Quṭbaddīn an-Nahrawālī’s (917-990/1511-1582) Treasure of Names and Other Ottoman-Era Arabic Treatises on the Art of the Muʿammā

In Timurid times Persian littérateurs devised a new kind of logogriph (muʿammā) that differed considerably from the muʿammā as was known in the Arabic tradition. The most salient feature of the new, Persianate muʿammā, which is normally a couplet, is that it has two levels of meaning: an obvious or surface meaning (‘the poetic meaning’), and an encoded ‘riddle meaning’, which gives the clues to the solution of the riddle. Since the 16 th century the new, Persianate muʿammā became very popular with Ottoman Turkish and Arabic littérateurs as well. In fact, to judge by the available evidence, it appears that the new muʿammā gradually became the most popular kind of literary riddle in Arabic. The present paper presents Quṭbaddīn an-Nahrawālī’s (917-990/1511-1582) Treasure of Names, the most influential Arabic treatise on the new muʿammā, which is modelled on earlier Persian treatises. It discusses the growth of the new muʿammā in Arabic, describes the rules applying to it as presented in an-Nahrawālī’s treatise, analyses several Arabic muʿammayāt cited in that treatise and concludes by mentioning some additional Ottoman-era Arabic treatises on the riddle that testify to the great popularity of the new muʿammā in Arabic until the late 19 th century.

The rise of the riddle as a literary form in Arabic started in the Buyid era-apparently as a result of the growth of descriptive poetry and especially allusive and enigmatic descriptions that typically made extensive use of figures of speech. Since the fifth/eleventh century riddle exchanges between littérateurs and scholars became increasingly popular. The riddle gradually became a favourite pastime in elite gatherings as well as at the various Muslim royal courts. 1 The Arabic literary riddle flourished especially in the Ayyubid and Mamluk periods and continued to thrive throughout Ottoman times. In the course of time all three Arabic words initially used to denote a 'riddle' in general, to wit luġz (pl. alġāz), muʿammā (pl. muʿammayāt) and uḥjiyya (pl. aḥājī), have acquired specific meanings and have come to signify particular kinds of riddles, even though not all premodern Arab authors were aware of the specific meanings of these words nor applied them accurately. In its special meaning the word luġz signified a 'notional riddle', a riddle whose topic (the encoded entity) is a notion, a word's meaning (in Saussurian terms, the signified), not a word as such (signifier). The clues given in 'notional riddles' refer to properties and features of the encoded notion and are mostly puzzling and contradictory. A typical example of such a riddle is the following (metre sarīʿ): 2 ‫َدَيفَطاعةَالباري‬ ‫جمهت‬ َ ََََََََََ ََ ََ َ َ َ I recall somebody scrawny, bowing down, putting his forehead on the ground, blind but discerning-his tears flowing, He keeps to the five (prayers) in their appointed times and exerts himself in obeying the Creator (/the sharpener).
The solution to this riddle, which operates with metaphors and double entendres, is the pen. The five to which it keeps are the fingers-not the prayers as the context, which refers to a pious person and to 'appointed times', suggests. The context also misleads one to think of God, the Creator (al-bāriʾ), even though al-bārī is the sharpener.
Besides being used for 'riddle' as well as for 'encrypted text' in general, the word muʿammā, on the other hand, has been applied from early on to logogriphs or 'word riddles', namely riddles whose topic is a word as such (in Saussurian terms, the signifier) and most often a person's name in particular. The clues given in logogriphs hint at the letters of the hidden word, their sequence, way of writing, vocalization, numerical value, etc. 3 Consider, for example, the following poem on the name Saʿīd (metre munsariḥ): 4 My sweetheart smiles revealing the beginning of his name, then captivates one with its second letter.
Then two more letters-had he shown up with them, he would've done me a favour (lit. hand, yad) that builds the form of his name.
TOMASEK, Das deutsche Rätsel im Mittelalter, Four letters, the half of which equals their sum in number-it doesn't fall short nor exceeds it.
Moreover, there's a day in his name, of which both Arabs and non-Arabs are proud.
Think about it and ponder and ride with it every intractable mount! (i.e. brave the difficulties).
Because of its shape, the letter sīn is typically likened to teeth; likewise, the letter ʿayn is likened to a lovelock. 5 Yāʾ and dal are hinted at in verse 2 (the word yad, which 'builds the form of his name'), whereas the glorious day is the ʿīd, the Feast of the Sacrifice. Because of its three teeth (nabras) the letter sīn may be analysed as three letters of the form of bāʾ, tāʾ, ṯāʾ etc., that is, composed of one nabra. Thus, half of the total number of the word's letters, the sīn and the ʿayn, can be said to be four letters, equalling the sum of the word's letters in number when the sīn is counted as one letter. As in this case, the great majority of logogriphs are on persons' names. The reason why this is so, as Friedrich Rückert has ingeniously remarked, is that a person's name is not a notion and the clues regarding it can only refer to its letters, the only other possibility being to refer to famous bearers of the name (which would make the riddle too obvious). 6 Certainly, several riddles encode elements of both the signifier and the signified and can therefore be called 'mixed', but there was no term used for them, even though some authors commenting on the difference between luġz and muʿammā point to the existence of 'mixed riddles'. The following is such a 'mixed riddle' (metre sarīʿ): 7 O druggist, tell us plainly the name of a thing that you offer for sale for a bargain price, You see it with your eyes while you are awake just as it's seen with the heart (qalb) in your sleep (nawmik).
The solution to this riddle, which offers clues as to both the sought word's meaning and its way of writing, is cumin (kammūn). Cumin is a cheap druggist's ware, seen with one's eyes when one is awake. Even though it can also be seen with the heart when one is asleep, namely in a dream, at the riddle level the word qalb means 'inversion': inverting the letters of nawmik ('your sleep') one obtains kammūn (the vocalization and the doubling of consonants are ignored).
Finally, uḥjiyya denoted specifically a charade. This special kind of 'word riddles' divides a relatively long word into two or very rarely three parts, each of which is a word in its own right, and substitutes each part with a synonym. The solver must guess what the synonyms stand for and thus reconstruct the encoded word. Arabic charades are for the first time attested in al-Ḥarīrī's al-Maqāma al-Malaṭiyya, who was also the first to apply the word to this special type of riddle, thus narrowing its meaning and making it a specialist term, but there are also several collections of aḥājī by later authors. Suffice it to mention one example by al-Ḥarīrī (metre mujtaṯṯ): 8 O you who soars high in excellence with his sparkling intelligence, What equals my saying 'a hunger that has been supplied with provisions'?
As in this case, the uḥjiyya is normally a couplet or, very rarely, a three-liner. Its first line praises the excellence and sagacity of the addressee, who in the second verse is challenged to solve the riddle. The solution to the present uḥjiyya is ṭawāmīr ('scrolls', sg. ṭūmār). For ṭawā (= ṭawan) is a synonym of jūʿ ('hunger') and mīr(a) is a synonym of umidda bi-zād ('it was supplied with provisions').
Pre-Ottoman Arabic authors did not generally bother to distinguish between luġz and muʿammā. Abū l-Maʿālī al-Ḥaẓīrī (d. 568/1172), the author of Kitāb al-Iʿjāz fī l-aḥājī wa-lalġāz (The Inimitable Book on Quizzes and Riddles), the first surviving Arabic treatise cum anthology on riddles, is a notable exception. For even though he does not give a definition of muʿammā and only notes that 'it mostly relates to the way of writing' (wa-huwa yakūnu fī l-ḫaṭṭi akṯar), he uses the term eight times to describe riddles cited in the anthology. All eight riddles thus labelled are logogriphs, whereas seven of them are on persons' names. Al-Ḥaẓīrī never uses this term to describe notional riddles. 9 Later authors, however, including Mamluk authors who took a vivid interest in riddles, such as Ṣalāḥaddīn aṣ-Ṣafadī, may refer to logogriphs on persons' names as luġz and vice versa dub a notional riddle a muʿammā. 10 This was probably due to the growing numbers of 'mixed riddles' in Mamluk times, when the use of encoding techniques typical of logogriphs, namely miswriting (taṣḥīf), changes in the vocalization (taḥrīf), inversion (qalb, ʿaks) and omission of letters (ḥaḏf), and computations based on the letters' numerical value (ḥisāb), became noticeably more widespread in riddles encoding notions.
Despite the long tradition of the Arabic literary riddle, a clear distinction between luġẓ and muʿammā was first made in early-Ottoman Arabic works. This is no coincidence. It is related to the rise and unprecedented efflorescence of the art of the muʿammā in Persian and the composition of several specialist treatises setting and describing the rules to be followed in composing and solving muʿammayāt by Persian authors of the fifteenth century, starting with the Timurid historian and littérateur Šarafaddīn al-Yazdī (d. 858/1454). 11 Al-Yazdī was probably the first (or one of the first) to describe the rules of this art and the operations involved in solving muʿammayāt in his treatise entitled Ḥolal-e moṭarraz (composed in 832/1428 for his patron Sultan Ibrāhīm b. Šāhrūḫ). 12 The next great figure to compose treatises expounding the rules and operations of the muʿammā was ʿAbdarraḥmān Jāmī (817-898/1414-1492), 13 who was followed by several other writers, a most acclaimed one being the slightly later poet Mīr Ḥusayn al-Muʿammāʾī (d. 904/1498-9 or 912/1506-7) of Nīsābūr. 14 Another figure that must be mentioned in this context is the great Chaghatay Turkish poet ʿAlī Šīr Nawāʾī (844-906/1441-1501), a close companion and adviser of the Timurid Sultan Ḥusayn Bāyqarā of Herat (r. 873-911/1469-1506) and a lifelong friend of Jāmī, who also penned a treatise on the muʿammā in Persian and as the literary doyen of Herat and a great patron of literature and the arts encouraged contemporary poets to cultivate the genre. 15 Ottoman Turkish and Arabic authors followed suit. Throughout Ottoman times the muʿammā witnessed an extraordinary efflorescence in all three great Islamic literatures. 16 But this was a new, Persianate muʿammā. The rules set by al-Yazdī and his successors and reproduced in the Ottoman Turkish and Arabic specialist treatises describe a kind of muʿammā that differs from the 'old', Arabic muʿammā in many respects. In what follows I shall present the new, Persianate muʿammā in Arabic and the rules applying to it as given in Quṭbaddīn an-Nahrawālī's (917-990/1511-1582) Treasure of Names, the most influential Arabic treatise on this art, and conclude by mentioning some additional Ottoman-era Arabic treatises on the riddle that testify to its great popularity throughout the Ottoman period.

