conciseness, I refer to the sū ras numbers in the standard text indicate that approximatively three folios are missing, more than
edition. the two existing folios.
The S ̣anʿā ʾ Palimpsest ✦ 25

Table 6—The Order of sū ras in the S ̣anʿāʾ primary codex and the Companions codices. The sū ras’ numbering in boldface type are those
present in the manuscript.
Companion According to Ibn al Nadı̄m, According to As-Suyū tı̣̄ ,
Position In the primary codex codex Fihrista al-Itqā nb
Quire 1 2, 7 Ibn Masʿū d 2, 4, 3, 7 2, 4, 3, 7
Ubayy 2, 4, 3, 6, 7 2, 4, 3, 6, 7
Quire 2 5, 10, 11 Ibn Masʿū d 5, 10, 9, 16, 11 5, 10, 9, 16, 11
Ubayy 5, 10, 8, 9, 11 5, 10, 8, 9, 11
Quire 4 + Quire 8 11, 8, 9, 19 Ibn Masʿū d 9, 16, 11, 12, 17, 21, 23, 26, 77, 9, 16, 11, 12, 18, 17, 21, 20,
38, 28, 24, 8, 19 23, 26, 37, 33, 22, 28, 27,
24, 8, 19
Ubayy 8, 9, 11, 19, 26, 22 8, 9, 11, 19, 26, 22
Quire 11 12, 18 Ibn Masʿū d Sū ra 18 is missing in the list 12, 18
Ubayy 12, 18 12, 18
Quire 7 15, 25 Ibn Masʿū d Sū ra 15 is missing from the list 25, 15
15, 42, 30, 43, 41, 32, 14, 35,
Ubayy 48, 47, 57, 58, 25 Sū ra 25 is missing from the list
Quire 6 + Quire 10 16, [17], 20, 21, [23/26/ Masʿū d 16, 11, 12, 17, 21 (sū ra 20 is 16, 11, 12, 18, 17, 21, 20, 23,
27/29 or 31+32/32 missing), 23, 26, 37 26, 37
+35/ 32+36], 37 16, 33, 17, 39, 45, 20, 21, 24, 16, 33, 17, 39, 20, 21, 24, 23,
Ubayy 23, 40, 13, 28, 27, 37 34, 29, 40, 13, 28, 27, 37
Quire 5 22, 24, 33 Masʿū d 33, 28, 24, 8, 19, 29, 30, 36, 25,
22 33, 22, 28, 27, 24
22, 12, 18, 16, 33, 17, 39, 45, 22, 12, 15, 16, 33,17, 39, 20,
Ubayy 20, 21, 24 21, 24
Quire 9 30, [35/36/38], 39, 40 Ibn Masʿū d 30, 36, 25, 22, 13, 34, 35, 14, 30, 36, 25, 15, 13, 34, 35, 14,
47,31, 39, 40 38, 47, 31, 39, 40
39, 45, 20, 21, 24, 23, 40, nine 39, 41/43, 20, 21, 24, 23, 34,
Ubayy sū ras, 30 29, 40, eight sū ras, 30
Quire 12 34, 13 Ibn Masʿū d 13, 34 13, 34
Ubayy Sū ra 34 is missing from the list 34, 29, 40, 13
Quire 3 78, 76, 75, 81, 79, 64, [. . .], Ibn Masʿū d 64, 63, 62, 61, 72, 71, 58, 60, 64, 63, 62, 61, 72, 71, 58, 60,
80, 83, [. . .], 63, 62, 89, 66, 55, 53, 51, 52, 54, 69, 56, 66, 55, 53, 52, 51, 54, 56,
90 68, 79, 70, 74, 73, 83, 80, 79, 70, 74, 73, 83, 80, 76,
76, 75, 77, 78,81, 82, 88, 87, 77, 75, 78, 81, 82, 88, 87,
92, 89, 84, 85, 96, 90 92, 89, 85, 84, 96, 90
Ubayy 77, 78,76, 75, 81, 79,80, 83, 77, 78, 75, 81, 65, 79, 64,
84, 95, 96, 49, 63, 62, 65, 80, 83, 84, 95, 96, 49, 63,
89, 67, 92, 82, 91, 85, 86, 87, 62, 66, 89, 90, 92, 82
88, 64, 98 (Sū ra 90 is missing
in the list)
a
Nadı̄m, Fihrist (1971), 29.
