The Medieval Arabic Nights
The first thing to be said is that The Thousand and One Nights is a rather exceptional work in the context of medieval Arabic literature. It happens sometimes that a person takes up the study of a language because of his love for a single work, but if someone were tempted to begin the task of learning Arabic because of his love of The Thousand and One Nights, he should be forewarned that the book is sui generis. He will really find nothing else like it in the literature, one reason being that the Nights seems to have absorbed a number of once independent medieval Arabic fictions; the story of “Sindbad” is probably the most famous example. The borders of this text were not, it seems, ever very well defined. Hence the size of the Nights. Unfortunately, in the case of the Nights its marginality in this respect has also worked to veil its history in a good deal of obscurity. Indeed, in recounting its history in the medieval period, there is no need to summarize; a fairly complete account will read like a summary, since most of its medieval history is unknown and is likely to remain unknown. To retell the story, let us think of it for the moment as a piece of architecture—a palace, as Borges calls it. “To erect the palace of The Thousand and One Nights, it took generations of men, and those men are our benefactors, as we have inherited this inexhaustible book, this book capable of so much metamorphosis,” Borges said of one of his favorite books.
The ‘palace’ of the Nights, in the form in which we now know and enjoy it—which is to say the nineteenth century Arabic editions printed in Egypt, India and Germany that have served as the basis for all but one of the translations since Galland—that structure with its spacious pavilions, its charming recesses, its secret chambers and mysterious passages must have been the work of many literary hands, and it must have been the result of a development in Arabic alone that lasted seven or eight centuries. Evidence about the development of the book may be divided into two categories: testimony by medieval writers and what in a legal hearing would be called ‘material evidence’—here, textual evidence—those stories in The Thousand and One Nights that show some relation to other works in medieval Arabic literature and to works in other literatures.
Earliest mention of a text with a clearly antecedent relation to the contemporary work is found in a papyrus dating from the ninth century A.D. The papyrus mentions two characters, Dînâzâd and Shîrâzâd—later to become Dunyâzâd and Shahrazâd—and has a few lines of narrative in which the former asks the latter to tell a story. There is also mention of a title that anticipates the title we now know: “The Book of Stories From the Thousand Nights.”
About a century later two writers in Baghdad, al-Mas‘ûdî and Ibn an-Nadîm, mention the same work. In his book Meadows of Gold (Murûj adh-dhahab) the historian Mas‘udi (d. 956) states that among the translations made in Baghdad of stories from Persian, Indian and Greek sources was a book called “A Thousand Tales” (Alf khurâfah), also known as “A Thousand Nights” (Alf laylah). Mas‘udi says that it is the story of a king, his vizier, the vizier’s daughter and her slave, and that the last two are called Shirazad and Dinazad. But they are not yet sisters, as they will be later. In his bibliographic work The Catalogue (Al-Fihrist), Ibn an-Nadim (d. circa 995) mentions a work translated from Persian called “The Thousand Stories.” He also gives a summary of the frame story, but he criticizes it as “a coarse book, without warmth in the telling.”
For the next seven centuries there are only two even briefer references to its existence. In the twelfth century a loan record for a Jewish bookseller in Cairo mentions the title “The Thousand and One Nights”—the earliest mention of its present title: “Majd ibn al-‘Azîzî has The One Thousand and One Nights.” And in the early 15th century the Egyptian historian al-Maqrîzî (d. 1442) cites authors who indicate the work was in circulation in Cairo in the late 11th century. These brief references separated by long periods when the book all but vanishes from view suggest an analogy with an unconscious thought that only infrequently and briefly makes its presence known in conscious thought and then quickly vanishes again beneath the force of repression—and repression is not mere metaphor here, as we shall see.
On the basis of these facts, we know that there are three main layers to the book: a translation of a group of Persian stories (which themselves incorporated Indian stories), a Baghdad layer and a Cairo layer. D.B. Macdonald, in an important article published in 1924, “The Earlier History of the Arabian Nights,” further divided its development into five stages: i. a Persian core “The Thousand Stories” (Hazâr Afsânah); ii. an Arabic version of this; iii. the frame story of “The Thousand Stories” with new Arabic stories added to it; iv. a late Fatimid version (twelfth century); v. the Syrian recension whose sixteenth century manuscript was the basis of the first European translation, that of Galland. Stages ii and iii correspond to the Baghdad layer, while iv and v are part of the later Egyptian layer. To this we may add a sixth and final stage, suggested by Nabia Abbott, a stage which extends into the sixteenth century and introduces the materials from popular epics.
The infrequency of the medieval references to the work cited above, their brevity and their tone all point to the insignificance of the work in the eyes of recognized practitioners of medieval Arabic literature. Yet it seems to have been a popular work—the work has not completely shed this paradoxical reputation in Arab countries even today.
In addition to these testimonies, there is also the textual evidence, stories in the Nights that bear unmistakable links other works in medieval Arabic literature and to works in other literatures. Here it will be useful to distinguish between three categories.
