The Paper Trade on the Red Sea (19th c.-first half of the 20th c.): ‘local’ vs. Italian paper

Share this:

The case of an ‘Ottoman’ watermarked paper used in Yemen and in Ethiopia¹

Anne Regourd
Austrian Academy of Sciences, Institute for Social Anthropology, Vienna CNRS, UMR 7192)

Article contents: Introduction Overview of paper-making in 19th & 20th c. Turkey Methodology, codicology & history An Ottoman paper I. Beyāḍ Abū Ṧubbāk papers II. The Waraq Abū Ṧibbāk/Ṧubbāk papers Posterity & longevity of Abū Ṧubbāk paper From the Circulation of Paper to a Few Thoughts on Trade: An Abū Ṧubbāk / Abū Ṧibbāk / Abū Ṧubbāk ‘Zone’? Production and trade: Towards an Ottoman Paper vs. Italian Paper Competition Conclusion

Introduction

There is no doubt for the codicologist as to the supremacy of Italian paper in the 19^th c., particularly from Veneto (Friuli), as the material of choice for manuscripts in the Arab or African world. Works by Terence Waltz have demonstrated this for Egypt and Sudan². In Yemen and in Ethiopia, the study of manuscript paper confirms the strong presence of Italian papers³. In Sub-Saharan Africa, there is evidence of much the same for Nigeria or thanks to manuscript collections gathered in West Africa⁴.

All along the 19^th c., as the Industrial Revolution was in full swing, production techniques evolved and the movement began to affect paper production, starting from the 1820s, be it for wallpaper or writing paper⁵. The balance of power also changed in the region as the achievements of engineers turned the Red Sea into the preferred means of communication between the Mediterranean Sea and the Indian Ocean: the British presence, especially in Aden, from 1839 to 1967, has to be taken into account, as well as the consecutive appearance of the steam boat, requiring its own logistics, and the opening of the Suez Canal, in 1869; the second Ottoman occupation of Yemen takes place from 1872 to 1919 and meets some resistance in the Tihāmah, the part of  Yemen which runs along the Red Sea⁶; finally, the Sykes-Picot Treaty led to the dividing up of the Ottoman Empire between the French and the British, in 1916.  In the Red Sea zone, which will be the focus of this paper, Italian supremacy however, did not prevent on the one hand, the import of other European papers, from France, Britain, or Austria, nor on the other hand, those papers referred to as ‘local’ (as opposed to the former). Among these, and during the same period, are papers produced in Egypt, India, whether watermarked or not⁷, or in Turkey. By way of example, “Ledger” paper is found in the manuscripts of Zabīd in the Tihāmah (m/ḥ 29, copied in 1369/1949-1950), as well as in Ta‘izz at the library of the Hā’il Sa‘īd Foundation, a robust paper, much valued for accounting books. The case is different for India: because its commercial and community ties with Yemen date back centuries, the presence of Indian paper is linked to the capacity to locally produce paper or not. Several Indian papers have been recorded in Ṣan‘ā’ and in Zabīd, e.g. a paper refers to Bombay in its countermark (Mumbay, m/ḥ 38, copied in 1334/1915-1916). In the second half of the 20^th c., an Egyptian paper factory is said to have exported its paper to India, the Hijaz, Yemen, Northern Africa and even Europe⁸.

As for Turkish paper, over the last few years, it has gained a lot of attention from researchers. Generally speaking, the presence, in Arab manuscripts, of watermarked paper made in Ottoman Turkey according to an Italian technique, from the early 19^th c. to the mid-20^th c., has already been verified⁹. Ş. Tekin reproduces several of them, at least one of which is dated 1219/1804-1805. This one is an ensemble made up of a crescent enclosing an eight-pointed star and of a countermark in osmanlı, Islāmbôl sene 1219¹⁰. More recently, Abū Ṧubbāk paper has been found in Yemen¹¹. However, our knowledge of such papers is far from complete.

Overview of paper-making in 19^th & 20^th c. Turkey

To date, few studies relying on primary or secondary sources have been devoted to the paper-making industry in Turkey. Based on the few well-known publications, it appears production was irregular, for various reasons, not necessarily linked to the international context or to economic competition¹². For the 19^th and the beginning of the 20^th c., and the types of paper which concern us,¹³ O. Ersoy and M. A. Kāğıtçı mention the following paper mills and factories:

>> the Hünkār Iskelesi mill, located in Beykoz, on the eastern shore of the Bosphorus in 1803, and fully operational at least until 1832. As the watermarked paper bearing the crescent and star with a countermark in osmanlı, has just been mentioned, we will not go into it anymore¹⁴;

>> the steam-powered paper mill of Halkapınar in Izmir (Smyrna), established in 1846, equipped with machinery for continuous paper production from the British Bryan Donkin Company¹⁵. Its raw material was rags. However, it does not seem to have survived very long and appears to have succumbed to importers dumping paper on the Ottoman market;

>> the Hamidiye (Hamidiye Kāğıt Fabrikası) paper mill in Beykoz, named after the river running through it. It was born at the very end of the 19^th c. officially 14 September 1886. Permission to print was granted (imtiyaz) for fifty years to Osman Bey, Sultan Abdülħamid II's chamberlain. The paper mill was considerable in size and powerfully equipped with 4 English paper machines by Masson Scott & Co, a large workshop for cutting and finishing paper, as well as a thermal power plant. Straw and rags were its raw materials.¹⁶ It ceased to function around 1915 when the Ottoman Empire sided with the Germans, causing the British engineers to depart;

>> finally, Anatolia, bastion of the modern papermaking industry, took over the manufacture of cellulose pulp and paper. The five-year plan of 1933-1938, called for the creation of a great paper mill¹⁷. An initial paper mill opened in Izmit in 1936¹⁸; others would appear in Aksu, Dalaman, Afyon, Balikesir, Kastamonu.

