• 1 For the first edition of this work see Ibn Bâdîs, 1971. A different edition of the chapter on paper (...)
• 2 For more information on the copies of this manuscript see Gacek, 1997: 58, n. 1.

1Our knowledge of die methods of papermaking in the Arab world is largely based on the well-known chapter from ʻUmdat al-kuttâb wa-uddat dhawî al-albâb ascribed to the Zirid ruler of North Africa al-Muʻizz ibn Bâdîs (d. 454/1062).1Recently,howewer, another work containing a recipe for papermaking has come to light. The work in question is al-Mukhtard fî funûn min al-ṣunà, a copy of which is preserved in al-Khizâna (al-kutubkhâna) al-Âṣafîya, Hyderabad (India). Although the Hyderabad copy is anonymous, this title has been attributed to the Rasulid ruler of Yemen, al-Malik al-Muẓaffar Yûsuf al-Ghassânî (d. 694-1294).2

• 3 I am grateful to Mufti Muḥammad Azeemuddin, Editor-in-Chief of Dâ'irat al-Ma'ârifal-'Uthmânîya (Hyd (...)

4The Mukhtardʻ, in its original form, is a compilation of various accounts of art and craft techniques by artisans who were themselves expert in their fields. According to the preface the author ordered each craftsman to express himself in his own words, a policy which may explain the different styles of writing found in the text. The existence of this manuscript was first signalled by ʻAbd al-Quddûs al-Hâshimî (Hâshimî, 1358/1939) who provided a short codicological and palaeographical description. According to him the copy consists of 118 leaves as opposed to 142 in an earlier foliation and measures 30 x 18/24 x 15 cm, with 43 lines per page. It has an illuminated headpiece, the rule border (jadwal) is executed in red and blue ink, and the chapter headings are rubricated. Furthermore, the manuscript contains numerous diagrams. It was copied by Muḥammad ibn Qawâm (Qiwâm) ibn Ṣafî ibn Muḥammad Ḍiyâʼ Turk Nâgûrî (Nâgawrî), known as Qâdîdkân, in Rajab 876/1471, who wrote in a small, but clear naskh hand. The copyist was a Persian speaker, most probably of Indian (more specifically Nâgawrî) descent, as evidenced by his nisba and the fact thatboth Arabic and Persian are used in die colophon. The manuscript itself has a number of lacunae and unfortunately the chapter containing the recipe and the recipe itself are incomplete. To judge from a later foliation there is only one leaf missing. However, it appears that only a small portion of the recipe is wanting. The text is partially pointed and in places illegible due to water stains. The letters dâl, dhâl, râʼ and zây, as well as medial ʻayn, ghayn, qâf and fâʼ, are often difficult to distinguish. Thus, for example, such words as dahaka, dahk and dahîk can be read as rahaka, rahk and rahîk (lin. 46-49), and daqqa as raqqa (lin. 49), just as mikhrash and mikhdash can stand for the same word (lin. 27-31). In addition, the letter kâf issometimes written without the line above (shaqq), which makes it look like lâm (e. g. lin. 17, 19), while the hamzah after the alif of prolongation at the end of words is rarely supplied.3

• 4 Moshe Piamenta (1991: II, 461) makes a reference from mudaḥ to mudakh. Both hadah and madaḥ (the la (...)
• 5 For a description of markh, see Lane, 1984: II, 2705. This word does not appear in Piamenta, 1991 o (...)
• 6 For marjandA mizj seeDînawarî, 1973: 52, 268, 269, 283 and Dimyâṭî, 1965: 144.

