SOURCE: "The Unity of the Ghazals of Hafiz," Der Islam, Vol. 51, 1974, pp. 55-96.

[In the following essay, Rehder critiques A. J. Arberry's analysis (see Further Reading) of the unity of Hafiz's ghazals and discusses his own conclusions on the subject.]

The study of the poetry of Hafiz is important not only in its own right and for the understanding of Persian (and Islamic) literature, but also for what it contributes to poetics, to the understanding of all poetry. Persian literature has only very rarely been looked at as literature, and this is true of Hafiz's poems as well, but one subject which has attracted some attention is the problem of the unity of his ghazals. The first discussion in any detail of this subject is A. J. Arberry's 'Orient Pearls at Random Strung,' Bulletin of the School of Oriental and African Studies [BSOAS], xi (1946), 699-712, and I will take this article as my point of departure. I will (I) analyze Arberry's conclusions about the unity of the ghazals of Hafiz, (II) discuss the text of the poem he uses as an example (the famous and beautiful poem beginning: Agar an Turk-i Shirazi …), (III) examine Arberry's specific comments on this poem, and (IV) analyze the poem myself and make some general statements about the unity of the ghazals of Hafiz.

I

The following summary statement is made by Arberry: 'Hafiz' technique is fundamentally thematic; by which is meant, that he constructs each lyric upon the basis of a limited number of themes selected from a repertory which is itself definitely restricted, and to a great extent conventional. Having chosen his themes—as a rule not more than two or three whole themes, with fragments of others so familiar as to be immediately recognisable—he then proceeds to work out his pattern. It is supremely important to understand how vital and inevitable pattern is to the Persian poet: a people which produced craftsmen of unsurpassed skill in the arts of line and colour might indeed have been expected to throw up men of equal parts in the marshalling of verbal images and sounds; and it was natural that they should work their materials into forms essentially similar to those invented by their fellow-craftsmen, the creators of mosaics and miniature paintings. So it is as a mosaic of sounds and symbols that the Hafizian lyric is to be appreciated; and its artistry, including its unity, is to be understood as being of the order of artistic unity that is found in the finest mosaic pattern.' (Orient Pearls, 704-705).

Arberry's statement as to how Hafiz constructed each lyric has no historical evidence to support it. In the oldest sources for the biography of Hafiz (those texts composed or copied within about 140 years of his death) there is neither any reference to his habits of composition nor anything which clearly indicates his attitude to his poems.¹ There is nothing to suggest that Hafiz conceived of a ghazal as a presentation of, or as a variation on, several themes, or that he first chose his themes and then worked out his pattern. What for Arberry is two or three themes may have been for Hafiz one subject, but there is nothing in the biographical evidence to prove that Hafiz saw disunity or unity in his poetry.

The oldest manuscripts reveal neither the genesis nor the development of the poems. There are some variant readings which seem to represent different versions, or corrections made by Hafiz. An example is the two final bayts preserved for QG 440, but there is not enough in what is known at present even to begin to document a description of how Hafiz composed his ghazals.²

Arberry also speaks 'of the complete works of dozens of poets certainly studied by Hafiz in his formative period and always kept by him close at hand.'³ The anonymous old Preface to the Divan-i Hafiz states that the poet spent time in 'scrutiny of the works of the Arab poets (tajassus-i davanin-i arab),' and the Tadhkirat al-shuara asserts that Kamal Khujandi was a friend of Hafiz and that they exchanged poems.⁴ This is all that is known from the biographical sources about Hafiz's reading of poetry. There is no evidence that all the poets he alludes to, or makes use of, in his works were read at the same time early in his life in any 'formative period,' and nothing suggests that he composed with certain books 'always kept by him close at hand.'⁵ From his laqab, hafiz, and the importance of rote learning in his culture, one assumes that he probably had many poems and bayts in his memory, but the sources do not help us beyond this point. There is also the textual problem of whether the bayts and misras which contain these literary allusions are authentic.

That Hafiz's poems are traditional and conventional is clear when they are read in conjunction with the poets who preceded him, but Hafiz's exact place in the tradition of Persian lyric poetry is still obscure and careful study of the poems themselves may only provide a limited amount of illumination.⁶ That 'Hafiz was understandably most ambitious to surpass' Sadi cannot be proven, however interesting it may be as an hypothesis.⁷ The fact of an allusion is usually not enough to describe the attitude of the poet making the allusion. To show that Hafiz echoed Sadi's poems, as Arberry has, is important, but it is not the same thing as showing that Hafiz was 'most ambitious to surpass' him. In addition, this idea must be considered side by side with the sense of superiority and the great confidence in his own work which Hafiz expresses at the end of many of his ghazals.⁸

A number of serious criticisms must be made of the statement that 'Hafiz's technique is fundamentally thematic.' There is the problem as to what Arberry means by theme. How does one decide where one theme ends and another theme begins? The opening bayt of the first poem in the Qazvini-Ghani edition may be translated:

Hey, saqi, pass the cup and hand it on,
For love first seemed easy, but difficulties
  appeared.

Is this to be identified as the theme of "drinking" or as "the difficulties of love," or as "drinking to forget love's difficulties"? Is part of the bayt to be considered the theme and the remainder the fragment of another theme? The address to the saqi and the circulation of the wine-cup are extremely common in Persian poetry. In this context are they whole themes or fragmentary themes? This raises the question of how much space a theme may take up and stay a theme. If the first misra read:

Hey, let us drink until the daybreak,

would the number or names of the themes and theme fragments in the bayt be changed, that is, does the definition of a theme depend on the general idea expressed or on the specific images or language in which it is expressed, or both?

However one deals with these questions, the effect of Arberry's analysis is to shift the attention from the poem to the analysis itself, from the words of the poet to the words of the critic. The definite and particular language of the poem is subsumed to alien and abstract categories. The poems of Hafiz, like all poems, should be described in common language, as simply as possible, and should be analyzed in their own terms and in the terms of their author. To show that they are conventional and traditional one does not need the category of themes, one has only to list the apposite parallels with other poets, as Arberry himself has done when he cites bayts in the Ghazaliyat of Sadi which appear to have been echoed in QG 3, in what is the most valuable part of his article (Orient Pearls, 707-709).

In speaking of 'how vital and inevitable pattern is to the [my italics] Persian poet' Arberry glosses over the fact that not all Persian poets have the same attitude to pattern and form. The short lyric poems of Rudaki, Khayyam, Rumi and Hafiz, for example, differ in this respect. At the same time Arberry does not distinguish the peculiarities of Persian practice from that of other poets. Inevitably pattern is vital to all poets, and to assert this fact does not bring us very much closer to Hafiz.

Similarly, to consider Hafiz's ghazals primarily in terms of their themes stereotypes them unnecessarily, and fragments them so that there is a tendency to look more at the affiliations of parts of a poem with parts of other poems by other poets, and less at the relationships and functions of the parts within the poem, and at the poem as a whole. Such an emphasis substitutes a simple historical labour for the more complex process of understanding the poem as a poem, as the unique creation of a unique individual.

Thus Arberry writes: 'No complete understanding and appreciation of this poem [his example, QG 3] is attainable, as we have already indicated, until all prior treatments of the themes have been examined, and all images and verbal pictures drawn by earlier poets have been compared.'⁹ If one does not quibble on the word complete, this judgement denies that an individual poem has any significant existence by itself, which certainly goes against the grain of experience, and mistakes the nature of poetry. Are the only satisfactory interpretors of a poem those scholars who have examined 'all prior treatments of the themes' it contains? Do we understand and appreciate Shakespeare when we have not read all his predecessors? Overlooking the obvious limitations of human life and memory, I would prefer to say that good poems are usually those which can be understood and appreciated by themselves.

There is also a curious historical problem in all this. Hafiz himself was probably not aware of 'all prior treatments of the themes' which he took up. How then is one to assess the influence upon him of that of which he was ignorant? There is the further problem of distinguishing between those prior treatments of which he was conscious and those of which he was unconscious, and among those of which he was aware, to distinguish those which intrigued him from those in which he took no interest. The question dissolves into myriad shadows and abysses such that no analysis worthy of the name can make the accumulation of parallels its major purpose.

Analogies between the arts are sometimes very interesting, and they are necessary if one is to form any good idea of a culture as a whole, but extreme care must be taken in making them, and it cannot be forgotten that they are only analogies, suggestions of partial, putative, and commonly unessential, similarity. Painting and poetry, for example, are so different in their ways and means, in the modes of the mind which they engage, that the more one is concerned with the essential aspects of one of them, the less the help which is provided by analogies with the other. The paintings of Turner do not explain the poems of Wordsworth, and the more one tries to describe the essential and characteristic qualities of either artist the more it is necessary to forsake any analogy, and to deal with the given work or the given art on its own terms.¹⁰ The most interesting problems: meaning, style, form, unity, development, are radically different in the different arts. The patterns and forms of mosaic making and miniature painting cannot help us very much in understanding the unity of the ghazals of Hafiz, or what they are as poems. Literary problems must be solved in literary terms.

Furthermore, there is no reason to believe that 'a people which produced craftsmen of unsurpassed skill in the arts of line and colour might indeed have been expected to throw up men of equal parts in the marshalling of verbal images and sounds.' This assumes a correspondence between painting (and/or craftsmanship) and literature which does not exist. The great period of Dutch painting in the seventeenth century threw up no poets or writers at all comparable to the painters. English history is unusually rich in great poets but comparatively poor in great painters, and the best English poets do not regularly appear in the periods of the best painters. The ancient Egyptian culture 'which produced people of unsurpassed skill in the arts of line and colour' never produced a literature commensurate with its art. The archaic Arabic poetry appears in a culture in which there seems to have been no painting and very little art work.

As I have already stated above it is virtually impossible, because of the different ways, means and modes of the different arts, for poems to be made 'into forms essentially similar' to those 'of mosaics and miniature paintings.' It would be difficult enough, perhaps impossible, to show that Persian mosaics exhibit 'forms essentially similar' to those of Persian miniature painting. The majority of mosaics are geometric designs, while the majority of miniatures are representations of human and animal figures. Consequently, they very often have a center, a subject, and a meaning to which there is nothing comparable in most mosaics, where the principle of composition is commonly the repetition of a geometric or calligraphic form. There are also great differences in size, colour and quality of design between the two arts.

With mosaics and paintings we usually behold the form or pattern at a glance. There is a sense of unity inherent in the very act of vision which is absent when we hear or read a poem. The form of the ghazal is primarily determined by meter, monorhyme, the bayt, and the system of rhetorical figures. They have no counterparts in mosaics and miniature paintings. Are all the instances of the colour blue in a design to be said to rhyme? If so, then how can an analogy be made between the single rhyme of the ghazal and designs in which there are several colours? Are the outlines which the miniature painters filled in to be compared to the bayts, although they are, unlike the bayts, of different shapes and sizes? Or is a bayt to be compared to a brick or a tile? Also does 'pattern' refer to the transformation and combination of ideas within the ghazal or to the formal elements which I have listed above? The more one thinks about different works of art the more their individuality becomes apparent. As for calling Hafiz's poetry a 'mosaic of sounds and symbols,' this, in so far as it has any meaning, could be said not only of all poetry but of all literature and all speech.

