2.1 The Translation Movement and Its Context

After the Arab-Islamic conquests, the old administrative languages remained in use in the conquered areas, and Arabic was made the language of administration only towards the end of the seventh century. At first the scientific tradition also remained in the hands of its former, mainly Christian and Jewish, practitioners, and Greek, Syriac, and Pahlavi were the languages of science until towards the mid-eighth century.

From the early ʿAbbāsid period onward Persian culture, itself influenced by Greek culture,¹ influenced various fields of Arab-Islamic culture, including language, court etiquette, the organization of the Empire’s administration, historical tradition, and literature that we would label belles lettres, although such a concept was more or less unknown to the Arabs themselves, who saw in, e.g., Kalīla wa-Dimna a book of practical philosophy or a Fürstenspiegel, rather than a book of animal fables for entertainment.²

A massive translation movement started in the mid-eighth century and continued until the eleventh century. In a short time, a huge amount of originally Greek scientific and philosophical literature was translated into Arabic, either directly or through intermediate Syriac or, sometimes, Middle Persian translations.³

This Arabic translation movement began with the translation of some Middle Persian texts (whether of Greek origin or not) already at the end of the Umayyad period (i.e., before 750),⁴ but it gained strength only with the ʿAbbāsids who, rather paradoxically, both internationalized and Arabicized their culture. The Umayyads had, at least until around 700, been content with using Greek, Syriac, Middle Persian, and Coptic as their administrative languages, but had otherwise been distinctly Arab, so much so that their dynasty has, since the Middle Ages, often been called “an Arab kingdom”.⁵ The ʿAbbāsids turned the policy around. Their culture was heavily international but they worked in Arabic and sponsored translations into the language of the ruling elite, though some Caliphs are even said to have known Greek.⁶

Around 1000 ad, almost the whole extant and available Greek scientific legacy had been translated into Arabic. The works of Aristotle and his main commentaries; the works of Galen; and a huge number of other works ranging from agronomy to zoology were available and in use in Arabic at the beginning of the second millennium.⁷

Many Greek books were not only translated once into Arabic (or Syriac), but were retranslated several times. A good example of this variety of translations is found in the translation history of Aristotle’s Topics, which in the first millennium was translated five times into Syriac or Arabic, namely:⁸

1. by Athanasios of Balad (d. 686): Greek → Syriac.

2. around 782 by the Patriarch Timothy i (d. 823) and Abū Nūḥ al-Anbārī: Syriac → Arabic.

3. late ninth century by Abū ʿUthmān al-Dimashqī (d. after 914): Greek → Arabic.

4. by Isḥāq ibn Ḥunayn⁹ (d. 910): Greek → Syriac.

5. by Yaḥyā ibn ʿAdī (d. 974): Syriac (of Isḥāq) → Arabic.

As we shall see in Chapter 3, the translation history of this book may resemble that of the Khwadāynāmag.

2.2 Translations of Middle Persian Texts

The translations of the Khwadāynāmag will be studied in detail in Chapter 3, but a general overview of what was translated from Middle Persian into Arabic will give some background for understanding the specific case of the Khwadāynāmag.

Whereas translations from Greek and Syriac into Arabic are well documented and we can even occasionally see how the translators worked, the translations from Middle Persian are still very imperfectly known and only very rarely do we have both the original and the translation at our disposal and all too often neither, merely a reference in Ibn al-Nadīm’s Fihrist or some such source.¹⁰

Our main source of information for the translations from Middle Persian is the bibliographical work of Ibn al-Nadīm, Fihrist, supported by occasional references to, or quotations from, translated books in the works of early authors such as Ibn Qutayba, al-Masʿūdī, and al-Jāḥiẓ. Middle Persian literature itself very rarely refers to any booktitles.