The first category would contain stories that reveal its links with a few specific works in medieval Arabic literature. For example, ‘The Story of the Steward’ told within the longer story of ‘The Hunchback’ is roughly the same story recounted as fact by the tenth century author at-Tanûkhî (d. 994) in his book Happiness after Hardship (Al-Faraj ba‘d ash-shiddah). It is the story of a husband who offends the delicate sensibility of his wife on their wedding night by forgetting to wash his hands after having eaten a certain spicy stew called zîrbâjah. Years later the same dish is served at a banquet and the other diners demand that he eat some of it; the unfortunate man washes his hands one hundred and twenty times, then tells the story of his disastrous wedding night.
Another story narrated within ‘The Hunchback,’ ‘The Story of the Lame Young Man,’ also shares its plot with stories and anecdotes found elsewhere in the literature. In the other versions this plot makes use of a certain stew called madîrah, which similarly provokes the recounting of a painful story. However, in ‘The Story of the Lame Young Man,’ as regards its plot function, the dish of stew is transformed and split into two human characters, into a judge and his daughter. A purportedly factual madirah anecdote which focuses on a clash between guest and host over the provocative dish is fairly widespread in medieval anecdotal literature. An example can be found in the anecdotal work The Misers (Al-Bukhalâ’) by the eleventh century writer al-Khatîb al-Baghdâdî (d. 1071). Interestingly, the anecdote there is attributed to the same author, Tanukhi, who has the analogue of the previous story, ‘The Story of the Steward,’ in one of his books. The story ‘The Madirah Maqamah,’ found in a late tenth or early eleventh century work Al-Maqâmât by al-Hamadhânî (d. 1008), makes use of the same madirah plot. The eleventh century physician Ibn Butlân (d. 1066) also makes use of this plot in his comic work The Physicians’ Banquet (Da‘wat al-atibbâ’). And, finally, it is also found in an anonymous thirteenth century work Wonderful Stories (al-Hikayât al-‘ajîbah).
Finally, more such textual links are found in the story ‘The Sleeper Awakened,’ a story which Galland translates, but which is not found in two of the major 19th century Arabic editions, the Bûlâq or the Second Calcutta editions. The story is comprised of two parts; in the first part, the main character, Abû ’l-Hasan, is drugged by the caliph Hârûn ar-Rashîd and tricked into believing that he is the caliph. In the second part, Abu ’l-Hasan tricks Harun into thinking that he and his wife have died. The first part of the story is also found in a work by a 17th century Egyptian author al-Ishâqî (d. 1651). Ishaqi writes when Egypt is a province in the Ottoman Empire, and in his work Accounts of Previous Rulers of Egypt, he gives the first half of the story as fact in the section devoted to Harun ar-Rashid. The second half of the story seems to be an elaboration of an anecdote told of the Abbasid poet/buffoon Abû Dulâmah that is found in the famous work of Abû ’l-Faraj al-Isfahânî (d. 967), The Songs (Al-Aghânî ). The relative lateness of Ishaqi makes the occurrence there more interesting from the perspective of literary history, for with him we are within a few decades of Galland’s discovery of the manuscript of the Nights in Istanbul. Another brief passage in Ishaqi is also of interest in this context. In an account of a visit to a graveyard by an anonymous narrator said only to be one of the ‘people of refinement,’ the narrator states that the purpose of his visit is “to visit the dead and reflect on the lessons of what has passed... and to remember the destroyer of delights, the separator of societies, he who makes orphans of son and daughters.” The phrase “the destroyer of delights” (hâdim al-ladhdhât) is found at the end of numerous stories in The Thousand and One Nights, and indeed the version here with mention of the ‘maker of orphans’ (muyattim al-banîn wa ’l-banât) is, word for word, the same as the version of this sentence that comes at the end of the last story Shahrazad tells, the story of ‘Ma‘rûf the Cobbler.’ The nature of Ishaqi’s work is also telling; much of it is patent fiction passed off as historical and edifying anecdote. These things suggest that Ishaqi knows the Nights in a version very much like the one we know, and we are only a few decades removed from the date when Antoine Galland, the first European translator of the Nights, purchases his Syrian manuscript in Istanbul.
The second category of textual evidence contains stories in The Thousand and One Nights that make use of and revise the plots of extremely well-known stories in Islamic culture; stories about figures like Abraham, Joseph and Solomon are examples.
Finally, in the third category are the stories which make use of plots which are even more widely spread, the ‘Cinderella’ plot, the ‘Phaedra’ plot and so on.
What do these textual links suggest about the development of the text of The Thousand and One Nights at any particular stage in the Middle Ages? Unfortunately, not very much than I have already said. The latter two categories of material tell us nothing at all, since the material is pervasive throughout the period. With regard to the first, more specific, category, the safest assumption would seem to be that the storytellers involved in the creation of the Nights sometimes made use of ‘factual’ anecdotal collections for plots, revising them freely to suit their purposes. But even this modest conclusion must carry a caveat.