So far I have followed Arberry's usage of mosaic even though it is not a very satisfactory term to use for Iranian art. It suggests to a European or American St. Vitale in Ravenna or Aya Sofya and the Church of the Chora in Istanbul, and while there are similar mosaics in the Umayyad mosque in Damascus, there are none, as far as I know, in Iran. It is necessary to distinguish between the architectural designs made with plain bricks laid in different directions, or indented or set-out from the surface of a building, and designs in coloured bricks (which may also incorporate direction and surface irregularity). Both of these are different from the so-called faience mosaics where the pieces are coloured and of different sizes and shapes (unlike the more-or-less uniform modules of most Byzantine mosaics). They must not be confused with the designs composed of painted tiles. There are also the more modern mirror mosaics, as well as the inlay work known as khatamkari. All of these techniques may be said to produce mosaics, and it is questionable whether they, compared to each other, display the same sense of pattern or essentially similar forms.

'The creators of mosaics and miniature paintings' are from the same hypothetical, hypostatic, and ahistorical realm as 'the Persian poet.' Just as all 'the arts of line and colour' cannot be run together, the Isfahan of Shah 'Abbas cannot be used to gloss the Shiraz of Hafiz. The buildings of the Seljuqs are not like those of the Timurids, nor do Bihzad's illustrations for Sadi's Bustan (dated 894/1489) resemble the miniature paintings of the Tabriz school (dated to the end of the fourteenth century) in the Kevorkian MS of Rashid al-din's Jami al-tavarikh.¹¹ These disparate things may be compared, and ought to be compared, but particular care must be taken to appreciate their individuality and to understand the uniqueness of their historical context. This cannot be done by the wave of a hand or the flourish of a phrase.

If one wants to assume that artists of different kinds share certain inarticulate notions or are informed by the same spirit, the easiest and simplest place to begin is with contemporaries in a given culture. This possibility is passed over completely by Arberry. Unfortunately, the art of fourteenth century Shiraz, or for that matter, of Iran, has not been studied in anything approaching detail.¹²

Arberry, however, seems to postulate more a Kulturgeist than a Zeitgeist, a Persian Islamic spirit which is the genii and genius of the whole historical existence of the culture. It is, as he states it, a variation of the familiar description of Islamic civilization as atomistic. Bernard Lewis, for example, sees atomism as one of the four major features of medieval Islamic civilization. 'By this is meant the tendency to view life and the universe as a series of static, concrete and disjunct entities loosely linked in a sort of mechanical or even casual association by circumstances or the mind of an individual, but having no organic interrelation of their own … Arabic literature, devoid of epic or drama, achieves its effects by a series of separate observations or characterisations, minute and vivid, but fragmentary, linked by the subjective associations of author and reader, rarely by an overriding plan. The Arabic poem is a set of separate and detachable lines, strung pearls that are perfect in themselves, usually interchangeable.… Arabic art—mainly applied and decorative—is distinguished by its minuteness and perfection of detail rather than by composition or perspective. The historians and biographers, like the fiction writers, present their narrative as a series of loosely connected incidents. Even the individual is drawn as a sum of attributes, often listed, as a recent writer remarks, like the description of a passport."¹³

Kowalski refers to the 'molecular structure' of Arabic and Persian poetry, and Rypka writes, 'Thus, a Persian poem should be read in a different manner from that customary to the European, less as a whole, more as filigree work, for it is full of finely-wrought details, with no strictly logical sequence of verses in any given poem as is common in the West. It is as if the poets exhausted themselves to such an extent by giving form to such refinement that the fitting together as a whole escaped them."¹⁴

I do not propose here to discuss all the difficulties inherent in these statements, but as far as the atomism of Islamic poetry is concerned it is important to note that the atom is different for each author. For Arberry it is the theme which is the fundamental unit, for Lewis it is the line (does he mean misra or bayt?), for Rypka it is the individual detail or point (nuktah), although I would say the two are not synonymous.

The atomism of Islamic literature appears to be more a matter of emphases than of essences.¹⁵ Some of Lewis' observations are true, but they misrepresent the subject as a whole. There is unity in the longevity and vitality of forms such as the qasidah and the ghazal, and in the nature of the literary tradition. Men for hundreds of years wrote ghazals not out of mere habit, but because the ghazal expressed what they wanted to say in a way which pleased them. The many conventions, the common store of ideas and subjects, the repeated images, the many allusions to the work of other poets, as they testify to the power of the literary tradition they testify to its unity. That so many historians, even if their interest was in recent or contemporary history, began their works with the creation and the early history of the world, and then commonly went through all the periods of Islamic history, bespeaks their sense of the wholeness of the past and of their own culture. Chronological order, the order of virtually all Arabic and Persian histories, is a logical sequence, and possesses an obvious and simple unity. Thus the histories can be used as illustrations of atomistic thinking and of a feeling of wholeness, which suggests certainly that the best summary is one which gives up the dichotomy and includes both aspects.

Islamic philosophical writing exhibits unity in that it can all be said to be the elaboration and exploration of one philosophy, a body of thought permeated and transformed by Greek ideas. Wolfson has described the Hebrew, Latin and Arabic philosophical literatures to which Spinoza was the heir as 'a common tradition.' 'They were all based upon Greek philosophy, at the centre of which stood Aristotle. The same Greek terminology lay behind the Arabic, Hebrew, and Latin terminology, and the same scientific and philosophical conceptions formed the intellectual background of all those who philosophized in Arabic, Hebrew, or Latin. The three philosophic literatures were in fact one philosophy expressed in different languages, translatable almost literally into one another.¹⁶ This unity has its literary expression in that many native speakers of Persian and Turkish felt obliged to write in Arabic when they wrote on philosophical subjects. Very often their works are in the same form and they take up not only the same problems but discuss them in the same order, dividing the question, and arguing the points in a similar manner. As in the case of history, an overriding plan is clearly visible. That original work was presened in the form of commentaries on the works of others, and that so many commentaries were written, is good evidence that the authors felt the unity of the tradition and the continuity of the problems.

Finally, Islamic literature abounds in encyclopaedias, epitomes, anthologies, handbooks and surveys. How does one explain this number and this variety of general and comprehensive works as the expression of an atomistic spirit?

Atomism, then, is not a satisfactory theory of Islamic culture, and theories of culture, such as atomism, the notions of a Kulturgeist and a Zeitgeist, and analogies between various arts, neither explain the unity of Hafiz's ghazals nor illuminate the beauty of his style.

II

Criticism of a literary work can proceed only after the state of the text has been considered. This means only that textual problems must be taken into account, because in many cases they cannot be solved. It appears that the text of the Divan-i Hafiz will never be absolutely fixed. The student, therefore, has no choice but to deal with probabilities, possibilities, hypotheses and conjectures instead of certainties.¹⁷

Qazvini has shown how the number of poems in the Divan-i Hafiz began to increase rapidly after the ninth century A.H., and Rempis has demonstrated the existence of spurious poems in two Hafiz MSS of the early part of the ninth century A.H. (Muzaffar Husayn MS 87 dated 810 A.H. and the New Delhi MS dated 818 A.H.).¹⁸ A comparison of old and new MSS and editions reveals that as the number of poems in the Divan increased, individual poems became longer, and the order of the bayts was sometimes changed.¹⁹ The preface to a recension of the Divan made in Harat in 907/1502-1503 proves that the text was thought of as corrupt at that time. The statement 'many ghazals … which because of the indolence and misapplication of the copyist have remained forsaken and unknown on the page of time were put in order (dar silk rabt dar amad)' suggests that the order of the bayts was considered a problem. In the next phrase this process is spoken of as tanqih, cleaning or purging, so it may be that the problem was seen as the addition of spurious bayts. My guess is that both problems were recognized, but in either case it shows that the integrity of individual poems was in doubt.²⁰

To illustrate his remarks on Hafiz, Arberry employs the ghazal beginning Agar an Turk-i Shirazi … (QG 3). Since the Qazvini-Ghani edition is easily available (and as Arberry reproduces the text of the poem from that edition), I will give the text as it appears in Aya. Sofya 3945 (without changing the orthography) and then collate it with the other old MSS available to me:²¹

Akar an Turk-i Shirazi ba-dast aradh dil-i ma-ra
   Ba-khal-i Hindu-yash bakhsham Samarqand
 u Bukhara-ra
Bi-dih saqi may baqi kih dar jannat nakhvahi
  [y]aft
   Kanar-i ab-i Ruknabadh va kulkasht-i
  Musalla-ra
Fighan k-in luliyan-i shukh-i shir[i]n-kar-i
  shahr-ashub
   Chunan burdand sabr az di] kih Turkan
  khvan-i yaghma-ra
Man az an husn-i ruz-afzun kih Yusuf dasht
 danastam
  Kih ashq az pardah-yi ismat birun arad
 Zulaykha-ra
Bad-am kufti va khursandam afaka Allah niku
 kufti
  Javab-i talkh mizibad lab-i la I-i shakar-
 kha-ra
Nasihat kush kun jana kih az jan dusttar
  darand
   Javanan-i sa adatmand pand-i pir-i dana-ra
Ghazal kufti va dur sufti biya va khush ba-
 khvan Hafiz
  Kih bar nazm-i tu afshanad falak aqd-i
 surayya-ra

Aya Sofya 3945 is the oldest, largest known Hafiz MS, dated 813/14 A.H., and copied in Shiraz. Of it Ritter declared: 'Dieser alteste, nur 22 Jahre nach des Dichters Tode in seiner Vaterstadt fur den damaligen Herrscher dieser Stadt geschriebene Textzeuge darf sicherlich die hochste Autoritat beanspruchen.²²

I do not have very detailed information on all the known MSS which are possibly older than the Khalkhali MS, but on the basis of what is available to me, this poem does not appear in the Mujam MS, the Isfahan Municipal Library MS, in British Museum 261/27 or in the Tarbiyat MS. In addition to Aya Sofya 3945, it appears in the New Delhi MS, Saray Revan Kosk 947 (dated 822 A.H.; f. lb-2a) Nuru Osmaniye 3822 (dated 825 A.H.; f. 2b), and in the Stalinabad MS.²³ …

There are not very many variant readings. In her list of the first lines of the poems in the Stalinabad MS Galimova omits the, between Samarqand and Bukhara in the first bayt, but this is likely to be an error either in transcription or printing.²⁴ In the second bayt Saray Revan Kosk 947 reads may-yi safi and nakhvahi did (although some letters are effaced in this MS), and in the fourth bayt it reads ruy instead of rang. The major difference between QG and the older MSS is in the first misra of bayt six. Kamaliyan reports a reading identical to the one in Aya Sofya 3945 as the only significant variant from QG for this poem in the New Delhi MS.²⁵

Saray Revan Kosk 947 has:

Bad-am kufti va khursandam jazaka Allah
 niku kufti

Nuru Osmaniye 3822 reads:

Bad-am kufti va khursandam niku kardi va
 khvush kufti

It is interesting to note that these variations, important as they are, do not represent a change either in the purport of the bayt or in its images, and consequently do not affect in any significant way the discussion of the unity of the poem. The only other variant reading is in the first misra of bayt nine where Saray Revan Kosk 947 has kuy and kuy in place of gu and ju.

On the basis of the evidence the text of the first misra of bayt six should be emended so as to adopt the reading of Aya Sofya 3945 and the New Delhi MS. There is also the possibility that one or both of bayts four and eight are spurious, and if eight is genuine there remains a question as to its place in the sequence of the poem.