The chapter on Persian scripts (and languages) in the Fihrist, pp. 15–17/12–14//22–27, mainly derives from information attributed to Ibn al-Muqaffaʿ. Although the specimens of Middle Persian writing in this chapter are no longer quite accurate in the preserved manuscripts, they clearly show that the ultimate authority of Ibn al-Nadīm really could read several of the scripts. In addition, Ibn al-Nadīm seems to have relied on an authority of his own, Amād al-Mōbad.¹¹

The chapter lists seven differerent types of Persian scripts. In addition to various esoteric scripts, these include the Avestan script, called dīn-dafīrīh (← dēn-dibīrīh), Huzwārishn, and a variety of Pahlavi scripts. To point out just one example to show that the description of the scripts does go back to a person well acquainted with some of these scripts, one might quote the explanation of the Book Pahlavi script (called by Ibn al-Nadīm nāme-dabīrīh), speaking on letter writing: “Some of them are written in Ancient Syriac (…) but read in Persian”, which is an accurate description of Pahlavi, if we keep in mind that the language we call Aramaic was called Suryānī in Arabic. Likewise, he is able to tell correctly that the Pahlavi words for “meat” and “bread”, gōsht and nān, are written bsrʾ and lhmʾ and even the original scripts he adds are still easily recognizable.

In this section, Ibn al-Nadīm also mentions several letters that he claims, based on al-Jahshiyārī, have survived from remote Antiquity (e.g., Rustam’s manumission letter, see Chapter 5.1). Whether this refers to Pahlavi versions of such pseudepigraphical texts or is a mere legend cannot be decided.

In the case of quotations from Middle Persian texts in Arabic and Classical Persian books, it is unfortunately rare that the exact source is given and a lot of material that can safely be identified as deriving from Middle Persian literature is quoted in Arabic and Persian literature merely by the ultimate authority. Thus, e.g., maxims coming from andarz literature are often quoted solely on the authority of the king or sage to whom the saying is attributed, without reference to the Pahlavi book from which it has been translated.

Few of the translated texts are still extant and even they have usually been transmitted over centuries so that the text has undergone changes which make it impossible to reconstruct the original – such, e.g., is the case of Ibn al-Muqaffaʿ’s Kalīla wa-Dimna, where we have a great number of manuscripts, versions, and recensions, not to mention further translations, but only a very general idea of what the original translation of Ibn al-Muqaffaʿ may have looked like.¹²

Most of the translations that we know by title have been lost except for occasional fragments and the same goes for the original Middle Persian texts, so that no real comparison of the original and the translation is usually possible and we have to be satisfied with an overall list of such translations. This chapter does not aim at being a full history of translation from Middle Persian into Arabic, and I have endeavoured some kind of comprehensiveness only in the case of those works that are closely related to Persian national history.

A further complication in studying the texts is that the extant originals hardly ever overlap with the extant translations. The dwindling Zoroastrian community of the Islamic times was primarily interested in keeping up the tradition of their religious literature, and secular literature was to a large extent lost during the centuries, while the translators were less interested in Zoroastrian religious literature and mainly translated secular texts.

2.3 The Alexander Romance

The number of preserved Pahlavi texts is rather small in comparison to the number of texts that were once written in that language. In the centuries after the Arab conquest, the number of extant Middle Persian manuscripts quickly diminished, although in the tenth century many texts still existed that we nowadays lack. The reasons for the disappearance of Middle Persian texts are various: the Pahlavi script is very complicated, the scribal tradition was weakened by the lack of institutionalized support for Zoroastrian culture, and the majority of Persians soon converted to Islam and seem initially to have lost interest in pre-Islamic culture and its texts.

With the growth of Classical Persian literature from the ninth century onward, the old script soon became obsolete and the knowledge of the script and the by then archaic language was restricted to a diminishing population of Zoroastrians. They did produce some seemingly new texts in Middle Persian, but the majority of such “new” texts, such as the Dēnkard, were, in fact, largely compilations from older ones.

As the preservation of Middle Persian texts was left to a religious minority, it is understandable that their efforts mainly centred on religious texts, which were of great importance for the preservation of the old religion. In this situation, secular texts were no longer copied and the extant copies disappeared in time. This may have been precipitated by the translation movement from around 750 onward when the most interesting texts, the Khwadāynāmag among them, were translated into Arabic and these translations became more easily accessible to historians than the Pahlavi originals, which thus became, in a sense, superfluous.⁷⁶

This has left very little for modern scholars to work upon. However, we do know that secular literature existed in Middle Persian in the Sasanian period, and this has opened the doors for speculation. Scholarly literature is full of such speculation about texts that might once have existed. Very often, as in the case of the Khwadāynāmag, we have ample evidence for their one-time existence; Arabic texts, such as Ibn al-Nadīm’s Fihrist, contain information on what was translated and some translations are still extant, either completely or in fragments within other texts (cf. Chapter 2.2).