Muʿammā Old and New
The most salient feature of the new muʿammā is that the poem has a meaning of its own and can be read and appreciated for it, without the reader being aware of the fact that it is a muʿammā. In the relevant Arabic treatises this obvious or surface meaning is termed 'the poetic meaning' (al-maʿnā š-šiʿrī) to distinguish it from the meaning of the poem at the riddle level, which is termed 'the riddle meaning' (al-maʿnā l-muʿammāʾī). As a matter of fact, the Arabic treatises stress that a good muʿammā must have a second, 'poetic meaning' apart from its meaning as a riddle, even though this had not been expressly stipulated by the early Persian masters. Nonetheless, muʿammayāt put into circulation, sent to friends or published in anthologies, dīwāns, etc., were identified as such in the introductory rubrics, which usually also give the solution to the riddle. It is important to stress that the solution was mostly given in advance, the solver's task being to discover how the sought word can be extracted from the poem by identifying and following the hidden clues, that is, decoding 'the riddle meaning'. The various treatises deal with the different sorts of hidden clues and explain in detail how they can be identified and combined to extract the sought name from the muʿammayāt that are cited as examples. Obviously, having identified the clues and understood how the name is obtained, one admired the poet's ability to compose a poem with an obvious and a hidden meaning. According to Ṭāhir al-Jazāʾirī (1267Ṭāhir al-Jazāʾirī ( -1338Ṭāhir al-Jazāʾirī ( /1851Ṭāhir al-Jazāʾirī ( -1920, the author of Tashīl al-majāz ilā fann al-muʿammā wa-l-alġāz, the most recent Ottoman-era Arabic treatise on riddles (published 1303/1886), one must always give the solution of a muʿammā in advance, for otherwise it is impossible to solve it. 17 Even though explicit comments to this effect are absent from the earlier Arabic treatises I have examined, both Ibn al-Ḥanbalī (see here below) and an-Nahrawālī occasionally refer to the evidence (qarīna) that the hidden name affords, which means that the 'hidden' name was normally disclosed in advance and the solver was expected to use the evidence (indications, clues) it afforded in order to decode the 'riddle meaning'. The fact that the solution was Edebiyatı", İA. Apart from the study by Anwari-Alhosseyni and the brilliant comments of Losensky in Welcoming Fighānī, the Persian muʿammā seems to have been relatively little studied; see also STOREY, Bio-Bibliographical Survey, 218-35; F. RICHARD, "Quelques traités d'énigmes (Moʿammâ) en persan des XV e et XVI e siècles dans les collections de la Bibliothèque Nationale"; G.L. WINDFUHR, "Riddles", esp. 315-26. A.A. SEYED-GOHRAB's Courtly Riddles: Enigmatic Embellishments in Early Persian Poetry deals exclusively with the Persian loġaz (in the sense of 'notional riddle'). more often than not given in advance shows that what was primarily appreciated was the existence of two different levels of meaning in the text.
As is true of old-style logogriphs, most new muʿammayāt are on proper names, but there are also several muʿammayāt on common names. Formally, the new muʿammā is usually a couplet, a single verse, or very rarely a three-liner, even though, to cover all possibilities, most treatises state that a muʿammā may also be in prose. The clues ('the riddle meaning') are mostly found in the last verse of the poem. The Arabic treatises expound the rules applying to this art as set and described by the Persian masters. It is very remarkable that they make no reference to figures of speech known and discussed at length in the Arabic rhetorical tradition, such as the tawriya (double entendre) or the jinās murakkab (compound paronomasia), to name only the two most relevant figures for the new muʿammā. This indicates their dependence on Persian models. As to the genre of the poems at the surface level, they are mostly ġazals (love poems), but as becomes clear from the selection presented below, they may also treat any other subject (wine, wisdom, religious topics, etc.).
The first author to write about the new, Persianate muʿammā in Arabic was Raḍiyyaddīn Muḥammad b. Ibrāhīm, a polymath and historian of Aleppo known as Ibn al-Ḥanbalī (908-971/1502-3-1563), who authored several works on history, philology, the religious sciences, mathematics and the natural sciences. 18 Ibn al-Ḥanbalī composed a muzdawij poem (a poem in rhyming couplets) in the metre wāfir, which comprises 66 verses and deals with two riddle kinds, the (new) muʿammā and the uḥjiyya, as its title indicates: Kanz man ḥājā wa-ʿammā fī l-aḥājī wa-l-muʿammā (The Treasure of Him Who Poses Charades and Logogriphs: On the Uḥjiyya and the Muʿammā). This poem was inspired by Jāmī's poem on the muʿammā (see note 13, above) and was dedicated to the son of a Persian scholar and Sufi, Muḥammad al-Ḫālidī al-Kaššī as-Samarqandī known as Mullā Šāh, a second-generation disciple of Jāmī who died on his way to Mecca in Aleppo in 945/1538-9. 19 At some later date, an unnamed person requested Ibn al-Ḥanbalī to compose a commentary on the Kanz man ḥājā, which he completed in Šaʿbān 965 / May-June 1558 and titled Ġamz al-ʿayn ilā Kanz al-ʿayn (Winking With the Eye to the Golden Treasure). 20 The first nineteen verses of the poem are introductory, praising God, naming the subject of the work, and commending Jāmī and the dedicatee. Verses 20-54 are each a muʿammā exemplifying one or more of the various operations involved in solving muʿammayāt (see below), whereas verses 55-63 are each an uḥjiyya. The last three verses mark the end of the poem and express the author's wish that his friends and beloved ones pray for the forgiveness of his sins. Even though Ibn al-Ḥanbalī often expatiates on irrelevant matters, the Ġamz al-ʿayn, his extensive commentary on the Kanz man ḥājā, is a good introduction to the new muʿammā (and a useful discussion of the uḥjiyya, for that matter) and presents the various operations involved in solving muʿammayāt systematically, as was customary in Persian treatises following al-Yazdī. In the commentary to each of verses 20-54 Ibn al-Ḥanbalī discusses three issues: Firstly, he explains which operation is exemplified by the verse; secondly, he explains how the sought word is extracted, analysing the 'riddle meaning' of the verse and naming the additional operations required for its solution; and, finally, he explains the surface or 'poetic' meaning. In the commentary Ibn al-Ḥanbalī very rarely cites muʿammayāt by other authors as additional examples for the named operations. 21 The slightly later treatise Kanz al-asmā fī fann al-muʿammā (The Treasure of Names: On the Art of the Muʿammā) by Quṭbaddīn Muḥammad b. ʿAlāʾaddīn an-Nahrawālī (b. 917/1511-12, d. 26 April 990/20 May 1582), the famous Meccan religious scholar, historian and mufti of the Two Holy Cities, 22 however, serves its purpose better and was therefore more often referenced by later authors. For one thing, it is not a commentary on a poem dealing with two riddle kinds, but was conceived as a specialist treatise on the (new) muʿammā only, from the outset. Quṭbaddīn, who without a doubt was a much better poet than Ibn al-Ḥanbalī, very often cites muʿammayāt by other authors in addition to his own nice specimens and thus documents the growth of the genre in Arabic in his days better than his predecessor.

