[In the essay below, Davidson analyzes the oral tradition from which Ferdowsi drew the Shah-Nama and in which the text figured as a recitation piece. Davidson contends that the Shah-Nama was shaped by the creativity of its oral tradition.]

The composite picture of an assembly of mōbads, whose coming together literally constitutes Ferdowsi's “sourcebook” by way of their collective recitation, can be supplemented by individual pictures, recurring throughout the Shāhnāma, of individual recitation. Here too, as in the composite picture, the idea of an archetypal book can be combined with the idea of performance, wherever Ferdowsi claims that he heard a given story from a reciter, who in turn got it from an “ancient book”: …

Now, O aged singer,
What did the book of the Truthful
          say
return to the time of the Ash-
                    kanians.
that the reciter recollected from an-
                    cient times?

VII 115.46-47

In one case, the reciter is described as having special affinities not only with the archetypal book but also with the family of the main hero of the Shāhnāma, Rostam. The hero Sām, Rostam's grandfather, is described as an ancestor of the family of the reciter himself, so that the source of the oral tradition, the performer, is directly linked to the subject of that same tradition, the hero: …

There was an old man whose name
                    was Āzādsarv,
His heart was full of wisdom, his
                    head full of words,
Who had the Book of Kings,
He traced his ancestry back to Sām,
                    son of Narimān
I will now say what I found out
                    from him;
If I remain in this fleeting world,
I will finish this ancient book
who was in Marv with Ahmad son
                    of Sahl.
and his speech full of ancient tradi-
                    tions.
who had the images and portraits of
                    pahlavāns.
and knew much about the battles of
                    Rostam.
I will weave the words together,
                    one to another.
and if my soul and intellect guide
                    me,
and leave to this world a story.

VI 322.1-7

With reference to the second of these two examples, it has been argued on chronological grounds that Ferdowsi could not actually have heard his predecessor perform epic.¹ We have already seen, however, that the medium of performance, as an authority, is just as stylized in the Shāhnāma as is the medium of the book. What is essential, therefore, in the reference to Āzādsarv is that his authority is envisaged as a performance that was heard, just as any living oral performance can be heard. Such a stylized reference to authority, then, affirms that the medium of performance is intrinsic to Ferdowsi's own poetry, which presents itself as part of a continuum in oral tradition.²

The point remains that the Book of Kings, where it is described as the possession of a performer heard by Ferdowsi, is a visible sign of that performer's authority, parallel with other visible signs such as an ancestry actually shared with the lineage of a principal hero. The final authority is not in the book itself but in the actual performance of the poem. Even the preface of the prose Shāhnāma acknowledges the authoritativeness of performance, which is then immediately tied in with the concept of “book.”

Whatever we discuss of this book must be told from statements of the dehqāns, for this kingdom was in their hands and they know the affairs and proceedings, whether good or bad, and whether more or less. Therefore we must go by what they say. Consequently, whatever we learn concerning them has been collected from their books.³

Till recent times, in fact, the Shāhnāma has survived by way of performance in oral tradition—albeit indirectly, since it has assumed the format of prose. Mary Ellen Page has made a study of the professional storytelling, naqqāli, of the Shāhnāma as it is performed in the Iran of recent times.⁴ The word for the professional storyteller is naqqāl, meaning literally ‘transmitter’.⁵ Granted, the format of this transmission of the Shāhnāma is prose, but the ubiquitous conceit of the naqqāl is that he is indeed performing Ferdowsi's Shāhnāma.⁶

The traditional social context for such performances was a setting of coffee- and tea-houses. Adam Olearius describes such a coffee-house in his account, dated a.d. 1631-32, of the main square in Isfahan.

Kahweh chane is ein Krug in welchem die Tabackschmäucher and Kahweh Wassertrincker sich finden lassen. In solchen drenen Kruegen finden sich auch Poeten und Historici welche ich mitten im Gemache auff hohen Stuelen sitzen gesehen und allerhand Historien Fabeln und erdichtete Dinge erzehlenhoeren. Im erzehlen phantasiren sie wie mit einem Stoecklein gleich die so aus der Taschen spielen.