This preliminary overview, although extremely interesting, nonetheless remains incomplete, as the various Turkish local archives have not been the subject of extensive exploration yet. Authors who have addressed the history of Turkish papermaking figured competition pitting Turkish papermaking against European papermaking had led to the imitation or “borrowing” of Europeans watermarks (especially Italian and Polish) by the Turks, on, but regretted their inability to demonstrate this¹⁹.

Should someone now want to look into the trading of Turkish-made paper, and more specifically, its export, and then piece together the trade routes by which it was shipped, the problem will remain the same, as long as we do not have, as in this case, more data from sources such as the archives stored at the Baçbakanlık, the archives of paper-makers or of locals²⁰. To which must be added the country's archives under Ottoman rule and those of Italian paper-makers, which should be studied from that angle. As for the different memoirs in the wake of Napoleon's expedition, traveller's tales, studies, etc., regarding trading on the Red Sea, their sorting has, for the moment, yielded only haphazard information, often hard to interpret or comment on²¹.

It is however possible to proceed otherwise, whilst we wait for these future studies to be carried out, but also in order to increase and crosscheck our data. Indeed, a trace of imported papers is kept, ultimately, in the manuscripts of those countries importing the paper. Let us refer to these as ‘destination’ papers. Hard data can thus be obtained, informing about the production of a given paper, its export and arrival to a particular destination at one point in time. Dating the production of the paper based on such data should nevertheless take into account the fact that date brackets are based on its use by copyists. However, a paper can be stored for an extended period before being used²². The time taken by its transport should also be added on. A significant correction in dating the production of a paper based on this material can be obtained by adopting a statistical basis.

Methodology, codicology & history

As early as 1985, in his previously mentioned article, Terence Waltz opened the way to working on the paper trade based on watermarked papers, in his case, those he had documented predominantly in archive documents. This was initially a study meant for a collective work on Sudan. This remarkable work did not immediately attract much of a following. In “Les routes commerciales entre Zabīd et l'Europe: les papiers filigranés de fonds manuscrits de Zabīd (Yémen, fin 18e-milieu 20e s.)” [“Trade Routes between Zabīd & Europe: Watermarked Papers from Zabīd's Manuscript Collections (Yemen late-18^th-mid-20^th c.)”], we made an initial attempt at establishing this inaugural work as a method²³.

The method thus defined implies a considerable epistemological leap: papers from manuscripts are to be treated as a primary source. The status of papers makes them no longer just a means of dating and locating manuscript production in which they were used, but an actual datum in the history of the trade, of the migration of ideas and people. This posits that studying the papers allows us to identify the paper mill of origin and therefore the place where it was made, an operation which supposes the existence of means to document this. It requires a careful selection of the manuscript by making sure that, as initial criteria, the copy date is correct and that we are able to locate where the manuscript was copied. Only then can the paper be classified as "destination paper". Finally, it demands creating statistical clouds on the use of a given paper.

The task of identifying the papers, the time between the production of a paper and its use, is already part of the work carried out by a codicologist. What changes here is the end-goal: it is no longer solely about dating and locating the paper, but about studying the paper trade, i.e. identifying exported papers and their destinations, their prices, their processing, the networks and buyers, tradesmen, competition, how they were shipped and along which supply routes. In this search, manuscript papers are a seldom used and promising source, to which findings from other sources must be added.

If we are to follow this path, where should we turn to in order to find exported papers of Turkish manufacture? Inevitably, the Ottoman administration, well-known for being a "paper glutton", springs to mind. An initial set of data was provided by Nikolaev, Velkov and Andreev, who published four volumes of watermarked papers from Ottoman archives kept at the National Library in Sofia²⁴. The presence of papers of Ottoman origin, concurrently with papers of other origins, is confirmed for the period between the 16^th c. and the 18^th c.²⁵. Unfortunately for our study, the two authors' selection does not go beyond archive documents dated 1820²⁶.

At the same time, research on watermarked papers used in codices has made progress and identified a paper referred to as “Ottoman”, found in Yemen. It is this paper, present from the early 19^th c. to the mid-20^th c. that we will focus on.

Finally, generally speaking, working on the dating and provenance of modern and contemporary Arab codices becomes a necessity, as their predominance has been observed: on the one hand, they are the ones which have been passed down on to us, but for certain regions, such as Africa, manuscript production (be it codices or merchant accounting books) witnessed a boom in the 19^th c., as a result of the increased presence of paper²⁷. In Yemen, based on the samples we got to see in Zabīd, Ta‘izz and Shibām (Hadramawt), the bulk of manuscript production preserved today dates back to between the 17^th c. and the 20^th c.