Quṭbaddīn an-Nahrawālī's Kanz al-asmā fī fann al-muʿammā
In the introductory part of the Treasure Quṭbaddīn discusses the origins and rise of this genre, his reason for composing the treatise, the sources he relied upon, and the antecedents of the new, Persianate muʿammā in the Arabic tradition. 23 He starts by noting that the muʿammā, which is one of the many fields of adab, has been invented by the Persians and has gradually become a discipline in its own right, with its rules and its specialists. In his times, Arab littérateurs had started composing muʿammayāt (of the Persian kind) in Arabic, without however knowing the rules and conditions applying to this genre. The first to do so, according to his knowledge, was his teacher Aḥmad b. Mūsā b. ʿAbdalġaffār (d. 940/1533-4), who used to encourage his pupils to compose muʿammayāt. 24 Quṭbaddīn followed his advice and collected a large number of muʿammayāt by himself, his teachers and his fellow students during his studies, but that early collection had meanwhile been lost. Therefore, he decided to write down what he could remember of his own muʿammayāt and those by his friends in order to preserve them, and to expound the rules of this art based on what he had read in the Persian specialist treatises, his aim being to exercise people's minds and hone their taste. Claiming to be the first to write about this art in Arabic, Quṭbaddīn asks for the reader's indulgence. Despite this claim, he probably knew of Ibn al-Ḥanbalī's work, since the two are known to have met in Aleppo in 965, 25 whereas Quṭbaddīn composed his Treasure after 974. 26 Interestingly, he adds that he did not deny that it would have been better for him to keep to his prayer place and spend his time on things that would profit him in the Hereafter than to busy himself with this treatise, but for the likes of him it were enough to avoid sin and dubious activities.
Thereupon he gives a definition of the muʿammā, states the conditions it has to meet and explains them. Two of these conditions must be stressed: firstly, that the sought word must be hinted at obliquely and must not be obvious. Secondly, in his opinion, the muʿammā must without fail have a second, poetic meaning, even though muʿammā specialists (the Persian masters) do not stipulate this. Otherwise it is not nice and good taste rejects it. Another very important rule he mentions is that the vocalization and the doubling of consonants-as well as the hamza-are not taken into consideration in the (new) muʿammā. To offer clues as to these features of a word is supererogatory. Quṭbaddīn also stresses that more than one word may be extracted from a single muʿammā and subsequently explains the difference between luġz and muʿammā: Luġz is to hint at a thing by mentioning its distinguishing qualities, whereas muʿammā is to hint at a specific word as an utterance (iḏā dalla ʿalā smin ḫāṣṣin bi-mulāḥaẓati kawnihī lafẓan). 27 As Quṭbaddīn adds and shows by citing an apposite example (the riddle cited above on kammūn), there are also 'mixed' riddles, riddles that are both luġz and muʿammā, in that they hint at both the letters of a word and at the properties of the thing that this word signifies.
In addition, he stresses that the rules and operations described by muʿammā experts are not aimed at restricting the ways in which to hint at a word and that any clue agreeable to 25 IBN AL-ḤANBALĪ, Durr al-ḥabab, 2:439-41. Note also the word Kanz in both works' titles. 26 974 is the death year of Ibn Ḥajar al-Haytamī, to whom Quṭbaddīn refers in the Treasure as deceased.
Hence the Treasure must have been written between 974 and 990, the year of Quṭbaddīn's death. At the end of the treatise he actually gives the date of its completion, but does so in a cryptic way that I have not been able to decipher: ḫitāmuhu fī mustahalli l-Muḥarrami ṣabīḥati s-Sabti l-mubārak; iḫtitāmuhū ṯāliṯu ṯ-ṯāmini mina l-ʿaqdi ṯ-ṯānī baʿda ikmāli l-ʿaqdi ṯ-ṯāliṯi min sinī l-hijrati š-šarīfa (the last ʿaqd should stand for 300 years); the years in which the first of Muḥarram was a Saturday were 976, 979, 984 and 987.
27 Kanz al-asmā, MS Escorial, fol. 4b, cf. al-BAĠDĀDĪ, Ḫizānat al-adab, 6:452-3; ḫāṣṣ should here mean 'specific', rather than 'proper', since there are several muʿammayāt on common names, as the examples given below show, and despite the fact that most muʿammayāt are on proper names. In Arabic the word ism may mean 'proper name' or a 'noun' as such, i.e. the signifier, as opposed to musammā, which denotes a word's meaning, the signified. But Quṭbaddīn makes this very clear by referring to lafẓ ('spoken word, vocal sound'), which is implicitly opposed to maʿnā ('meaning'), the Arabic pair lafẓ and maʿnā corresponding to the Saussurian 'signifier' and 'signified'. -For similar definitions and distinctions in the Persian tradition before and after al-Yazdī, see ANWARI-ALHOSSEYNI, Loġaz und Moʿammā, 70-88. As Anwari-Alhosseyni points out, a clear distinction between loġaz and moʿammā was first made in al-Badīʿī and al-Yazdī.
good taste can be used. It is also possible that a poet composes a muʿammā on a certain name giving specific clues about it, but one may find some other way to solve it extracting the same or a different name. One may also extract names from 'simple' poetry, poetry that has not been composed as a riddle. He illustrates this point by citing a verse by al-Wawāʾ ad-Dimašqī (d. between 370/980-1 and 390/1000) from which his friend Yaḥyā al-Lāhijī (unidentified) had extracted two names, and a verse by Ibn al-Fāriḍ (576-632/1181-1235) from which he himself had extracted four names. 28 Nevertheless, he adds, most people are unfamiliar with the operations used to solve muʿammayāt and their taste may therefore reject them. One must not follow the opinion of such people, as they are loath to learn things they do not know and only value what they know. On the other hand, there are people who, using these operations, extract names from Quranic passages. Aḥmad b. Ḥajar al-Haytamī (909-974/1503-1567), the famed religious scholar of Mecca, disapproved of this and argued with Quṭbaddīn on this matter. Quṭbaddīn mentions the arguments he had adduced to defend the legitimacy of this practice, which to him was comparable to iqtibās (integrating Quranic passages in literary texts), and stresses that the muʿammā is not a game (talāʿub), as al-Haytamī claimed, but an exercise of the mind. He also names a few examples of miraculous predictions of dates of events based on the numerical value of the words of certain Quranic passages-predictions made by respected religious figures and which he considered as inspired by God. Subsequently Quṭbaddīn briefly presents the history of the (new) muʿammā as devised and developed by the Persians and mentions by name three great masters offering summary biographies of them: According to him, al-Yazdī was the first to describe the rules of this art in his al-Ḥulal al-muṭarraz. Quṭbaddīn remarks that al-Yazdī's work was very influential and that several authors after him composed similar treatises further specifying the features and rules of the genre. The next great figure he names is Jāmī, about whose several treatises he says that they were repeatedly commented upon by later writers. As opposed to Ibn al-Ḥanbalī, who was a great admirer of Jāmī and based his poem and commentary on the latter's treatises, Quṭbaddīn praises especially Mīr Ḥusayn al-Muʿammāʾī, who in his opinion produced 'licit magic', 'an almost inimitable' treatise (kataba fīhi risālatan takādu tabluġu ḥadda l-iʿjāz), and surpassed all other authors on this subject-so much so that Jāmī declared that had he read al-Muʿammāʾī's treatise he would not have composed his own ones. Quṭbaddīn adds that the number of later Persian authors on the muʿammā is so big that to write their biographies (tarājim) would require a whole volume. Therefore he only mentions those whom he had met personally, namely a certain ʿAnbar al-Harawī with whom he had read the treatise of Mīr Ḥusayn al-Muʿammāʾī and who had read that work with the poet; a certain Muḥammad ʿIyānī al-Harawī (d. after 940/1533-4), a pupil of Jāmī who had settled in Mecca and with whom he read Jāmī's long treatise; and one ʿAbdalwahhāb an-Nīsābūrī (d. 950/1543-4), who had composed a 'superb' commentary on al-Muʿammāʾī's risāla and with whom he read that commentary during his second trip to Egypt in 945. 29 After that Quṭbaddīn revisits the issue of the origins of the muʿammā and notes that there are several Arabic riddles that fit the definition of the muʿammā as given by the Persians. In his view, the Persians did not really invent it but ordered, systematised, refined and added subtleties to it, whereas the Arabs used to cast it in the mould of luġz. One should keep in mind that the Persian masters did not stipulate that a muʿammā must without fail have a second meaning, even though this was typically the case in Persian muʿammayāt as well. For Ibn al-Ḥanbalī, an-Nahrawālī and later Arabic authors, however, this was obligatory. Thus, all muʿammayāt that do not have a second meaning and can immediately be recognized as riddles, that is, make clear that a word must be sought in them-as Ibn Bābak's muʿammā on the name Saʿīd cited above-, were perceived by Quṭbaddīn and later Arabic authors as alġāz. Nevertheless, according to Quṭbaddīn, there were some rare logogriphs in the earlier Arabic tradition that fitted exactly the description of the (new) muʿammā, but the few examples he cites do not really prove this. He is however right in noting that several of the operations involved in the new muʿammā were earlier used by the Arabs in their old-style muʿammayāt, for which point he cites some apposite examples. 30 By saying that the Persians 'recorded, ordered, arranged and diversified' this art (dawwanūhu wa-rattabūhu wa-bawwabūhu wa-fannanūhu), Quṭbaddīn apparently means to say that they set and described the rules and, most importantly, the operations involved in solving muʿammayāt. Perhaps more than the existence of two different levels of meaning in the poem-something that the Persians did not deem mandatory but which for Arab authors became the distinguishing trait between luġz and muʿammā-it was the systematic exposition of these operations by the Persian masters, starting with al-Yazdī, that intrigued Quṭbaddīn and other Arabic and Turkish authors to the extent that they ascribed its invention to the Persians. 31 The subsequent, main part of Quṭbaddīn's Treasure is devoted to the systematic exposition of the muʿammā operations in Arabic following al-Yazdī's system. After explaining what each operation does, Quṭbaddīn cites a few muʿammayāt by himself and other contemporary authors to exemplify its use. In total, the treatise comprises seventy-one muʿammayāt by ten different persons. 32 me). These Arabic works offer further evidence of the popularity of the new muʿammā and deserve closer study.
30 Pace Quṭbaddīn no riddles by Ibn al-Fāriḍ fit the definition of the new muʿammā. 31 To judge by al-Muḥibbī's remarks on the origins of the (new) muʿammā in his note on the genre in Ḫulāṣat al-aṯar, 2:390-3, the issue was much debated. Al-Muḥibbī reports that he once had an argument with somebody in 'Rūm' (Rumelia and Anatolia) who asserted that the muʿammā was invented by the Persians and the Rūm-Turks (Arwām) and that the Arabs did not know it. To counter him, al-Muḥibbī cited several muʿammayāt in Arabic as well as a passage by Ibn Qutayba (sic! -he does not cite this passage in the Ḫulāṣa) that stated that the Arabs had invented it and the Persians and the Rūm-Turks took it from them-for all three words, muʿammā, uḥjiyya and luġz, are Arabic. Al-Muḥibbī concludes that actually the Persians and the Rūm-Turks contributed to it so much that they excelled. It is worth noting that, in the same note, al-Muḥibbī stresses that a muʿammā must have a second, 'poetic' meaning, this being the reason why it is counted among the funūn and the muḥassināt of the badīʿ. Obviously, to him as to later Arab authors a muʿammā was solely the new, Persianate muʿammā.
Persian, Turkish and Arabic authors adhering to al-Yazdī's system in discussing the various kinds of clues and hints used in the new muʿammā divide these into three basic categories dubbed 'operations': al-ʿamal at-taḥṣīlī ('productive operation'); al-ʿamal attakmīlī ('perfecting operation'); and al-ʿamal at-tashīlī ('facilitating operation'). Each of the three categories actually subsumes several subcategories; therefore, in what follows, I shall be speaking of 'productive operations', 'perfecting operations' etc. Al-ʿamal at-taḥṣīlī encompasses eight operations that are used to solely extract the letters of the sought word, irrespective of their order. The order of the letters is deduced by means of al-ʿamal attakmīlī, which includes three operations. Al-ʿamal at-tashīlī subsumes four operations that, as their name suggests, are meant to facilitate and support the productive and perfecting operations by offering additional clues regarding their implementation. Finally, there is a fourth operation, al-ʿamal at-taḏyīlī ('accessory operation'), which is just one operation and refers to supererogatory clues concerning the vocalization of the sought word.
It is important to stress that, as Quṭbaddīn and most other authors remark, in every muʿammā there are normally several-not just one-operations at work. 33 As a result, as one works through the list of the operations presented, one must also comment on the additional operations involved in the muʿammayāt that are cited to exemplify the given operation. 34 This complicates the presentation of the system. Besides, one may object to subsuming this or that operation to one of the first three categories or to the necessity of certain operations that require or overlap with others. 35 To be sure, one can explain how a muʿammā is solved, namely how the sought word is extracted from it, more simply. Discussing the muʿammā in his Grammatik, Poetik und Rhetorik der Perser, Rückert, for example, dispensed with presenting this elaborate and complex system of operations and explained how the cited muʿammayāt are solved avoiding unnecessary terminology and categorization. Moreover, as shown by Anwari-Alhosseini, there were Persian authors who did not adhere to al-Yazdī's system and did not distinguish four categories of operations. These authors list and discuss more or less the same operations and various subcategories of them but do not bother to group and hierarchise them. 36 But apparently al-Yazdī's system served well its purpose, which was twofold: to assist muʿammayāt authors in composing them and to help their recipients and the general public to solve them. In what follows I present this system based on Quṭbaddīn's presentation and cite two to three muʿammayāt for each operation, a selection of the muʿammayāt cited by Quṭbaddīn.
33 Kanz al-asmā, MS Escorial, fol. 11a; cf., e.g., IBN AL-ḤANBALĪ, Ġamz al-ʿayn, MS Munich, fol. 9b. 34 Given that I follow Quṭbaddīn's presentation, the same is true of the following presentation as well. 35 E.g., as will become clear below, the operation termed tasmiya presupposes and therefore partly overlaps with the operation termed tarāduf, whereas the operation called taḫṣīṣ obviously stands for several other operations. 36 See ANWARI-ALHOSSEINI, Loġaz und Moʿammā, 137-64, cf. 91-100. -It should be added that, even among the Persian authors who adher to al-Yazdī's system, there are minor differences in the way they present the fifteen operations and especially various subcategories of them. The same is true of Arabic authors.