III

To show how his theory of thematic technique and unity can be applied to a specific text, Arberry makes the following comments on the poem, Agar an Turki Shirazi …:

The principal theme—is the fair charmer, beautiful, proud, unapproachable, the human, this-worldly reflection of the immortal loveliness of the Divine spirit. This theme is stated in line 1; lines 3, 4, 5 and 6 develop it, introducing variations in the form of fragments and reminiscences of other themes from the general stock-in-trade: the tumult of love (line 3), the unworthiness of the lover and the self-sufficiency of the beloved (line 4), the story of Yusuf and Zalikha as a myth of divine and profane love (line 5), and the sweet-bitter tongue of the beloved, symbolizing the pleasure and pain of loving (line 6). Restriction of space prevents the quoting of parallel passages from Hafiz' other lyrics; but all familar with the Divan will have no difficulty in recognising these themes as favourites of the poet.

The subsidiary theme is—wine (and music) are the sole consolation of the lover, to compensate his sorrow over the incapacity of his love, and the transitory nature of mundane affairs, and to enable him to solve those mysteries of the spirit which baffle and defeat the reason. Line 2 introduces the theme; it is developed in line 7 (listen to the advice of the old man of experience who knows the way by having trodden it) and line 8. This is perhaps the most important and characteristic of all the themes used by Hafiz: it expresses supremely well his theory of the "intoxicated" lover (who has in his hand the Mirror of Alexander, the Cup of Jamshid), and symbolizes his rejection of all formal, "sober" life, whether it be the life of the cloistered Sufi, the orthodox theologian, or the philosopher.

The foregoing analysis, brief and inadequate as it is, demonstrates the superb skill and artistry with which Hafiz treats two typical themes separately and in close integration; it is a fair example of the technique which informs all his poetry, though it should be added that in this poem he is being comparatively simple and straightforward.

These eight lines complete the statement and development of the chosen themes, to the evident and not unjustified satisfaction of the poet: it only remains therefore to sign the poem. Hafiz has a number of different devices for appending his signature, and a close study of his final lines brings its reward in an increased appreciation of his poetic artistry. The "clasp" theme here used is a very common one, but its present treatment is scarcely surpassed for beauty in the whole Divan. The poet looks upon his handiwork and finds it very good: it is deserving of praise and reward.²⁶

That Arberry himself feels the difficulties (discussed in the first part of this essay) of delimiting individual themes is demonstrated by his remark that 'the principal theme' is developed and varied 'in the form of fragments and reminiscences of other themes.' The reasonably clear and coherent picture of the beloved's character is thus broken up into a rubble of subjects. These difficulties are compounded by the problem of deciding upon a 'principal,' 'subsidiary' and 'clasp' theme. In the poem the poet drinks wine because of the behaviour of his beloved. The two things are related. To say that one is 'subsidiary' to the other is to misrepresent the poem. He creates an hierarchy which is not in the poem. The distinction, moreover, appears to be merely one of space. The poem in QG has nine bayts; the principal theme fills five, the subsidiary, three, and the clasp theme, one. Any consideration of the force, beauty, or function of the bayts is excluded, except in the case of the so-called clasp theme which is defined on a different standard from the principal and subsidiary themes. Its position and function are considered, although Arberry does not note that it shares with bayt eight the subject of song and an allusion to a knot, and is, therefore, more than a mere appendage.

Thematic analysis is not subtle enough for connections of this kind. In addition, while perhaps one may speak of a clasp theme in a poem like this, where the poet names himself in the final bayt and makes a statement which does not seem to be explicitly tied to, or derived from, the rest of the poem, there are poems by Hafiz where this is not the case. In QG 1, QG 30 and QG 341, for example, the poet's name appears in the last bayt but that last bayt follows obviously from the rest of the poem. What, then, is the clasp theme? The final bayt of QG 20 is continuous with the rest of the poem, but nowhere in the poem does the poet's name appear. Is there a clasp theme? The meaning of the last bayt of QG 319 may be said to be different from those immediately preceding it, (although there are ideas in the last bayt, such as the poet's servitude and his obedience to the great, which echo other bayts in the poem) but its second misra imitates the first misra of the first bayt with words of the same form in the same syntactical order. The poet's name occurs in the next-to-the-last bayt. I do not think the idea of a clasp theme can be applied here.

Another shortcoming of thematic analysis can be seen in the statement that the principal theme is 'the fair charmer' herself, and not anything which she does or is in the poem.²⁷ The description, 'beautiful, proud, unapproachable,' is so general as to fit many very different poems. In this way all the particularities in the poem, the specific nature of the character of the Turk, of the mood of the poet, of their relationship, are glossed over. This is particularly damaging because the unity of Hafiz's poems depends in part upon these specifics, upon, in this case, for example, the several Turkish allusions. That the Turk's beauty is said to be the 'this-wordly reflection … of the Divine spirit' reveals the tendency of thematic criticism to amalgamate elements (and again to ignore everything which is unique about a poem). There is nothing in this poem to cause us to believe that this is a religious or Sufi poem, or that the beloved is in any way divine. In fact, the capricious and ruthless behaviour of the Turk, the blasphemous statement in bayt two, and the way in which the poet invokes God's help for the Turk, and the absence of any contrary evidence, suggest that this is a secular love lyric. It is worth noting that Sudi not only identifies the Turk as one of the descendents of the Turkish soldiers of Hulaku Khan who settled in Shiraz (the historicity of this story does not affect its value as evidence here), but makes a point of declaring that the Shirazi Turk is not to be understood in a metaphorical sense.²⁸ Similarly, the story of Yusuf and Zulaykha is in some contexts, 'a myth of divine and profane love,' but there is nothing in the poem to indicate that that is its meaning here.

There is no question that Persian poetry is very conventional, and that a limited number of conventions are used over and over. This, however, does not make all ghazals interchangeable. What is interesting is that within these narrow limits some poems seem artificial, trivial and insipid while others generate great power and beauty, and that whatever their quality there is a great variety of poems. It is very important to discriminate between the different ways in which a given poet made use of the tradition. That the story of Yusuf and Zulaykha was used to speak of divine love, that Hafiz was aware of the tradition (as he doubtless was), does not immediately establish what it means in this poem. The context of a word or phrase is probably more important than its history in determining its meaning. Thematic criticism, as it is usually practiced, ignores or takes very inadequate note of the context in which words appear. These problems can only be untangled by detailed analyses of individual poems and comparative studies of poems and poets. This is the most important work which students of Persian literature can perform at this time.

Also, in this poem there is no sign of 'the Mirror of Alexander, the Cup of Jamshid.' Wine provides the poet with consolation for his troubles, not with any glimpse of the future. It does not 'enable him to solve those mysteries of the spirit which baffle and defeat the reason.' He is uncertain of what the behaviour of the Turk will be, and the point of bayt eight is that a man cannot know his fate or understand the nature of things. It is certainly wrong to attribute a 'theory of the "intoxicated" lover' to Hafiz. Literature is the enemy of theory. Men whose major concern is theory rarely find self-expression in writing poetry, or prose stories. Here it is indeed all formal, sober life which the poet is rejecting. Hikmat, philosophy, theoretical speculation, is rejected by name, but one theory is not exchanged for another. Furthermore, intoxicated should not be in italics. It is real wine that the poet is drinking.

IV

The unity of any particular ghazal is dependant on what the ghazal is as a form. The essential unit of the ghazal is the bayt and each bayt is composed of two misraat. There is nothing in English poetry which corresponds to these units. Neither bayt nor misra may be correctly translated by line. Each ghazal is metrically regular, but a number of meters are used for ghazals. Every misra (and consequently every bayt) in a given ghazal is metrically identical. In QG 3 the same foot, /---, is repeated throughout the poem, but the meter is said to be: / - - -|/-- |/-/- - - / which comprises one misra. Usually a ghazal meter is made up of different feet. In QG 2 the meter is: / - / -| / / - - / -/ -/ / - which again comprises one misra. The meter, then, is defined in terms of the misra. From these two examples it can be seen that the bayts of one ghazal may be of a different length than those of another ghazal. Also, because two short syllables may be substituted for a long syllable (and vice versa), within a single ghazal some bayts may have one or two more syllables than the others, although they are metrically of equal length. In most ghazals such substitutions, if they occur at all, do not occur more than a few times. Frequently they are confined to a single foot in the meter. These small modulations, meters composed of different feet, and the slight irregularity in the number of syllables in combination with metrical regularity (as well as other things, such as that many syllables may be either long or short depending on the demands of the meter), help to preserve the ghazal from monotony. The meter is, nevertheless, one of the forces which holds the ghazal together and unifies it.

In a ghazal the end of every bayt rhymes in the same rhyme, and in the first bayt, the end of the first misra rhymes in this same rhyme with the end of the second misra. As a result the ghazal has not only the unity provided by a regular rhyme scheme, but a special cohesiveness because only one rhyme is used, and this rhyme is repeated twice in the first bayt, as if to clearly establish it, and then once in every subsequent bayt. This monorhyme is another aspect of the ghazal which has no counterpart in English poetry. In Persian it does not sound peculiar, ridiculous or forced as it does in English, when it is possible. One reason for this is that the rhymes are further apart in Persian than they are in English. In English it is difficult to successfully sustain a line which is regularly more than ten syllables. The units of the ghazal are much longer. In the examples above, in QG 2 each misra is fourteen or fifteen syllables and each bayt is between twenty-eight and thirty syllables. In QG 3 each misra has sixteen syllables and each bayt thirty-two. A rhyme at the end of thirty-two syllables is very different from a rhyme at the end of ten syllables.

The monorhyme of the ghazal is also a unifying factor in that it sets up a certain syntactical regularity within the poem. QG 3 is a good example. The rhyme is -a-ra in which the -ra is the suffix which denotes what may be called the object of attention in a sentence, commonly the definite direct object. It can also be used in a dative construction. The choice of -ra as part of the rhyme causes Hafiz in QG 3 to place a noun or noun with modifiers at the end of each bayt, and in all the bayts but one (QG 3/5b where the dative construction is employed) the noun is the definite direct object of the sentence. The first person singular of the past tense of the verb kardan (to do or make) forms the final part of the rhyme in QG 319 'which makes the poem a succession of assertions and heaps up a whole series of completed past actions.¹²⁹ Ibn Khaldun (1332–1406), who was perhaps Hafiz's greatest contemporary in the Islamic world, understood the power of rhyme as an organizing force in a poem. In giving directions to poets, he recommends: '(The poet) should have the rhyme (in mind), when the verse is first given shape and form. He should set it down and build (his) speech on it all the way through to the end, because if the poet neglects to have the rhyme (in mind) when he makes a verse, it may be difficult for him to get the rhyme into its proper place, for it often is loose and unstable.'³⁰

The ghazal is always a short poem, which in itself causes it to be felt as a whole, but it does not have to be a fixed length. Browne says that the ghazal 'seldom exceeds ten or a dozen bayts' and Rypka defines it as having from 'five to fifteen' bayts.³¹ Kamal Khujandi, a contemporary of Hafiz who died probably in 803/1400–1401, declared that he made a point of writing ghazals of seven bayts. This is presumably what he had in mind when he wrote in one of his ghazals: 'For the form of the ghazal Hafiz does not ride along with us,' as Hafiz's ghazals are usually more than seven bayts.³²

In the last bayt of a ghazal the poet very often mentions his own name or pen name. This does not happen invariably (although it is the ordinary case in Hafiz's ghazals), and appears to be a later development in the history of the form. Browne comments, 'In later days (but not, I think, before the Mongol Invasion) it became customary for the poet to introduce his takhallus, nom de guerre, or "pen-name" in the last bayt or maqta, of the ghazal which is not done in the qasida.'³³ Rypka states that 'this custom came into more general use only later (6th/12th century), though sporadically, it is true, at the end of a lyrical song as early as Daqiqi,' but that 'when it occurs in Rudaki it is certainly a forgery' (Rudaki died in 329/940-941 and Daqiqi between 366/976 and 370/981).³⁴ With the takhallus as with length there is flexibility possible in the form of the ghazal, but by the period of Hafiz, and in the work of Hafiz, the takhallus appears in the final bayt often enough that one may say that there was a sense of the ghazal as a form having both a beginning (the matla or initial bayt with its double rhyme) and an end (the takhallus). Any criticism of disunity in Hafiz must recognise this sense of the ghazal as a formal whole.