b
Suyū tı̣̄, Itqā n (1997), vol.1: 181–82.

Quire 9, which hypothetically followed Quire 5 in the sixth quire (ff.13–16), the arrangement of the sū ras
the primary codex, includes verses of sū ra 30 (f.25B: between sū ras 16 and 20 in the primary codex did not
Kor 30, 38–50) and parts of sū ra 39 (f.26A: begins at contain sū ras 18 and 19, as their presence is observed in
Kor 39, 25), followed by sū ra 40. Approximately three or two other quires. The five lost folios between ff.15 and
four folios are missing between ff.25 and 26. Calculations 16 are therefore sufficient to contain the end of sū ra 16
indicate that the missing sections of sū ras 30 and 39 rep- and the entire sū ra 17.
resent approximately 2.8 pages. The best-fitting sū ra is Quire 10 should follow Quire 6 in the primary
the thirty-sixth. However, due to possible rasm variants, codex, as it contains the end of sū ra 20. Based on my
sū ras 35 or 38—with some adjustments—may also fit. calculations, the reversed f.32 can be situated in the
It is similarly possible to identify the potential inter- first half of the primary quire, following a first missing
mediate sequence between two sū ras within a quire. In folio, where the primary text ran from Kor 20, 80 to
26 ✦ Journal of Near Eastern Studies

approximately Kor 20, 112 (beginning of f.32).109 In lio—followed by MS 22, indicates a new sequence of
this quire, the missing sequence between Kor 21, 19 six sū ras: Kor 78 → 76 → 75 → 81 → 79 → 64.112 In
(f.32A) and Kor 37, 15 (f.30A) is the equivalent of four the Christie’s and Louvre AD folios, the scriptio inferior
folios, which certainly contained the end of sū ra 21 is reversed. Consequently, as they are halves of bifolios,
(two folios), another unidentified sū ra, and the begin- they should be replaced in the second half of the primary
ning of sū ra 37. The question remains: which sū ra was quire, after MS 22. The Christie’s folio displays the se-
contained in the two folios between the sū ras 21 and quence of the sū ras Kor 63 → 62 → 89 → 90, while
37? It is already possible to exclude seven of the fifteen the verso of the Louvre AD folio has the sequence Kor
intermediate sū ras of the traditional order (22, 24, 25, 80 → 83. Based on their text content, the Louvre AD
28, 30, 33, 34), which are already present in other se- folio was probably placed after MS 22 and prior to the
quences/quires. There are thus seven remaining possi- Christie’s folio. Unfortunately, the pictures of MS 22
bilities that match calculations regarding the missing verso and Louvre AD recto (which was the verso in re-
text: one sū ra (23, 26, 27, or 29) or two sū ras (31 and gard to the primary quire) are missing.