A1. aṭ-Tanṣīṣ wa-t-taḫṣīṣ (Citation and Specification)
According to the common definition of this operation, tanṣīṣ (citation) is to cite the sought letters in the poem and taḫṣīṣ (specification) is to deal with them in one of the various acceptable ways in order to specify them. To judge by the given examples, however, tanṣīṣ seems to refer specifically to citations of segments of the sought word (as opposed to single letters), whereas taḫṣīṣ refers to practically any instructions offered as to how to handle these segments in order to solve the riddle, including the various operations described in the following pages.
Abū Bakr Muḥammad b. ʿUmar al-Yutayyim on Ismāʿīl (metre: ṭawīl) 37 My lord, you've become a devotee of Muhammad's family-may you attain glory, power and highness! Mention them to me again without haste, for, to me, listening (samāʿī) to (the mention of) the family (āl) of him whose intercession will be accepted on Judgment Day is sweet.
Quṭbaddīn an-Nahrawālī on Qāsim (metre: rajaz) 38 I enjoyed a period of time (lit. 'I donned the garb of life') with a full-moonlike beloved who used to keep his promises.
But now, after he is gone (lit. 'after him'), I don a garb of sickness.
Saqām, 'sickness' (written sīn-qāf-alif-mīm), is citation (tanṣīṣ) of the sought letters. At the riddle level, the garb of saqām are its first and last letters, i.e. the sīn and the mīm, that encompass as it were the word's core (see facilitating operation C1, intiqād, which is to hint at the position of specific letters within a word found in the poem). Putting sīn and mīm after the rest of its letters, i.e. after qāf-alif ('the first and last letters of saqām after it' being the taḫṣīṣ), yields Qāsim (qāf-alif-sīn-mīm).