The coffee-house is an inn in which smokers of tobacco and drinkers of coffee-water are found. In such inns are also found poets and historians whom I have seen sitting inside on high stools and have heard telling all manner of legends, fables, poeticized things. While narrating they conjure up images by gestures with a little wand, much as magicians play tricks.⁷

A typical performance by a naqqāl lasts for about ninety minutes.⁸ He uses a tumār ‘prompt-book’ which contains highly compressed thematic summaries of his own repertoire. A study of such a tumār shows that it is not a text to be adhered to but rather a skeletal outline of a story—a story that the naqqāl may expand or compress, even shift around with variations of theme; the decision is up to the performer, whose primary need is to keep his hold over the audience.⁹ The tumārs of different naqqāls covering a parallel stretch of narrative vary in much the same way.¹⁰ The naqqāl can of course diverge from his tumār as he or his audience wishes.¹¹

There are of course profound variations between Ferdowsi's verse Shāhnāma and the prose retellings of the Shāhnāma tradition. For example, whereas the reign of Bahman is covered in about two hundred distichs in Ferdowsi's Shāhnāma, it takes up forty-eight pages in one naqqāl's tumār and roughly a month's length of actual retelling in successive daily performances.¹² Yet the conceit of the naqqāl, as we have noted, is that he is indeed performing the Shāhnāma of Ferdowsi.

There is a comparable conceit among the South Slavic poets, the guslars studied by Parry and Lord. The guslar will say that he performs the song that he has learned exactly as it has always been performed. For example, Lord quotes from an interview with the guslar Suleyman Makić:

Interviewer: “Could you still pick up a song today?”

Suleyman: “I could.”

I: “For example, if you heard me sing a song, let's say, could you pick it up right away?”

S: “Yes, I could sing it for you right away the next day.”

I: “If you were to hear it just once?”

S: “Yes, by Allah, if I were to hear it once to the gusle.”

I: “Why not until the next day? … What do you think about in those two days? Isn't it better to sing it right away than later, when you might forget it after so long a time?”

S: “It has to come to one. One has to think … how it goes, and then little by little it comes to him, so that he won't leave anything out. … One couldn't sing it like that all the way right away.”

I: “Why couldn't you, when it is possible the second or third day afterwards?”

S: “Anybody who can't write can't do it.”

I: “All right, but when you've learned my song would … you sing it exactly as I do?”

S: “I would.”

I: “You wouldn't add anything … nor leave anything out?”

S: “I wouldn't … by Allah I would sing it just as I heard it. … It isn't good to change or add.”¹³

This conceit notwithstanding, the fieldwork of Parry and Lord has established that no two performances, even of the “same” song by the same guslar, are ever identical.

In the case of Persian poetic traditions, it is important to note that the narratives of Ferdowsi and of any given naqqāl can converge point-for-point—as well as diverge. And such thematic convergences between Ferdowsi and a naqqāl are in effect no different from the convergences between any two different naqqāls. It is as if Ferdowsi too were a naqqāl—the definitive naqqāl—of the Shāhnāma tradition.¹⁴

Occasionally a naqqāl will recite some of Ferdowsi's actual verses.¹⁵ Again this may be a matter of convergence, not derivation. After all, there can be found, in a given naqqāl's tumār, verses in mutaqārib meter that are not even attested in the canonical version of Ferdowsi's Shāhnāma.¹⁶ In fact, when the naqqāl introduces his story, he can use rhymed prose or a combination of poetry and rhymed prose,¹⁷ and sometimes there is even a melody.¹⁸ Thus the traditional format of a naqqāl's introduction may reveal vestigial aspects of an earlier stage in the art-form of the naqqāl when his medium was indeed all poetry.¹⁹

But the crucial indication of the naqqāl's independence from Ferdowsi's Shāhnāma lies in the fact that there is much narrative material attested in the naqqāl's oral traditions that is not attested in any of the literary epics so far known.²⁰ Many of the themes found in later literary epics such as the Garshāspnāma appear as an integral part of the naqqāl's narrative repertoire.²¹ For example, the Garshāsp stories will be included in the naqqāl's story-line where it would have been chronologically appropriate for the Shāhnāma to include it. Nöldeke condemns these later epics as not folklore but invention: “It is a common opinion that a great deal of popular epic has been preserved in those poems. It might sound a little bold if I flatly deny that and declare the contents of those narratives to be essentially a free fancy of the respective authors.”²² Similarly, Mohl claims that the epics after Ferdowsi's Shāhnāma are not only artless but also simply a matter of filling in lacunae left by Ferdowsi, with no pride in authorship.²³ There can only be a limited number of ultimate poets, however, and what is worthy of special note is the sheer mass of poetic compositions that deal with material beyond Ferdowsi and which usually take a verse from Ferdowsi as a point of departure.

These considerations bring us to one of the major problems confronting the “scientific method” of editing Ferdowsi's Shāhnāma. The manuscripts seem to be full of “interpolations,” sometimes massive ones, from other literary epics; at other times it is impossible to establish the provenience of the “interpolation.” But it can now be seen, from the perspective of studies centering on the naqqāl traditions of the Shāhnāma, that such interpolations may correspond to actual conventions of performance. In other words, the divergences of manuscripts in this regard may be parallel to divergences in performance, since the tumār allows the naqqāl to expand or compress during any performance in patterns of thematic variations that are clearly parallel to those of the manuscripts. And, as noted, the naqqāl can diverge from his tumār.