5a) It appears that, according to the recipe, paper was made of the white fibres of the inner bark, liḥâʼ (the liber of the bast, as opposed to the cortex), of the fig tree (ficus populifolia) or mudakh. This word (also written mudah) is of Yemeni origin and no longer used, the more common term being tin or balas.4This piece of information is significant since Chinese paper was made from the bast fibres of the paper mulberry (broussonetiapapyrifera), popularly known as kozo, and fig trees belong to the same family (moracea). The earlier reports on papermaking traditionally mention qunnah (hemp) and kattân (flax); usually hemp and flax products, such as cordage (ropes) and linen, respectively, were used, but unrefined hemp may have been the main ingredient of Ibn Bâdîs' recipe (Irigoin, 1993: 279, 281). Barwîn Badrî Tawfîq reads this word as markh and the same word is mentioned by Ḥabib Zayyât (Tawfîq, 1985: 277; Zayyât, 1992: 77).5It is recorded by Aḥmad ʻÎsâ as laptadenia pyrotechnica, cynan-cum pyrotechnicum and periploca ephedriformis, all belonging to the asclepiada-ceœ family (ʻÎsâ, 1981: 64, 108, 136). According to the same source the periploca ephedriformis is known in Yemen as marakh!Other possible readings of this word are marjzandmizj. Some sources link these words to the bitter almond tree6. However, in view of the above mentioned fact that the fig belongs to the same family as paper mulberry and it was used for bark paper in Mexico, Polynesia and Africa, and in papermaking in Burma (Bell, 1990: 68), the form mudakh (mudah) seems to be the more likely reading.

• 7 For the word jaffur see Asadî, 1988: III, 66.

6b) One cycle of paper production was calculated to produce 100 sheets. The initial stage of soaking and beating the bast fibres took 12 days. At the end of this process the paper pulp was so fine that it resembled tufts of puffy cotton fibres (jaffur, lin. 28).7 This may lend extra strength to the argument put forward by Karabacek (1991: 43) that Arab paper was never produced from cotton but rather the pulp was so fine as to ressemble cotton. According to Bell (1990: 59), however, paper was made from cotton rags in Xativa, Spain in 1151, while cotton was used in India in various combinations, for example, in pure cotton, cotton/wool and rice straw/cotton paper (e. g. Macfarlane, 1987).

• 8 Irigoin, 1993: 290. Note, however, that the sentence in the text of Ibn Bâdîs is much more explicit

7c) A supple, as opposed to a floating mould (qâlib) was used. The mould was dipped in the vat containing pulp. The sheet was then removed from the mould and put out to dry. However, just as in the description given by Ibn Bâdîs, it is not unlikely that the technique of drawing the pulp from the vat may have been that which pertains to a floating mould.8 The lines in question (31-32) read: wa-yaṭlaʻ fîhi al-shajar bi-qadr ḥattá yaṭlà mutasâwiyan wa-takûnu al-waraqa mutasâwiyatan min jamîjawânibihâ wa ʻalá qadr athkhânihâ allatî yurîd (“and he dips a mould into the vat and covers it with a certain amount of pulp until it is level and all the sides of the sheet are even according to the thickness he desires”).

• 9 Or « in al-Yaman they paste papers together and line the inside covers of their books with starch » (...)

10f) A part of line 53 reads: thumma yamudd ʻalá al-waraq al-munshâ waraqa ma'ahu(!) nashan (« and then he stretches (places) over the pasted sheet a sheet of paper with paste »). Although the sentence in Arabic is not as clear as one might wish, it appears that two sheets of paper may have been pasted together to form one thicker sheet. The sheets must have been very thin as the author of the text reports that one ball of fibre (kubba) the size of a lemon would yield five sheets, presumably from an average size mould. The practice of pasting two sheets together was first discussed by Karabacek (1991: 51) but later questioned by Don Baker and Malachi Beit-Arié (Baker, 1991: 31, Karabacek, 1991: 89, Beit-Arié, 1996: 12). Karabacek based himself on a passage from die Ihsân al-taqâsîm of al-Muqaddasî (4/9th cent.) (Muqaddasî, 1967: 100) where the Arabic text reads: wa-bi-al-Yaman yulazziqûna al-durûj wa-yubaṭṭinûna al-dafâtir bi-al-nashâ (« and in die Yemen they paste die sheets of paper togedier and line die book covers with wheat starch » (Karabacek, 1991: 51).9

• 10 One way to create lineation is to fold a sheet of paper along its width into a concertina file shap (...)