It is also true that each bayt in a ghazal is usually a syntactical whole and a complete thought. A misra is commonly a distinct syntactical unit, such as a clause, and it may even be able to stand as a complete sentence, although in the context of the poem it may not be the complete thought (a good example of this is QG 3/4a). That the misra is the basic unit for the meter, and that the bayt is the basic unit for the syntax and meaning is one of the things which makes the ghazal an interesting poetic form. Beyond being a complete sentence and a complete thought, each bayt is expected to have a particular point and beauty of its own. Ordinarily, the meaning is expressed epigrammatically and in the form of a rhetorical figure. The bayt has an independence which in some ways suggests an English stanza. This is an independence of form and imagery which is only in some contexts an independence of meaning.

Bayts three, five, and six in QG 3, for example, are all about the same subject, the relationship between the poet and his beloved, and they all may be said to express the same general meaning: the cruel behaviour of the beloved does not diminish the poet's passion.³⁵ If they seem unconnected, it is because in each bayt the thought is expressed in different images: the mischief-making and banditry of luliyan and Turkan in bayt three, the love story of Yusuf and Zulaykha in bayt five, and the sweetness of even bitter words from the beloved in bayt six. It is also because there are no connecting particles, or any other explicit connections to link the bayts. The lack of transitions and connectives, however, is not necessarily incoherence or disunity. What is absent is any explicit or obvious logical sequence: bayt five may precede bayt three or follow bayt six without substantially altering the meaning or the style of the ghazal, which is obviously another very interesting characteristic of the form. In addition, it should be noted that there is nothing illogical about any sequence in which these three bayts may be arranged, and no reason, beyond the change of imagery (which is, I would say, a small change in this context), to feel a jump from one to another. In a poem as short as the ghazal the number of changes of imagery alone is important. Many such changes can give the impression of disunity where there has been no change of subject or meaning.

How important the context is can be illustrated by Shakespeare's ninety-fourth sonnet:

6 They are the Lords and owners of their
  faces,
  Others, but stewards of their excellence:

8 The sommers flowre is to the sommer
  sweet,
  Though to it selfe, it onely liue and die,³⁶

If each pair of lines is thought of, momentarily, as a bayt one can feel here the same absence of any transition or explicit connection that one feels in the excerpt from Hafiz, especially if the colon is forgotten. The colon connects the lines in a way which was not available to Hafiz as his language had no punctuation marks. When the lines are restored to their context, one can see that the jump between lines seven and eight is the only one of its kind in the poem, and that it is, in terms of the whole poem, less of a jump not only because it is unique, but also because lines 1-8 develop the same idea in the same terms, and lines 6-14 present a similar unity, and because the second image, of the summer with its flowers and weeds, is developed so as to be analogous, and suggest parallels, with them 'that haue powre to hurt, and will doe none.' Between Hafiz and Shakespeare there is not so much a difference in essential technique as a difference in the way techniques are employed.

The ghazal is also defined by certain limitations as to subject matter. The Kitab al-mujam, one of the most important Persian books on prosody, rhyme and rhetoric, (which was begun in Marv in 614/1217 and completed in Shiraz in 630/1232-1233) states that: 'Any poem which has as its purpose the ways of love, the description of the curl and the mole, "and the story of union and separation, and the manifestation of passion by the mention of sweet-smelling herbs and flowers," and the winds and the rains, and the description of the traces of habitation and ruins, is called a ghazal. Originally the word ghazal meant evening conversation with girls and stories about them, and amorous, flattering talk with women. One says: a man is a ghazil, that is, a man is a gallant and one who enjoys listening. Consequently, the commentary on the states of the lover and the description of the beauty of the beloved is called a ghazal.³⁷ The ghazal, then, is a love lyric by definition, and this concentration on the single subject of love cannot but give unity to individual ghazals and it is, in part, why so many of them seem all of a piece. Here, as elsewhere, the sonnet offers many analogies. The ghazals which are religious poems and panegyrics nevertheless remain within the terms of the love lyric. They refer to 'the curl and the mole,' tell 'the story of union and separation,' comment 'on the states of the lover,' and describe 'the beauty of the beloved.'

The subject matter of the ghazal is restricted further in practice. It is very difficult to recover references to specific events, of any kind, from a poet's ghazals or even to distinguish personal experience. This is not to say that they may not contain many such references and represent the poet's experience, but that the facts have been transmuted by the poem, usually beyond recognition. This is particularly striking in love poems. As a love lyric the ghazal is both personal and impersonal. There is intimacy without autobiography. As a group, ghazals are ahistorical, which is not the case with either qasidahs or qitahs. The difficulties which Ghani had in dating Hafiz's ghazals (see his book cited in footnote 5) make this clear. This aspect of the ghazal as a form is tacitly noted by Browne when he remarks of Hafiz's contemporary, Kamal Khujandi: 'As is so often the case with Persian poets, Kamal's fragments [qitat] are much more intimate and personal, and contain more allusions to contemporary events and persons (though for lack of fuller knowledge these allusions must often remain obscure) than his odes [ghazaliyat] …'³⁸ This absence of historical detail from the ghazal has meant that it has been filled instead with emotion, genuine or spurious, and because of its shortness there was no room for complicated patterns or elaborate rhetoric.

They are associated with the qasidah.³⁹ Love of rhetoric usually finds expression in long speeches and long poems, and a short poem can sustain emotion in a way a long poem cannot. Compression means exerting pressure. It must intensify and, to a certain degree, simplify. Mere shortness in a poem makes every aspect of it more conspicuous, if not more significant. English haiku furnish an interesting illustration of this phenomenon (the more so because the form is simplified by its removal from its cultural context). The limitations of length and subject matter appear to have operated so as to make many ghazals bland and dull, but in the ghazals of Hafiz they have contributed to the depth and sharpness of the feelings.

In order to understand the ghazal as a form it is necessary to think of all these things together, the bayt and misra, the regular meter, the monorhyme, the syntactical regularity set up by the rhyme and the epigrammatic nature of the bayt, the rhetorical figures, the takhallus, the restrictions of length and of subject matter, and the body of conventional images and allusions. When they are all considered together, it can be seen that the ghazal is a very rigorous form. The uniformity, such as it is, of so many ghazals, the conformity of so many poets to this pattern, demonstrates that this definition of the form was understood and accepted. Moreover, poems such as this do not happen by themselves. There is no spontaneous poetic generation. Each part and every aspect of the poem is the result of a decision by the poet, although they all may not have been highly conscious or meditated decisions. The poet, furthermore, has the opportunity to revise his work after he has composed it, and revisions, like the decisions not to revise, are made in terms of some sense or ideal of what the poem ought to be. The greater the poet the stronger that sense of the ideal is likely to be.

If the ghazal is seen as a deliberate and deliberated creation, the question of its unity is seen in a new light. The poet must select a meter and words to fit it. He must decide whether to change the subject or the imagery. He must choose an order for the bayts and establish a length for the poem. If one bayt follows another, it is because he has so arranged it. If the poem has nine bayts, it is because he had decided that his poem demanded it and that his purposes could not be accomplished in eight. Each ghazal has the unity of the form, of being a ghazal, and of being the particular poem which it is, the result of countless definite decisions by a given poet. This does not mean that the reader or scholar, or even the poet himself, can recover all or most of these decisions, much less the order and atmosphere in which they were made. The texts of most Persian poems are such that we shall probably be permanently uncertain as to whether they are as their authors left them. Similarly, there does not seem to be any possibility of describing or reconstructing the process of composition for any of the older Persian poets. Any detailed biographical criticism is extremely difficult, if not next to impossible, because the biographical information about most Persian poets is at least as dubious and problematical as the texts, as well as, in most cases, very sparse.

That there was a sense of a poem (and of the ghazal) as an aesthetic whole can be seen in a number of old texts. In the final bayt of QG 3 the poet expresses consciousness of having completed his poem: 'You have spoken a ghazal and bored the pearls come and sweetly sing, Hafiz …' The metaphor appears to be that the ghazal is a string of pearls with each bayt probably corresponding to a pearl, each unit a unity, but all of a uniform colour, size and quality—with a hint of a vision of the whole as a necklace. The craftsman's skill which is emphasized is that of drilling holes in pearls, not, as Arberry says, sorting and grading. Furthermore, Arberry apparently understands durr sufti as 'the pearls are strung' which is incorrect.⁴⁰ In the poem it is almost as if the singing, the performance of the ghazal, strings the pearls. The metaphor of the pearls is continued in the play on nazm which means both verse and, according to Steingass, 'Joining (pearls) in a row' and 'a string (of pearls),' and is subtly echoed in the image of 'the knot of the Pleiades.' This constellation may be said to resemble a small handful of pearls ready for stringing (although this cannot definitely be said to be in the poem), and in the poem the stars are to be scattered, like a broken necklace. Images of the Pleiades have a long history in Islamic poetry and that this particular idea is present in the tradition is proved by Ibn al-Tazriyyah's bayt: 'When the Pleiades were in the sky as though they were pearls scattered from their (broken) string.'⁴¹ The image of the necklace is extremely common in Islamic literature. The anonymous friend of Hafiz who is the author of the old Preface to the Divan-i Hafiz there compares each poem to a pearl and all of them to a necklace.⁴²

Ibn Khaldun's discussion of poetry, although he is primarily concerned with the qasidah, may be used as a gloss here. He clearly asserts the independence of each bayt, but at the same time is very concerned with transitions, and behind his discussion is a notion of the poem as a whole ('the whole complex') which can be damaged by poor transitions or sharp contrasts: 'The whole complex is called a "poem" (qasidah or kalimah). Each verse, with its combinations of words, is by itself a meaningful unit. In a way, it is a statement by itself, and independent of what precedes and what follows. By itself it makes perfect sense, either as a laudatory or an erotic (statement), or as an elegy. It is the intention of the poet to give each verse an independent meaning. Then, in the next verse, he starts anew, in the same way, with some other (matter). He changes over from one (poetical) type to another, and from one topic to another, by preparing the first topic and the ideas expressing it in such a way that it becomes related to the next topic. Sharp contrasts are kept out of the poem. From a description of the desert and the traces of abandoned camps, he changes over to a description of camels on the march, or horses, or apparitions (of the beloved in a dream). From a description of the person to be praised, he changes over to a description of his people and his army. From (an expression of) grief and condolence in elegies, he changes over to praise of the deceased, and so on.'⁴³

A little further on he repeats himself, but with greater emphasis on the harmony that must exist between all the bayts: 'A poet must produce (a verse that) stands alone, and then make another verse in the same way, and again another, and thus go through all the different topics suitable to the thing he wants to express. Then, he establishes harmony among the verses as they follow upon each other in accordance with the different topics occurring in the poem.'⁴⁴ This argues a standard for the appropriateness of topics within a given subject matter, and for a stylistic 'harmony' between bayts based on their meaning. How 'they follow upon each other' is important to the poet. A sense of the poem as a whole appears very clearly when Ibn Khaldun states: 'If a verse is satisfactory but does not fit in its context, (the poet) should save it for a place more fitting to it.'⁴⁵ Throughout this discussion by Ibn Khaldun, as in the books of rhetoric, the making of a poem is a matter of craftsmanship and knowledge. The 'intention' of the poet is stressed. The decisions involved are thought of as conscious and deliberate.