Thus the Shāhnāma tradition has survived till recent times, albeit indirectly, as a medium still dynamic, still alive. It could theoretically generate an infinite number of performances—provided that the naqqāl is still there to perform and the audience is there to listen. In this light, we may call into doubt the theory that there were gaps in the story-line of the Book of Kings—gaps that had to be filled with Ferdowsi's “Phantasie.”²⁴ From what we can see even from the naqqāl traditions, there was a seemingly endless reservoir of narrative traditions standing ready to be filled in at any point in the retelling of the Book of Kings. And just as the naqqāl testifies that he is indeed performing Ferdowsi's Shāhnāma, so also Ferdowsi himself testifies that he is “translating” the Pahlavi Book of Kings.

After comparing what the poem says about itself with the external evidence about Middle and New Persian poetry, we may now isolate characteristics of oral poetry as formulated in current scholarship. The fieldwork on oral poetics by Parry and Lord corroborated Parry's earlier work on the crystallized traditions of ancient Greek epic as they survived in the Homeric Iliad and Odyssey. For our present purposes, the most important aspect of the findings of Parry and Lord is their observation that the formal building blocks of oral poetry consists of what they call formulas.

Parry's working definition of the formula had been as follows: “A group of words which is regularly employed under the same metrical conditions to express a given essential idea.”²⁵ This definition, devised by Parry on the basis of his work on Homeric poetry, before he even started work on the living poetry of the South Slavic tradition, has proved enduring despite the need of one small adjustment. Ironically, this adjustment has been prompted at least partly by the evidence of Homeric poetry itself: it has recently been shown that the metrical conditions of the formula can vary, although this variation itself is systematic.²⁶ Thus it may be useful to revise the phrase “under the same metrical conditions” to read instead “under fixed metrical conditions.” The phrase “to express a given essential idea” is crucial, since this aspect of Parry's definition has often been undervalued or even missed altogether. It is important to stress that, for Parry, the formula is not simply a phrase that is repeated for its metrical utility.²⁷ Rather, the formula is the expression of a traditional theme. To quote Parry, “The formulas in any poetry are due, so far as their ideas go, to the theme, their rhythm is fixed by the verse-form, but their art is that of the poets who made them and of the poets who kept them.”²⁸ The word “theme,” according to Lord's working definition is “a subject unit, a group of ideas, regularly employed by the singer, not merely in any given poem, but in the poetry as a whole.”²⁹ In other words, the Parry-Lord definition of oral poetry is founded on the proposition that the traditional formula is a direct expression of the traditional theme; in oral poetry, there is a formulaic system that corresponds to a thematic system.³⁰

In a 1977 book by Ruth Finnegan, however, which purports to present the overall subject of oral poetry to the general reader, this basic aspect of the Parry-Lord definition of the formula goes unmentioned.³¹ She consistently treats formula as if it were merely a repeated phrase, repeated simply for its metrical utility. In discussing Homeric epithets, for example, she writes that they “are often combined with other formulaic phrases—repeated word-groups—which have the right metrical qualities to fit the [given] part of the line.”³² In the same context, she quotes Parry for support: “In composing [the poet] will do no more than put together for his needs phrases which he has often heard or used himself, and which, grouping themselves in accordance with a fixed pattern of thought [emphasis mine], come naturally to make the sentence and the verse.”³³ We see here that Parry is saying much more than Finnegan, however; the formula is not just a phrase that the poet is free to choose according to his metrical needs,³⁴ since the formulas are regulated by the traditional themes of the poet's composition. By contrast, Finnegan seems to assume that formulas and themes are separate ingredients in the poet's repertoire: “As well as formulaic phrases and sequences [emphasis mine], the bard has in his repertoire a number of set themes which he can draw on to form the structure of his poem.”³⁵ Working on the assumption that formulas are simply stock phrases repeated to fill metrical needs, Finnegan offers the following criticism of the Parry-Lord theory of oral poetry: “Does it really add to our understanding of the style or process of composition in a given piece to name certain repeated patterns of words, sounds or meanings as ‘formulae’? Or to suggest that the characteristic of oral style is that such formulae are ‘all-pervasive’ (as in Lord 1960, p. 47)?”³⁶ In light of what I have adduced from the writings of Parry and Lord, I find this criticism unfounded; if the formula is the building-block of a system of traditional oral poetic expression, then I cannot find fault with Lord's observation that the formulas are “all-pervasive” in oral poetry.