11g) As mentioned earlier, one ball of fibre (kubba) was the size of an orange or lemon and yielded five sheets (lin. 30). The author uses three words in this connection: untrunja, lîmûn and nâranja. The utrunja is bigger than the nâranja and lîmûn. Sheets of newly-made paper were at first stacked in piles of 100, then separated into lots of five sheets each. These were later glazed, folded in the middle, stacked in groups of five bifolia (dast) and presumably sold to scribes in this form. This papermakers' practice may explain why, in the majority of cases, gatherings in Arabic codices consist often leaves (quinion). In the West the gathering, in general, consisted of eight leaves (i. e., four bifolia, known as a quaternion) which were formed by folding the original sheet twice. Why they were stacked in fives is interesting. One might be tempted here to ascribe this practice to the idea that odd numbers were preferred in the Islamic world following the prophetic tradition which says: inna Allâh witr, yuḥibbu al-witr (« God is one, he loves the odd number »). However, as in the case of the uneven number of lines per page, which appears to have its origin in the arrangement of the lines in the mistara (ruling board), the answer may lie somewhere else.10 The numbers 100 and 10 are moreover easily divisible by 5.

• 11 The word in the text appears to read hâyir. Another possibility could be jâ'ir from jîr, « lime »). (...)
• 12 This word is mentioned four times in the text (1. 5, 6 and 9). In the first instance it is not poin (...)

13[1] —The manner of making local paper. The bast (liber, phloem) of the fig tree (mudakh) is taken and dried. The outer layer (qishrd) is peeled off, and he (the papermaker) throws it away, but the white inner layer is kept. It looks like die fibres of a cord (matn) (Piamenta, 1991: II, 459). It (the bast fibre) is soaked (macerated) in fresh (?) water (mâʼḥâʼir)11such as (comes from) a pool (birka). It (the pool) is lined with palm leaves used for brooms (waraq al-miqashsh)12 in order to protect it from dirt and other impurities. The pool is filled with a good quantity of water and on top of the bast are placed palm leaves and, on top of them, small sticks which prevent it and the palm leaves from rising to the surface. The bast is soaked for four days and nights.

• 13 The word in the text is unpointed and can be read as majmûdan or makhmûran. The latter seems to mak (...)
• 14 Qadda (« plaster, cement ») concrete with qaḍâḍl' qaḍâḍa;muqaḍḍaḍ (« cemented, plastered ») (Piamen (...)
• 15 « Large pitcher-shaped vessel made of clay [put up in walls and cornices] as a bird's nest » (Piame (...)
• 16 See Piamenta, 1991: 1, 152 under dafrand ṭafr(« to push by force », « to flood »). The context show (...)

14[2] — It is then taken out of the pool (ḥawḍ), the water is squeezed out of it and it is left in the corner of the house, which is lined with the palm leaves to prevent it from being soiled with dirt, in the form of balls of fibre (kubab) stacked one on top of the other. It is left thus wet (fermenting) (makhmûrari),13and covered with palm leaves, for not more than three days and nights. For if it were left there in die corner for longer than this, it would be ruined and the palm leaves would break (désintégrate). (After the three days and nights) the bast is taken out of the house and dried thoroughly in the sun (yuḍaḥḥá), on rocks (ṣafan or ṣofâ) or a clean, plastered roof (saqf muqaḍḍaḍ)14 until there is no moisture left in it. Then, for the second time, he (the papermaker) soaks it in water, squeezes (presses) it so that the water comes out and makes it into small balls of fibre [...], each ball the size of an orange (nâranja). He empties the bast fibres (khiraq al-kubab) into a pitcher-shaped container (mirkan),15 one on top of the other, then takes one ball after another and, while they are moist, cleans them from any remaining thick green bast (layer) and other impurities [...] attached to them, until no husks (qashr) or dross (ghashsh) remains. The cleaning (of the bast) is done by means of pounding (ṭafr).16 There is no need to separate it into fibres (tansîl) as the beating (daqq) separates them, softens them and makes them blend together.

• 17 The verb shanna means « to sift », « strain » (Piamenta, 1991:I, 267) and the noun shanna presumabl (...)
• 18 For mikhrash and mijdaḥ see Lane, 1984: 1, 389 and 723.