A2. at-Tasmiya (Naming)
This is to mention the name of a letter and mean the letter itself or vice versa to mention a letter and mean its name. At the surface level, a letter's name is not used as such but bears some other meaning it may have as a word. It may also be a segment of a word found in the poem, as in the following example: O you who blames me for loving him (= a third person)-you overdid it, cut short your reproof! You've racked the heart of a lover whose beloved is absent from him.
But as soon as he sees him, he gets together with him and recovers his health.
At the riddle level, the phrase talāqā fa-ṭābā ('he gets together (with him) and recovers his health') can be analysed as talā qāfa ṭā bā, namely, '[the letters] ṭā [ʾ] and bā [ʾ] follow [the letter] qāf'. Placing ṭāʾ and bāʾ after qāf yields Quṭb (qāf-ṭāʾ-bāʾ). This reading of the text involves both uniting and dividing the speech flow differently than at the surface level and the written form. Uniting and dividing the speech flow differently is what they call synthesis (tarkīb) and analysis (taḥlīl)-the facilitating operations C2 and C3 (see below).
Abū Bakr Muḥammad b. ʿUmar al-Yutayyim on Isḥāq (metre: ḫafīf) 40 A Lord who provides for those who are in His care warrants sufficiency for His creatures.

A3. at-Tarāduf wa-l-Ištirāk (Synonymy and Homonymy)
Tarāduf (synonymy) is to use a word in the poem hinting at a synonym of it in Arabic, Persian or Turkish to be used in solving the riddle. For example, when you mention the Arabic word māʾ, 'water', and mean āb, the Persian word for it, or when you say laysa, 'is not', and mean the negative particle mā. Ištirāk (homonymy) is when you use a polysemous word, its meaning at the surface level being different from its meaning at the riddle level, as in the above example for tasmiya, where the word Qāf is a toponym at the surface level but stands for the letter qāf in the riddle. Or as in the following example, where the word qalb has two different meanings: My cupbearers, who serve me from the very best wine, I am thirsty. So keep filling my cup!
And have mercy on a desperate, destitute and distressed man, whose heart cheers up over the water of life! At the riddle level, qalb, 'heart', means 'inversion'-the inversion of all or some letters of a given word is a very common perfecting operation (see below B3). Inverting the word hašša, 'it cheers up' (written hāʾ-šīn), yields šīn-hāʾ; added to ('next to', ʿinda) 'water' in Persian, āb (written alif-bāʾ), this yields Šihāb (šīn-hāʾ-alif-bāʾ).

A4. al-Kināya (Indirect Expression)
This is to use a word or expression and mean some other word or expression conveying the same meaning figuratively or obliquely, as when one says, for example, 'the brightest star' meaning the sun, or 'the contrary of lesser sins' meaning 'the grave sins' and suchlike. As Quṭbaddīn rightly stresses and as the following examples show, the possibilities one has for oblique expression are countless.
Quṭbaddīn an-Nahrawālī on Ḥusām (dūbayt) 43 The manager of this hammam is a bath-keeper twin to slender twigs, a fair gazelle.
His body is so soft that it flows like water. How delicate is this person in the hammam! As Quṭbaddīn explains, a person, insān (written alif-nūn-sīn-alif-nūn), entering the hammam (written ḥāʾ-mīm-alif-mīm), takes off his turban and clothes. This means that the first letter of insān, alif, its 'turban' (see facilitating operation C1, intiqād), and the two nūns that surround the central part of the word like a shirt, must be dropped, yielding sīnalif. In this case, kināya is the oblique reference to the dropping of these letters (an operation which is otherwise called isqāṭ, see below B2). Sīn-alif must then be inserted in the word ḥammām ('this person in the hammam'), whose water, mā[ʾ], written mīm-alif, must be shed, because water is shed in the hammam, as Quṭbaddīn explains. Inserting sīnalif between ḥāʾ and mīm (what is left when mīm and alif are dropped from the word ḥammām) yields Ḥusām (ḥāʾ-sīn-alif-mīm).
Quṭbaddīn an-Nahrawālī on Jamāl (metre: muḫallaʿ al-basīṭ) 45 Do not wonder at the leanness of my body when the one I love has been rude to me.
Even if he were a mountain, the heart of him who suffers what I suffer would melt.