Another important point of disagreement between Finnegan and Lord is her insistence that, on the basis of what we know of oral poetry in such cultures as that of the Bantu of South Africa (both Zulu and !Xhosa), the oral poet can both compose poetry and write it down.³⁷ It is tempting, of course, to extend such findings to medieval European poetry, where the fundamentals of what is freely acknowledged as oral poetry are preserved and transmitted by literati in the context of a vigorous scribal tradition. Finnegan's point of contention with Lord provides ammunition for medievalists who have argued that an Old English poem like the Beowulf cannot be considered oral poetry on the basis of the formulas that we find as its building-blocks, simply because we can find comparable levels of formulaic behavior in other Old English poems which were clearly written compositions and some of which were even translations from Latin originals. As one expert concludes, “To prove that an Old English poem is formulaic is only to prove that it is an Old English poem, and to show that such work has high or low percentage of formulas reveals nothing about whether or not it is literate composition, though it may tell us something about the skill with which a particular poet uses tradition.”³⁸

An important challenge to such a position has been proposed by Michael Zwettler: applying the work of the medievalist H. J. Chaytor,³⁹ Zwettler suggests that, even when an Old English poem is written down, it is not meant to be read by an individual but performed before an audience.⁴⁰ In other words, as Zwettler points out, there is no such thing as an “audience of readers” in medieval European poetry.⁴¹ To quote Chaytor: “The whole technique … presupposed … a hearing, not reading public.”⁴² The rules of this poetry, written or not, are those of oral poetry. Zwettler extends this principle to pre-Islamic Arabic poetry, and I for my part hope to extend it here to the New Persian poetry of Ferdowsi. So long as I can argue that the building-blocks of his Shāhnāma are functional formulas, I can also argue that his poetry is based on the rules of oral poetry.

In the appendix to this book, using as a test case a randomly selected passage, I show that every word in this given passage can be generated on the basis of parallel phraseology expressing parallel themes. The degree of regularity and economy⁴³ in the arrangement of phraseology is clearly suggestive of formulaic language. Moreover, the regularity extends to the actual variation of phraseology. This factor may well be an important additional clue to the formulaic nature of Ferdowsi's Shāhnāma. As Parry and Lord had noticed in their South Slavic fieldwork each new performance/recomposition of a song involved variation in the deployment of formulas. This principle has been applied successfully by Michael Zwettler in his study of classical Arabic poetry.⁴⁴ He extends the observations of the Romance philologist Ramón Menéndez-Pidal, who has drawn attention to the fact that three of the earliest manuscript versions of the Chanson de Roland do not share a single identical verse with each other.⁴⁵ He had inferred from this and other such facts that this kind of poetry, which Zwettler rightly equates with oral poetry, is “a poetry that lives through variants.”⁴⁶ “How ironic,” Zwettler writes, “that scholars of Arabic poetry have so often cast doubt upon the ‘authenticity’ or ‘genuineness’ of this or that verse, poem, or body of poems, or, sometimes, of pre-Islamic poetry in general, because they have found it impossible to establish an ‘original version’.”⁴⁷ The same irony, as we shall see, applies to scholars of Persian poetry. Zwettler goes on to say that

the multiplicity of variants and attributions and of formulaic phrases and elements attested for the great majority of classical Arabic poems may undermine our confidence in ever establishing an “author's original version”—as indeed it should! But they ought to convince us that we do have voluminous record of a genuine and on-going oral poetic tradition (even if in its latest stages), such as no other nation can match in breadth of content and scrupulosity of collection and documentation.⁴⁸

The conscientiousness of those who preserved all these variants in their editions is a reflection of an attitude that we also witness in the context of Islamic oral transmission, or hadīth, and Zwettler insists that the editors' quest for authenticity by way of examining and collecting all variants was due not so much to any need of determining the author but to the desire of recovering the authentic poetic traditions of Bedouin poetry.⁴⁹

An analogous principle of variation, I propose, can also be applied to the text tradition of the Shāhnāma.⁵⁰ A systematic and exhaustive application, of course, is at this point impossible, since there is no available centralized collection of all the variants as could be collected from the entire textual tradition. Such a collection would be a monumental task indeed. Still, the limited experiment of formulaic analysis that I present in the appendix illustrates the principle of compositional variation as reflected by textual variation. The examples could be multiplied by the hundreds, even thousands, and by then we would start to see clearly that there are legitimate formulaic variants attested for vast portions of the Shāhnāma. We may postpone for later any questions about how these considerations may affect our evaluation of the standard Moscow edition. What is important is that even a preliminary test reveals such patterns of variation in the text of the Shāhnāma—the surest available sign that we are dealing with the heritage of oral poetry.