32, 32 and 35, or 32 and 36). Another manuscript from Before concluding, it is worth noting that the hypo-
S ̣anʿāʾ—DAM 01-32.1—may offer evidence in favor of thetical sū ras sequence of the S ̣anʿāʾ primary codex re-
the sequence 21 → 26 → 37, because it shows sū ra 26 flects a strong connection to the so-called “Compan-
followed by 37 on the same page.110 ions’ codices” described by medieval authors. Those
Last, there are three additional observations regarding codices were attributed to several members of the com-
new sequences of the sū ras, based on Hythem’s identifi- munity, such as Miqdād ibn al-Aswād, Ibn Masʿū d, and
cation of the folios from the Eastern Library (see Table 3). Ubayy ibn Kaʿb. With the expansion of the Muslim em-
For the present, these points should be considered with pire, some of these collections began to be considered
caution, given that the observations are primarily based on as authoritative to a certain extent in different centers,
the Ḥ amdū n photographs. The first occurs in Quire 1, before the establishment of a standard text by the caliph
Ḥ amdū n MS 7B. Here, the scriptio inferior displays the ʿUṯ mān. Little information has reached us regarding
end of sū ra 2 followed by the beginning of sū ra 7 on the two codices of Ibn Mas’ū d and Ubayy ibn Kaʿb.
the same page.111 The second appears on Quire 2. The Two principal sources—Ibn al-Nadı̄m (d. 380/995
partial decipherment of Ḥ amdū n MS 10B, 15A, and or 998)113 and al-Suyū tı̄ (d. 911/1505)114 —record
18A reveals two new transitions between the sū ras. The variations in the arrangement of the sū ras as a feature
first transition occurs between f.10 and f.18. At the end of these codices. They also preserve lists of the sū ras
of f.10B, it is possible to identify the end of sū ra 5 and, for both codices, with some discrepancies between
at the beginning of f.18A, the beginning of sū ra 10. Ac- the two authors. Ibn al-Nadı̄m also insists on the het-
cording to my reconstruction, these two folios consti- erogeneity of Ibn Masʿū d codices he saw (“no two of
tute a single bifolio that should be placed at the center the Qurʾā nic copies were in agreement”115). Despite
of the primary quire. Regarding f.15A, it displays the their differences from the standard ʿUṯ mānic text, how-
transition between sū ras 10 and 11 on the same page. ever, the arrangement of the sū ras in these codices fol-
We can therefore assume that the primary Quire 2 dis- lows the same principle of arrangement of the sū ras ac-
played the sequence 5 → 10 → 11. cording to decreasing length.116
The third is related to Quire 3, which now comprises
six folios, including four from the Eastern Library
(Ḥ amdū n MS 19A to 22B), the Louvre AD folio, and
112
This sequence partially records another sequence recently
the Christie’s 2008 folio. The decipherment of Ḥ amdū n
identified in another S ̣anʿāʾ manuscript, the DAM 01-29.1, where
MS 20 and 21—which certainly formed a central bifo- the sequence is Kor 80 → 75 → 81. See Cellard, “Nouveau té mo-
ignage” (2020).
113
Nadı̄m, Fihrist (1971).
109 114
However, my calculations indicate that this final folio retained Suyū tı̣̄, Itqā n (1997).
115
a shorter sequence of the Cairo edition (only thirty-eight lines). Dodge, Fihrist (1970), vol. 1, 57.
