A major concern of 20th-century Hafez scholarship has been the establishment of a reliable text of his poems. While the texts of other poets have also been the subject of critical inquiry and debate, the case of Hafez seems exceptional. This may be partly because his relatively small poetic output appears more susceptible to control than a large divān or a lengthy narrative maṯnawi, partly because many manuscripts exist that were produced close to the poet’s own time, and partly because the density of Hafez’s poetic language inspires a demand for accurate readings; but it also reflects the poet’s iconic status as a symbol of Persian cultural and literary identity.

Hafez’s poems are found in numerous manuscript sources, ranging from anthologies and other works to redactions bearing the title divān. It is estimated that there are at least 1,000 known manuscripts of the Divān in Iran and other parts of the world, and perhaps two or three times that many that are as yet unknown (Matini, p. 600); there is no doubt that many have disappeared. Rather than simplifying the task of establishing a textus receptus, this proliferation of sources has only served to complicate it. Since a discussion of the manuscript tradition as a whole would be an impossible task, this article will rather address some of the issues that have preoccupied scholars and critics with respect to establishing the text of Hafez’s Divān.

It is, by and large, only the ghazals that have received sig-nificant attention; Hafez’s poems in other forms are generally regarded as marginal (cf. Neysāri, p. 206). Efforts to establish a reliable text of the ghazals focus largely on four main issues: the number of ghazals in the Divān; the number of verses in each ghazal; the correct order of verses; and the correct reading of each verse (Neysāri, p. 210). Underlying these issues is the assumption that there is an “original” Divān, an authentic, authorial (or at least authoritative) redaction, that can be recovered, so that the Divān can be reconstructed in the form it would have taken had Hafez himself compiled it (cf. Neysāri, p. 214).

Historically, lyric poems (qaṣidas, ghazals, and so on) were transmitted in a variety of ways, and were not routinely collected in divāns (see DIVĀN iii de Bruijn; Flemming, esp. pp. 8-9; Lewis, pp. 295-97). This is especially true of the ghazal, which only acquired major importance from the 12th century onwards, and was transmitted primarily in oral form (de Bruijn, pp. 27-28; Flemming, p. 9). A few early poets compiled their own divāns; others refer in their poetry to a divān; but poets continually revised their poems, and the term divān simply indicates a selection of poems compiled for a patron, for circulation among friends, or for other purposes (including, perhaps, self-advertisement). A poet’s work might be compiled, during his lifetime or posthumously, by someone else (again, usually for some specific purpose); but there was no standard procedure for publishing a poet’s work in written form.

Hafez is said to have edited his Divān in 770/1368; but there is no evidence to support this, and in any case it would not have been a complete text (Wickens, p. 56). The “Golandām preface” found in many manuscripts of the Divān (the authenticity of which was once disputed, but which can now be accepted; see Hafez, ed. Ḵānlari, II, pp. 1146-48) states that after Hafez’s death he, Moḥammad Golandām, collected and recorded the poet’s scattered ghazals. This indicates that the earliest sources, oral and/or written, were multiple, and that the hope of reconstructing the “true divān” (divān-e sÂaḥihá) is indeed slim.

The nature of the early manuscript sources further supports this. Even in redactions of the Divān as such, the number of ghazals varies considerably; and while it is generally assumed that later redactions became infl;ated through the addition of poems by other poets mistakenly or falsely attributed to Hafez, the exact number of ghazals in the “original” divān—the exact number of ghazals (and other poems) Hafez composed—cannot be securely ascertained. Nor can we be certain that all of his poetic output has survived. It is a remarkably small output: if we assume around 500 ghazals, composed over a poetic career lasting roughly forty years (if not longer), the poet would have composed, on average, one ghazalper month. (This contrasts with the far greater number of ghazals in the divāns of his close contemporaries—for example Ḵᵛāju Kermāni, Salmān Sāvaji, and Kamāl Ḵojandi the latter poet, like Hafez, specialized in the ghazal.) The number of qaṣidas preserved in the sources varies; and one manuscript (Nur Osmania Library, Istanbul, MS 3822; an anthology dated 825/1422) is said to include several elegies (marāṯi), among them one on Šāh Šojāʿ (see ed. Ḵānlari, II, pp. 1134-35; Rašid ʿAyvazi and Akbar Behruzi, who based their edition on this manuscript, do not mention marāṯi in their introduction).