We must note, however, an essential difference between the patterns of variation in the textual tradition of the Shāhnāma, as revealed by its textual transmission, and in the Arabic poetry studied by Zwettler. In the case of the Arabic evidence, the variants seem to have been collected while the given poem was evolving into a fixed text in the process of continual performance/recomposition. In the case of the Shāhnāma, on the other hand, the variants seem to have gone on accumulating even after the composition had become a fixed text by way of writing. This fact alone suggests that, side by side with the written transmission of the text, the oral transmission of poetry continued as well. Each new performance must have entailed recomposition, and the oral poetry must have continually influenced the text.

This means that we cannot reconstruct with any absolute certainty the original composition of Ferdowsi, because of its susceptibility to recomposition with each new performance in a living oral tradition. All we can say about the original is that, if it is capable of being recomposed, it too must be a product of oral composition. And the continual recomposition on the level of form was matched by recomposition on the level of content, leading to new accretions that are anachronistic to the patterns of earlier layers.⁵¹ We may even compare the accretion of Muslim elements in the Arab pre-Islamic poetic traditions studied by Zwettler:

We must consider the alleged “inconsistencies,” “anachronisms” and “Islamic emendations” that do crop up in our received texts and have so frequently been adduced as proof of the “corruption” of the tradition. Such phenomena as the introduction of post-Islamic expressions or neologisms into archaic poems, elimination of pagan theophoric names or substitution of the name Allāh, allusions to Qur'ānic passages or Islamic concepts or rituals, and so on, can all legitimately be seen as a natural result of the circumstance that versions of those poems were derived from oral renditions performed by Muslim renderants conditioned now to the sensibilities of Muslim audiences.⁵²

Similarly, we find the accretion and eventual dominance of Shī'ite elements in the poetry of Ferdowsi, which seems to show traces of an earlier Sunni patronage.⁵³ But even if we cannot reconstruct the original composition, its authority or authenticity as tradition could survive the countless accretions and reshapings of each recomposition in performance. That is the true nature of oral poetry.

Let us imagine going back in time to a point where the oral tradition of the Shāhnāma was still in mutaqārib verse. At such a point, the formal and thematic variations of performance/composition in oral poetry would surely have affected the manuscript tradition of Ferdowsi's Shāhnāma. Even as late as in the era of Bāysonghor Mirza (died a.d. 1433), it seems that no definitive edition was possible, because the extant manuscripts were clearly no better than tumārs would be in recent times for the purpose of deciding what is definitive and what is not. The song must have existed in performance. Even though Nöldeke yearns for the attestation of the original “critically revised” copy of the Shāhnāma commissioned by Bāysonghor in a.d. 1425,⁵⁴ he realizes that he would be disappointed if it suddenly came to light: “How,” he asks, “could the Persian bel-esprit in those times—only such can be thought of, if it were not simply copyists—have managed to accomplish a great and purely philological work somewhat critically?”⁵⁵ The point is, if no copy could be definitive and preemptive even as late as 1425, it may be that each copy was, at least in part, a reflection of traditions in performance.

Nöldeke says that redactors in the era of Bāysonghor could not be “scientific” about consulting other manuscripts, for they had no Aristophanes, no Zenodotus, no Aristarchus in the Timurid court.⁵⁶ The princely libraries were full of manuscripts of the Shāhnāma such as the one described as “a fine looking, beautifully written, and very defective copy” (to quote from a contemporary evaluation).⁵⁷ How, we may ask, were such copies of the Shāhnāma “defective”? Is there a trace here of a contemptuous attitude on the part of those better versed than others in the performance of poetry? In Nöldeke's own words, there exists no “final touch” for Ferdowsi's Shāhnāma.⁵⁸

As if to console himself, Nöldeke adds: “We are not really worse off than with the text of Homer.”⁵⁹ But the Homeric analogy in fact leads back to the factor of performance in the constitution of the text of any poetry that is built on an oral tradition. Since Nöldeke's days, new discoveries have emerged about the factor of performance as it affects the canonical text of Homer. It now appears that even the Homeric text is replete with variants that are to be attributed not to textual inconsistencies but rather to actual formulaic alternatives.⁶⁰ So also with the textual tradition of Ferdowsi's Shāhnāma in its present form: as even Nöldeke concedes, it is replete with “various genuine versions” of given passages.⁶¹ In other words, a given passage may have two or more textual variants that are not a matter of one genuine reading and one or more corrupt or interpolated readings, but rather of two or more traditional alternatives, either or any of which would be acceptable to the discerning audience of Ferdowsi's Shāhnāma.