16[4] — Afterwards, he places it (the dough-like mass) in a pool (ḥawḍ) filled with clean water and stirs it (yukhîḍuhu) so that it is well mixed with water. He strains the pulp (slurry), while stirring the water, so that all coarse fibres (shanna, lîfal-shajar)17 come out. He thus gathers one ball of fibre (kubbd) after another and places them in another clean, plastered pool (vat) (ḥawḍ muqaḍḍaḍ), of a greater length and breadth than the mould (qâlib), and fills it with clean water. The water should not contain any impurities (waʻth, kadar, qushûr). He then places all the balls in the vat and beats (stirs) them well with a wooden tool (mikhrash) at the end of which are two crossed pieces of wood, like a churn-staff (mijdaḥ)18, so that they are well mixed and become like tufts of puffy fibres teased from cotton (al-jaffur alladhî yandifmin al-qutn). After this, he strains the pulp, for the second time, with a cloth (khirqa) and makes balls of the size of a citron (utrunja) or as he wishes. He places the balls on the edge of the vat (ḥawḍ) and takes one chunk after another, each the size of a lemon (lîmûn murakkab) or orange (nâranja), and returns them to the vat. He makes approximately five sheets of paper from one ball. He beats die balls, for the second time, using the mikhrash so that the fine pulp (shajar) is well mixed with water.

• 19 This is the reading given by Varisco (1994: 173). The other varieties of sorghum are not given in t (...)
• 20 The zabadî (pi. azbûd) is a dry measure used for crops (see Varisco, 1994: 164-165). According to P (...)

19[7] — Subsequently, a quantity of white moist sorghum (dhura), which is called alʻ- ayyâshî(?) or al-azraq (?) or al-shurayjî,19 and which does not exceed one half of the zabadî al-nafarîmeasure20 for one hundred sheets, is crushed (ground). It is soaked accordingly until all the husks (qishrd) are eliminated and is gently mashed with a spatula (marra) up to seven times. Then it is left in a container (inâ) till the next day so that it becomes sour. It is then passed (strained) through a coarse cloth until the fine mash comes out of it and it remains (?) [...]. The mash (muqill) is left as it is and it increases in volume (?). It is then cooked until it is well done, just as one cooks starch (nashan) for glue (gharan, ghirâ). Next, it is placed in another receptacle (inâ). A cloth, in the shape of a ball (tukabbib kubbd) and having a handle (miqbaḍ), is taken. He (the papermaker) places it in a receptacle of paste (minshâ), takes the paste which adheres to the cloth and with it pastes the front (wajh) of the sheet, turns it upside down and smears its back (qafan). He does this for all the sheets, recto and verso, in an even way. He stacks the sheets one on top of the other up to twenty or thirty sheets. He thenstretches (places) over the pasted sheet a sheet with paste. Then he picks it (them?) up and places it (them?) on a headcloth (ghifâra) to dry in the sun on a clean plastered roof (saqfmuqaḍḍaḍ).

• 21 Mishraṭ, a cutter. For a description of this tool see al-Malik al-Muẓaffar, 1471: f. 21.
• 22 This is an interesting usage of the word dast (from the Persian « hand »). Comp. Dozy, 1967:1, 441: (...)
• 23 Although the word'in the text looks like nawḍ (« ploughshare », see Piamenta, 1991: II, 499), the m (...)

20[8] — He sticks (attaches) the edges of the sheets onto a plastered area (qiḍâḍ), while the paste is still moist, so that the wind will not carry them away and so that they will not become crinkled and so that the paste (nashan) which is on them will dry. Thereafter, he takes a sharp cutter [...], such as a mishraṭ,21and cuts the glued edges off the plastered area so that they are easily separated and do not tear [...]. After that, he picks up the dry sheets five at a time (dusûtan)22and burnishes them one at a time on a smooth stone, such as a marble slab (rukhâma), with another smooth and round stone such as a miṣqala or with a glass bead (kharazat al-maṣaqil al-zujâj) or a wooden tablet (lawḥ)23 of the size that the papermaker (ṣâniʻ) can handle. He burnishes the sheet lengthwise, back and front, until all sheets are done. Then, he bends a sheet (yaʻṭifuhâ) on its front (wajh) into two halves and holds it by its edges on die left side so that they (edges) do not move. He folds (yaksir) its middle with a burnisher (miṣqala), polishes and folds all sheets and, when finished, places them one on top of the other. Subsequently, he groups them five by five and lays them on their folds (makâsirihâ) underneath a wooden board with a stone on top to press them down for one night after the burnishing. Then he takes [...?].