The earliest manuscripts of The Thousand and One Nights date from the fourteenth century, and most of the works cited above for containing analogues are earlier. Yet as we have seen, on the basis of the testimony of earlier authors we know that some sort of collection with a title The Thousand Nights existed centuries prior to the fourteenth century. So lacking any knowledge of what the book contained in, say, the tenth or eleventh centuries, it is not absolutely certain that material moved only in one direction, out of established writers and works and into the Nights. For, a priori, it is not impossible that an author like Tanukhi and a One Thousand and One Nights storyteller drew upon a common store of material. Or a third possibility: it is not out of the question that some material might have moved in the other direction, out of the Nights and into the ‘factual’ works of a writer like Tanukhi. After all, writers do lie sometimes. In fact, in some instances in Tanukhi’s Happiness after Hardship I think this third possibility is the most likely one, that it is Tanukhi who is borrowing from some version of The Thousand and One Nights. For example, in the first chapter Tanukhi tells as fact a brief story about a traveler who is shipwrecked. The story has two parts. In the first, a voice from the sky shouts at the man and his fellow travelers that they should throw their money overboard in order to gain a piece of knowledge that will be of spiritual benefit to them. But only the one man does so. Then, when a storm destroys the ship, he is saved while all the others drown. In the second part, the saved man washes ashore on a desert island and finds a subterranean chamber which contains a treasure and a beautiful girl (also shipwrecked). While these elements are found in various combinations in many stories in The Thousand and One Nights, this story is basically that of ‘The Story of the Second Qalandâr’ told in ‘The Porter and the Three Ladies of Baghdad.’ In this case, since the story is an obvious fiction for which he gives no sources, it seems likely that Tanukhi has borrowed the second part from one of two places, either from some version of the Nights or from a store of narratives (perhaps as yet unwritten) which the Nights also uses; this he joined to a Sufi conversion tale. In view of its similarity to ‘The Story of the Second Qalandar,’ it is tempting to say that the first case is what happened, and that, therefore, ‘The Story of the Second Qalandar’ must have existed in the tenth century. But given the ubiquity of these plots and motifs, given the instability of the texts, especially in a manuscript culture, and given the question of oral and written versions and their possible interaction—given all these factors, who can say for certain what happened in any particular case?
Thus, while the textual evidence – its demonstrated links with other works – may tell us something about the sorts of literature that furnished the raw materials for the storytellers of The Thousand and One Nights and how they reworked it, that evidence does not, in my opinion, tell us much about the specific content and shape of the text at any particular stage prior to the fourteenth century, roughly the date of the earliest Arabic manuscript.
The European Nights
The first appearance of The Thousand and One Nights in 1704 in Europe was not unlike the uncanny appearance of the jinni in the first story Shahrazad tells. The jinni is a paradoxical being, now tiny and now enormous; he towers over the merchant and yet his son is so small that the merchant’s date pit has killed him. A similar sort of paradox attached to Antoine Galland’s French translation of an Arabic manuscript. The book immediately enjoyed huge popular success, yet as Georges May points out, it drew scarcely any critical or scholarly attention. The disparity between popular success and critical attention, which recalls its status in medieval Arabic literature, has waxed and waned over the course of almost three centuries but has never entirely disappeared—a fact all the more problematic when one considers that the publication of Galland’s translation is an early landmark in what Raymond Schwab was to call ‘The Oriental Renaissance.’ And there are other complications and paradoxes, for ever since Galland’s translation the book has led an unusual double life in Europe and the Arab world. Indeed, to speak of ‘doubles’ here merely hints at the complicated relations between various European translations and Arabic editions.
In 1704 Antoine Galland began to publish his translation of a manuscript he had purchased in Istanbul while serving there as an assistant to the French ambassador to the Ottoman Empire. The first six volumes of Galland’s translation contained 234 Nights, after which he abandoned the divisions of Nights in the last six volumes. Given the title however, some of his readers may well have wondered when the rest of the Nights would appear.
And at this point things get a little complicated.
There are basically two opposed explanations of what happened next. The first and simplest was that Europeans seeking the rest of the book more or less found it, either in manuscript form or in the form of persons—Arabs—who knew more stories that had made their way into the work by this time. The result was the ‘complete editions’ published in the nineteenth century, the Arabic editions of Bulaq and Second Calcutta and Breslau—the latter two we should note were the work of Europeans. According to this version of its history, the late medieval Nights looked much like the texts of Bulaq and the First and Second Calcutta and Breslau. This version has its contemporary defenders who have made some important arguments (and revisions) to support it.
Those adherents must deal in one way or another with a ‘revisionist’ version put forth recently by Muhsin Mahdi that opposes the first explanation on almost all the important points. Mahdi devoted years to the study of Galland’s manuscript sources, some of which are now lost, and to the reconstruction of a prototype manuscript for The Thousand and One Nights. The results of those labors were a reconstructed Arabic text, faithful, in Mahdi’s view, to the presumed antecedent of Galland’s manuscript, a fourteenth century Syrian manuscript, and a book detailing his views on the history of the Nights. His conclusions as to which stories belong in The Thousand and One Nights are based on this research. Mahdi argues that European demand for a ‘complete version’ of the work distorted the Mamluk-era original. Europeans wanted a book with literally one thousand and one nights of stories, which the work, in Mahdi’s view, did not have through most of its existence. The result was the creation of Arabic manuscripts in the eighteenth and nineteenth centuries that delivered more nights and more stories, and, as Mahdi puts it, the book came to be a “catch all” for popular narratives to meet European demand. In his view, the medieval Nights was a much smaller and more coherent work, perhaps one quarter the length of the nineteenth century editions. In this view, the number ‘one thousand and one’ for much of its history simply meant ‘a lot,’ in accordance with the rather free use of numbers in medieval Arabic literature. But after Galland the number proved fateful, and, if Mahdi is right, the subsequent history The Thousand and One Nights followed ‘the path of the signifier,’ so to speak. From that point on, the work’s dual versions in European translations and in Arabic manuscripts became intertwined in a very complicated relation, with its popularity in Europe creating a demand for a ‘complete version.’ Mahdi’s comparison of his presumed ‘original’ with the Bulaq edition and Macnaghten’s Second Calcutta, which are based on later Egyptian manuscripts, leads him to conclude that a huge number of the stories in modern versions of The Thousand and One Nights do not ‘belong’ in it. Thus Mahdi’s version of the Nights and a nineteenth century Arabic edition like the Second Calcutta confront each other as uncanny doubles with a claim to the same name. It is rather like the Amphitryon-like moment in ‘The False Caliph’ when Harun ar-Rashid is confronted with his double in the form of the false caliph.