With a few exceptions (noted below), most of the early sources for the ghazals are not redactions of the Divān as such. Scattered verses and poems may be found in sources composed during the poet’s lifetime (see Moḥiṭ-Ṭabāṭabāʾi Aḥmad, pp. 130-36). One of the earliest sources is the Majmuʿa-ye laṭāyef o safina-ye ẓarāyef, a handbook on rhetoric by Sayf-e Jām Haravi, which, although completed around 803/1400-01, was begun much earlier, in the reign of the Delhi sultan Firuzšāh b. Moḥammad b. Toḡloq (1351-88). (The manuscript, which was probably re-copied around the end of the 15th century [introd., p. 42], was formerly in the library of the Kabul University Faculty of Literature; an incomplete manuscript in the British Museum known as the Dastur al-šoʾarā, dated 803 [Rieu, Persian Manuscripts, supp., p. 232, no. 374], is another copy of the same work.) The book’s main section, on the poetic art, is followed by an anthology of poetry by Persian poets (plus a section of poetry by Indian poets) containing 127 ghazals by Hafez, arranged non-alphabetically. These were published by Naḏir Aḥmad (New Delhi, 1991); one wishes he had published the entire text, as the manuscript—now undoubtedly lost to scholars—is an important document for the history of the reception of Hafez.

Indeed, the question of reception has received relatively little attention. Indicative of the early stages of “publication” of Hafez’s poems is the fact that they are found chiefl;y in anthologies, the varied nature of which may be shown by a few examples. (1) Library of the Academy of Arts, Dushanbe, MS 545, dated 807/1404-05, a miscellany of varied texts in prose and verse, with 41 ghazals and two moqaṭṭaʿs by Hafez on the margins (see ed. Ḵānlari, II, pp. 1130-31; Neysāri, pp. 73-74; Rehder; the ghazals were published by K. Galimova, Dushanbe, 1971). (2) Köprülü Library, Istanbul, MS 1589, dated 811/1408, an anthology of mystical treatises with selections by various poets, including Hafez, on the margins (see Ḵānlari, II, pp. 1127-28). (3) Aya Sofya Library, Istanbul, MS 3945, dated Ṣafar 813/June 1410, containing the divāns of twenty poets including Hafez (see ed. Ḵānlari, II, p. 1128; Neysāri, pp. 80-84; Rehder; published by Elisabeth Boelke in Zum Text des Ḥāfiẓ, Cologne, 1958). (4) Salar Jung Museum, Hyderabad, MS Pers. poetry 36/2289, dated Moḥarram 813/May 1410, an anthology containing a prose abridgement of the Ḵoffi-e ʿAlāʾi, a medical text by Esmāʿil Jorjāni (d. ca. 531/1136), and poetry, including excerpts from Ferdowsi’s Šāh-nāma and three of Neẓāmi’s verse romances, with more selected poetry on the margins, including 47 of Hafez’s ghazals in non-alphabetical order. The ghazals were published by Naḏir Aḥmad (1988a), with facsimile reproductions of the relevant folios. Aḥmad’s conclusion, on the evidence of this manuscript and another anthology from the same period (probably that made for Eskandar Solṭān; see below) in which the ghazals are in alphabetical order, that the Divān must have existed in two recensions, alphabetical and non-alphabetical (although the original recension was probably non-alphabetical; pp. 5-6), clearly rests on the assumption of a single “original” on which later copyists drew, and ignores the possibility of multiple sources, or of alphabetical or non-alphabetical ordering being a matter of choice.