An Ottoman paper

In 1997, Geneviève Humbert pointed out for the first time, the presence of an Ottoman paper in Yemen. Two of the fourteen manuscripts she had studied, at the Ašā‘ir library in Zabīd, were written on a watermarked paper with a lunar profile, bearing a countermark, in osmanlı writing: “Beyaz Abū (sic) Shabbāk Stambūlī” (“paper by Abū Ṧabbāk of Istanbul”)²⁸.

Since then, in the same town, examples of watermarked papers, sporting a crescent moon face inserted in a double-rimmed shield and bearing the countermark:

“[1st line] Beyāḍ Abū (sic) Ṧubbāk Isṭambūlī (بياض ابوُ شبّك اسطمبولي);

[2^nd line] ‘ālī aṣīlī” (عالي اصيلي)²⁹ have been recorded.

Zabīd's wealth of ulama libraries can be explained by the important part it played in religious transmission, from the 11^th c. to the 20^th c³⁰. In 2006, we pointed out four examples of it³¹:

1. The first one can be found in manuscript m/ḥ 1 in ‘Abd al-Rahman al-Hadhrami's library, colophon dated Dhū al-Qa‘ada 1375/1956 by copyist Muḥammad Ismā‘īl Khalīl.

2. The second is a single folio from codex 11 of collection (majmū‘) m/ḥ 7, 25 x 17.5 cm in size, belonging to the same library and written mostly on an Italian paper (Galvani); in the margin a comment dated 1303/1885-1886 can be found.

3. The third example appears on a single folio a chiromancy text was recorded on, belonging to Aḥmad Ja‘far's library, Zabīd. Its size making it possible to see, at the same time, the watermark and its countermark³². The presence of the countermark also makes it possible to specify the orientation of the crescent, facing right.

4. Finally, we indicated that in 2006, the most ancient copy on Isṭambūlī paper we had found in Zabīd dated back to 1214/1799-1800.

In 2008, we published a fifth example of this paper, used as a flyleaf for codex m/ḥ 17 in ‘Abd al-Raḥman al-Hadhrami's library, colophon dated Ḏū al-ḥijja 1192/1777-1778³³. As the cataloguing of manuscripts from this library progressed, occurrences of it kept on multiplying: codex m/ḥ 35 (copy dated 1358/1939-1940), 47 (undated copy), 12 codices out of the 13 in collection (majmū‘) 50 (copies dated 1330/1911-1912 to 1361/1942-1943), codex 60 (copy dated 1263/1846-1847)³⁴, 66/1 (copy dated 1328/1910-1911)³⁵. Today, the presence of “Abū Ṧubbāk” paper is confirmed in manuscripts from several regions of Yemen. In 2007, a book manuscript consisting of 84 folios, written on “Abū Ṧubbāk” paper dated 1316/1899, was documented in Ṣan‘ā’, in the private library of ‘Abd al-Khāliq b. Ḥusayn al-Maġribī, the descendant of a well-known Yemeni family of scholars³⁶. About a hundred papers have been sketched free-hand at Dār al-makhṭūṭāt, Ṣan‘ā’, between 2009 and 2010, among which the paper which interests us. This public library acquires its manuscripts mostly through donations and purchases. They may thus originate from different parts of Yemen, but, to be sure, the capital's surrounding area make for a regular source. In October 2010, twelve of the 97 manuscripts examined by me, out of a total 334 in the possession of the Hā’il Sa‘id Foundation, in Ta‘izz, contained Abū Ṧubbāk paper; they bear the following shelf marks: ms. 27, 51, 57, 70, 71 à 75, 97 (Qur’ān), 106 and 107; only two of the texts show a colophon date, these are ms. 27, copied in 1073/1662-1663, and ms. 107, copied in 1328/1910-1911³⁷. In the first case, the presence of this paper at that early period can be explained by the fact that it was used as flyleaf. This initial data already points to a trend which growing interest in the study of Yemeni papers should reinforce, showing an increasing number of examples in the future and demonstrating, eventually, that it was in fact a very widespread product.

Aside from these statistical indications, already quite eloquent, the name of the copyists or the copy place of at least 4, probably 5, manuscript codices written on Abū Ṧubbāk and belonging to ‘Abd al-Rahman al-Hadhrami's library, show that this paper was indeed imported into Yemen and reached Zabīd³⁸.

Outside of Yemen, a study based on scriptural sources, mentions as early as the late 60's, the existence of an "abu shibbāk" paper among European imports to eastern Sudan, at the end of the 19^th c.³⁹. Very recently, Abū Ṧubbāk paper was mentioned in the Far East: firstly, in January 2010, in a fiqh manuscript from the Regency of Aceh Besar, North of Sumatra, acephalous and with the last part of the text missing; then, between January and October 2010, in a codex (several codices?) kept in Indonesia. These are manuscripts in Arabic which, according to today's research, are not prior to the 19^th c.⁴⁰. Such manuscripts possibly having been imported, particularly from Yemen and more specifically from Hadramawt, it is too soon to assume this paper was circulated in the specified zone⁴¹. Abū Ṧubbāk paper can still be found in codices in Arabic kept in Ethiopia⁴². Ms. BNF 7264 contains two codices, probably copied in Ethiopia, in 1345/1926-1927⁴³. Indications at hand regarding the provenance of these collections, to which palaeographical data is added, allow us to posit, at least for some of the codices concerned, that they were copied in Ethiopia, more often than not in Harar.