A5. at-Taṣḥīf (Miswriting)
In the new muʿammā, taṣḥīf is restricted to adding, omitting or changing the position of the diacritical dots over or under a given letter, as opposed to taṣḥīf in the earlier Arabic tradition, which meant any kind of miswriting, e.g., changing lām into kāf, or final ʿayn into final jīm or ḥāʾ, or even making sīn into three letters with a single nabra (tooth) each, etc. The diacritical dots may be hinted at by words denoting things similar to them visually or metaphorically, like tears, drops, points, moles, decoration, ornaments, etc., whereas the context indicates the necessary change in their number and/or position. This is termed taṣḥīf jaʿlī ('artificial(?) taṣḥīf'). This operation may also be hinted at by words or particles denoting similarity, for example ka-, miṯl, etc., meaning 'similar to' such and such a word. This is termed taṣḥīf waḍʿī ('conventional(?) taṣḥīf').
Muḥammad aṭ-Ṭayr al-Yamanī on ʿUmar (metre: ḫafīf) 46 A youth with captivating, languid eyes has shot at me with a shot (i.e. cast at me a captivating glance), He allured me without having any ornaments on-imagine how he would affect me, were he to wear a belt! Ġarra, 'he allured' (written ġayn-rāʾ) min ġayri ḥilyatin, 'without any ornaments', means to delete the dot of ġayn, which yields ʿayn-rāʾ. Because of its shape, the letter mīm is often likened to a belt, minṭaqa (operation A7, comparison or metaphor). As such, in the present context, it should be placed in the middle of the segment ʿayn-rāʾ, which corresponds to the young man's waist, yielding ʿUmar (ʿayn-mīm-rāʾ).
Quṭbaddīn an-Nahrawālī on Ḥabīb (metre: muḫallaʿ al-basīṭ) 47 The stature and buttocks of my beloved are (like) a bending twig on a hillock. Because of its shape, the letter ḥāʾ is often likened to a lock of hair or a bending twig (operation A7, comparison or metaphor). Ḥāʾ must be placed 'on top of', i.e. in front of kaṯīb. The word kaṯīb, 'hillock', is analysed as consisting of two words (operation C2, analysis), the preposition ka-, 'like', and the grapheme ṯīb, meaning 'what resembles (in writing) ṯīb', i.e. the miswriting of the word segment ṯīb (written ṯāʾ-yāʾ-bāʾ)', namely, bīb (written bāʾ-yāʾ-bāʾ). For both ṯāʾ and bāʾ consist of a single nabra (tooth) but differ in the number and the position of the diacritical dots. Putting ḥāʾ in front of bāʾ-yāʾ-bāʾ yields Ḥabīb (ḥāʾ-bāʾ-yāʾ-bāʾ).

A6. at-Talmīḥ (Allusion /Hint)
This is to allude to widely known things and to conventions of the various sciences: for example, the astronomers' convention to name the planets after the final letter of their name in Arabic (qamar, 'moon', hints at rāʾ, šams, 'sun', at sīn, ʿutārid, 'Mercury', at dāl, etc.). Likewise, the word al-muʿarrif may be used to allude to the letters alif-lām, given that in the grammarians' terminology it means the definite article al-(in the poem, however, this word may have any of the possible meanings of the active participle of the verb ʿarrafa).
Quṭbaddīn an-Nahrawālī on ʿĪsā (metre: kāmil) 48 She covered her eyebrows with a protecting hand. Look at her two eyes without the eyebrows! As Quṭbaddīn explains, by her 'two eyes' he means the word ʿayn and, further, the word šams, given that ʿayn is a synonym of it (operation A3, ištirāk, homonymy, ʿayn may mean 'sun'). The word šams hints at the letter sīn, which is used by astronomers to indicate the sun (šams); here the poet means to indicate the letter's name, the word sīn itself. The name ʿĪsā (ʿayn-yāʾ-sīn-yāʾ) is produced by uniting the words ʿayn (ʿayn-yāʾ-nūn) and sīn (sīn-yāʾnūn), having dropped the nūns or 'eyebrows' at their end-since due to its shape the letter nūn is often likened to an eyebrow (productive operation A7, comparison or metaphor, and perfecting operation B2, elimination).
Raḍiyyaddīn al-Qāzānī on Ṣāliḥ (metre: ḫafīf) 49 Your love has killed your slave. If you reject him, you will ruin him.
Then, after he perishes because of you, the legal witness will rightly identify (you) as the culprit at court.
In legal terminology, muʿarrif is a special witness identifying the persons involved in a court case. As a grammatical term, however, it denotes, as said, the definite article al-, written alif-lām. At the surface level, ṣaḥḥa fīhi l-muʿarrifu means 'the witness will give sound testimony concerning him (the slain slave)', but at the riddle level, it means 'aliflām [must be inserted] in ṣaḥḥa (written ṣād-ḥāʾ)' (perfective operation B1, taʾlīf, composition). Inserting alif-lām between the letters ṣād and ḥāʾ yields Ṣāliḥ (ṣād-aliflām-ḥāʾ).
Muḥibbaddīn b. Mullā Ḥājjī on Masīḥ (metre: muḫallaʿ al-basīṭ) 50 May I be the ransom of the one whose love is deeply rooted and anchored in my heart and my heart-blood! I seek his mouth and his ear lock, with which he has turned my temples and my head grey.
A handsome youth whom I love and who loves me-a little belt is tied around his supple waist.

A8. al-ʿAmal al-Ḥisābī (Computations)
This involves several kinds of clues and computational operations based on the numerical value of the letters of the alphabet.
Quṭbaddīn an-Nahrawālī on Šams and Kamāl (metre: ṭawīl) 52 The censurer says: "The sun is a dearer being than the person whom you love", but he is wrong in this.
I only see the face of my beloved once a year, but what he named is obtained daily.
At the surface level Quṭbaddīn plays with the meanings of aʿazz. The censurer refers to the preciousness of the sun, while the lover supposedly understands the word to refer to its 'dearness', the rareness of its appearance. At the riddle level, wajh man, lit. 'the face of man (mīm-nūn)', means the first letter of the word man, i.e. mīm (facilitating operation C1, intiqād). ʿĀm, 'a year', stands for the number 360, the number of days of the 'natural' solar year. 360 equals 300 + 60, the numerical value of the letters šīn and sīn respectively. Wajh man fī l-ʿām means that mīm must be inserted between šīn and sīn, which yields Šams (šīnmīm-sīn). Inserting mā (mīm-alif) between kāf and lām, the letters of the word kull (mā fī kull, 'mīm-alif in kāf-lām'), yields Kamāl (kāf-mīm-alif-lām).

B1. at-Taʾlīf (Assembling/Composition)
This is to bring together the various segments of the sought word that are mentioned in various places of the riddle and either add one to another (taʾlīf ittiṣālī) or insert it into it (taʾlīf imtizājī).
How noble is he, pious and devout! You always see him revered among people.
He joyfully throws himself prostrate before the Lord, so that you see the lines of his face touch the soil. This is an example of taʾlīf ittiṣālī. The face (wajh) of a word is its first letter (facilitating operation C1, intiqād). Here the face refers to the word bišran ('joyfully, with joy'), written bāʾ-šīn-rāʾ-alif. Bāʾ touches the soil, i.e. it must be placed at the very end of the word, thus yielding šarāb (šīn-rāʾ-alif-bāʾ). This is because, in the new muʿammā, letters standing at the beginning of a word are referred to by various words indicating things placed high, whereas letters standing at the end are referred to by words indicating things placed low (see below C1, intiqād).
Muḥibbaddīn b. Mullā Ḥājjī on Jawhar (metre: wāfir) 56 A handsome youth avoids me and puts me off more and more often, but my heart cannot resist him.
I have discarded love from my existence because of his forsaking me that tore apart what my ribs enclose.
Slender twig, spare me the deferrals and the excuses so long as you love me (lit. your love of me lasts), my hope!

B3. al-Qalb (Inversion)
This is to invert all or some of the letters of words occurring in the poem as part of the operations required to solve the riddle. The inversion is indicated by words like qalb ('inversion'), ʿaks ('reversal'), dawr ('turning, round'), fatl ('twisting'), dāra ('to turn', 'to revolve'), etc. 58 Quṭbaddīn an-Nahrawālī on Aḥmad (metre: wāfir) 59 If the wine cup goes round (for other people), for us blood goes round on spear-tips.
Abū Bakr Muḥammad b. ʿUmar al-Yutayyim on Rustam (metre: mujtaṯṯ) 60 I only disclose my secret to him who kindly bonds with me.
How could it be otherwise? The secret is in my heart and makes my bond complete.