Other anthologies support the argument that the anthologists’ aim was not to provide a definitive text of Hafez’s Divān but to furnish selections of poetry chosen for various reasons (cf. Mahdavi, p. 5496). In another early anthology (British Museum, London, add. 27.272; dated 813-14/1411-12), Hafez’s ghazals appear in two places (145 arranged alphabetically, another 9 arranged non-alphabetically) on the margins of Neẓāmi Eskandar-nāma (see ed. Ḵānlari, II, p. 1129; the ghazals were published by Ḵānlari in 1337 Š./1959). This anthology, compiled for the governor of Fārs Eskandar Solṭān b. ʿOmar Šayḵ (q.v.), may have had more than mere “literary” significance, as it was prepared only a few years before Eskandar rebelled against his uncle Šāhroḵ. An anthology dated between 817/1415 and 838/1434, containing 43 ghazals (Ketābḵāna-ye Majles, Tehran; see ed. Ḵānlari, II, pp. 1136-37), was copied for Eskandar Solṭān’s successor Ebrāhim Solṭān by Moḥammad Golandām, and includes a qaṣida by the latter in praise of the prince (excerpt quoted by Neysāri, p. 21). In a manuscript in the Asafiya Library, Hyderabad (MS Aḵlāq 508), dated Rabiʿ I 818/November 1415, the main text is Kalila wa Dimna with, on the margins, ʿAṭṭār’s Manṭeq al-ṭayr and the Divān of Hafez (357 ghazals and 13 qaṭaʿāt; arranged alphabetically). Naḏir Aḥmad, who published the ghazals (1988b), argues that since in many of the panegyric ghazals the name of the mamduḥ is omitted, as are encomiastic verses or passages, the Divān was later revised, either by the poet himself or by someone else (p. kāf); the suggestion is that the “original” version lacked the encomiastic elements, which were later added, but it is unclear why the situation might not be the reverse (cf. Neysāri, pp. 215-19, and, on the manuscript, see also Rehder, p. 110; Kamāliān).

The fact that so many of the early sources are anthologies refl;ects the vogue for such works during the Timurid period (see Lentz and Lowry, p. 116). The extent and lavishness of Timurid patronage is well known, as is their interest in learning and literature; but Timurid bibliophilia was especially connected with their patronage of the arts of the book, which, under them, became highly professionalized (see Losensky, pp. 145-49; Lentz and Lowry). Books were not merely records of liter-ary greatness, but objects of value, of gift-giving and exchange, as well as important components of military spoils, to be placed in the royal treasuries and, most probably, only rarely read. (Some later manuscripts belonging to the Mughals bear inscriptions noting the dates on which they were viewed by the current ruler; see e.g. Qāsemi, pp. 139-40.) Many manuscripts of this period, especially those which were highly decorated, were not public texts whose purpose was to disseminate the work of a writer or poet, but symbols of the ruler’s magnificence.

In 907/1501-02 the Timurid prince Faridun b. Ḥosayn Mirzā Bāyqarā produced a revision of Hafez’s Divān (see Losensky, p. 145; Lentz and Lowry, p. 369; ed. Ḵānlari, II, pp. 1148-49; a copy of this recension is in the British Library, London, MS Or. 3247; Ḵalḵāli in his edition [p. 396] notes another in a private collection in Tehran). In the preface the calligrapher, ʿAbd-Allāh Morvārid, states that over the course of time and at the hands of ignorant copyists many errors and distortions had entered the text. Over five hundred copies of the Divān were collected on the prince’s order; then he, together with a number of learned friends and boon-companions, set about collating and editing the poems, which were compared with anthologies and ghazals written down before the poet’s death. Many poems were discovered which had previously been unknown. The resulting divān was given the title lesān al-ḡayb, “the language of the unseen” (Meyḵāna, pp. 84-85; the entire text is reproduced in facsimile and translated in Roemer, fols. 99a-101a [text] and pp. 134-41 [trans.]). Aḥmad Golčin-e Maʿāni, citing this passage (which appears on the margins of the Meyḵāna), observes, “It is evidently at this point that poems by other poets found their way into the . . . divān, and later caused trouble for scholars and researchers” (Meyḵāna, p. 85, note).

As Morvārid’s statement makes clear, there were early copies of the (or a) divān. The earliest to be identified so far is a manuscript in the Biruni Library of the Institute of Oriental Studies, Tashkent, dated 803/1400-01, copied by Borhān b. Ḡiāsò Kermāni for a certain Majd-al-Dawla wa’l-Din (published by Ṣādeq Sajjādi and ʿAli Bahrā-miān, Tehran, 2001[?]; for a review see Bahār, who provides no more details concerning the manuscript). Another (Topkapi Saray, Istanbul, Ravan Library, MS 142/497), dated 822/1419, was copied by the famous calligrapher Jaʿfar-e Ḥāfeẓ (Jaʿfar Bāysonḡori), possibly for Bāysonḡor (q.v.). Especially well-known is the “Ḵal-ḵāli manuscript,” dated 827/1423-24, which was published by ʿAbd-al-Raḥim Ḵalḵāli in 1306 Š./1927 and was the basis for the Qazvini-Ḡani edition (1320 Š./1941); it contains 496 ghazals, 29 qeṭʿas, 2 maṯnawis, and 42 robāʿis, but no qaṣidas. A number of anthologies and divāns which antedate the revision of 907 were produced at the court of Ḥosayn Mirzā Bāyqarā. One, dated 895/1489-90 (Ketābḵāna-ye Majles, Tehran, MS 969), copied by Solṭān-ʿAli Mašhadi (d. 926/1520; one of the master calligraphers employed in the atelier of Ḥosayn Mirzā), contains 211 ghazals together with other poems. Another, undated but bearing the autograph of Ḥosayn Mirzā on its end fl;yleaf (Khuda Bakhsh Oriental Pub-lic Library, Patna; published as Diwan-e-Hafez, Royal Mughal Copy, Patna, 1992), also contains marginal notes by the Mughal rulers Homāyun and Jahāngir which indicate that it passed to the Mughal court and that it was used for bibliomancy or fālgiri (see DIVINATION); and, indeed, many copies of the Divān appear to have been prepared specifically for this purpose.