In the geographical area we have just covered, Abū Ṧubbāk paper is present over a period from the very beginning of the 19th c. through to the mid-20th c. The brand "Abū Ṧubbāk" has lived on in Yemeni memories to this day, as a synonym for excellent paper, sturdy and long-lasting, recommended as such when drawing up legal documents. This is what the term beyaz (instead of kāğıt), which in osmanlı is used to speak of paper used for the final versions of documents as opposed to draft paper⁴⁴, would lead us to believe. Its sturdiness also explains its frequent use as a flyleaf for manuscripts⁴⁵. In the 2000's, we got to buy, at the souk in Zabīd, “an Abū Ṧubbāk”, i.e. a complete sheet of non-watermarked paper⁴⁶.

First attempt at classifying Abū Ṧubbāk papers: Beyāḍ Abū Ṧubbāk and Waraq Abū Ṧilubbāk papers

As demonstrated, Abū Ṧubbāk paper was widespread, from the 19^th c. to the 20^th c., at least in Yemen and in Ethiopia. Fortunately, it is much easier to classify than the Tre Lune, a watermarked paper very frequently found in codices as well as in archives⁴⁷.

To this day, two main groups of Abū Ṧubbāk papers have been detected, the Beyāḍ Abū Ṧubbāk papers, and the Waraq Abū Ṧilubbāk papers. This is only an initial attempt: neither this list, nor, more specifically, the use periods worked out for the paper, can be considered definitive.

I. Beyāḍ Abū Ṧubbāk papers

The first group of papers is the one for which we have much the most statistical information at present.

Type A

Paper watermarked with a crescent moon face inside a double-rimmed shield and bearing the countermark: “[1^st line] Beyāḍ Abū (sic) Ṧubbāk Isṭambūlī (بياض ابوُ شبّك اسطمبولي); [2^nd line] ‘ālī aṣīlī (عالي اصيلي)”⁴⁸.

Type A.1.

This is by far the most common. It differs from type A.2. by the dimensions of the shield bearing a crescent moon face, as well as by the length taken up by the 2^nd line of the countermark. Another difference, very useful for sorting paper types at a glance, resides in a sort of “defect” located on the last term of the countermark “aṣīlī (اصيلي)”: the ligature between the first yā’ and the lām is only partially present. Given the frequency of the defect, it is unlikely to be due to some pulp adherence, more probably to how the countermark was set up on the grid.

This type comprises ms. m/ḥ 1⁴⁹ and 5 codices out of the 13 in collection (majmū‘) m/ḥ 50 from ‘Abd al-Rahman al-Hadhrami’s library (ill. 1.a, 1.b, 1.c; ill. 1bis).

# of manuscript in collection (m/ḥ)	Copy date in chronological order crescendo	Maximum sheet dimensions (cm)	20 laid lines measured from the outer edge (cm)	Spacing between 2 chain lines measured from the inside edge (cm)
m/ḥ 50/8	1351/1932-1933	24 x 18	3-3.1	2.5
m/ḥ 50/11	1351/1932-1933	24 x 18	3-3.1	2.5
m/ḥ 50/12	1355/1936-1937	23.5 x 18	3-3.1	2.4
m/ḥ 50/4	1356/1937-1938	24 x 18	3-3.1	Alternating double chain, 2.4-2.5
m/ḥ 50/3	1358/1939-1940	24 x 18.5	3-3.1	Alternating double chain, 2.4-2.5
m/ḥ 1	1375/1956	24 x 17	2.8	Alternating double chain, 2.6 [See CCBMZ, I. 1. Les papiers filigranés, see under no. 126 and 127. No mention to double chain lines means only that it was not observed.]

The document of Aḥmad Ja‘far (Zabīd) should be added to this list, with the same spacing between two chain lines and the same measure for 20 laid lines as m/ḥ 1.

Based on the copyist's identity for m/ḥ 1 and 50/3, the paper had made its way to Zabīd.

The paper in manuscript ms. Arabe BnF 7264, previously mentioned, can be added to this category. Maximum sheet dimensions are 35.2 cm x 23.1 cm. It has a yellow appearance and its edges have gone brown. The entire manuscript was copied out on this A.1 type of paper. The first text, untitled, is made up of 3 folios, and bears the colophon date of 1345/1926-1927. The copying of the second work, the Kitāb Tanbīh al-Anām wa-Shifāʾ al-Asqām fī Bayān ‘Ulū Maqām Nabīnā Muhammad ‘Alayhi Afḍal al-Ṣalāt wa-Azkā al-Salām by ‘Abd al-Jalīl ibn Muḥammad b. ‘Aṭūm al-Qayrawānī al-Muradī, was completed, on 185 folios, in Dhū al-Qa‘ada 1345/1927⁵⁰, i.e. the same year as the first text. The manuscript circulated in Ethiopia, as indicated by a note in Amharic. This is not sufficient to make it the place where it was copied⁵¹. Common traits with Yemeni manuscripts can be noticed, such as the taṣliya perpendicular to the triangular colophon⁵². But it does seem more likely that it was copied out in Ethiopia, especially given its dimensions.