It is now possible to imagine how Ferdowsi's Shāhnāma could survive and prevail, albeit with accretions and modifications, if we allow that everything in it is traditional. In his own lifetime, Mohl had an ever-growing intuition that Ferdowsi invented nothing, and he says so most forcefully: “The more one studies the work of Ferdowsi, the more one is convinced, I believe, that he invented nothing and that he was content to restore in brilliant hues the traditions that formed the popular stories of Persia.”⁶²

For Ferdowsi, the writing down of his composition would make permanent his appropriation of living poetic traditions. For a typical oral poet, by contrast, appropriation could ordinarily be achieved only in the context of performance. But even if Ferdowsi's book, his Book of Kings, constitutes a more lasting way of establishing appropriation of his composition, it is nevertheless not a frozen text, like some Pahlavi book. To put it schematically, we could say that the survival of Ferdowsi's Shāhnāma depends not so much on the writing down of the composition as on the Persian nation's general approval of the writing down of the performance of the composition. And the approval of Persian audiences through the ages could happen only if the Shāhnāma were traditional, that is, if it conformed to the rules of composition-in-performance. It could even be claimed that the survival of the Shāhnāma, in the context of countless performances for countless audiences steeped in oral poetry, is the best argument for its own essence as oral poetry. It is also, of course, the best argument against the notion that the poetic form and overall content of the Shāhnāma were in any sense an invention of Ferdowsi.

If, however, we accept the idea that the medium of the Shāhnāma is that of traditional oral poetry, we should expect that it will be subject to accretions and modifications in the context of each new performance. Thus the recording of an original composition by Ferdowsi, in the process of its textual transmission, would be continually subject to interference from the concurrent process of oral transmission in performance, since each performance in oral poetry entails recomposition. Thus the Shāhnāma really could not ever become a completely fixed text until the oral tradition died altogether. The manuscripts of the Shāhnāma seem to reflect a period when oral poetry had not yet died, so that editors are left to struggle with the textual variants that are not just a matter of textual transmission. As we have seen, it seems that some variants are also a matter of oral transmission.

The archetypal fixed text of Ferdowsi's Shāhnāma can never be recreated, since it would be impossible to decide in any given instance which of, say, two “genuine” variants was actually composed by Ferdowsi. To understand the full creative range of the Shāhnāma tradition, it would be more important to have an edition that lists all variants, since many of these will be a matter of composition/performance, not text.

If indeed textual variants arise from the perpetuation of the Shāhnāma in performance, we need just the opposite of the so-called critical Moscow edition (1960-1971) of Y. E. Bertels and his colleagues.⁶³ This edition strips Ferdowsi's Shāhnāma to its bare bones (50,000-odd distichs), selecting variants essentially on textual grounds by comparing “superior” and “inferior” manuscripts. It is based essentially on five manuscripts:

(1) L. = ms. Add. Or. 21103 of the British Museum, London, dated a.d. 1276, the oldest extant ms. at the time that the work on the Moscow edition was proceeding; contains the preface of Abu Mansur.

(2) I = ms. 329 of the National Public Library of St. Petersburg, dated a.d. 1333 and the second oldest ms. after L.

(3) IV = ms. S.1654 of the Oriental Institute of the Academy of Sciences, St. Petersburg, dated a.d. 1445; contains the preface of Abu Mansur.

(4) VI = ms. S.822 of the Oriental Institute of the Academy of Sciences, St. Petersburg, dated a.d. 1450.

(5) K = ms. S.40 of the Dār al-Kutub al-Misrīya, Cairo, dated a.d. 1394,⁶⁴ utilized only in volumes IV-IX of the Moscow edition; contains the preface of Abu Mansur.

In view of the fact that there are about 500 extant manuscripts of Ferdowsi's Shāhnāma,⁶⁵ and especially in view of all the variations in manuscript readings, the restriction of the editorial field of vision to five manuscripts is a bold move indeed. The Moscow editors' confidence in this particular 1 percent of manuscript evidence was based primarily on two facts: that this particular manuscript family was singularly old and that this family inherits the “older preface,” that is, the preface of Abu Mansur.

But we have already seen that the “older” preface of Abu Mansur, no matter how valuable it is for understanding the history of early Persian prose, cannot be directly linked to the composition of Ferdowsi's Shāhnāma. Even on textual grounds, there is a contextual gap between the poetry of Ferdowsi and this particular preface, in marked contrast with the preface of the recension of Bāysonghor.⁶⁶