110 116
UNESCO photographs. In DAM 01-28.1, sū ra 37 is omitted Nö ldeke observed that this “cannot be accidental but must be
between sū ras 36 and 38, meaning that it should appear in another the result of textual relationships” between these Companions’ codi-
sequence, which has been lost. ces and the first collection of Zayd ibn Ṯ abit, referring to the leaves/
111
The basmala is visible between lines 11 and 12 of the upper suh ̣uf of Ḥ afsạ (Nö ldeke, History [2013], 246). Blachère stated that
text. the principle of decreasing length was perhaps more rigorous in Ibn
The S ̣anʿā ʾ Palimpsest ✦ 27

Compared to reports about these codices, the hypo- Conclusion


thetical sequence of sū ras in the S ̣anʿāʾ primary codex
Was the Codex S ̣anʿāʾ 1 a codex/musḥ ̣af that was dis-
shows connections with the Companions’ codices, al-
mantled and re-used to produce another codex? Codi-
though there is no perfect match with either report. In
cological observations of the eighty fragmentary folios
several instances, the similarity with the Companions’ co-
from the S ̣anʿāʾ palimpsest offer evidence of bifolios
dices is obvious: the sequence of sū ras 5 → 10, 12 → 18,
and quires for both successive texts. In addition, the
63 → 62, and 76 → 75 are identical to both Ibn Masʿū d
current state of conservation of both sets of folios from
and Ubayy codices. The potential sequence of sū ras 26 →
the Eastern Library and DAM reflects the secondary
37 is noted in Ibn Masʿū d’s codex, while the long se-
structure of the codex because the DAM folios be-
quence of sū ras 78 → 76 → 75 → 81 → 79 → 64 is al-
longed to nine consecutive quires. Further physical
most identical to Ubayy’s codex, such as the sequence
analysis of the folios in the Eastern Library is clearly
80 → 83 (see the sequences in Table 6). The palimpsest
needed to definitively address these questions. For
most often exhibits the same gatherings of sū ras as the
the present, close study of the DAM folios reveals a
Companions codices, but in a different sequence. Accord-
clear connection between these secondary quires and
ingly, the gathering of sū ras 11 → 8 → 9 → 19 in the pa-
the scriptio inferior, while also offering clues regarding
limpsest is reported for Ubayy’s codex, but in a different
the palimpsesting process itself. It appears highly prob-
order (8 → 9 → 11 → 19). The same observation goes
able that the primary codex—which had perhaps al-
for the gathering of sū ras 78 → 76 → 75 → 81 → 79 →
ready lost its bindings and become fragmented—was
64, which stands prior to the gathering 63 → 62 →
re-used quire by quire. Its status as a codex is now de-
89 → 90, such as in Ubayy’s codex. The gathering of
scribed and can thus be compared to other codices in
sū ras 15 → 25, 22 → 24, and 34 → 13 is closer to the
order to situate it in the history of the written transmis-
sequences of Ibn Mas’ū d’s codex (sū ras 25 → 15; sū ras
sion of the Qurʾā n.
33 → 22 → two sū ras → 24; sū ras 13 → 34).
Material analysis has proven significant because it
Despite these connections to the two Companions’ co-
provides a snapshot of the initial appearance and layout
dices, the order of sū ras of the Codex S ̣anʿāʾ 1 shows a ten-
of the fragmentary folios. Several of the characteristics
dency towards the ʿUṯ mānic order. Several sequences are
of the primary codex point to a professional and planned
indeed congruent with both the ʿUṯ mānic order and one
work. The collaboration of two experienced scribes,
of the two Companions codices, as sū ras 8 → 9, 20 → 21
probably copying from a written exemplar, could reflect
and 89 → 90 (like Ubayy’s codex) and 39 → 40 (like Ibn
the context of a local scriptorium active during the early
Masʿū d’s codex). Certain other sequences, such as sū ras
Umayyad period. This analysis also opens new perspec-
16 → 17 → 20 or 30 → [35/36/38] → 39, are far closer
tives on the study of the text and the sequencing of sū -
to the ʿUṯ mānic order than to the Companion codices.