The manuscripts mentioned above are only a small sample of the early sources for Hafez’s poetry. Over 80 manuscripts from the 9th/15th century, at least 35 of which date from the first half of the century, have so far been identified. The varied nature and wide geographical distribution of these manuscripts gives some indication of the problems surrounding the establishment of a tex-tus receptus. The traditional approach has been to search for manuscripts closest to the poet’s own time (see Neysāri Māyel Haravi, pp. 202-9); but this approach, which is similar to that employed in editing Greek, Latin, and medieval European texts, assumes a “fairly uncontaminated textual tradition,” a situation that does not apply in the case of Hafez (cf. Flemming, p. 7). A re-cent effort to address the issue of filiation (cf. Māyel Haravi, pp. 315-19), also invoking traditional European methodologies, demonstrates the scope of the problem. Aṣḡar Mahdavi, in a study analyzing seventeen early manuscripts, argues that “in producing critical editions of Persian texts . . . one ought to be able to establish, if possible, the relationship of each manuscript to the others with respect to descent and filiatioŋin other words, to establish if possible a chain of descent [šajara, i.e., a stemma] for each manuscript,” or to determine “whether that manuscript was unique” (Mahdavi, p. 5494). Mahdavi asks: “Is it possible, by comparing the extant early manuscripts, to arrive at an understanding of the original organization [tarkib-e aṣli] of Hafez’s collected poems or not?” He argues that copyists, whatever their individual goals, would surely not have felt free to contradict an original source (nosḵa-ye aṣl; p. 5495).

The “seventeen oldest manuscripts” (Mahdavi used only those available to him, most of which are mentioned above; for the details see pp. 5495-96) were painstakingly collated (see the tables in the Appendix to the article). Mahdavi concluded that the copyists’ aim “was not to collect all of Hafez’s poems;” what they copied constitutes selections, made for themselves or for someone else, “or perhaps a text compiled on the basis of scattered copies of Hafez’s ghazals” (p. 5496). Only ten of the manuscripts can be described as “Hafez’s Divān,” and of these several are defective (p. 5496). The oldest manuscripts (the 807 Dushanbe ms., with 48 ghazals, and the 811 Köprülü ms, with 36 ghazals) are considered the most reliable “from the point of view of age and/or of the persons who copied them”; four others (the 813 Aya Sofya anthology, the 813-14 Eskandar Solṭān anthology, the 816 Aya Sofya anthology, and the 817-38 Ebrāhim Solṭān anthology), which were copied in Shiraz at dates relatively close to Hafez’s own time, are said to have “a particular and limited value and reliability” because of their dates and their Shiraz provenance (pp. 5496-97).

Mahdavi concludes that at the time the 807 (Dushanbe), 811 (Köprülü), and 813 (Aya Sofya) manuscripts were copied “the compilation of Hafez’s poems in the form of a divān had not yet reached a complete and final stage.” Alphabetical ordering by rhyme may or may not be observed; and none of the early manuscripts are organized in the “familiar” form of modern editions. Moreover, the first ghazal in the Divān as we know it today is found in only thirteen of the manuscripts, and is the first only in twelve of these. (Mahdavi argues that, on the basis of the 807 manuscript, “we can assume that the original [aṣli] manuscripts of Hafez’s ghazals had not yet accepted the present first ghazal as the first” [p. 5497]; but we should note that the manuscript is an anthology and the ghazals are arranged non-alphabetically.) While several manuscripts appear to be related (see pp. 5497-99), it is difficult to divide even such a small sample into groups on the basis of filiation; and, given the large number of extant early manuscripts, there would appear to be little hope of establishing meaningful relationships among them.