We have little concrete evidence for any translations from Greek into Pahlavi, not to speak of extant texts or fragments. Admittedly, the majority of non-religious (and even religious) Pahlavi books were lost when most Persians switched over to the more practical Arabic script and the dwindling Zoroastrian community mainly cared for their religious inheritance.

This admitted, it remains a disturbing fact that Pahlavi secular literature is mostly hypothetical and the little information we have on it usually comes from much later Arabic sources.⁷⁷ There is no question that non-religious books in Pahlavi existed during the Sasanian period, and there probably were among them some translations from Greek. But one should beware of speculating on their existence in cases where the evidence is purely hypothetical.⁷⁸

As a brief case study, let us consider the Alexander Romance,⁷⁹ which ties up with the Book of Kings tradition. This book is commonly thought to have existed in Middle Persian translation, although there is little tangible evidence for this. Rubin (2008b): 31, goes even as far as to speak about “the Alexander Romance which was popular in Sasanian Iran during the 6th century.” As we shall see, it is very dubious whether the Romance was translated into Middle Persian in the first place and there is absolutely no evidence whatsoever for it having been popular in sixth-century Iran.

The idea of a lost Middle Persian translation of the Alexander Romance originally comes from Theodor Nöldeke’s Beiträge zur Geschichte des Alexanderromans (1890). None of our Pahlavi, Arabic, or Classical Persian sources mentions its existence. Nöldeke only postulated it on the basis of his analysis of one of the Syriac versions.⁸⁰

The Syriac manuscript A, dated to 1708–09, is the oldest of five manuscripts which contain one version of the Syriac Romance of Alexander (Ciancaglini 1998: 55). The History of Alexander was edited by E.A. Wallis Budge in 1889 and in the following year Nöldeke published his study in which he claimed that the Syriac text is based not on an earlier Arabic version of the text, as had hitherto been thought, but on a lost Pahlavi version. Since then, a Pahlavi Alexander Romance is usually presumed to have existed in the Sasanian period.

Nöldeke (1890: 11–12) lists some cases in the Syriac text which might be taken as evidence for it having been translated from Arabic,⁸¹ as Budge had proposed. Then he continues with a list of cases which can be interpreted as examples of Persian or Pahlavi influence (1890: 13–17). Nöldeke’s evidence is hypothetical, consisting of the occurrence of Persian terms in the text and word forms more easily explicable through Pahlavi orthography. The former are inconclusive as they could equally well be explained by the Syriac translator having himself been under Persian influence (e.g., living in an area where Persian was spoken). Nöldeke’s list of the latter is long and seems impressive, but it remains problematic: even though some forms would be explicable through Pahlavi, yet as the Pahlavi script is notorious for its inadequacy to represent the sounds of its own language, not to speak of unknown names in other languages, how come the names are no more corrupt than they actually are?

It remains a fact that Nöldeke assumes a corruption caused by Pahlavi script when it suits him, but silently accepts astonishing fidelity in other names or even in other parts of the same name. His case is far from conclusive.

Though generally accepted as fact, Nöldeke’s theory has also been criticized. While reviewing Budge’s edition Siegmund Fraenkel (1891), expressed some doubts as to Nöldeke’s conclusion and Richard N. Frye declined the suggestion of the existence of a Pahlavi version in his Two Iranian Notes (1985), though he gave little evidence for his opinion.⁸² Frye based his argument merely on a general improbability of a text celebrating Alexander the Great having been translated into Pahlavi during the Sasanian period, Alexander having become the archenemy of pre-Islamic Persia, as is well documented in a variety of Pahlavi texts.⁸³