C1. al-Intiqād (Selection, Picking Out)
This is to hint at letters that occur in the poem and contribute to the solution of the riddle by indicating their place in the words in which they occur. The first letter of a word is indicated by words meaning 'the beginning, the front, the face, the first, the head, the crown, the uppermost, the pinnacle, the froth', etc. Words meaning 'the end, the edge, the back, the rear, the last, the foot, the sediment, the lowest, the tail', etc., indicate the last letter of a word. Words meaning 'the heart, the guts, the belly, the filling, the middle, the core, the innermost, the pith, the seed, the heart', etc., indicate the middle letter. The first and last letters of a word may be referred to by words meaning 'the two sides', 'the two wings ', 'shell', (aṭ-ṭarafān, al-janāḥān, qišr), etc.
Abū Bakr Muḥammad b. ʿUmar al-Yutayyim on ḥammāma, 'dove' (metre: muḫallaʿ albasīṭ) I recall a party (as lovely) as a garden-I plucked a rose there (i.e. I kissed a cheek) that greeted me with cheer.
Abū Bakr (b. Sālim) al-Qunāwī al-Makkī on šorba, a Turkish loanword meaning 'soup' (metre: sarīʿ) 63 How often have I told the censurer, when he reproved me because of a full moon abiding in my heart: "Explaining my passion has no end and my love of him has no beginning"! Dropping the ḥāʾ at the end of the word šarḥ, 'explaining' (šīn-rāʾ-ḥāʾ), and the ḥāʾ at the beginning of ḥubbuhū, 'the love of him' (ḥāʾ-bāʾ-hāʾ) and uniting the two remaining segments yields šorba (šīn-rāʾ-bāʾ-hāʾ).

C2. at-Taḥlīl (Analysis)
As in some of the above examples, both this and the next operation (tarkīb, synthesis) involve analysing the text at the riddle level in a different way than at the level of the surface meaning of the poem. Taḥlīl is to read as two or more words what at the surface level of the text is one word. Tarkīb is the opposite, namely, to read as one word what at the surface level of the text is two or more words. The two operations are mostly combined. Quṭbaddīn an-Nahrawālī on Nūr (metre: wāfir) 64 Do not fret over the times when the times are grim, because mild times will follow.
Quṭbaddīn an-Nahrawālī on Nāṣir (metre: ṭawīl) 65 We took patience and as Patience saw our courage, it fell behind us brokenhearted.
'Broken-hearted' ṣabr, 'patience', is ṣād-bāʾ-rāʾ without bāʾ, its 'heart' or middle letter, namely ṣād-rāʾ. At the riddle level taʾaḫḫara ʿannā ('it fell behind us') is analysed as taʾaḫḫara ʿan nā, 'falls behind (i.e. must be put behind) nā (nūn-alif)'. Ṣād-rāʾ placed after nūn-alif yields Nāṣir (nūn-alif-ṣād-rāʾ). You took away my heart-blood-but how often have you, too, suffered your innermost to be taken away from you as I did! At the riddle level, the possessive suffix in aḥšāhu ('his innermost') is understood to refer to the word jaluda (jīm-lām-dāl) (at the surface level it refers to the beloved). If its 'innermost', i.e. lām, its middle letter, is 'taken away' (maʾḫūḏah), there remain jīm and dāl. Likewise, the suffix in ʿalayhi, which also refers to the beloved at the surface level of the poem, refers to what remains from jaluda after the elimination of lām. In the context of the muʿammā, ʿalayhi means 'on top of it', that is, 'in front of it'. Analysing maʾḫūḏah as two words, mā, mīm-alif (hamza is not taken into account), and ḫūḏah, mā is construed as the subject of a nominal clause, namely ʿalayhi mā, 'mā is (i.e. must be placed) in front of it'. Mīm-alif placed in front of jīm-dāl yields Mājid (mīm-alif-jīm-dāl). Quṭbaddīn praises the poet for using the word māḫūḏah twice in the clues and comments that this is very subtle (huwa fī ġāyati l-luṭfi).

C3. at-Tarkīb (Synthesis)
Quṭbaddīn an-Nahrawālī on ʿĀʾiša (metre: wāfir) 67 A matured wine (looking) like a sun in a crescent moon (= the goblet)-the bubble stars on its surface resemble pearl necklaces.
Take the sun and leave 'no!' and you will gain, along with the old wine, new joy! Šams in the second verse stands for ʿayn, its synonym (operation A3, tarāduf), which hints at the letter ʿayn. By 'leave 'no!'' Quṭbaddīn hints at its opposite, namely ī, 'yea' (written alif-yāʾ), meaning 'take ʿayn and take alif-yāʾ', ʿayn-alif-yāʾ being the first three letters of the sought name. Applying the facilitating operations of tarkīb and taḥlīl, the phrase šarābi ʿatīqihim ('their old wine') should be read as ša-rābiʿa tīqihim. This indicates two things: firstly, that 'šīn is the fourth [letter]' of the sought name (š-rābiʿ); and, secondly, read together with the word ḫuḏ that stands at the beginning of the verse and ignoring the šīn, i.e. read as ḫuḏ rābiʿa tīqihim, it means 'take the fourth [letter] of the segment tīqihim (tāʾ-yāʾ-qāf-hāʾ-mīm), which is a hāʾ, the last letter of the name ʿĀʾiša (ʿayn-alif-yāʾ-šīn-hāʾ).
Quṭbaddīn an-Nahrawālī on Ḫalīl (metre: mujtaṯṯ) 68 I fell in love with his forehead, which shines like the crescent moon.
My body grew thin like a toothpick for love of my crescent moon.

C4. at-Tabdīl (Substitution)
This is to substitute one or more letters of a word with some other letter or letters.
Abū Bakr Muḥammad b. ʿUmar al-Yutayyim on ʿImād and ʿĀbid (metre: basīṭ) 69 I swear by the mouth that tastes like honey and honeycomb and swear by a rose on top of a blushed cheek- The heart of your slave is not out of stone, master, so that you make it bear the unbearable! At the riddle level, the phrase mā qalbu ʿabdika ('the heart of your slave is not') must be understood as meaning 'mā (mīm-alif) is the heart, i.e. the core or middle letters, of the word ʿabd (ʿayn-bāʾ-dāl)', namely, mīm-alif must take the place of bāʾ, thus yielding ʿImād (ʿayn-mīm-alif-dāl). If the same is done with āb (alif-bāʾ), the Persian word for water and a synonym of mā [ʾ] in this sense, the name produced is ʿĀbid (ʿayn-alif-bāʾ-dāl).