Mahdi’s investigations have done much to clarify the modern history of the text. It seems clear that European demand influenced, to some degree, the shape and content of subsequent Arabic editions. He has also produced an Arabic text that is probably closer in style to the medieval work than any of the nineteenth century editions. But recent criticism of his ‘revisionist’ version raises some important objections to it. Robert Irwin argues that Mahdi’s conclusions proceed from mistaken assumptions about what sort of book The Thousand and One Nights was in the medieval period. Mahdi’s work on the transmission of the manuscripts is based on theories about manuscript transmission developed from the study of texts by known and esteemed classical authors, but as Irwin says, the Nights never was accorded the sort of respect that such texts enjoyed, hence we cannot assume that its manuscripts were copied and transmitted with anything like the same sort of care. Moreover, there is manuscript evidence for medieval versions of the Nights with many more nights than Mahdi would allow. In her recent book Eva Sallis points out that there are manuscripts predating Galland with many more nights. Moreover these manuscripts contain stories, ‘Umar ibn Nu‘mân for example, that must be regarded as relatively late additions. Hence, she argues that the ‘expansion’ beyond Mahdi’s ‘core’ of stories “was a feature of Nights compilation earlier than the eighteenth century…” In other words, well before the ‘European demand’ created by Galland’s work, Middle Eastern writers had already created a text much longer than Mahdi’s ‘original’ Mamluk-era text.
Although Mahdi’s argument does not bear directly on the analyses that follow, my position on it is largely that of Irwin and Sallis. Even were Mahdi right about what constituted the real medieval Nights, only the logic of the specialist could impel someone to reach the conclusion that such stories as ‘The Seven Voyages of Sindbad’ or ‘Aladdin and His Lamp’ do not ‘belong’ in The Thousand and One Nights. At this point in its history, this seems rather like mounting a campaign to change the name of the West Indies to rectify Columbus’ error. The most recent English translation of the work, which is based on Mahdi’s Arabic text, illustrates my point.
In 1990 Husain Haddawy published a translation called The Arabian Nights based on Mahdi’s reconstructed Arabic text. In his introduction, Haddawy naturally enough subscribes to Mahdi’s position; he writes of the Egyptian manuscript tradition that it “produced an abundance of poisonous fruits that almost proved fatal to the original”.  ‘Aladdin and His Lamp,’ he tells us, must be regarded as a “forgery” (xiii). Five years later, however, Haddawy brought out a second volume, The Arabian Nights II containing—yes, the ‘forgeries’ and ‘poisonous fruits.’
For my part, the diversity of stories in the later manuscripts does not bother me, and I think, in any case, there is rather more unity to be found in the nineteenth century Arabic editions than Mahdi is ready to allow—a point I will argue in Chapter 8. Even later stories like ‘Jûdar and His Brothers’ show features that relate them stylistically and thematically to earlier stories. But more importantly, given the general aversion of the medieval Arabic literary elite for outright fictions, we should be happy if the book did act as a kind of ‘catch-all’ for stories. For, setting aside the popular sîrahs like ‘‘Antar’ and his literary kin (which are admittedly are not to my taste), if a story did not find its way into The Thousand and One Nights, it likely did not survive.  Beginning probably from a core of translated and reworked stories, the book must have grown by process of accretion—the Ottoman royal palace Topkapi may furnish an architectural image for the process; what were originally independent structures are gradually joined together. And what we know of the later history does not seem to depart from this pattern. Insofar as the later additions preserve more stories, I think they are welcome additions to the palace. After all, in such a vast structure, if one does not care for a certain passage, one can move along.
For the reader who is interested in a more detailed history of the text there is a fairly extensive literature. Unfortunately, much of it is printed in old and obscure journals. Besides Mahdi’s work, a few of the more important works in English and French may be mentioned here. The article in the new edition of The Encyclopedia of Islam under its Arabic title Alf laylah wa laylah is a good place to start. Irwin and Sallis offer the best and most recent ‘anti-revisionist’ accounts. Irwin devotes a chapter in The Arabian Nights: A Companion to both the development of the Arabic work and its European editions. I would recommend Irwin’s whole book without hesitation; it is an excellent work. The second chapter of Sallis’ book offers a very readable and current account of the history of the Arabic text with a good discussion of various manuscripts. The first chapter by Andre Miquel in the work Les mille et un contes de la nuits (Paris 1991) speculates in a very interesting way on the reasons for the ‘eclipse’ of the work in Arabic in the Middle Ages. Mia Gerhardt’s book, The Art of Storytelling, published in 1963, also discusses the history of the work, though her account is superseded in some degree by recent work. Lastly I will mention a book by David Pinault, Story-telling Techniques in The Arabian Nights, which tries to show us how the storyteller worked; Pinault’s analysis is based on a detailed examination of differences between manuscripts.