The Harat preface of 907 A.H. (cited in footnote 20) is evidence that the order of the bayts in the ghazals of Hafiz was felt to be important. It is, in fact, the reason given for the new recension, and vehement invective is used against those responsible for the disorder. That cleaning or purging of the ghazals was necessary suggests that the members of the court of Faridun Husayn Khan who engaged in this task had a sense of each ghazal as a whole, as I suspect that this process was accomplished more by literary than by textual criticism.

Khvandamir, in his account of Shah Mansur's arrival in Shiraz, introduces into the narrative two bayts from a ghazal by Hafiz. On that occasion, he says Hafiz 'arranged a ghazal on the thread of style (ghazali dar silk-i insha intizam dad).'⁴⁶ This conventional phrase may indicate that Khvandamir had a sense of 'context' or 'harmony' similar to that of Ibn Khaldun, and that he felt that Hafiz's ghazals were unified by style. A theory cannot be built on a single example such as this, but together with the other evidence it is important, and it helps to show that the ideas and feelings which form the thinking about poetry in a culture (and in the Islamic culture for the period of Hafiz) are not always fully articulated.

This appreciation of the form and nature of the ghazal does enable us to see that it is not necessary to read ghazals, or any Persian poem, in any way radically different from the way in which we read sonnets, or any European poem, or any other poems. The ghazal must be read differently than a European poem in so far as it is part of a different culture, but as a poem it must be read in the same way. We see also, that the lack of an explicit logical sequence similar to that in certain European poems is not the same thing as disunity, incoherence, or a total want of logic. It becomes clear that it is not correct to assert of this lack of explicit logical sequence in the ghazal, that, 'It is as if the poets exhausted themselves to such an extent by giving form to such refinement that the fitting together as a whole escaped them.'⁴⁷ This is incorrect, not only because in many instances it is not true, and in other instances other techniques of order and 'fitting together' are to be found, but also because it is unreasonable to assume that great poets capable of solving the formal problems of the ghazal could not have successfully put their bayts in an explicit logical sequence if they had wanted to. This notion of them exhausting their abilities and energies suggests Yeats' 'Three Movements':

Shakespearean fish swam the sea, far away
  from land;
Romantic fish swam in nets coming to the
 hand;
What are all those fish that lie gasping on the
 strand?

but those that 'lie gasping on the strand' are more often the critics than the poets, and many of the charges of disunity in the ghazal must be dismissed as failures of analysis and understanding.

When one considers the rigorousness of the ghazal as a form and the strength of the literary tradition of which it was a part, it seems more reasonable to assume that many explicit logical connections were not made because they were not felt to be necessary, and that they were not felt to be necessary because the plot, logic and connections were provided (in varying degrees depending on the poem) by the tradition, by the many conventions. In writing ghazals the Persian poets may have taken for granted much of what a European lyric poet would have felt obliged to spell out. Or it may be that what is now sometimes felt as disorder and disunity is merely the expression of their individuality, that the very rigorousness of the ghazal created a desire for freedom and innovation, or a certain impatience with what was customary, which was satisfied within the tradition by including more subjects in a single poem, by glossing over or ignoring transitions, and by making tacit and allusive connections between the bayts. By the time of Hafiz it would have been difficult to write ghazals and to do anything new within the limits of the conventions in any other way. That Hafiz may have felt something of all this is suggested not only by his ghazals, but also by the story referred to above in the first footnote in the Tarikh-i habib al-siyar.

If Khvandamir is to be trusted, one day Shah Shuja, in a faultfinding mood, said to Hafiz: 'The bayts—not one—in your ghazals, from the opening to the closing, do not happen to be of one kind, instead in each ghazal there are three or four bayts about wine and two or three bayts about sufism and one or two bayts about the characteristics of the beloved. The changeableness of each ghazal is contrary to the way of the eloquent. The Khavajah said: That which the blessed tongue of the Shah discerns is the essence of truth and of unalloyed accuracy; however, the poetry of Hafiz has found consummate fame in all regions of the world and the verse of his various rivals has not set foot beyond the gate of Shiraz.' Even allowing for the situation and the temper of the two speakers it is interesting that Hafiz completely accepts the truth of Shuja's criticism. This criticism, that the ghazals of Hafiz contain too many different subjects (however they may be connected?!), depends upon an idea of the ghazal as a whole. The objection to many subjects is presumably that they destroy the unity which tradition ('the way of the eloquent') demanded in the ghazal. Shuja thinks of the ghazal as having between six and nine bayts, and of three distinct subjects, even if treated in two or three bayts each as too many for one ghazal. It would seem that he means each subject is treated in consecutive bayts, but this is not absolutely clear.⁴⁸) The views of Shuja are similar to those of Ibn Khaldun cited above. He, too, evidently feels that sharp contrasts should be kept out of a poem, and that a harmony should be established among the bayts of a poem.⁴⁹) Thus, the evidence suggests that the ghazals of Hafiz were seen by his contemporaries as a departure from tradition, and that the poet himself knew very well what he was doing.

Arberry cites the following remarks by William Leaf: 'We have learnt from our Greek masters to seek the unity of a poem in the thought or mood developed in it. Whether sensuous or intellectual, the unity is internal and essential. To a Persian poet this is not so; and that is a hard lesson which we must learn before we can do full justice to Eastern art. In the Persian ode we find a succession of couplets often startling in their independence, in their giddy transitions from grave to gay, from thought to mood. To the Persian each couplet is a whole in itself, a nukta, or "point," sufficiently beautiful if it be adequately expressed, and not of necessity owing anything or adding anything to that which comes before or after. It is from the common metre and common rhyme alone that the ode gains a formal unity … The lyric poetry of Persia is indeed a reflection of the minds of those who sang it—sensual, mystic, recalling the voluptuous dreams of Hashish, the flashes of intuition wherein the Godhead reveals himself in moments of blinding visions to the ecstatic drunk with wine, be it of Heaven or of Earth.'⁵⁰ Leaf is attacking the unity of Hafiz's ghazals, but when this passage is put side-by-side with QG 3, one can see that the poem does exhibit an obvious unity of thought and mood.

In the poem, the poet declares his passion for the Shirazi Turk and the cruelty of the Turk in not requiting his passion causes him to reflect on the meaning of life. The unexpressed logic of the poem seems to be that the Turk in being difficult, capricious and incalculable suggests to the poet that every man is treated by life as he is treated by the Shirazi Turk. Nevertheless, it is too much to say that the poet discovers in his love of the Turk an image of every man's predicament, because the two ideas are not explicitly connected and the particular occasion is not merely an excuse for a general lesson. The poem is neither allegorical nor an example of simple moralizing, nor an account of how the poet has become wise. Its beauty derives from its elegance and subtlety, from the fact that these connections are suggested and not directly stated. In a sense, the first bayt states the situation or problem and the second bayt, the solution, and the whole poem may be seen as a development of those two statements, but this, as will be seen, ignores the other ways in which the bayts are connected. Then, too, this successful parataxis cannot be called disunity if the first two bayts are recognised as the statement and resolution of a single problem.

Despite the troubles which are mentioned the mood of the poem is joyous. The extravagance of the first bayt is matched by the last. The Shirazi Turk may not return his love, but the poet is drinking wine and singing, two activities on which in the poem he places the highest value, and he is in Shiraz, which he obviously considers the best place to be. The hard words of the beloved satisfy him and make the beloved seem sweeter. The poet says he must sing sweetly (not sadly), and there are no sighs, tears or lamentations, all of which are common in Hafiz's poems. The inscrutable and knotted problems of life cause him to think of the beauty of the Pleiades, the immortality of his work, and that he should be well paid for his art. Thus, the poem is of a single mood and the character of the poet in the poem does not vary. None of his feelings and thoughts are incongruous in this context. Similarly, the character of the beloved is a unified and consistent whole. The Shirazi Turk does not share the poet's passion, but is compared to the flirtatious, trouble-making luliyan and the raiding, plundering Turks of Central Asia. The Turk is more beautiful every day, with lips as red as rubies and as sweet as sugar, and the bitter reply which those lips make to the poet is said to be completely in character. Everything in the poem suggests that the Turk will not change.

There is a further unity to this poem in that, like so many of Hafiz's ghazals, it is the discourse of a single speaker. Those ghazals which are not the discourse of a single speaker are usually dialogues. The poet is speaking, but his call to the saqi for the last of the wine gives the poem the effect of conversation rather than an interior monologue. Bayt two, consequently, should be understood as an aside in his speech instead of the beginning of a new and disconnected subject. It indicates the setting of the poem, revealing indirectly that the poet is drinking in a wine-house. The successive imperatives in the three final bayts also strengthen the impression that the poem is being spoken aloud or sung, as does the statement that the poet would like his poem crowned with stars, which contains a hint to a patron that the poet would like money as a reward for his labour. This hint makes sense only if the ghazal is addressed to an audience other than the poet himself. There is perhaps a slight problem as to who is speaking in bayt eight, which is the advice of the pir-i dana. In bayt seven the poet asks his soul to listen to the advice of the pir, which is a good reason for thinking that the poet and the pir are two different persons, although jana is a simple vocative without a possessive pronoun and may be understood as directed to the poet's audience. The poem as a whole causes me to believe that the poet is speaking to himself, as well as his audience, and that he is repeating the advice he has heard from the pir, the immediacy and applicability of which he now feels very keenly.

This ghazal is the embodiment of the poet's emotion. Hafiz is not describing something which he sees either in his mind or around him. The beloved is no more than a beautiful Turk with red lips and a black mole. The emphasis is on the beloved's behaviour and the poet's reaction to it. There is no precise or specific visual description, no concentration on the uniqueness of things. Definite places are named, but the name is left to conjure up by itself the qualities of the place. Nothing distracts the attention from the poet's feelings, but this generality is neither vague nor abstract. The emotion is personal and particular. Yusuf and Zulaykha, and the luliyan and Turkan, are as specific as Samarqand and Bukhara or the Pleiades, and even the pir-i dana and the javanan-i saadatmand must count here as specific and definite.

Emotion is, of course, more a matter of mood than of a necessary sequence, and a certain discontinuity or illogicity of thought in its literary representation may be the most accurate way to render it. A rush of feeling does not have or need a plot, except perhaps its place in the plot which is every man's life. Here there is nothing which can be identified as autobiographical or historical. How the Shirazi Turk actually fitted into the poet's life is unknown. There is no story or anecdote. Beloved and lover confront each other in what may barely be called a situation and they can barely be said to act out their parts. There are two characters and one mood. The whole poem is the brief testimony of only one witness. The philosophy, if it may be called that, in the eighth bayt has the quality of emotion. Wisdom is a better name for it, and wisdom is often felt to be virtually spontaneous and in many ways like emotion. It is significant that in the poem the advice of the pir is opposed to philosophy as a formal study (hikmat) and presented as a result of experience and old age. The advice is a simple assertion. There is no reasoning and no argument.