The paper from ms. Arabe BnF 7084, a collection of poems from the Colin collection, also belongs to this category⁵³. The title page bears a purchase title: “Ṣan‘āʾ Déc. 1929 / bought 18 Riyāl = 18x12 = 216 frcs / about 800 pages (400 folios)⁵⁴”. Moreover, on folios 133v-135r, a poem of mad'ḥ type congratulates the Ottomans, i. e. Mustafa Kemal and the Sultan, for their victory on the Greeks, which took place in 1340/1921-1922. This poem is followed by a controversy on qat consumption from folios 135r to 138v dated 1340/1921-1922. Finally, between folios 226 and 227, a loose calendar sheet can be found, dated 8^th Rabī‘ al-ʾAkhar 1330, i.e. 1912, and, between fol. 328 and 329, a loose printed form both in French and in Ottoman used to send telegrams: the pre-printed date, without its unit, to be filled in on the form is 192... The manuscript could be a holograph. All these data allow us to admit that it could not have been written on a short term. Folio dimensions are 245 x 180 mm. Spacing between two chain lines measured from the inner edges is 2.5 cm, the chains are simple. Twenty laid lines measured from the outer edges cover 3.4 cm.

The use range obtained for this paper encompasses a period from 1345/1926-1927 –maybe 1330/1912– to 1375/1956 for Yemen, with maybe a peak in 1345/1927 for Ethiopia.

Type A.2.

Regarding type A.2., for the moment we have instances of it going from 1263/1846-1847 to 1355/1936-1937, i.e. a range of almost one century.

Found in ‘Abd al-Rahman al-Hadhrami's library, in Zabīd (ill. 2, 3):

# of manuscript in collection (m/ḥ)	Copy date in chronological order crescendo	Maximum sheet dimensions (cm)	20 laid lines measured from the outer edge (cm)	Spacing between 2 chain lines measured from the inside edge (cm)
m/ḥ 60	1263/1846-1847	24.5 x 18	3-3.1	2.5-2.7
m/ḥ 50/9	1330/1911-1912	24 x 18	3-3.1	Alternating double chain, 2.5
m/ḥ 50/12	1355/1936-1937	23.5 x 18	2.8-2.9	2.5
m/ḥ 47	undated	24.5 x 18,5	2.9-3.1	Alternating double chain, 2.5-2.6 [No mention to double chain lines means only that it was not observed.]

A difference between papers put together under type A.2 can be noticed. It is more of a sensory nature. Paper m/ḥ 60 is dark and brittle, whereas the other occurrences are slightly yellow. Pending more in-depth investigation, we would emphasize the fact that ms. m/ḥ 60 is the most ancient of the lot. This could be the result of more advanced decomposition. Walz notes: “Glazed papers preserved in the Maḥkama dating from the sixteenth century look today as if they had been written on only very recently. Unglazed paper has turned brown and brittle⁵⁵”. This is based on the fact that imported paper could be prepared on site, as was the case in Egypt in the 19th c. Although it was used for codex production, A.2. therefore, seems to be of lesser quality.

Given the dating bracket for type A.2. as it has been established, based on a limited statistical basis for the moment, and for the first two examples, due to their sensory appearance, we would tend to put in this category the papers from the following manuscripts, until a full examination is carried out:

- Zabīd, m/ḥ 66/1, copied in 1328/1910-1911, maximum folio dimensions = 23 x 17.5 cm (82 fol.), copyist born around Ta‘izz;

- Ta‘izz, manuscript 107, copied in 1328/1910-1911, and kept at the Hā’il Sa‘īd Foundation;

- Ṣan‘ā’, manuscript copied in 1316/1899 and kept at ‘Abd al-Khāliq b. Ḥusayn al-Maġribī's private library, maximum folio dimensions = 24.5 x 17.5 cm.

Type B. Paper from the al-Khizāna al-‘Āmira Catalogue⁵⁶

Al-Khizāna al-‘Āmira is a library founded by Imam Yaḥyā b. Muhammad b. Yaḥyā Ḥamīd al-dīn (1869-1948) in the Great Mosque in Ṣan‘ā’. Its catalogue, the Fihrist kutub al-Khizāna al-mutawakkiliyya al-‘Āmira bi-al-jāmi‘ al-muqaddas bi-San‘ā’ al-muḥammiyya, has been printed out several times (ill. 4)⁵⁷. We personally know of the existence of printouts in 1312/1894-1895 and in 1361/1942, both coming from the same –Yemeni– printing presses.

Generally speaking, it is interesting to note that the print copy in our possession, that of 1312/1894-1895, was dealt with as if it were a manuscript.  First of all, a black border frames the area for writing. If we are to believe what is left of the binding, a printed piece of cloth was affixed on the back of the book and kept in place by a glued paper strip, on top of which was sewn a headband made of two coloured threads (red & white). It is the paper of this initial printout which we propose to study⁵⁸.