In my view, despite the antiquity of many of the plots in The Thousand and One Nights, the stories as we have them now seem to wear the garb of the late medieval period in the Arab-Islamic world, that is, the eras of the Mamluks and the Ottomans. Thus, stories about the Abbasid caliph Harun ar-Rashid, who ruled about six centuries before the Ottoman Turks established their power, may reflect popular notions about how a Turkish sultan lived in the fifteenth century rather more than they reflect such images of the way Harun lived six hundred years earlier. Not that ornate palaces, gardens, wine drinking, cup bearers and slave consorts were not features of the Abbasid court, but as found here these features often seem like those of the Ottomans. An obvious example comes in the frame story, in the scene in which King Shahriyâr’s wife commits adultery in the garden with a black slave; the lover and the enclosed garden are stock elements in any Ottoman Turkish gazel or love poem. On the general question of the work’s historical context I might finally remark that the Ottoman period is viewed in the Arab world now as one of cultural decadence, a view propounded by Arab nationalism. However, the very existence of this book is testimony, I think, to cultural vitality in that period.
On the basis of the preceding discussion of the various versions of the Nights, my reader may well wonder at this point which Arabic text or texts (and which translations) are in my view the ‘real’ Nights. My answer is simple: all of them. The Thousand and One Nights is a multiple text, and I see no reason to exclude any standard Arabic edition or any of the translations based on it, no matter how controversial they may be. I have for the most part relied on Macnaghten’s Second Calcutta edition, an edition based, it seems, on the Bulaq edition and a now lost manuscript probably copied sometime shortly after 1830. The manuscript belonged to the late Egyptian recension on which the Bulaq edition was also based. Those manuscripts are now known as Zotenburg’s Egyptian Recension (ZER) after the scholar who studied them. But I have made free use of other editions (and of various translations) as it suited my purposes and in accordance with my ‘maximalist’ position.
So much for literary history. Much more important for this study are the reasons why the book had such a shadowy existence in the pre-modern Arabic literature. That is to say, the historical problems posed by the book’s orphan-like existence in Arabic literary culture are of less importance for what follows than the characteristics of the book that made it an orphan. These call for some discussion here, for, as I have said, my readers should know that The Thousand and One Nights is an exceptional, even aberrant work with respect to some of the most important conventions of medieval Arabic literature. Hence, that the stories say the things I contend they do stems in many ways from the fact that this work is the ‘repressed’ of the literature. The factors that contribute to this status may be discussed under three headings: its genre, its linguistic style and its content. Foremost is genre, and this raises the question of the place of narrative in medieval Arabic literature.
Narrative in Medieval Arabic Literature
The early development of medieval Arabic literature was part and parcel with the development of Islam. While there existed a rich poetic tradition in pre-Islamic Arabia and a body of narratives that accompanied it and purported to provide the factual background of the poetry, this material seems to have been preserved in oral form until it began to be written down in the late eighth and early ninth centuries—that is, at the same time that, with one exception, the other early texts of Arabic literature are being written. The one exception, of course, is the Quran.
The death of Muhammad in 632 A.D. furnishes a reference point. The Quran is the only text that we know to have existed in some written form in the century after his death. It is, we should note, a work that contains little narrative as compared to the Bible.
After the Quran, the next major work we possess is The Life of the Prophet (Sîrat an-nabî), composed in the first half of the eighth century by Ibn Ishâq (d.767), but available to us in the recension of Ibn Hishâm (d. 828 or 833) who made his own cuts and additions. Close on its heels comes a book by al-Wâqidî (d. 823) called The Raids (Al-Maghâzî) describing the raids Muhammad and his followers made on the pagan tribes of Western Arabia.
Because of their priority and the importance of their subject matter, these works established precedents for the use of narrative that would have lasting effects throughout the Islamic Middle Ages. They are made up of the earliest narratives in Arabic, traditions about what Muhammad and his followers said and did in the course of conquering western Arabia. These narratives are known as akhbâr; al-akhbar in modern vernacular simply means “the news.” The singular khabar simply means a piece of information recounted for someone—the same term is used in Arabic grammar to mean the predicate of a sentence. In the early literature a khabar is a short narrative, usually a half a page or less in length, and hence confined to the pithy recounting of a single incident. Because of its religious significance and to buttress its factual claims, each khabar-narrative came to be preceded by a feature known as the isnâd or ‘chain’, a series of names representing a kind of bucket brigade of tradents who, it is claimed, have passed the narrative along to the writer from an eye-witness to the original event. Both with respect to narrative form and the uses to which narrative may be put, these traditions exerted enormous influence on subsequent narrative literature. Thus, in its beginning, narrative literature in Arabic purports to confine itself to fact, to the recounting of real events by eye-witnesses, real people. If one accepts everything at face value in works such as Ibn Ishaq’s Life of the Prophet and Waqidi’s Raids, then Ibn Ishaq and Waqidi, the ‘authors,’ have composed very little of their books’ texts, having simply copied verbatim the ‘accounts’ from other people and other written sources.