ras, underscoring the existence of alternative sequences
Moreover, the sequence 10 → 11 is specific to the ʿUṯ -
of sū ras and confirming that several sequences did not
mānic order. It is thus possible to consider this order
follow the standard ʿUṯ mānic order. These findings sug-
of sū ras as falling between that of the two Companions
gest the need to reconsider the rules behind the organi-
codices—as the narratives reported them—and the ʿUṯ -
zation of the sū ras.118 The original sequence in the Co-
mānic order. Finally, the Codex S ̣anʿāʾ 1 agrees with
dex S ̣anʿāʾ 1 resonates with that of the two Companion
these codices (i.e., those of both ʿUṯ mān and the Com-
codices, although without being perfectly consistent
panions) concerning the arrangement of the sū ras by
with either of them. This is an insufficient basis for ex-
decreasing length, which obviously implies an original
cluding affiliation with these textual traditions, however,
complete text.117
because we lack an exact idea of the order of the sū ras
or text variants in the original Companion codices. Un-
like the ʿUṯ mānic text, these written recensions, which
may still have existed in the 2nd and 3rd centuries AH,
Masʿū d’s codex than in Ubayy’s, which was itself stricter than in the
Ḥ afsạ leaves (Blachère, Introduction [2002], 46). Regarding the never experienced official standardization. They remain
chronological arrangement of the sū ras that would have character-
ized a supposed ʿAlı̄’s codex, Nö ldeke affirmed that this cannot “be 118
Several studies of the Qurʾā n conclude that the canonical sū ra
expected of a contemporary of the Prophet” (Nö ldeke, History sequencing—at least certain clusters of sū ras—is inherent in the com-
[2013], 250). position of the text, because sū ras beginning with similar “mysterious
117
This could explain why sū ra 7 follows sū ra 2. letters” are grouped together. See S ̣aleh, “End of Hope” (2016).
28 ✦ Journal of Near Eastern Studies

plural and heterogenous, as they were depicted in later ——— “Un nouveau té moignage sur la fixation du canon
centuries.119 Ultimately, different iterations of these co- coranique dans les dé buts de l’Islam: le manuscrit S ̣anʿāʾ
dices may have existed throughout the centuries, possi- DAM 01-29.1,” Comptes-Rendus de l’Academ ́ ie des Inscrip-
bly with locally-specific variations. They may also have tions et Belles-Lettres 2018/2 (Paris, 2020): 1107–1119.
Cellard, É lé onore and Catherine Louis, “ A New Palimp-
interacted with the ʿUṯ mānic tradition to give rise to
sest and the Early Transmission History of the Qurʾan,”
eclectic codices.120
Journal of the International Qurʾanic Studies Association
The only way to increase our certainty about the tex- 5 (2021) forthcoming.
tual affiliation and status of the Codex S ̣anʿāʾ 1 will be to Christie’s, Art of the Islamic and Indian Worlds (London,
extend this study to other early manuscripts, including April, 2008).
but not limited to the S ̣anʿāʾ collection,121 in which at Cohen, Zina, et al., “Composition of the primary inks in
least three other fragmentary codices indicate the exis- medieval palimpsests—effects of ink removal,” Opuscula
tence of different sū ra sequences.122 An initial survey musealia 23 (2015): 75–82
of these codices reveals a stronger connection to the Costa, Paolo, “La Moschea Grande di S ̣anʿāʾ,” Annali
ʿUṯ mānic text, at least in the existing folios. The central del’Instituto Orientale di Napoli 34 (1974): 487–506.
question for future studies will be the relationship be- Declercq, Georges, “Introduction,” in Early Medieval Pal-
tween these manuscripts and the S ̣anʿāʾ palimpsest. Is impsests, ed. G. Declercq (Turnhout, 2007), 7–22.
Déroche, François, Catalogue des manuscrits arabes. Deuxiem ̀ e
it possible that they belong to the same scribal tradition?
partie: Manuscrits musulmans, Tome I, 1, (Paris, 1983).
And if so, what does this living tradition tell us about the
——— La transmission manuscrite du Coran dans les deb́ uts
status and evolution of the ʿUṯ mānic text? de l’islam: Le Codex parisino-petropolitanus (Leiden, 2009).
——— Qurʾans of the Umayyads (Leiden, 2014).