This is a paper with a crescent moon face inserted into a double-rimmed shield and bearing the countermark: “[1^st line] Beyāḍ Abū (sic) Ṧubbāk Isṭambūlī بياض ابوُ شبّك اسطمبولي) [);

 [2^nd line] ‘ālī aṣīlī (عالي اصيلي)”. A loose bifolio allows us to see that the crescent is facing right. Spacing between two chain lines, measured between the inner edges, is 2.4 cm, while spacing between 20 laid lines is 3 cm, between the two outer edges. There is a system of alternating double chains. The paper has a slightly yellow appearance (ill. 5, 6).

Traces in the shape of a series of short, oblique lines all along the top edge of the folios, show up regularly on this paper (ill. 7.a, 7.b). They suggest it was made as continuous paper, by means of a cylinder, on which was installed the grid: they represent the butt joint system of both edges of the metallic grid. The chain lines are installed perpendicularly to the axis of the cylinder. The first documented machines with cylinders for the making of watermarked paper date back to 1826, then went on to evolve⁵⁹.

The distance between two consecutive traces left by the butt joint of both edges of the metallic grid gives us the circumference of the cylinder. Dimensions of a bifolio from the catalogue are 36.4 x 46.9 cm height by width. The watermark is located on the right-hand side of the bifolio and the countermark on the left-hand side, the first line of text is placed slightly higher than the top of the watermark (ca. 2 cm). On each sheet of the work, either the watermark or the countermark is visible, i.e. no folio is exempt of either one or the other mark. The butt joint trace, when it does show up, is either located north of the watermark and countermark, or south of them; otherwise, watermark and countermark are present without the trace. This allows us to work out that the circumference of the cylinder is approximately 36.4 x 3 = 109.2 cm, which means a diameter of about 35 cm.

Moreover, we can observe a series of dotted lines on a considerable number of sheets, sometimes with a diagonal line. These marks were made when the pulp was still damp, in much the same way as the watermark, countermark and butt joint marks. They are very close to the sheet edges –top / bottom– and run parallel to them or merge. They always start from the outer edge of the sheet and are only visible for a short distance, 2 cm at the very most (ill. 8). They could represent pre-cutting of the paper roll into sheets. On one of the edges, top or bottom, of a certain number of folios, we can observe butt joint marks and dotted and oblique lines on the opposite edge (see for example p. 46). If our interpretation of the dotted and obliques lines is correct, this would confirm that, for the length of watermarked paper made by the turn of a roll, more than one sheet can be obtained. Moreover, these marks would be an additional clue to the mechanized production of this paper made as a continuous paper. Finally, the format of the Catalogue would be that of a pre-cut sheet.

The pulp was the subject of an analysis by optical microscopy⁶⁰. It is a bleached chemical pulp with a resiniferous base, and a high proportion of spruce, and deciduous trees, particularly poplar and beech. The presence of annuals is also confirmed (straw, linen, hemp) by their strong traces⁶¹. The pulp underwent a mechanical treatment, observable in terms of fibre cut, fibrillation, swelling of the fibrous wall and by the presence of numerous fine elements (ill. 9). Damage to the fibrous wall makes identifying the ingredients difficult. The paper is soft to the touch.

II. The Waraq Abū Ṧibbāk/Ṧubbāk papers

Compared to the former, the second group is, to this day, the least well represented of the two.

Type C. Le papier Waraq Abū (sic) Ṧibbāk al-Aṣīlī

Manuscripts m/ḥ 35 and 50 from ‘Abd al-Rahman al-Hadhrami’s library were written on a watermarked paper, sporting a crescent moon face, both left and right of the top half, two six-pointed stars, inserted into a double-rimmed shield, and bearing, on a single line, the countermark in Arabic and Arab writing: “Waraq Abū (sic) Ṧibbāk al-aṣīlī (ورق ابوُ شِباك الأصيلي)⁶²”. Watermark and countermark appear on the same bifolio, the dimensions of which are indicated for each example. Chains are double and alternating, 2.4 cm can be counted between the inner edges of two chains and 2.7 cm between the outer edges of 20 laid lines.

Ms. m/ḥ 35 contains a manuscript copy dated 1358/1939-1940. It is made up of a total of 7 folios, and consists of one ternion + 1 folio. Maximum sheet dimensions are 24.5 x 18 cm⁶³.

Three of the 13 codices in collection (majmū‘) 50 were written on the same paper, 50/6, s. d., 50/7, colophon dated 1356/1937-1938, and 50/10, colophon dated 1361/1942-1943. Manuscripts 50/6 and 50/10 were most certainly copied in Zabīd, judging by the copyists' names.

Codex 50/6 is made with a quaternion, codex 50/7, of 2 quinions + 1 folio, finally, codex 50/10 is made with a quinion + 1 ternion + 2 loose folios. Maximum folio dimensions are the same, 24 x 18 cm (ill. 9.a, 9.b)⁶⁴.

We can observe that, none of the codices have type A papers used concurrently with this one, whereas it's not rare to find codices written on different papers.

At the Hā’il Sa‘īd Foundation's collection, ms. 51 was written on a paper bearing the countermark “Waraq Abū (sic) Ṧibbāk al-Aṣīlī (ورق ابوُ شِباك الأصيلي)”. However, information concerning the laid lines and chain lines was not recorded. The manuscript is not colophon dated⁶⁵.