For various historical and ideological reasons, it proved very difficult for narrative literature to free itself of these conventions. The theoretical and practical problems these conventions pose for the development of fictional narrative should not be underestimated. First of all, narratives are always about ‘real events.’ Within the literature proper, a generic space for fiction never really opened up. That which never happened is simply a lie. Curiously, The Thousand and One Nights is in agreement with this attitude—it has ‘internalized’ the prejudice, so to speak. Shahrazad does not make up stories. As Abdelfattah Kilito writes: “It sometimes happens that characters in the Nights make up a story, that is to say, they affirm that they know it to be false. Invention is then synonymous with lying.”
Secondly, the device of the isnad poses a formal obstacle for beginning a fiction. It bars the way to the space where a fiction unfolds. This can be seen in some of the few fictional narratives composed by recognized authors. Thus Ibn Butlan, deferring to the need for an isnad, begins his Dinner Party of the Physicians, “One of them said...” One of whom? We can imagine a novel beginning with those words, but we would expect shortly to be given at least a hint as to the identity of ‘one of them’ within the fiction. That never happens in Ibn Butlan’s work. We never learn anything about ‘one of them,’ for this ‘one of them’ exists in some extratextual place. The French Symbolist Paul Valéry remarks somewhere that he could not write fiction because he could never bring himself to write a sentence like, “The marquise went out at five.” Being so used to the conventions of fiction, we may underestimate the amount of literary development that prepares the way for someone to begin with a sentence like that.
The isnad can also pose a problem with respect to the length of a narrative if one must represent a fiction as fact. Here I might remark that speculation about length per se as a factor that distinguishes The Thousand and One Nights narrative from the anecdotal works of adab like those of Tanukhi misses the larger point. It is not a question of length in and of itself, but of length as an index of fictionality. A second glance at a work like Tanukhi’s Happiness after Hardship, which, as we have seen, contains some analogues with stories found in The Thousand and One Nights, shows this. The stories in Tanukhi are almost entirely anecdotal, by which I mean most are a page or less in length. They rarely have the scope or the detail found in the full-blown fictions of The Thousand and One Nights. The very length of a narrative would pose a problem vis à vis the assertion of fact staked out by the isnad. To focus on only one aspect, any khabar is likely to contain direct speech, but if such an account extends for many pages, one must grant the original eye-witness/reporter extraordinary powers of recollection to think that the reported speeches are the verbatim words of the various persons involved. The question must inevitably arise, ‘How could he (and all of the other persons in the isnad) have remembered all of this exactly as it was said?’
Other conventions pose other sorts of difficulties for the development of fictional narrative. Being supposedly a literature of hard fact, in a khabar–narrative no one would presume to make a statement about what someone else was thinking or feeling, for how would he know such a thing? Thus, only the grossest sorts of emotions are registered—‘He was very angry’—and thought, unless it is spoken, is absent. The continuing influence of these features will be seen in The Thousand and One Nights.
There may also have been ideological constraints. M. Arkoun has written: “The theological and philosophical tradition imposed an ontological weakness on the imagination. The Koran contributed to this weakening with its attacks against ‘the poets whom the erring follow, who wander in every valley and who say what they do not do’ (26:224-226).” And a well-known episode in the sirah and maghazi literature shows Muhammad’s anger with a storyteller who claimed his stories were ‘better’ than Muhammad’s. Nadr ibn al-Hârith so vexed Muhammad that the latter ordered him killed.
In time, to be sure, some obvious fictions came to exist within the literary canon; Kalîlah wa Dimnah, the fables translated by Ibn al-Muqaffa‘ (d. 756), and the Maqamat or ‘Seances’ of al-Hamadhani (mentioned above) and al-Harîrî (d. 1122) are the most notable examples. But such works are also exceptional. Kalilah wa Dimnah is a translation of a Persian work based on the Panchatantra, and Hariri was attacked for having made up his stories by Ibn al-Khashshâb (d. 1172).
As a result of all of these developments, in the general view of the literate minority inventing stories was not distinguishable from simply telling lies. And on all these accounts, The Thousand and One Nights just goes too far. Its excessively fictional character in this respect can be seen in the way the Nights’ version of ‘The Sleeper Awakened’ adds a second part to the story that Ishaqi has. Thus its genre, its clearly fictional nature, is the first major reason why The Thousand and One Nights is an exceptional work; there is no generic space in the medieval literary canon for it.
Another important factor has to do with its style; the sort of Arabic found in The Thousand and One Nights also counted against it. As far back as linguists can determine, Arabic seems to have been characterized by a considerable divergence between the written and spoken forms, a divergence in both vocabulary and grammar. The difference is perhaps most visible in the case endings that are preserved in the written language, but almost completely ignored in spoken dialects. With such a divergence, literacy is a rather more difficult and elusive goal to attain—something that remains the case today in the Arab world. In the Middle Ages, the spoken dialects were not even considered real languages by the literate minority. The power of that minority rested in their grasp of al-‘arabîyah, the classical written language in which the official discourse of the culture was carried on, while the illiterate were excluded from speech, in their view, since they had no language proper. They were, in a sense, ‘unspeakable’—one could say ‘repressed’ in the strict sense that they were denied the words to express themselves.