——— Le Coran, une histoire plurielle: Essai sur la formation
Works Cited
du texte coranique (Paris, 2019).
al-Aqwaʿ, Ibrā hı̄m, “Ğāmi’ S ̣anʿāʾ, abrāz maʿālim al-h ̣ad ̣āra Dodge, Bayard, ed. and trans., The Fihrist of al-Nadı̄m: A
al-islamiyya fı̄ l-yaman,” in Masạ̄ h ̣if S ̣anʿā ʾ, 19 March – Tenth-Century Survey of Muslim Culture, 2 vols., (New
19 May 1985 (Kuwait, 1985), 9–23. York, 1970).
Bergmann, Uwe, et al., “Chemical mapping of paleontolog- Dreibholz, Ursula, “Preserving a Treasure — The Sanaʿa
ical and archeological artifacts with synchroton X-rays,” in Manuscripts,” in Museum International, UNESCO 203:
Annual Review of Analytical Chemistry (2012): 361–89. 51/3 (Paris, 1999), 21–25.
Blachère, Ré gis, Introduction au Coran, 2nd ed. (Paris, 2002). ——— “Treatment of early Islamic manuscript fragments
Blair, Sheila, Islamic Calligraphy, 2nd ed. (Edinburgh, 2008). on parchment. A case history: the find of Sanaʿa, Yemen,”
Bonhams, Islamic And Indian Works Of Art (London, Oc- in Proceedings of the 3rd conference of al-Furqan Islamic Her-
tober 2000). itage Foundation, 2nd ed., (London, 1423/2003), 131–
Cellard, É léonore, Transmission (2015): La Transmission 45.
manuscrite du Coran. Etude d’un corpus de manuscrits data- Easton, Roger and Keith T. Knox, “Digital Restoration
bles du 2e s. H./8e s. ap. J.-C. (Ph.D. dissertation, INALCO, of Erased and Damaged Manuscripts,” in Proceedings of
Paris, 2015). the 39th Annual Convention of the Association of Jewish Li-
——— Le Codex Amrensis 1 (Leiden, 2018). braries, ed. E. Gensler and J. Biella, (New York, 2004), 1–
——— “Les manuscrits coraniques anciens,” in Le Coran des 8.
historiens, ed. Ali Amir-Moezzi and Guillaume Dye (Paris, Fedeli, Alba, “Early Evidences of Variant Readings in Qurʾanic
2019), 667–69. Manuscripts,” in Die dunklen Anfä nge, neue Forshungen
zur Entstehung und frühen Geschichte des Islam, ed. Karl-H. Ohlig
119
Ibn al-Nadı̄m emphasized that he never found two similar and Gerd-R. Puin (Berlin, 2005), 293–316.
copies of Ibn Masʿū d’s recension. ——— “I manoscritti di Sanaa: fogli sparsi che diventano
120
We have little information about the “secondary codices,” at- corani,” in Il mio cuore e ̀ a Oriente: Studi di linguistica storica,
tributed to the tā biʿı̄na. For some of them, such as the codex of Saʿı̄d
filologia e cultura ebraica dedicati a Maria Luisa Mayer
ibn al-Ğubayr (d. 94/692), the data highlight the mixed nature of
Modena, ed. F. Aspesi, et al. (Milan, 2008), 25–48.
the text. See Jeffery, Materials, 245. See also Cellard, “Nouveau
Fendall, Ramsay, Islamic Calligraphy (London, 2003).
té moignage” (2020): 1117.
121
See Karimi-Nia, “A New Document” (2019). Fraser, Marcus and William Kwiatkowski, Ink and Gold:
122
A. Fedeli reports some instances of deviation in the order of Islamic Calligraphy (London, 2006).
sū ras in DAM 01-32.1 and numerous deviations in DAM 01-28.1. Jeffery, Arthur, ed., ʿAbd Allā h ibn Sulaymā n Sijistā nı̄, Ma-
See Fedeli, “Manoscritti” (2008). Another example of deviation is terials for the history of the text of the Qurʾan: the old codices
in DAM 01-29.1, see Cellard, “Nouveau té moignage” (2020). (Leiden, 1937).