The pleasure which the poet takes in his own emotions is expressed in the extravagant statements which open the ghazal. If his beloved will accept his love, the poet will give away Samarqand and Bukhara, two of the richest and most famous cities in the whole Iranian world, in order to possess the mole on his beloved's face. That the poet does not possess the cities and cannot in any sense give them away is part of the extravagance (and humour) of this declaration. The gesture was such as to capture the imagination of Timur and of many of Hafiz's biographers.⁵¹ The second bayt is identical to the first in its extravagance, for in it the poet clearly implies that Shiraz is better than Paradise, and that the pleasures of this world are better than those of the next. May-yi baqi probably refers to the durd, the lees or dregs of the wine (the durdkashan or dregs-drinkers are often met with in the Divan-i Hafiz, see, for example, QG 9/5, QG 26/5, QG 110/6, and QG 131/6) so the poet is not only celebrating the drinking of wine, prohibited by Islam, but also suggesting that the dregs of the wine of this world may be superior to anything to be enjoyed in Paradise. All this is blasphemous and even dangerous in an intolerant, religiously orthodox society, and is more than rhetorical daring.⁵² Orthodoxy is challenged and mocked again in the poem with the same spirit in bayt eight when the poet refers to the hadis of music and wine. He appears to enjoy allowing his emotion to carry him away. This pitch of feeling is maintained either by grammatical forms, such as vocatives and imperatives, or by the subject matter: by the imperative, bidih, to the saqi (bayt 2), by the vocative fighan (bayt 3), and by the vocative jana, and the imperatives, gush kun (bayt 7), qu and ju (bayt 8) and biya and bikhvan (bayt 9), as well as by the luliyan stirring up trouble and the Turkan plundering (bayt 3) and Zulaykha forsaking the veil of chastity (bayt 5). That uncontrollable emotion is expressed in a rigorously controlled form is not felt at all to be a paradox in Hafiz, it seems rather to be a purpose of poetry.

This ghazal of Hafiz, like his other ghazals, shows a further form of unity in the rhetorical figures which may be traced throughout the poem. These have sometimes been seen as fragmenting the ghazal, but it must also be understood that because each bayt embodies one or more rhetorical figures, the whole ghazal is a single poetic texture. The figures are a poetic algebra in which a whole range of values may be substituted in essential forms. In Islamic literature they are not an adjunct to poetry, but of its substance. One feels in Hafiz that their use might be barely conscious, that they are a mode of his mind and moulds of his thought. The unobtrusiveness of his figures as figures is a reason for the greatness of his poetry, its smoothness and elegance. They shape the discourse and participate in it without interrupting it.

Form and meaning (content is something more) are two points of view from which a poem may be considered. In poetry both aspects must be considered together. There is no meaning without words and there are no words without form. A thorough knowledge of rhetorical figures would be a pressure on the poet not only to compose in figures, but also to vary his figures—to play on all the keys of the repertory. The demand to cast each bayt as a new and separate figure might also force a change of subject. There is, however, no need to have a certain number of figures or to have any particular figure, or have them in any particular order. They determine neither the length nor the order of the poem. The figures, as is the case with meter, both inhibit and inspire the poet. They might limit him, or serve as a store of ready possibilities, or give his thought a particular bias and momentum.…

That a poem can be described in terms of rhetorical figures is not, of course, proof that the system was known to the poet. Because rhetorical, like grammatical, analysis can be used for all language, such proof cannot consist merely of the fact that figures can be discovered in a poem. It must mean an analysis of the figures and of the way in which they figure in a poem, and historical evidence about the poet and the literary culture of the period of the poem. Everything about this poem causes one to think that Hafiz was familiar with the system of rhetorical figures, and this is confirmed by the statement in the old Preface that Hafiz studied the Miftah al-'ulum of Sakkaki and by the variety of evidence which indicates that he was a learned man.⁵⁴

That identifying the figures is not enough to describe the way in which they are employed can be shown by a single example. Mutadadd does not denote the complexities of the relationship of the Turk-i Shirazi and the khal-i Hindu. Essentially the antithesis is between black and white, although no colour is named and no abstract, general words are used. (In a similar way the antithesis in bayt two is essentially between heaven and earth. Both cases offer examples of the specificity and definiteness mentioned above.) The two terms are not equal, a part is opposed to the whole. There is perhaps a similarity in shape: the circular moon-face of the Turk and the small black circle of the mole, and both are thought to be beautiful. Two contrasting forms of beauty, however, are present. The traditional idea is expressed by Kay Kaus in the Qabus-namah: 'It is known to all that their [Turkish] and Indian [Hindu] beauty and ugliness are opposite. Look, for example, at the Turk in detail—the large head, wide face, narrow eyes, flat nose, and inelegant lips and mouth—although when you look in detail he is not beautiful, nevertheless consider everything and see it together and he is beautiful. The Indian face is the opposite of this. When one by one you examine its features, each in and of itself appears beautiful, but when you look at the whole it does not appear like a Turkish face. First, the Turk has a natural freshness and clearness of complexion which the Indian does not. The Turks have surpassed all races in freshness.'⁵⁵ Interestingly, in the poem the whole is Turkish and the individual feature is Indian. The adjective Shirazi does not enter into any of the antitheses generated by the two terms. Turk is set off against both khal and Hindu. Turk and Hindu not only stand for colours and for two types of beauty, but they represent geographical extremes and styles of civilization: the nomadic culture of the north and the settled culture of the south. Turk also belongs to the group of Turkish references which give a distinctive quality to the opening of the poem, while Shirazi fits with Bukhara and Samarqand, and with Ruknabad and Musalla. The proliferation of subdivisions and subtypes within the classification of figures represents the efforts of the rhetoricians to capture what they felt was eluding them.

The lesson of the rhetorical figures is what they suggest in the way of forms and poetic techniques. To compose poetry with the system of rhetorical figures in mind, consciously or half-consciously, means producing poetry of a particular kind. The enumeration above illustrates how the presence of many rhetorical figures creates a single poetic texture (a form of unity), and helps to make clear some of the effects which Hafiz sought as well as the nature of his control over the poem. The texture, the effects and the control are all complex because of the disparate natures of the rhetorical figures. Mutadadd, for example, is a matter of meaning, while muzdavaj depends only upon the repetition of sounds. Both may be contrasted to the three forms of mubalaghah, classified according to their truth-value, as possible, improbable and impossible.⁵⁶ Their mixed quality as a group is embodied in the terminology of the rhetoricians which is an amalgam of grammar, philosophy, and an attempt to describe in abstract terms the play of meaning and connotation in all possible poems. By means of the rhetorical figures, notions of form and meaning are tangled up together in the theory of Islamic poetry. There are, moreover, other forms and techniques in the poem which are not, in the terms of Hafiz and his contemporaries, rhetorical figures, but which may be said to be analogous to them. It is difficult to describe and classify them, because their effect is in their combination and, like the rhetorical figures, they are a mixed and disparate group.

The most common figure in the poem is mutadadd (antithesis). The bayt, as a poetic unit composed of two equal parts each long enough to contain a statement, lends itself easily to antithesis, and indeed promotes it. (There are reasons for supposing that this binary form may be a vestige of oral poetic composition.) The binary aspect of the bayt may also promote the many symmetries and correspondences of grammatical forms, the pairings and balancings, which help to unify the ghazals of Hafiz and which seem analogous to antithesis but in no way the figure mutadadd. The continuity of a long literary tradition can be felt here as the same ends and means can be observed in ancient Near Eastern texts, in Gilgamesh and in the poetry of the Old Testament. It is part of Hafiz's skill that he uses these symmetries and correspondences to complicate the structure of his ghazals and to prevent them from resolving into simple bipartite forms.

The first bayt is a conditional sentence where the first misra is the protasis and the second misra the apodosis. (The fourth, fifth, sixth and eighth bayts, although not conditional, have the same structure. The ninth bayt is similar except that the first misra is made up of short sentences.) The syntax of the first misra, however, cannot be said to be parallel with that of the second, although there are many correspondences between the two misraat. There is one verb in each half of the bayt. They are in a similar position and each is followed only by its direct object. That they are opposite in meaning is a further correspondence. The antithesis acts as a bond. Bukhara and Samarqand balance each other and are set off against Shirazi while dil-i ma is set off against khal-i … ash, Turk against Hindu. Almost every word or phrase in the first misra of the first bayt has one or more counterparts (of some kind) in the second misra. The correspondence is sometimes of linguistic form, sometimes of meaning, and not infrequently involves both. A word may be balanced by a phrase, a noun by an adjective and the corresponding words are arranged in different orders within their respective misra.

Correspondences such as these can be discovered between the first and second misra of every bayt in this poem, but, more important for the unity of the ghazal, they also exist between bayts. The whole second misra of the second bayt can be said to be parallel to Samarqand u Bukhara-ra. Each unit contains two proper nouns (both place names) joined by the conjunction va, and is the direct object of its respective sentence (and bayt). The members of each pair of place names are congruous. Samarqand and Bukhara are comparable cities in Turkestan, the banks of the Ruknabad and the gardens of Musalla are two comparable pleasances in Shiraz. This same combination of Turkestan and Shiraz exists in the name of the beloved Turk-i Shirazi, and also between Turk-i Shirazi and Samarqand u Bukhara. There are probably more correspondences of this order between the first two bayts than any other two bayts in this poem, and some other bayts (for example, three and four, or three and five) are not strongly connected in this way at all.

Antitheses established by the poem cannot be classed as examples of mutadadd. Every bayt except perhaps the ninth is built on such antithesis. The whole poem, in fact, is an expression of the antithesis between the Shirazi Turk and the poet.

What may be called groups of references and allusions are another aspect of the unity of the ghazals of Hafiz. I am not certain that this phenomenon can be delimited by any clear, comprehensive, abstract definition. I do believe that it can be defined by example and that this definition can be applied by analogy. I have attempted, therefore, to elucidate this aspect of the poem without insisting upon any classification or terminology.

There are five Turkish references in the first three bayts of QG 3. In the first bayt there is the Shirazi Turk, and Bukhara and Samarqand, the names of two famous cities of Turkestan. Hafiz in another ghazal speaks of 'the Turks of Samarqand.'⁵⁷ The third bayt refers to the Turkan at the khvan-i yaghma. Yaghma is both the name of a Turkish tribe of Turkestan and a word meaning plunder.⁵⁸ This double meaning conveys clearly the reputation of the Turks. Under yaghma, Steingass records that it is the 'name of a city in Turkistan celebrated for the beauty of its inhabitants.' The phrase may be translated either as 'the feast of the Yaghma' or 'the feast of the plunder.'⁵⁹ I am uncertain as to which sense Hafiz intends or whether he is playing on both. In any case, yaghma is a Turkish reference, and particularly apt in this poem as some of the qualities the Yaghma Turks were reputed to possess are those of the Shirazi Turk.

This repetition of meaning and connotation has its effect like the repetition of phrases and words, sounds (assonance, consonance—and their special cases, alliteration and rhyme) and meter. The Turkish references, and the other groups like this one in the poem, help hold the ghazal together. They do not form any special pattern or occur with any fixed frequency. The bond which they supply does not depend on any order or sequence. There is no mention of the Turks in the second bayt, but the allusions in the other two bayts contribute to making a whole of the initial four or five bayts. Other arrangements of the five Turkish references would give analogous effects, for example, transferring one of the two references in bayt three to bayt four or five. Neither as a group nor individually are they necessary to the action, ideas or mood of the poem, although the poem would, of course, change if they were changed. They are not themes in any sense of the word, and not all of them are traditional. As far as I know, there is no convention for alluding to the khvan-i yaghma, or mentioning Samarqand and Bukhara in the same poem with either Shiraz or the Yaghma.