In such a vast work as The Thousand and One Nights there are many registers of language, but apart from perhaps conscious stylistic variations, the different manuscripts of The Thousand and One Nights all are salted with significant amounts of colloquial usages. These usages have led various scholars to speak of it as being written in ‘Middle Arabic,’ but that term, coined by a modern Western scholar, had no status at all among the authors who wrote ‘Middle Arabic’; those men were attempting to adhere to the classical grammar—they simply no longer mastered it. In the case of the Nights the writers may simply not have thought enough of the Nights to go to the bother of adhering to al-‘arabiyah with the result being ‘Middle Arabic.’ It may also have been in part the result, as Sallis suggests, of “the accretions of a random textual history.” In any case, the language of The Thousand and One Nights works against it. The influence of popular tongues, Syrian and Egyptian, is heard in it, and those sounds were, no doubt, repellent to the ears of the litterateurs who wrote in the ‘pure language’ of al-‘arabiyah.
Finally there is the subject matter. Illicit sex and wine stain the pages. Hashish is used frequently. Despite the way the book is used as a stock reference for certain sorts of events (‘It was like something out of The Arabian Nights’), many people are surprised by the real contents of the book. There is a widespread assumption that it is something like children’s literature, but it is not.
On the other hand, whether the book is ‘a faithful mirror of medieval Islam,’ as one edition touts it, is difficult to say. In such a view, one of the reasons the Nights is so valuable is that it depicts many things that are not often treated in the literature proper. That literature, as I have said, was the product of an elite who seldom wrote about such homely subjects. But since the Nights is in many ways a unique source, the argument threatens to become circular if one claims it is a faithful picture of medieval Muslim society. Moreover, the book is a ‘catch-all’ or omnium gatherum—to use Irwin’s phrase—and as Irwin says, “one can use its texts, through selective quotation from stories, to support the argument that homosexuality was widely approved of, or to argue that it was indifferently accepted, or to argue that it was absolutely abominated.” In other words, one would have to take a Hegelian stance—(‘the truth is the whole’)—to make it a ‘faithful mirror.’ It is some sort of reflection no doubt, but some literary refraction and distortion must be involved. The fantastic elements that abound also seem to pose some problems for this view of the book: Jinn fly back and forth between China and the Near East in the course of a single night; humans are transformed into animals and back into humans; islands turn out to be whales, and mountains pull the nails from ships’ hulls by means of mysterious, magnetic powers. So, even while one ‘feels’ the truth of its stories, as a source for social history, The Thousand and One Nights must be used carefully. Yet this much is clear: its subject matter could be an affront to the pious writers who always made up a considerable portion of the literary elite, men who wrote books with titles like The Condemnation of Fun (Dhamm al-lahû).
For all of these reasons The Thousand and One Nights was a marginalized work in medieval Arabic literature, and that is one reason why it is so valuable. It escaped the self-censorship of more typical narrative works. From the outset its pages are filled with desires and ideas that are rarely articulated elsewhere in the literature.
 Jorge Luis Borges, Seven Nights, translated by Eliot Weinberger (New York: New Directions, 1980), p. 54.
 First described by Nabia Abbott in her article “A Ninth-century Fragment of the Thousand Nights: New Light on the Early History of the Arabian Nights,” Journal of Near Eastern Studies 8 (1949), pp. 129-164.
 Al-Mas‘ûdî, Murûj adh-dhahab, edited and translated by C. Barbier de Meynard and Pavet de Courteille, Les prairies d’or (Paris: Imprimerie imperiale, 1861-1877), Vol. 4, pp. 89-90.
 Ibn an-Nadîm, The Fihrist of Ibn an-Nadim, translated by Bayard Dodge (New York: Columbia University Press, 1970), Vol. 2, p. 714.
 Samuel Goitein, “The Oldest Documentary Evidence for the title Alf layla wa layla,” Journal of the American Oriental Society 78 (1958), p. 301.
 Al-Maqrîzî, Al-Khitat (Cairo: Bulaq, 1854), Vol. 1, p. 484.
 D.B. Macdonald, “The Earlier History of the Arabian Nights,” Journal of the Royal Asiatic Society (1924), pp. 353 ff.
 The similarities were first noted and discussed by Amedroz in his article “A Tale of the Arabian Nights Told as History in the ‘Muntazam’ of Ibn al-Jawzî” in Journal of the Royal Asiatic Society (1904), pp. 273-293. Muhsin Mahdi has recently discussed them in his book The Thousand and One Nights (Leiden: E.J. Brill, 1995), in Appendix 3, “From History to Fiction”, pp. 164-180.
 This and other transformations of the madirah plot tell us something of the plasticity of the material in the hands of the medieval storyteller, matters I discuss in an article called “A Mighty and Never Ending Affair,” Journal of Arabic Literature Vol. XXIV, Pt. 2 (July 1993) pp. 139-159. One finds other echoes in the Maqamat of al-Hamadhani: in ‘The Hunchback’ the barber brings the apparently dead hunchback back to life, and in al-Hamadhani’s ‘The Maqamah of Mosul’; the trickster Abu ’l-Fath comes upon a corpse and also tells an astonished crowd, “This man is not dead!” He promises to raise him within two days, but he fails—the man is dead after all.