The S ̣anʿā ʾ Palimpsest ✦ 29

Ḥ amdū n, Razan Ghassan, Al-maḫ tụ̄ tạ̄ t al-qurʾā niyya fı̄ ——— “Koranpalimpsest V” (2014): “Ein frü her Koran-
S ̣anʿā ʾ min al-qarn al-awwal al-hiğrı̄ (M.A. thesis, Al- palimpsest aus Sanʿaʾ (DAM 01-27.1) – Teil V: Die scriptio
Yemenia University, 2004). inferior auf den Blättern 14 und 15 sowie Auseinander-
Hilali, Asma, “Le palimpseste de S ̣anʿāʾ et la canonisation du setzung mit den Thesen und der Edition des Koranpalimp-
Coran: nouveaux é lé ments,” in Les Cahiers Gustave Glotz sests von Behnam Sadeghi und Mohsen Goudarzi,” in Die
21 (2011): 443–48. Entstehung einer Weltreligion II: Die heilige Stadt Mekka -
——— “Was the S ̣anʿāʾ Qur’ān Palimpsest a Work in Pro- eine literarische Fiktion, ed. Markus Groß and Karl.-H.
gress?” in The Yemeni Manuscript Tradition, ed. D. Hollen- Ohlig, (Berlin, 2014), 477–618.
berg, Christoph. Rauch, and Sabine Schmidtke (Leiden, Puin, Gerd-Rü diger, “Methods of Research on Qurʾanic
2015), 12–27. Manuscripts — A Few Ideas,” in Masạ̄ h ̣if Sanʿā ̣ ʾ, 19 March –
——— The Sanaa palimpsest (London, 2017). 19 May 1985 (Kuwait, 1985), 9–17.
——— “The Writing Process in a 7th century Qurʾān Manu- ——— “Observations on Early Qurʾān Manuscripts in S ̣anʿāʾ,”
script: The Upper text of the ‘Sanaa Palimpsest’ as exam- in The Qurʾan as Text, ed. S. Wild (Leiden, 1996), 107–
ple,” in Palaeography Between East and West. Proceedings 111.
of the Seminars on Arabic Palaeography at Sapienza Uni- Sadan, Joseph, “Genizah and Genizah-like Practices in Is-
versity of Rome, ed. A. D’Ottone Rambach (Rome, 2018), lamic and Jewish Traditions: Customs Concerning the Dis-
141–53. posal of Worn-out Sacred Books in the Middle Ages, Ac-
Irigoin, Jean, “Le palimpseste de Sophocle,” Revue des cording to an Ottoman Source,” Bibliotheca Orientalis
́
Etudes Grecques, 64/302–303 (1951): 443–55. 43 (1986): 37–58.
Karimi-Nia, Morteza, “A New Document in the Early His- Sadeghi, Behnam and Mohsen Goudarzi, “S ̣anʿāʾ 1 and
tory of the Qurʾān,” Journal of Islamic Manuscripts 10 the Origins of the Qur’ān,” Der Islam 87/1–2 (2012):
(2019): 292–326. 1–129.
Masạ̄ hif
̣ : Masạ̄ h ̣if S ̣anʿā ʾ, 19 March – 19 May 1985 (Kuwait : Sadeghi, Behnam and Uwe Bergmann, “The Codex of a
Dār al-Athār al-Islamiyya, 1985). Companion of the Prophet and the Qurʾan of the Prophet,”
Muzerelle, Denis, Vocabulaire codicologique: Reṕ ertoire Arabica 57 (2010): 343–436.
met́ hodique des termes français relatifs aux manuscrits (Paris, Saleh
̣ ̣, Walid A., “End of Hope: Sū ras 10–15, Despair and
1985). a Way Out of Mecca” in Qurʾanic Studies Today, ed. An-
al-Nadı̄m, Ibn, Kitā b al-Fihrist, ed. R. Tajaddud (Teheran, gelika Neuwirth and Michael A. Sells (Abington 2016),
1971). 105–123.