These voices were given little heed, just as Ciancaglini’s (1998) painstaking analysis of the question has not received the attention it deserves.⁸⁴ Ciancaglini analyses several Persian calques indicated by Nöldeke and shows that the words are not attested in Middle Persian but only in Classical Persian (p. 68), having thus had almost a millennium to creep into the text before the manuscript of 1708–09 was copied and providing no real evidence for a hypothetical Pahlavi original. She also shows how few the graphemic variations explicable through Pahlavi script are (pp. 75–76), against which one can put the numerous cases where the Greek sounds are properly represented in the Syriac manuscript, which should have been equally prone to corruption had the texts gone through a Pahlavi intermediate text. This is especially clear in the many Grecisisms,⁸⁵ where especially L and R are correctly represented as against the few cases of wrong representations Nöldeke is able to point out. As Ciancaglini states (1998): 78:

Si dovrebbe presupporre che il redattore siriaco sia stato capace in quasi tutti i casi di nomi comuni presi in prestito dal greco di risalire al modello, nonostante le ambiguità della scrittura pahlavica. Questo non sembra molto verosimile.

Nöldeke himself had noted this (1890: 16), but he underestimated the number of the correct forms, and Ciancaglini’s detailed study shows that many of Nöldeke’s counter examples are, in fact, untenable for various reasons.

If Ciancaglini’s arguments are valid, as they seem to be, how should we explain the evident Persianisms in the text, including many marginal notes that identify nouns and names with their Persian equivalents? As Ciancaglini points out (1998: 87–90), the Syriac text was written by Nestorian Christians and the oldest preserved manuscripts come from Northern Iraq, where the culture was heavily Persianized at least from 1500 onward, thus explaining the Persianisms, which, moreover, are more often Classical than Middle Persian.

Ciancaglini’s study shows forcefully that Nöldeke’s speculation is based on the slightest of evidence and as the existence of a Pahlavi Alexander Romance is not only undocumented but also counters what we might expect from a dynasty which saw Alexander as their archenemy, it becomes rather improbable.

Ciancaglini’s study has, however, been almost routinely ignored. It has also been countered by van Bladel (2007): 61–64. Van Bladel draws attention to the fact that approximately 18% of the L/R cases are transmitted wrongly in the text, which, according to him “require[s] a real explanation and cannot be merely dismissed as accident or as ‘weak’ evidence” (p. 62). While certainly a relatively high number, van Bladel fails to consider that, if the text were to come through Pahlavi script, it would also need a real explanation of how 82% of the words were transmitted correctly through a script that does not properly distinguish between R and L.⁸⁶ The fact that most of the cases refer to personal or geographical names that were not familiar from elsewhere calls even more strongly for an explanation. Speaking about distorted Greek names, van Bladel also notes (p. 62) that “[n]ormally, however, translations directly from Greek into Syriac do not entail such bizarre distortions.” Here van Bladel is basically right, but forgets that most works that were translated from Greek into Syriac were either scholarly or religious texts, both of which were usually more carefully translated and contained fewer unknown names than the Alexander Romance.

Finally (pp. 62–64), van Bladel is able to point out a few words where the Syriac text seems to keep a Pahlavi orthography (Balkh/bhly; the ending -īg in Sūndīqāyē “Sogdians”; plhy’ and plwhy’ for “Parthian”) as well as the Persian gloss Wahrām (Classical Persian Bahrām) and the mention of pagan Iranian divine names. These five cases do deserve our attention, but they hardly match Ciancaglini’s much more extensive material that would point in the other direction: many of the Persian glosses are not Middle Persian forms and the majority of graphemic representations are correct. Although unable to counter Ciancaglini’s arguments convincingly enough, it has to be admitted that van Bladel is able to keep the discussion alive.

Recently, the first real piece of evidence for the possible existence of a Middle Persian Alexander Romance has been brought to light. Dieter Weber (2009) has discussed a small Pahlavi fragment, which might derive from this lost Alexander Romance. This small piece of parchment (P.Pehl. 371), datable to around 600, measures only 18x15 cm. and has 8 partially preserved lines of writing on the recto, while the verso is blank. What is curious is that the lines both above and below the eight lines of writing do not seem to have contained any writing (see Table xiv in Weber 2009).