 Al-Khatîb al-Baghdâdî, al-Bukhalâ’ (Baghdad: al-Majma’ al-‘Ilmî al-‘Iraqî, 1964), p. 148.
 The Maqâmât are written in a style of rhymed prose peculiar to Arabic known as saj‘, and are rich in word play and rhetorical tricks. Al-Hamadhani’s eleventh century work al-Maqâmât (Beirut: Dâr al-Mashriq, 1986) is the first example.
 The story in al-Ishâqî’s Accounts of…Egypt (Akhbâr al-uwal fî man tasarrafa fî Misra min arbâb ad-duwal) (Cairo: al-Matba’ât al-fakhrah, 1859) begins on p. 129. The anecdote in al-Isfahânî’s Al-Aghânî (Cairo: Bulaq, 1868-69) is in Vol. 9, p. 131.
 Al-Ishâqî, Akhbâr al-uwal fî man tasarrafa fî Misra min arbâb ad-duwal, p. 91.
 Book of the Thousand and One Nights Commonly Known as ‘The Arabian Nights Entertainments’ Now for the First Time Published Complete in the Original Arabic, ed. W.H. Macnaghten (Calcutta: Thacker, 1839-1842), Vol. IV, p. 730. Unless otherwise noted, all references to an Arabic text will be to this edition, the so-called Second Calcutta, giving volume and page numbers in the text.
 In Al-Faraj ba‘d ash-shiddah (Cairo: Khanji, 1956), pp. 23-24. I discuss this and other examples in “In the Second Degree: Fictional Technique in Tanukhi’s Al-Faraj ba‘d ash-shiddah,” Journal of Arabic and Middle Eastern Literatures Vol. 1, No. 2 (July 1998), pp. 125-139.
 Georges May, Les Mille et une nuits d’Antoine Galland (Paris: Presses universitaire de France, 1986), pp. 8-23.
 Kitâb Alf layla wa layla, edited by Muhsin Mahdi (Leiden: E.J. Brill, 1984), and The Thousand and One Nights, (Leiden: Brill 1995).
 The Mamluk dynasty ruled Egypt from 1258 to 1517.
 Robert Irwin, The Arabian Nights: A Companion (New York: Penguin, 1995), pp. 55-56.
 Ibid., pp. 57-62.
 Eva Sallis, Sheherazade Through the Looking Glass (London: Curzon, 1999), p. 34. Sallis even speculates that by the close of the twelfth century that work “incorporated most probably literally one thousand and one nights… and around two hundred tales” (p. 27). It is certainly possible, but then again—who knows?
 Hussain Haddawy, The Arabian Nights (New York: Norton, 1990), p. xii.
 Hussain Haddaway, The Arabian Nights II (New York: Norton, 1995).
 They are called ‘popular’ sirahs to distinguish them generically from works like the first biography of Muhammad, also a called a sirah. They are long fictions of chivalry, relating the adventures of a hero/knight; ‘Antar and Sultan Baybars are notable examples of which at least parts are now available in English and French. In my view, these works do not have the formal brilliance of The Thousand and One Nights, and while individual episodes may be entertaining enough, in the aggregate, they take on a monotonous character.
 Even measured against other works of a hitherto denigrated ‘popular literature,’ The Thousand and One Nights is a rather singular work. More typical representatives of that literature are the ‘popular sîrahs’ mentioned above.
 S.A. Bonebakker considers the problematic status of fiction in his essay “Nihil obstat in Storytelling?” found in the recent volume The Thousand and One Nights in Arabic Literature and Society , edited by Richard C. Hovannisian and Georges Sabagh (New York, USA and Cambridge, UK: Cambridge University Press, 1997) pp. 56-77. But he seems to think the question is an open one. I think I have shown that something does tend to obstruct it in my articles on parody and lying, on Tanukhi, and on early Muslim historical traditions. I am more or less going over the same ground here.
 This, despite the fact that the greater portion of ‘legal’ traditions or hadîths and much of the ‘historical’ ones are now known to be ‘fictions.’
 Abdelfattah Kilito, L’oeil et l’aiguille (Paris: Éditons la Découverte, 1992), p. 14.
 The same can be said of attempts to distinguish between different forms of the fantastic that would act as a criteria. It is not a question of forms of the fantastic per se, but of forms that would be unmistakable indices of fictionality. Joseph Sadan’s review, in Journal of Arabic Literature (March 1994), pp. 81-83, of a book by Wiebke Walther discusses some of these questions.
 Cited by Mohammed Arkoun in L’Islam, morale et politique (Paris: Desclée de Brouwer, 1986 ), p. 12.
 Ibn Ishâq, Sîrat an-nabî, translated by Alfred Guillaume, The Life of Muhammad (London: Oxford University Press, 1955), pp. 135-6, p. 308.
 Eva Sallis, Sheherazade through the Looking Glass, p. 40.
 Robert Irwin, The Arabian Nights: a Companion, p. 169.