Nö ldeke, Theodor, et al., The History of the Qurʾā n, trans. Al-Suyū tı̣̄ , Al-Suyū tı̣̄ , al-Itqā n fi ʿulū m al-Qurʾā n, (Beirut:
Wolfgang H. Behn (Leiden, 2013). al-maktaba al-’asriyya,
̣ 1418/1997).
Noseda, Sergio Noja, “La Mia Visita a Sanaa e il Corano Schmidt, Thomas, “Les palimpsestes litté raires grecs sur pa-
Palinsesto,” Rendiconti dell’Instituto Lombardo Academia pyrus,” in Palimpsestes et ed́ ition de textes. Les textes
di scienze e lettere. Rendiconti, Classe di lettere 137/1 (Milan, litteŕ aires, ed. Vé ronique Somers (Leuven, 2009), 83–
2003): 43–60. 100.
Puin, Elisabeth, “Ein frü her Koranpalimpsest aus Sanʿaʾ Sokolov, Michael and Joseph Yahalom, “Christian palimp-
(DAM 01-27.1),” in Schlaglichter: Die beiden ersten sests from the Cairo Geniza,” Revue d’histoire des textes
islamischen Jahrunderte, ed. Markus Groß and Karl-H. 8 (1979): 109–132.
Ohlig (Berlin, 2008), 461–93. Sotheby’s, Oriental Manuscripts and Miniatures, 23 Octo-
——— “Koranpalimpsest II” (2009): “Ein frü her Koranpa- ber 1992.
limpsest aus Sanʿaʾ (DAM 01-27.1) – Teil II,” in Vom Koran Tchernetska, Nathalia, “Cambridge University Library
zum Islam, ed. Markus Groß and Karl.-H. Ohlig, (Berlin, Nn.4.8, a Greek Palimpsest described and deciphered,”
2009), 521–81. Transactions of the Cambridge Bibliographical Society
——— “Koranpalimpsest III” (2010): “Ein frü her Koranpa- 12/3 (2002): 313–21.
limpsest aus Sanʿaʾ (DAM 01-27.1) – Teil III: Ein nicht- Rabin, Ira, “Building a Bridge from the Dead Sea Scrolls to
ʿutmanischer Koran,” in Die Entstehung einer Weltreligion Medieval Hebrew Manuscripts,” in Jewish Manuscript Cul-
I: Von der koranischen Bewegung zum Frü hislam, ed. Markus tures, ed I. Wandrey (Berlin, 2017), 309–322.
Groß and Karl.-H. Ohlig, (Berlin, 2010), 233–305. Reed, Ronald, Ancient Skins, Parchments and Leathers
——— “Koranpalimpsest IV” (2011): “Ein frü her Koran- (London, 1972).
palimpsest aus Sanʿaʾ (DAM 01-27.1) – Teil IV: Die scriptio Robin, Christian J., “L’Arabie dans le Coran” in Les origines
inferior auf den Blättern 17, 18 und 19,” in Die Entstehung du Coran, le Coran des origines, ed. F. Dé roche, Ch.
einer Weltreligion II: Von der koranischen Bewegung zum J. Robin, and M. Zink (Paris, 2015), 27–74.
Frü hislam, ed. Markus Groß and Karl.-H. Ohlig, (Berlin, Turner, Eric G., The Typology of the Early Codex (Eugene,
2011), 311–402. OR, 1977)
30 ✦ Journal of Near Eastern Studies

von Bothmer, Hans-Caspar, “Masterworks of Islamic bia Felix, ed. Werner Daum (Innsbruck, 1987), 178–
Book Art: Koranic Calligraphy and Illumination in 93.
the Manuscripts found in the Great Mosque in Sanaa,” Witkam, Jan Just, review of Masạ̄ h ̣if S ̣anʿā ʾ, 19 March – 19 May
in Yemen: 3000 Years of Art and Civilisation in Ara- 1985, Manuscripts of the Middle East 1 (1986): 123–24.

You might also like