Finally, the “AbūṦibbāk”noticed by T. Walz is mentioned for Eastern Sudan at the end of the 19^th c., however we have no additional indication about it.

Type C paper has, for Yemen, a use range, going from 1356/1937-1938 to 1361/1942-1943.

Type D. The Waraq Abū Ṧubbāk/Andrea Galvani paper

This is a paper watermarked with a crescent moon face inside a double-rimmed shield and bearing two countermarks: on the right-hand side, over two lines, in Italian and with Western letters, [1^st line] “Andrea Galvani [2^nd line] Pordenone”; on the left-hand side, on one line, in Arabic and with Arab letters: “Waraq Abū (sic) Ṧubbāk (وَرَق ابوُ شِبَاك)⁶⁶”. The crescent moon face is facing left. Both countermarks are in front of the crescent, the text in Arabic located at the same height as the interval between the two lines of Italian text. The inner gap between two chain lines is 2.5 cm and the distance measured from the outer edges of 20 laid lines, 3.1 cm. Double-chains are noted.

Only two occurrences of this paper have been recorded to this day, both of them found in the manuscripts consulted at the Institute for Ethiopian Studies (IES), in Addis Ababa (Ethiopia), ms. 4574 (18) and 4627.

Ms. 4574 (18) is a collection made up of two codices. The first one, consisting of a single quaternion, has a text without any title, nor colophon; numerous commentaries appear in the margins. This is the one written on type D paper (ill. 10.a.; 10.b.). Maximum folios dimensions are 24.5 x 18.5cm.

 It is difficult to tell the number of the quires in the second bulkier codex⁶⁷. The text bears a colophon, indicating precisely the copyist's name, Muḥammad b. ‘Arab (‘Arib?) al-A.r.k.(b.)ī Nāhiyatān al-Aykbarī Qariyatān al-Ṣāfi‘ī Maḏhabān al-Aš‘arī I‘tiqādān al-Qādirī Ṭarīqān, as well as the copy date, Jumāda al-Thānī 1355/1936. This second codex was copied onto type A Abū Ṧubbāk paper (most probably A.2., due to the dimensions of the shield and its current colour). Maximum sheet dimensions are 24 x 18 cm. The proximity of the two Abū Ṧubbāk have allowed us to compare them. They have not evolved in the same way: type D tends to be closer to white, whereas type A has gone brown.

Ms. 4627 has no colophon. The library card indicates it should be 128 years-old: admitting that estimate is accurate, we would still need to know starting from when. For conservation reasons, we were not able to count the different quires. Maximum sheet dimensions are 24.5 x 18 cm. It should be noted that bifolios have often fallen out along the centrefold.

The countermark mentions the Galvani house, via one of its sons, Andrea, who lived during the first half of the 19^th c. (1797-1855)⁶⁸. According to Heawood, this paper was produced starting from 1836⁶⁹. In any case, it remained in use until the beginning of the 20th c., well after his death: e.g. among the manuscripts of Zabīd, between 1291/1874-1875 (m/ḥ 14, copied out in the ribāṭ of Yaḥyā b. ‘Umar al-Ahdal in Zabīd) and 1337/1918-1919⁷⁰. It is one of the features of paper D which allows us to guess its dating: it is produced on a cylinder (ill. 10.c.). However, the mechanisation of Galvani's paper-making process took place around 1883-1895⁷¹.

Numerous manuscripts from the IES come from Harar. Beyond the history of the collection, a palaeographic study could certainly tell if these two books were copied in Ethiopia.

It is not the first instance of a countermark in different languages. Waltz mentions two in Nigeria, one in Arabic and Western letters, the other with Arabic letters but in Hawsa language⁷². According to Walz, this last example corroborates the production of a paper to satisfy a given market.

For type D paper, one can suppose it being in use starting from 1883-1895.

Posterity & longevity of Abū Ṧubbāk paper

Type E. Abū Ṧubbāk papers from the souk

This is a non-watermarked paper, of a yellowy colour (ill. 11). Original sheet dimensions are 50 x 35 cm. Chains are perpendicular to the sheet, taken in its length. The distance between the inner edges of two chains is 2.3 cm. The spacing covered by 20 laid lines, measured from the outer edges, represents between 3.2 and 3.3 cm.

A laboratory analysis of the type of paper pulp and its components was carried out on a sheet bought in the souk in Zabīd, at the beginning of the year 2000's. The paper was produced from a bleached chemical pulp from resiniferous trees (Scotch pine) but mostly from deciduous trees (birch, poplar, hornbeam…). No presence of annual plant fibres was detected. Minor damage to the fibrous walls, as well as few fine elements, likens it to industrial pulp in the 20^th c⁷³.

From the Circulation of Paper to a Few Thoughts on Trade: An Abū Ṧubbāk / Abū Ṧibbāk / Abū Ṧubbāk ‘Zone’?