The latter recension of Bāysonghor, transmitted in a vast family of manuscripts, represents our “vulgate”: the Calcutta edition (Macan 1829) follows it closely, and this edition, collated with the eclectic Paris edition of Mohl (1838-1878), is the basis for the incomplete Leiden edition of Vullers (1877-1884) and the completed Tehran edition of Nafisi-Vullers (1934-1936). But the recension of Bāysonghor is late: the preface is dated a.d. 1425, in marked contrast to the preface of Abu Mansur, dated a.d. 957. In view of this contrast, the Moscow editors of the Shāhnāma considered the Bāysonghor recension inferior, as opposed to the recension represented by the family of manuscripts L, I, IV, VI, and K, a recension that seems to have had affinities to the preface of Abu Mansur. Guided by the reasoning that a more recent recension must be inferior to an older recension, the Moscow editors as a matter of policy rejected variant readings stemming from the Bāysonghor recension. They also eliminated readings that could not be verified from the collective testimony of the old family of manuscripts that they had isolated, thereby reducing the corpus of the Shāhnāma to 48,617 distichs, to which are added in the appendix another 1,486 distichs, deemed probably spurious. We may appreciate the extent of this textual reduction by comparing the number of distichs in the Calcutta edition, 55,204.

But the basic principle of the Moscow edition, that the older group of manuscripts is by necessity closer to the “original,” is open to question. If, as I claim, the many variations in the textual transmission of the Shāhnāma are due at least in part to the rich repertoire of concurrent oral poetic traditions, then each attested variation must be judged on its own merits, regardless of its textual provenience.

Moreover, the Moscow edition's dependence on the manuscript family L, I, IV, VI, and K, to the exclusion of others, must now be brought in line with the discovery of yet another manuscript of the Shāhnāma:

F = ms. Cl.III.24 (G.F.3) of the Biblioteca Nazionale Centrale, Florence, dated a.d. 1217.

Here, then, is a document considerably older than L, which in turn is dated a.d. 1276 and which had been for the Moscow editors the oldest extant manuscript of the Shāhnāma. As Angelo Michele Piemontese, the discoverer of F, has demonstrated, this manuscript, two centuries away from the traditional date of the completion of Ferdowsi's Shāhnāma, is replete with valuable new readings that are not to be found in the manuscript family of L;⁶⁷ there are also about two hundred “new” distichs attested—distichs that have not been known to exist before.⁶⁸ This is not to say that F is closer to the “original” than L simply by virtue of being older than L. Moreover, this is not to discredit L and its family, as opposed to F. Rather, the point is simply that the editorial field of vision cannot be restricted to the family of L.

In fact, the preface of F is clearly in the same tradition that we find attested in the much later preface of the Bāysonghor recension.⁶⁹ Even more important, the actual variants that we find in F correspond far more closely to those of the Bāysonghor recension than to those of L and its family.⁷⁰ Thus, ironically, the Calcutta edition and its offshoots, most notably the Paris edition of Mohl and the Tehran edition of Nafisi-Vullers, contain “genuine” aspects of the Shāhnāma tradition that have been neglected by the “critical” Moscow edition.⁷¹ What we need is an edition of the Shāhnāma that accounts for all the variants, each of which may be a reflex of variation in the oral tradition. In addition, we need a concordance that would include all variants that are demonstrably not just a matter of textual corruption or editorial tampering. With the aid of such an ideal edition and ideal concordance, we could demonstrate more rigorously both the power and the flexibility of the oral tradition as it was kept alive in Ferdowsi's Shāhnāma.

Even without such ideal aids, however, we can already begin to appreciate the qualities of oral tradition in the poetry of the Shāhnāma. As I hope I have demonstrated, there is enough evidence, both in the Shāhnāma and in the history of Persian poetry before and after this monumental composition, to show that the creative power of a rich oral tradition produced and then maintained the authority of the national poem of the Persians.

Notes

1. See Shahbazi 1991.133.

2. I therefore disagree with Shahbazi's inference, ibid., on the basis of such examples as the anachronism just mentioned, that Ferdowsi was literally copying from earlier sources in earlier books whenever he introduces a narrative with such phrases as “So I have heard from the aged dehqān.” Such an inference simply displaces to a previous poet what Ferdowsi is doing in the present, that is, claiming a previous performance as the authority for what is “now” written in his book. Even in terms of the inference, the hypothetical previous poet would still be doing exactly the same thing. I maintain that, even if Ferdowsi follows the tradition of earlier books when he bases the authority of his written word on the authority of a continuum of performances that precede him, we have no reason to doubt that he could have direct access to that same continuum—and in fact that he could be part of it.

3. Minorsky 1964.269.

4. Page 1977.

5. Ibid. 4. For a survey of the contexts of the root naqala, see ibid. 16.

6. Ibid. 224. She adds at p. 2, paraphrasing Maranda and Maranda 1971.12: “Once the reworking of the tradition ceases to be meaningful to the audience, the tradition will disappear despite written versions.”