According to Weber’s reading (2009: 310), the fragment mentions a certain Timeus(?) of Samos, speaking to Alexander the Great (‘lksndlkysl “Alexander the Caesar”). Such a person is not known from the Alexander Romance, and the episode (of which, due to the state of the fragment, we know unfortunately little) cannot be located in any of the various Alexander Romances (Weber 2009: 313), although Weber still puts forth the idea that it might yet be attested in some unknown variant version.

While this is not impossible, it seems rather speculative and there are features in the fragment that make one doubt this. The little we can get out of the text could equally well be a piece of wisdom literature. That the text only contains eight lines implies a short text, such as a maxim, and the blank verso speaks for the same.⁸⁷ If the text came from the Alexander Romance, the copyist would have had to copy only a small fragment of it separately, and there is no obvious reason why he should have done so. What the fragment may prove, though, is that there was an undercurrent of a less hostile attitude towards Alexander already in the Sasanid period.

All evidence considered, the existence of a Pahlavi Alexander Romance remains a hypothesis and cannot be taken as an established fact until more evidence is produced.⁸⁸ It seems much more probable that the story found its way first into Arabic literature and only from there to the Persians, who used Arabic as their literary language down to the tenth century and even later, and finally to written Persian sources. This must have happened, at the latest, in the Prose Shāhnāme (Chapter 4.2) as the story of Alexander is found in both al-Thaʿālibī and Firdawsī, both drawing on this source. It may also have found a place in some earlier Persian compilations of the tenth century, as elements of the Alexander Romance are already found in Ḥamza’s Ta⁠ʾrīkh, pp. 33–34. Ḥamza probably draws on earlier Persian sources here, although, of course, contamination from Arabic material is quite possible. As he knew the Arabic tradition, it is possible that Ḥamza has here fleshed out his Persian material with material derived from Arabic literature.

2.4 Translation in the First Millennium

This chapter introduces some theoretical considerations on translation in the latter half of the first millennium.

In the first millennium, exactness was sometimes the ideal in translation, but it was restricted to certain genres. Basically, one can distinguish between four major categories of texts as to how they were handled in translation, namely:

1. religious, especially sacred texts;

2. scientific (including philosophical) texts;

3. historical texts;

4. literary texts.

In religious and scientific⁸⁹ texts, one easily finds cases where great effort is put into reproducing the text as exactly as possible. In the case of scientific texts, this was mostly functional: a formula obviously has to be translated as it is, otherwise the medicine will not work or works in a wrong way: a grain should not be changed into an ounce, however much that might entice the translator. In religion, it is the sanctity of the source text that demands exactitude in translation. The Word of God is not lightly to be tampered with.

In these cases, and especially in religious texts (but there are also scientific translations made according to the same principles), pseudo-translation is common, translating each word by its equivalent in the target language, in the worst of cases in the form of an interlinear translation, such as we find in many Persian or Turkish “translations” of the Qurʾān.⁹⁰ When read in connection with the original, such translations may be used as auxiliaries for comprehending the original, but when served separately, one can hardly make any sense of them.⁹¹ Such translations may have been “exact” in the eyes of their perpetrators, as they still sometimes are in the popular mind (“literal” translation being understood as word-for-word equivalence), but they were already criticized at the time, as the famous comments by Ḥunayn ibn Isḥāq in his Risāla show us, as well as the fact that such translations form a clear minority.⁹²

Both religious translations, especially those of the Bible and the Qurʾān, and scientific translations, especially those made during the ʿAbbāsid translation movement,⁹³ have been extensively studied, whereas historical and literary texts, the remaining two major categories, have received much scarcer attention. In these groups, the translation strategy hardly ever aims at reproducing the text in an exact form, whether word-for-word or dynamically. We may have occasional passages of a text translated very exactly, showing that the translator had the ability to do so when he was willing to, but this rarely extends over several pages before we find major alterations vis-à-vis the original.