Other groups of references knit this ghazal together. Three cities are named in the first bayt and the luliyan are described as city-disturbing (shahr-ashub) in the third bayt. Shiraz is referred to in the first bayt and two places in Shiraz are mentioned in the second bayt. These groups can be described as reinforcing the Turkish allusions in that they connect the same bayts with different words and meanings, and in different combinations. The fourth and fifth bayts do not share in any groups of this order.⁶⁰ The beloved's lips are mentioned in the sixth bayt, and emphasized by two strong adjectives. This glimpse of the beloved's face forms a weak but palpable connection with the opening bayt, the only other bayt in the poem in which the Turk's features are described.

The advice of the wise man forms the subject matter of the seventh and eighth bayts. Hadis and hikmat, as traditional forms of learning, help to join bayt eight with seven (which also contains two words for wisdom: nasihat and pand) even though they are antithetical in terms of the meaning of the poem. This group of words for knowledge (nasihat, pand, hadis and hikmat) is not of the same order as the Turkish references, and does not function in the ghazal in exactly the same way. Although both are unifying factors, the latter group is integral to the poem in a way the former is not. Criticism can, perhaps, go no further than this. The point is delicate and important. The descriptive phrases of a distinction such as this should not be used as categories.

The interconnections of the final three bayts are particularly subtle. The various forms of knowledge are seen as embodying different attitudes or solutions to the problem of fate, which is first described as raz-i dahr, the secret or mystery of time, and then as in mu 'amma, this enigma, riddle or puzzle. In combination with the verbs (na-gushavad u na-gushayad) with their sense of to loosen as well as to open and to resolve, there is a hint here of the image of the knot and the necklace which is so clear in the final bayt. The bored pearls, the thread of verse and the knot of the Pleiades (as has been already pointed out) all suggest the image of a necklace. Furthermore, the knot of the Pleiades may be about to be opened and the stars scattered, and the pearls of the poem may be seen as loose and ready to be strung. Aqd is thereby related to raz and muamma, but surayyara, too, has its connection with dahr. The sky is an old metaphor in Persian poetry for fate.⁶¹ The Pleiades are in the sky and in the final bayt it is the sky (falak) which is virtually personified as a cosmic power able to disperse them—at the prompting of the poet. Thus the poet can succeed where the philosopher failed. Hafiz has made a knot of the ideas of fate, sky, puzzle and necklace.

Bayt eight then can be taken as the advice promised in bayt seven, and Hafiz in bayt nine appears to obey the command in bayt eight to sing a story or hadis of music and wine. Bayt nine is not only joined to bayt eight by the reference to music but also to bayt two, as making music and drinking wine are the activities of the wine-house. Moreover, the reference is at the same time to the ghazal as a whole. The whole poem is Hafiz's song.⁶² It should be noted that bayts eight and nine evoke the scene of the wine house set in bayt two in a way similar to the manner in which bayt six brings back the picture of the beloved's face. I have also discussed above how the poet's reflections on fate are related to his passion for the Shirazi Turk. The close of the ghazal is, in a variety of ways, a true conclusion to the poem as a whole.

I do not wish to exaggerate the unity of the ghazals of Hafiz. That there are forces which hold his poems together does not mean that there are not others which, simultaneously, tug them apart, and certainly all his ghazals do not possess the same degree of unity. Each ghazal, nevertheless, must be considered as a whole, and Hafiz's ghazals cannot be described as 'a bunch of motifs only lightly tied together.'⁶³ As a rigorous poetic form in a conventional, slowlychanging literature, the ghazal had a formal unity apart from whatever meaning it expressed, and what it was as a form included matters of content. Analysis of Agar an Turki Shirazi (a representative ghazal by Hafiz) reveals the web of interrelationships which exist within the poem—beyond its unity as a ghazal. It exhibits a unity of mood and thought, which might be expected of a single speech on a given occasion, and demonstrates that changes of images do not invariably mean a change of subject matter or of the basic idea in a poem, and that the lack of a necessary logical sequence is not the same thing as disunity and need not make a poem illogical. The unity of the ghazal is created also by patterns of grammatical form and of meaning which are necessary neither to the subject nor to the mood of the poem, although they are essential to the fabric of the poem and a characteristic of the ghazals of Hafiz. That they are random, partial patterns, which are more-or-less independent of each other, and not total patterns, does not make them any the less unifying factors, and in the poem they belong at the same time, to other patterns of different orders. The distinctions of analysis should not be confused with the indivisible whole which is the poem, a whole greater than the sum of its parts.⁶⁴

Notes

¹ These texts are translated and analyzed in my book Hafiz: An Introduction [Hafiz] which I hope will be published in the near future. The Roman numerals in the references to this work refer to the chapters. There is one passage in the Tarikh-i habib al-siyar (Tihran, 1333 A.H.S.; iii, 315-316) which gives us a hint of Hafiz's attitude (see Hafiz, iii, for a translation and for an analysis), but it is only a hint and even then the story may not be authentic. Moreover, the text is late. Hafiz died in 792 A.H. and Khvandamir wrote his book between 927 and 930 A.H. Arberry did not consider this passage; see loc. cit.

² Hafiz, iii; for an inventory of the oldest, known MSS, see ii. QG stands for the edition by M. Qazvini and Q. Ghani, Tihran, 1320 A.H.S. of the Divan-i Khvajah-yi … Hafiz Shirazi. Thus, QG 440/8a refers to the first misra of the eighth bayt of ghazal 440 in QG.

³ Orient Pearls, 705.

⁴ QG, qu. See Hafiz, iii, for a discussion of the authenticity of the Preface, a translation, and an analysis of its biographical value. Dawlatshah, Tadhkirat al-shuara, edited by M. Abbasi (Tihran, 1337 A.H.S.), 365-366. The references to Hafiz in the Divan-i Kamal Khujandi are discussed in Hafiz, iii, as is the linkage of the names of Kamal and Hafiz in Jami's Baharistan.

⁵ This idea of a 'formative period' in Hafiz's life when he was influenced by other poets, especially Sadi, is developed by Arberry in his Fifty Poems of Hafiz (Cambridge, 1962), 28-33. This book was first published in 1947, a year after 'Orient Pearls at Random Strung.' The problems presented by this periodization are discussed in Hafiz, Bibliographical Note [BN]. Despite the excellent and fundamental work of Ghani in his Bahs dar asar va ajkar va ahval-i Hafiz, i (Tihran, 1321 A.H.S.), the chronology of Hafiz's poems is still largely problematical. Rypka's statement 'that important progress has been made which has led to the chronological determination of a considerable number of the poems' is too optimistic; J. Rypka, History of Iranian Literature (Dordrecht, Holland, 1968), 266. Lescot's article, 'Essai d'une Chronologie de l'œuvre de Hafiz,' Bulletin d'études orientales, x (1944), 57-100, as is not always noticed, is a review of Ghani's book (and others), and much of his reasoning is extremely tenuous; cf. Hafiz, BN.

⁶ The Arabic and Persian tadmin of Hafiz are discussed by Qazvini in 'Badi-yi tadminha-yi Hafiz,' Yadgar, i/5 (1323 A.H.S.), 67-72; i/6, 62-71; i/8, 60-71 and i/g, 65-78. Tadmin, as Arberry understands (Orient Pearls, 706, n. 1), is only a small part of what is traditional and of what is borrowed or echoed in Hafiz.

⁷ Orient Pearls, 706.

⁸ See, for example, QG 4/8, QG 42/7, QG 233/6, QG 319/10, and QG 447/7, and the passage in the Tarikhi habib al-siyar referred to above.

⁹ Orient Pearls, 707.

¹⁰ Wordsworth (1770-1850) and Turner (1775-1851), although almost exact contemporaries, do not seem to have taken any notice of each other, which is especially curious as poetry was very important to Turner. He both read a considerable amount of poetry and wrote poetry. Wordsworth and Turner were the greatest English artists of their time and great innovators. Each achieved a large body of work. They shared a taste for ordinary life and a deep love of the countryside. Nature was the subject of their best work and both turned to it for similar reasons and both felt in it a suspernatural power. The comparison is extremely interesting, so much so that it is surprising that no one has written on it, nevertheless, the detailed study of the works of Wordsworth does not help very much with the detailed study of the works of Turner and vice versa.

Jack Lindsay in his very good book, J. M W. Turner, His Life and Work (London, 1966), uses Turner's own poems and his reading of other poets to help explain his development. This is, of course, a different case than the one discussed above in the text as Turner himself, his mind and character, is usually the subject, and not style or a particular painting. Even when paintings are involved the poetic parallels help us most with meaning and least with the style, form, unity and colour, that is, with those aspects which distinguish them as paintings. Turner's writing and reading, in so far as they can be recovered, are a gloss on what he was thinking at the time he was doing a painting, and, therefore, only indirectly, upon the painting itself. The book makes the limitations of such evidence as clear as its uses. The assumption that every fact we know about an artist aids us in understanding his works must be applied with intelligence and caution. That Dr. Johnson put orange peels in his pocket (Boswell's Life of Johnson, Oxford Standard Authors edition, 1960, 602-603, and 1222) does not advance us much in understanding his greatness. This same assumption extended, by analogy, that every fact we know about a culture reveals its personality is more radically problematical, even if one accepts the idea that cultures have a personality or character. As one moves from the study of a man to the study of a culture the problems change their nature as well as increase in number and difficulty.

¹¹ Compare plates 886 and 887 with plates 847, 848 and 849 in A. U. Pope, A Survey of Persian Art, (London and New York, 1964-1965), ix.

¹² See, for example, J. D. Pearson, Index Islamicus 1906-1955, (Cambridge, 1958), and his Index Islamicus Supplement 1956-1960 (Cambridge, 1962). Under the heading of 'FINE ART: LOCAL FORMS' for Iran there is nothing specifically on the fourteenth century in either volume, nor is there anything on the period's 'MOSAICS.' Under 'ARCHITECTURE' in the two volumes there is only: M. E. Crane, 'A fourteenth-century mihrab from Isfahan,' Ars Islamica, 7 (1940), 96-100. 'MINIATURE PAINTING' is better represented. In the Index in twelve double-column pages there are ten items whose titles indicate that they are about miniature painting in Iran in the fourteenth century. Most of them are short notices of new discoveries rather than long studies. There is one item in the Supplement on this subject: R. Ettinghausen, 'Persian ascension miniatures of the fourteenth century,' Convengno di scienze morali, storiche efilologiche (1956), 360-383. Judging from the titles of the various items, which are somewhat deceptive, the lives and thoughts of the artists themselves, in any period, have rarely been a subject of study. The study of these arts is, of course, limited by what has survived. There is, as I remember, only one major surviving building in Shiraz which Hafiz may be assumed to have seen. This is the Masjid-i Jumah, which like so many of its kind, is a composite of many periods, and it has often been repaired, altered and restored since his death.

O. Grabar in his survey, 'The Visual Arts 1050-1350,' writes that 'the disastrous lack of proper monographic studies—except in the case of a very few objects and buildings—makes any generalization somewhat hazardous,' and that 'it is, at this stage of our research, still almost impossible to co-ordinate properly the monuments with the events of the time; and often in trying to explain the monuments one misses the human and spiritual context in which they were made and used.' He goes on to say that, 'almost no attempt has yet been made by archaeologists or historians to separate pan-Iranian trends from local ones or to assess the exact character of any one provincial development;' The Cambridge History of Iran, (Cambridge, 1968), v, 627-628.

¹³ B. Lewis, The Arabs in History (London, Grey Arrow edition, 1958), 141-142. The book was first published in 1950. The whole relevant passage is 139-143. Lewis makes atomism an Arab rather than a Persian phenomenon, but the distinction between what is Arabic and what is Islamic is sometimes blurred in the chapter. There is also the problem of whether, 'The atomistic outlook on life received its complete expression in the scholastic theology of Al-Ashari …' (142).