7. Olearius 1656 [1971].558.

8. Page 1977.55.

9. Cf. also the work of Smith 1979 on Indian oral epic traditions where pictures are used to prompt as well as explain the narrative.

10. Page 1977.134-151.

11. For examples, see ibid. 140, 142, 146, 277.

12. Ibid. 143.

13. Lord 1960.26-27.

14. Cf. Page 1977.150, where she describes Ferdowsi's Shāhnāma as “one recension of a work which continues being told today.”

15. Ibid. 67.

16. See, e.g., ibid. 73.

17. See ibid. 67.

18. Ibid. 118, n1. For comparable forms of expression in other cultures, where song, poetry, and prose are combined, see, e.g., Nagy 1990a.47, esp. n47.

19. Cf. Zumthor 1972.99-100 on the phenomenon of dérimage in Old French poetic traditions.

20. Page 1977.128; see, e.g., the details of the Sohrāb story at pp. 135-139.

21. Ibid. 128.

22. Nöldeke 1930.209.

23. Mohl 1838.liv.

24. Hansen, 1954, 116.

25. Parry 1971.272.

26. See Nagy 1990b.18-35.

27. Cf. Parry 1971.304.

28. Ibid. 272.

29. Lord 1938.440; see also Lord 1974.206-207.

30. For a useful survey of recent scholarship on the interrelationship of formula and theme, see Cantilena 1982.41-73. On p. 56, he offers this summary: “Ogni formula, dalla più stereotipa alla più consapevolmente usata, è motivata semanticamente.”

31. Finnegan 1977.

32. Ibid. 59.

33. Parry 1971.270.

34. Cf. Finnegan 1977.62: “He can select what he wishes from the common stock of formulae, and can choose slightly different terms that fit his metre … and vary the details.”

35. Ibid. 64.

36. Ibid. 71.

37. Ibid. 70, citing the work on Bantu oral poetry by Opland 1971.

38. Benson 1966.336.

39. Chaytor 1967.10-13 and chapters 4 and 6.

40. Zwettler 1978.15-19.

41. Ibid.

42. Chaytor 1967.13.

43. For this concept, see Lord 1960.53.

44. Zwettler 1978.

45. Menéndez-Pidal 1960.60-63; cf. Zumthor 1972.40-41.

46. Zwettler 1978.189.

47. Ibid.

48. Ibid. 212.

49. Ibid. 203.

50. Cf. Pearsall 1984.126-127, with reference to fifteenth-century English manuscript production: “The surviving manuscripts of a poem like Beves of Hamptoun make it clear that each act of copying was to a large extent an act of recomposition, and not an episode in a process of decomposition from an ideal form.” On the notion of mouvance, where the text is reconstituted with the production of each new copy, see Zumthor 1984.160. For an editorial application of this principle, see the exemplary work of Pickens 1978.

51. Such a possibility is emotionally and sarcastically resisted by Minavi 1972.110.

52. Zwettler 1978.221.

53. Cf. Shahbazi 1991.52.

54. Nöldeke 1930.125-126.

55. Ibid. 126.

56. Ibid.

57. Testimony by way of M. Lumsden, appendix 5, Nöldeke 1930.126.

58. Ibid.

59. Ibid. 127.

60. Cf. Muellner 1976.57-62.

61. Nöldeke 1930.125.

62. Mohl IV p. ii: “Plus on étudiera l'oeuvre de Firdousi, plus on se convaincra, je crois, qu'il n'a rien inventé, et qu'il s'est contenté de revêtir de son brillant coloris les traditions qui formaient l'histoire populaire de la Perse.”

63. For a brief history of this edition, see Yarshater 1988.viii-ix. On the principles governing the new edition of Khaleghi-Motlagh (1988), see Yarshater 1988.x-xi.

64. On the problems in dating this manuscript, see Yarshater 1988.ix.

65. See Piemontese 1980.11-12, n27.

66. See ibid. 32-34.

67. See ibid. esp. 218-221.

68. Ibid. 222-226.

69. Ibid. 31-34; cf. Shahbazi 1991.4, n9 (paraphrase of this preface at 5-6).

70. Piemontese 1980.218-219.

71. Ibid. 194, 219. In n145 of p. 219, for example, Piemontese cites cases where the readings of the Florence manuscript vindicate the adopted readings of (1) the Calcutta edition as against the Moscow and Tehran editions, (2) the Paris edition of Mohl as against the Moscow and Tehran editions, and (3) the Paris edition and the Manuscripts I, IV, VI of the Moscow edition as against the Tehran edition and the preferred manuscript L of the Moscow edition. Khaleghi-Motlagh (1988) takes into account the evidence of the Florence manuscript. See also Yarshater 1988.viii-xi.