Ḥāfeẓ, Divān, ed. ʿAbd-al-Raḥim Ḵal-ḵāli, Tehran, 1306 Š./1927; ed. Moḥammad Qazvini and Qāsem Ḡani, Tehran, 1320 Š./1941; ed. Rašid ʿAyvażi and Akbar Behruzi, Tabriz, 1354 Š./1975; ed. Parviz Nātel Ḵānlari, 2nd ed., 2 vols., Tehran, 1362 Š./1983; ed. Naḏir Aḥmad as A Critical Edition of the Ghazaliyat-i-Hafiz, New Delhi, 1988a; ed. Naḏir Aḥmad as Diwan-i-Hafiz, New Delhi, 1988b; ed. Naḏir Aḥmad as Ḡazaliāt-e Ḥāfeẓ bar asās-e Majmuʿa-ye laṭāyef o safina-ye ẓarāyef, New Delhi, 1991.

Naḏir Aḥmad, “Naẓari dar Divān-e Ḥāfeẓ čāp-e Doktor Ḵānlari,” Irān-nāma 7, 1367 Š./1988-89, pp. 126-41.

Esfandiār Bahār, “Kohantarin Divān-e Ḥāfeẓ,” Boḵārā 17, 1380 Š./2001, pp. 271-75.

Dār al-kotob al-Meṣriya, Fehres al-maḵṭuṭāt al-fārisiya I, Cairo, 1966.

J. T. P. de Bruijn, “The Transmission of Early Persian Ghazals (with special reference to the Dīvān of Sanāʾī),” Manuscripts of the Middle East 3, 1988, pp. 27-31.

Barbara Flemming, “From Archetype to Oral Tradition: Editing Persian and Turkish Literary Texts,” Manuscripts of the Middle East 3, 1988, pp. 7-11.

Thomas W. Lentz and Glenn D. Lowry, Timur and the Princely Vision, Washington, D.C. and Los Angeles, 1989.

Franklin D. Lewis, Rumi: Past and Present, East and West, Oxford, 2000.

Paul E. Losensky, Welcoming Fighānī: Imitation and Poetic Individuality in the Safavid-MughalGhazal, Costa Mesa, Calif., 1998.

Najib Māyel-Haravi, Naqd o taṣḥiḥ-e motun, Mašhad, 1990.

Mehdi Kamāliān, “Kušešhā-ye jadid dar šenāḵt-e Divān-e ṣaḥiḥ-e Ḥāfeẓ,” FIZ 6, 1337 Š./1958-59, pp. 204-72.

Aṣḡar Mahdavi, “Fehrest-e taṭbiqi-e nosḵahā-ye kohan-e ḡazaliāt-e Ḥāfeẓ,” in Iraj Afšār, ed., Nām-vāra-ye Doktor Maḥmud Afšār IX, Tehran, 1985, pp. 5494-5507.

Jalāl Matini, “Divān-e Ḥāfeẓ: Mirāṯ-e gerānqadr-e farhangi-e mā,” Irān-nāma 6, 1988, pp. 597-641.

Moḥammad Moḥiṭ-Ṭabāṭabāʾi, “Qadimtarin maʾḵaḏ-e kotobi-e šeʿr-e Ḥāfez,” in Akbar Ḵodāparast, ed., Majmuʿa-ye maqālāt dar bāra-ye Ḥāfeẓ, Tehran, 1363 Š./1984, pp. 15-27.

Mehrdād Niknām, Ketāb-šenāsi-e Ḥāfeẓ, Tehran, 1367 Š./1989.

Salim Neysāri, Moqaddema-i bar tadwin-e azalhā-ye Ḥāfeẓ, Tehran, 1367 Š./1988.

Šarif Ḥosayn Qāsemi, Fehrest-e nos-ḵahā-ye ḵaṭṭi o čāpi-e divān-e Ḥāfeẓ dar Hend, New Delhi, 1988.

Robert M. Rehder, “New Material for the Text of Ḥāfiẓ,” Iran 3, 1965, pp. 109-19.

Hans Robert Roemer, Staatschreiben der Timuridenzeit, Wiesbaden, 1952.

G. M. Wickens, “Ḥāfiẓ,” in EI2 III, pp. 55-57.

(Julie Scott Meisami)

Originally Published: December 15, 2002

Last Updated: March 1, 2012

This article is available in print.
Vol. XI, Fasc. 5, pp. 476-479