¹⁴ T. Kowalski, Na sz akach Islamu, (Kraków, 1935), 109 as quoted by Rypka, op. cit., 102; and Rypka, ibid., 102; see also 99-100.

¹⁵ By literature I mean everything written, with a special emphasis on the best writing. I believe it distorts the forms, values and ideas of the culture to refuse or fail to consider history and philosophy as literature, or to isolate, as is often done, certain works as belles-lettres, (which always carries with it at least a slightly pejorative sense) or to impose any foreign and artificial classification. It is a characteristic of Islamic literature that there are many works composed of alternate passages of prose and poetry. There are Arabic and Persian philological works wholly in rhyme, and there are histories written in an elaborate style akin to that of the maqamat and others written as masnaviyat, while books of stories, like the Gulistan, are stylistically similar to the tadhkirat. The masnaviyat of Attar, like many of the poems of Mawlana, must be considered both as philosophy and as poetry. That most of the best Persian poets were learned men and the learned nature of much of Persian poetry has not been fully appreciated. For a discussion of Hafiz's learning see Hafiz, iii.

¹⁶ Harry Austryn Wolfson, The Philosophy of Spinoza (New York, 1960), i, 10. See also the whole first chapter, 3-31 and the opening of the same author's Philo (Harvard, 1947), 2v.

¹⁷ This point is made in my 'New Material for the Text of Hafiz,' Iran, viii (1965), 114, and again with a fuller explanation in Hafiz, ii. For a history of the text of the Divan and an inventory of the known MSS older than the Khalkhali MS (dated 827/1424), the base MS of QG, and on QG as a basis for literary criticism, see Hafiz, ii.

¹⁸ QG, kt-lh and C. Rempis, 'Beiträge zur Hayyam-Forschung,' Abhandlungen für die Kunde des Morgenlandes, xxii (1, 1937), 126-127. On the two MSS see Hafiz, ii.

¹⁹ Hafiz, ii.

²⁰ Istanbul University MS F87, f. 99b—100a. I have used the facsimile published at the end of H. R. Roemer's Staatsschreiben der Timuridenzeit, (Wiesbaden, 1952). On the recension see also Hafiz, ii.

²¹ Aya Sofya 3945, f. 402a.

²² H. Ritter, 'Philologika XI: Maulana Galaladdin Rumi und sein Kreis (Fortsetzung und Schluß),' Der Islam, xxvi (1942), 239. See also Hafiz, ii.

²³ On the various MSS, see Hafiz, ii.

²⁴ G. Galimova, 'The Oldest Manuscript of the Poems of Hafiz,' [in Russian] Sovetskoe Vostokovedeniye (1959), 109. Unfortunately the first line of QG 3 is all she gives in her article. This undated MS is not 'the oldest manuscript.' For a full discussion of its date, see my 'New Material for the Text of Hafiz,' op. cit., or Hafiz, ii where the discussion is repeated with some proofreader's corrections.

²⁵ Mahdi Kamaliyan, 'Nuskhah-yi badalha-yi Divan-i Hafiz,' Farhang-i Iran Zamin, vi (1337 A.H.S.), 206. Unfortunately Kamaliyan in his collection of the New Delhi MS with QG does not give the order of the bayts and reports only those variants which he believes have value and merit ('vajid-i maziyat va rujhan'); 204.

²⁶ Orient Pearls, 706–707. When he says line he means bayt. The Persian bayt and the English line are not commensurate.

²⁷ Sudi refers to the Turk as mahbubah rather than mahbub; Sudi, Sharh-i Sudi bar Hafiz (Tihran, 1341 A.H.S.), i, 25, but compare E. Yarshater, 'The Theme of Wine-drinking and the Concept of the Beloved in Early Persian Poetry,' Studia Islamica, xiii (1960), 43–53.

²⁸ibid., i, 24.

²⁹ Hafiz, iv, where the whole poem is discussed.

³⁰ Ibn Khaldun, The Muqaddimah, translated by F. Rosenthal (New York, 1958), iii, 385. It is worth noting that he also wrote some poetry, although he did not have a very high opinion of his own efforts; see iii, 396, and also i; xliii, xlv.

³¹ E. G. Browne, A Literary History of Persia, (Cambridge, 1951), ii, 27; Rypka, op. cit., 95.

³² Hafiz, iii. For the poem see Aziz Dawlatabadi's edition of the Divan-i Kamal (Tihran, 1337 A.H.S.), 160-161. The poem imitates QG 167.

³³ Browne, op. cit., ii, 27.

³⁴ Rypka, op. cit., 99 and 124, n. 94 and 96; for the death dates see 144 and 167, n. 16 and 153. In another place (123, n. 75), Rypka writes that the regular insertion of the takhallus in the last bayt of a ghazal is a 'custom being first introduced by Hafiz.' The context is not absolutely clear here, instead of giving his own view he may be summarizing A. M. Mirzoyev, Abu Abdullo Rudaki, (Stalinabad, 1958). There were two editions of this work, one in Russian, the other in Arabic characters.

³⁵ As noted above, throughout this essay I have referred to the bayts by number according to their order in QG (as Arberry also does). In considering here bayts three, five and six as a sequence I am, of course, following, for this example, the text of Aya Sofya 3945. The point of bayt four is not very different from that of the others. The poet's love is said to be incomplete and his beloved is independent of his feelings. The poet is as unnecessary to the beloved as cosmetics. The detachment and aloofness of the beautiful one is made clear in both bayts four and five. In bayts three and six the beloved appears directly concerned with the poet. In bayt four, the poet does not assert his enduring passion, and less action is depicted. It is abstract, general and bland in a way the other bayts are not.

³⁶Shake-Speares Sonnets (London, 1609), Scholar Press facsimile (Menston, Yorkshire, 1970). This sonnet is a hybrid in form, both Shakespearean and Italian. By the rhyme scheme it divides into three quatrains and a couplet, by subject matter and imagery it divides into an octave and sestet.

³⁷ Shams al-din Muhammad al-Razi, Kitab al-mujam fi maayir-i ashar al-ajam, edited by Qazvini and Mudarris-i Radavi (Tihran, n.d.), t-y, 201-202, compare also 413-414.

³⁸ Browne, op. cit., iii, 330.

³⁹ibid., ii. 84.

⁴⁰ Orient Pearls, 703. Arberry also speaks of 'the double string' of pearls which is in no way suggested by the poem.

⁴¹ G. von Grunebaum, A Tenth-Century Document of Arabic Literary Theory and Criticism (Chicago, 1950), 75. This citation from Ibn al-Tazriyyah was widely circulated. Von Grunebaum refers to eight Arabic books in which it occurs. See also Hafiz, iii.

⁴² QG, p. qi.

⁴³ Ibn Khaldun, op. cit., iii, 373-374 (my italics).

⁴⁴ibid., iii, 375 (my italics).

⁴⁵ibid., iii, 385.

⁴⁶ Khvandamir, op. cit., iii, 321. The metaphor of the necklace is present not only in silk (thread) but in intizam which according to Steingass also means 'strung in a line,' and is from the same root as nazm mentioned above. In another passage about Hafiz, Khvandamir employs the phrase: ghazali dar silk-i nazm kashidah; ibid., iii, 316. These metaphors, with their sense that the ghazal is linear, help to show the untenableness of Wicken's idea of 'radial symmetry;' see his two articles, BSOAS, xiv (1952), 239-243 and xv (1952), 627-638.

⁴⁷ Rypka, ibid., 102.

⁴⁸ For a detailed analysis of the whole passage, see Hafiz, iii.

⁴⁹ Ibn Khaldun (op. cit., iii, 385-386) warns against 'putting too many ideas into one verse … If there are many ideas, the verse becomes crowded. The mind examines the (ideas) and is distracted. As a result, (the listener's literary) taste is prevented from fully understanding, as it should, the eloquence (of the verse). A poem is easy only when its ideas are more quickly grasped by the mind than its words. Thus our shaykhs used to criticize the poetry of the poet of eastern Spain, [Abu Bakr] b. Khafajah, for crowding too many ideas into one verse.' Abu Bakr's dates are given by F. Rosenthal as c. 451/1059-1060-533/1139.

⁵⁰ W. Leaf, Versions from Hafiz, 5–6, as cited in Orient Pearls, 703–704. Arberry properly calls Leaf's comments 'high-flown nonsense,' but does not refute their main point.

⁵¹ Hafiz, iii. The story of Timur's comment on this bayt appears in the Anis al-nas, Tadhkirah al-shuara and Lataif al-tavaif.

⁵² Ibn Battutah speaks of the piety of the inhabitants of Shiraz; The Travels of Ibn Battutah, edited by H. A. R. Gibb (Cambridge, 1962), ii, 300. The rule of Mubariz al-din is described in terms of oppressive orthodoxy by the Matla al-sadayn and the Tarikh-i al-i Muzaffar. Hafiz's feelings on this 'time of abstinence' are expressed in QG 41. On the dating of this poem, see the section on the Matla al-sadayn in Hafiz, iii. Shuja, whatever his motives, persecuted Hafiz on the grounds of his religious views according to the Tarikh-i habib-al-siyar; Hafiz, iii.…

⁵⁴ QG p. qv; Hafiz, iii. Ibn Khaldun (op. cit., iii, 337) records that: 'Contemporary Easterners are more concerned with commenting on and teaching (the Miftah) than any other (work).'

⁵⁵ Kay Kaus, The Nasihat-Nama known as Qabus-Nama …, edited by R. Levy (London, 1951), 64. This text is also evidence of a sense of an aesthetic whole independent of, or greater than, its parts.

⁵⁶ E. G. Browne, op. cit. ii, 69.

⁵⁷ QG 440/8. The text of this bayt is discussed by Qazvini in his footnotes, and also in Hafiz, iii.

⁵⁸ On the Yaghma Turks, see Hudud al-Alam, translated and explained by V. Minorsky (London, 1937), 95-98. Compare Kay Kaus, op. cit., 64.

⁵⁹ Sudi (op. cit., i, 27) understands it as 'the feast of the Yaghma,' food spread out for the Yaghma on holidays and important occasions. The dictionaries do not preserve any hint of it being a particular occasion or ceremony. Steingass defines it as 'A public feast to which all are invited, an open table,' and Redhouse, in his Turkish-English Lexicon, as 'one's board, a tray of food set out for the poor to scramble for.'

⁶⁰ For this discussion I am following, as can be seen, the text of QG. Because the text of Hafiz is not fixed, I have deliberately moved back and forth between Aya Sofya 3945 and QG in this essay so as to try and take account of as many probable textual variations as possible, so that the analysis and conclusions do not depend upon a single version of the text. I believe this way of working is necessary in cases of this kind.

⁶¹ Helmer Ringgren, 'Fatalism in Persian Epics,' Uppsala Universitets Årsskrift (1952), 1–134.

⁶² Because of the close connection of bayt eight to bayts seven and nine, and the easy and simple way in which the three can be seen to form a sequence, I prefer the sequence of QG for the final three bayts. For these reasons bayt eight seems out of place in Saray Revan Kösk 947 and Nuru Osmaniye 3822. The unifying network of references is changed, not destroyed, by these alternate sequences.

⁶³ A. Bausani, 'Ghazal: ii—In Persian Literature,' NEI, ii, 1036.

⁶⁴ I wish to express my gratitude to the American Philosophical Society and the National Endowment for the Humanities, with whose support I have finished this essay.