D. al-ʿAmal at-Taḏyīlī or Accessory Operation
Given that the vocalization and the doubling of consonants are not taken into account in the new muʿammā, clues relating to these features count as supererogatory. They are indeed very rare and Quṭbaddīn cites only one muʿammā in which such clues are given.
Quṭbaddīn an-Nahrawālī on Hilāl or hilāl, 'crescent moon' (metre: ḫafīf) 70 You have accused your lovers of trespassing against you, so that they seek another love.
You have broken their hearts, so that they change. Have you seen any change by breaking them?
For the purposes of the riddle, the word raʾaytum in the last hemistich is redundant. The clues to the solution are given in the phrase hal bi-kasrihā taḥwīlā, which must be analysed as hal bi-kasri hā taḥwī lā, namely 'hal (hāʾ-lām), the hāʾ being vocalized with kasra, encompasses lā (lām-alif)'. In other words, the subject of taḥwī is hal, hāʾ-lām, about which segment is said that it must encompass lām-alif, yielding hāʾ-lām-alif-lām. This is how the word hilāl is written. The hā in bi-kasri-hā is the letter hāʾ of hal, which must be vocalized with kasra, this being the supererogatory clue indicating the vocalization.
I hope that the above sample gives the reader an idea of the complexity as well as of the wittiness and beauty of the new genre-traits that justify the fondness of Ottoman-era Arab intellectuals for the Persianate muʿammā. To appreciate and enjoy this poetry, one has to familiarize oneself with the various kinds of clues offered in the poems. To be sure, in several cases, some steps in the solution procedure as described by Quṭbaddīn may seem arbitrary to new initiates. The deeper one delves into this world, however, the more one admires the poets' skill and ingenuity in creating poems with an obvious and a hidden meaning. Various later sources-treatises on the (new) muʿammā or the riddle in general, dīwāns, anthologies, as well as other literary works-testify to the great popularity that this genre enjoyed throughout the Ottoman period. 71 It is very significant that, starting with Ibn al-Ḥanbalī's Kanz man ḥājā and Ġamz al-ʿayn, specialist treatises on the riddle in general (or on more than one riddle kind, as Ibn al-Ḥanbalī's works) treat the muʿammā first and devote to it noticeably more space and attention. This fact suggests that it gradually became the most popular kind of literary riddle in Arabic. Another indication of the genre's growth are the rising numbers of authors whose muʿammayāt are cited in specialist treatises and anthologies over time. Nonetheless, given that the Arabic literature of the Ottoman period has been very little studied and most sources are still unpublished, a more thorough examination of the relevant material must be undertaken in order to show the efflorescence of the genre in its full extent. In the following Appendix, I list those specialist treatises, both extant and lost, which I know of. I have not been able to examine all the extant works and therefore comment only on the contents of those available to me. The list is most probably not exhaustive; additional titles will surface in the future, as our knowledge of the Arabic literature of the Ottoman period improves. Finally, it is important to stress that, as becomes clear from the identity of the authors of the works listed below, due to its complexity the new muʿammā was a genre that attracted elite writers in particular. ʿamal al-muʿammayāt wa-l-alġāz, a copy of which survives in the Maktabat al-Awqāf al-ʿĀmma in Baghdad (not seen by me). 78 Two muʿammayāt of his are cited in al-Bakrajī (no. 5).
3 -A seven-folio Risāla fī ʿamal al-muʿammayāt written in 1016/1607-8 by a disciple of one Waḥīdaddīn ʿAbdarraḥmān b. Bākīr survives in the Maktabat al-Awqāf al-ʿĀmma in Baghdad (not seen by me). 79 4 -Ṣalāḥaddīn b. Muḥammad al-Kūrānī (d. after 1049/1639), 80 an Aleppine littérateur who earned his living as a court secretary and, as his dīwān shows, took a vivid interest in the riddle in general, is the author of a twenty-folio treatise entitled Nūr maṣābiḥ ad-dayājī fī l-muʿammā wa-l-aḥājī (The Light of the Nightly Stars: On Muʿammayāt and Aḥājī), a copy of which survives in the Ẓāhiriyya Library in Damascus (not seen by me). The treatise, which first deals with the muʿammā and then with the uḥjiyya, contains 103 muʿammayāt by al-Kūrānī that exemplify the various operations. 81 This treatise is referenced a few times in al-Bakrajī.
The first part of the work, which is devoted to the (new) muʿammā (pp. 4,4-40,13), is divided into four sections preceded by an introduction. The introduction and the first three sections-one for each category of operations (al-ʿamal at-taḥṣīlī, al-ʿamal at-takmīlī, al-ʿamal at-tashīlī)-rely heavily on an-Nahrawālī and Ibn al-Ḥanbalī, whose treatises al-Bakrajī praises highly. Al-Bakrajī illustrates the various operations by citing one or two examples drawn from these works, but, as he notes in the introduction, he also knew the works by Ibn al-Bakkāʾ and Ṣalāḥaddīn al-Kūrānī as well as by others that he does not name, from which he draws muʿammayāt that he cites in the anthology (ḫātima). In the last section of the first part (pp. 22,11-40,13) he cites and explains sixty-two muʿammayāt of his own composition, arranged in alphabetic order according to the sought name.
The section of the anthology that is devoted to the (new) muʿammā comprises three subsections: the first cites and explains forty-one muʿammayāt in Arabic, the second eleven muʿammayāt in Persian, and the third forty-two muʿammayāt in Ottoman Turkish, ten of which by al-Bakrajī himself. 85 The section of the anthology that is devoted to alġāz comprises two subsections, one on Arabic and one on Turkish alġāz (the last three by al-Bakrajī).
As in Ibn al-Ḥanbalī (and al-Kūrānī), here too, the (new) muʿammā is given more space and attention. The forty-one Arabic muʿammayāt cited and explained in the first section of the anthology are by twenty-five different authors, which, too, shows how popular the genre had become since an-Nahrawālī's times. It is also very important that, as becomes clear from the poems cited in the section of the anthology on alġāz, old-style muʿammayāt-logogriphs that state more or less clearly that a name is sought and lacking a second, poetic meaning beyond their meaning as riddles-were considered as alġāz. Al-Bakrajī would classify the muʿammā of Ibn Bābak on the name Saʿīd which is cited at the beginning of this paper and which al-Ḥaẓīrī had expressly labelled thus, as luġz. Another very important point is al-Bakrajī's citation of Persian and Turkish muʿammayāt as well as Turkish alġāz. The knowledge of all three languages and the ability to compose in more than one was not uncommon among Ottoman-era Arab intellectuals. In fact, the spread of the new muʿammā from Persian into Turkish and Arabic is a result of the increased contacts and exchanges among these literatures during the same period. 86 85 Al-BAKRAJĪ possibly wanted to add a few more muʿammayāt in Persian, because there are two pages left blank between the Persian and the Turkish subsection. A page is left blank at the end of the Turkish subsection as well. 86 Another genre that the Arabs adopted from the Persians in early Ottoman times was the chronogram (taʾrīḫ), on which see T. BAUER, "Vom Sinn der Zeit: Aus der Geschichte des arabischen Chronogramms".
ʿamal at-tashīlī-in that order. Al-ʿamal at-taḏyīlī is mentioned briefly at the end of the third section. Al-Jazāʾirī gives numerous examples for each operation, several of which are of his own composition, while others are drawn from the treatises of an-Nahrawālī and Ibn al-Bakkāʾ and presumably other works that he does not name. In sum, the three sections comprise 133 muʿammayāt by seventeen authors (these include al-Jazāʾirī, Ibn al-Bakkāʾ, five authors cited in an-Nahrawālī, four authors cited in al-Bakrajī and six additional authors). Section Three is followed by six lengthy remarks (tanbīhāt), three of which concern the muʿammā (one should not compose too difficult muʿammayāt, so as not to discourage others from engaging in this art; on the possibility of extracting names from texts that have not been composed as muʿammayāt; every littérateur worth his name should know and engage in this art, but one should not despise those who ignore it). Two remarks concern complicated chronograms (taʾrīḫ) and dating events in puzzling ways, whereas the last one concerns a Persian work of admonitions that makes extensive use of puns, of which al-Jazāʾirī translates some examples. The fourth section of the book is devoted to alġāz and is divided into two subsections: the first is on 'notional riddles' (alġāz maʿnawiyya) and the second on 'word riddles' (alġāz lafẓiyya or ismiyya), which are what I call 'old-style muʿammayāt'. Al-Jazāʾirī remarks that 'notional riddles' refer to the properties of the encoded notion, whereas 'word riddles' refer to the letters of a given word. 'Professional riddles' (alġāz fanniyya), i.e. grammatical and juridical riddles, fall under the first category. It is further worth noting that several of the examples that al-Jazāʾirī cites in the second subsection are 'mixed riddles'. Finally, the last part of the book (ḫātima) contains ten 'useful lessons' (fawāʾid) on aḥājī, 'professional riddles', and some other issues traditionally discussed in Arabic works on the riddle.