No “Abū Ṧubbāk” was recorded in Egypt by Walz, who based his study on a survey of both archive documents and codices. However, some doubt remains as his study runs until 1880, a date he highlights as a turning point in the balance of power for the Egypt-Sudan region⁷⁴. He notes one “abu shibbāk” for Eastern Sudan, at the end of the 20^th c., which is beyond the use range we have worked out. Out of the 50 manuscripts of the Archinard collection (French National Library), coming from West Africa, 24 were written on watermarked papers, over a timespan going from the 18^th c. to the first half of the 20^th c. None of these correspond to the paper studied here⁷⁵. Lastly, in Nigeria, the watermarked papers containing Arabic are those described by Walz⁷⁶. Needless to say, research in Northern Africa needs to be carried out.

So, to this day, Abū Ṧubbāk paper has been confirmed in manuscript codices and printed books in Yemen, manuscript codices in Ethiopia, as well as the far East, in manuscripts in Arabic. Yet, the spread of knowledge, scholars and manuscripts between these various countries for the period which concerns us, are documented facts, even if the spread of knowledge between Yemen and Ethiopia is a little-known subject for the moment: Ethiopian scholars went to Yemen to study, especially in Zabīd⁷⁷; Yemeni sheikhs settled in Ethiopia; Sayyid families, therefore scholars, from Hadramawt, emigrated to Indonesia and Malaysia, at least as of the 17^th c⁷⁸.. Between Yemen and Ethiopia, manuscripts did circulate but, quite possibly, so did paper too. Regarding Indonesia, the presence of manuscripts from Hadramawt cannot be discarded⁷⁹. Piecing together the history of libraries and manuscripts, as well as a stronger statistical basis reflecting the presence of such a paper in Ethiopian codices should tell us more. The same issues arise with the papers used in the manuscripts kept at the Riyadha mosque, in Lamu (Kenya), as these manuscripts are evidence of links to Yemen⁸⁰.

For the moment, among these areas, the most likely centre for the import of this paper is Yemen.

As for type D, only two occurrences of it have been spotted, and both only in Ethiopia.

Production and trade: Towards an Ottoman Paper vs. Italian Paper Competition

 We have drawn up 2 groups of “Abū Ṧubbāk”papers, the Abū Ṧubbāk per seand the Abū Ṧibbāk/Abū Ṧubbāk which would point to two distinct production locations, according to our hypothesis.

The first group, comprising types A and B, have a countermark in osmanlı, mentioning Istanbul. Type A.2. was used to make codices from 1846-1847 to 1936-1937, i.e. a use range of almost one century, then was followed by type A.1., for a period going from 1926-1927 –1912?– to 1956, in Yemen, with possibly a peak in 1345/1926-1927 for Ethiopia. Type B was reported in 1312/1894-95 (maybe until 1942?), this one was a type of continous paper made industrially for printing.

The second group, comprising types C and D, have a countermark in Arabic and do not mention Istanbul anymore. As the countermark on type D refers to the house of Galvani, it may have been used as of 1883-1895. The information given by Walz quoting Shuqayr, who mentions a paper called “abū šibbāk” among European imports to the Sudan at the end of the 19^th c., does add up, even if the vocalisation of the šin in type D is not with an i, but most likely with a u, as in “Abū Ṧubbāk”, “abū šibbāk” corresponding more to type C. Finally, for Yemen, the use range for type C, present at a later date, stretches from 1937-1938 to 1942-1943. We may thus assert that type D was made in Friuli, but the maker of type C has yet to be identified.

Conclusion

The primary interest of this work is to show that there are several papers of the Abū Ṧubbāk type. The identification and classification of the examples known to date provide a means of more accurately dating and locating the production of manuscripts in which they were used.

But their documentary value for understanding the paper trade along the shores of the Red Sea during the 19^th and early 20^th centuries is also considered. It is with the aim of systematically treating the paper of manuscripts itself as a source, to clarify the methodology to follow and to demonstrate its epistemological interest.

From a historical perspective, this material provides a glimpse of the continuing competition in the Yemeni and Ethiopian markets from the mid-nineteenth century to the beginning of the 20^th century between a Turkish manufacturer (probably private), and the famous Galvani, who dominated the market in Egypt and Sudan during the 19^th century.

Although the question of imitation or borrowing of watermarks by Italians and Turkish producers has been raised, particularly by Turkish historians, the ‘traditional’ sources referred to have so far provided no proof. According to our analysis, type D paper provides an example of Italians borrowing from Turks (the countermark in Arabic), after the probable borrowing of the Italian watermark (the shield and crescent) by the Turkish.

Abu Ṧubbāk papers appear, though not exclusively, to be an excellent marker of the emigration or movement of the Yemeni population, who were often both scholars and merchants. The fact that the name “Abu Ṧubbāk” became a generic term indicates the affection towards this paper by those who use it. The routes of goods and those of men (scholars) overlap in this case, but a map of the circulation of papers reflecting people's customs and human preferences or tastes also calls for the attention of researchers. Because it is not certain that these routes always overlap.

It is hoped that the statistical information on which these findings rest may be expanded in the near future and that, in general, this type of study will proliferate.

---

Source note: This was published in: Research Articles and Studies in honour of Iraj Afshar, 2018, Al-Furqān Islamic Heritage Foundation, London, UK, pp. 195-228. Please note that some of the images used in this online version of this article might not be part of the published version of this article within the respective book