As an example of the different ideas of exactness in transmitting a text, let us consider a case of monolingual transmission (Arabic → Arabic) where both texts are, moreover, available in reliable editions, so that we may accept the passage as genuinely representing the quotation technique of the author. In his Ghurar, pp. 26–27, al-Thaʿālibī (wrote around 412/1022)⁹⁴ claims to be quoting the historian al-Ṭabarī (d. 310/923) (wa-dhakara al-Ṭabarī), yet only a minimal part of the passage is actually a quotation. Exactly quoted words are marked in the following excerpt from al-Ṭabarī in boldface, slightly changed ones in italics, and normal type indicates passages either very freely transmitted or with no equivalence whatsoever in al-Thaʿālibī’s text:⁹⁵

wa-zaʿamū annahū lam yusmaʿ min umūri l-Ḍaḥḥāk shayʾun yustaḥsanu ghayru shayʾin wāḥidin wa-huwa anna baliyyatahū lammā shtaddat wa-dāma jawruhū wa-ṭālat ayyāmuhū ʿaẓuma ʿalā l-nāsi mā laqū minhu fa-tarāsala l-wujūhu fī amrihī fa-ajmaʿū ʿalā l-maṣīri ilā bābihī fa-wāfā bābahū l-wujūhu wa’l-ʿuẓamāʾu min al-kuwari wa’l-nawāḥī fa-tanāẓarū fī l-dukhūli ʿalayhi wa’l-taẓallumi ilayhi wa’l-ta⁠ʾattī li-stiʿṭāfihī fa-ttafaqū ʿalā an yuqaddimū li’l-khiṭābi ʿanhum Kābī al-Iṣbahāniyya fa-lammā ṣārū ilā bābihī uʿlima bi-makānihim fa-adhina lahum fa-dakhalū wa-Kābī mutaqaddimun lahum fa-mathula bayna yadayhi wa-amsaka ʿan-i l-salāmi thumma qāla: ayyuhā l-maliku ayya l-salāmi usallimu ʿalayka? a-salāma man yamliku hādhihī l-aqālīma kullahā am salāma man yamliku hādhā l-iqlīma l-wāḥida – yaʿnī Bābila. fa-qāla lahu l-Ḍaḥḥāku: bal salāma man yamliku hādhihī l-aqālīma kullahā li-innī maliku l-arḍi. fa-qāla lahū l-Iṣbahāniyyu: fa-idhā kunta tamliku l-aqālīma kullahā wa-kānat yaduka tanāluhā ajmaʿa fa-mā bālunā qad khuṣiṣnā bi-ma’ūnatika wa-taḥāmulika wa-isāʾatika min bayni ahli l-aqālīmi wa-kayfa lam taqsim amra kadhā-wa-kadhā baynanā wa-bayna l-aqālīmi. wa-ʿaddada ʿalayhi ashyāʾa kāna yumkinuhū takhfīfuhā ʿanhum.

al-Ṭabarī, Taʾrīkh I: 208–209//II: 8–9

They assert that only one thing that could be considered good was ever said of al-Ḍaḥḥāk. When his affliction became great, his tyranny prolonged, and his days lengthened, the people felt that they were suffering so badly under his rule that their notables discussed the situation and agreed to travel to al-Ḍaḥḥāk’s gate. When the notables and powerful men from various districts and regions reached his gate, they argued among themselves about coming into his presence and complaining to him and achieving reconciliation with him. They agreed that Kābī al-Iṣbahānī would approach him to speak on their behalf. When they were traveling toward al-Ḍaḥḥāk’s gate, al-Ḍaḥḥāk was told that they were coming and permitted them to enter, which they did, with Kābī leading them. The latter appeared before al-Ḍaḥḥāk but refrained from greeting him. He said, “O king! What greeting should one give you? The greeting for one who rules all the climes or the greeting for one who rules only this clime – meaning Babylon?” Al-Ḍaḥḥāk replied, “Nay, but the greeting for one who rules all these climes, for I am king of the earth.” Then al-Iṣbahānī said to him, “If you rule all the climes and your sway extends to all of them, why then have we in particular been assigned the burden of you, your intolerance, and your misdeeds out of all the peoples of the climes? Why then do you not divide such-and-such a matter between us and the other regions?” Speaking the truth boldly, he addressed the issue and enumerated to al-Ḍaḥḥāk the ways in which the latter would be able to lighten their burdens.

Translation by William M. Brinner

As no translation is involved, we cannot say that al-Thaʿālibī would not have understood the text correctly or that he would have been unable to transmit it into another language in an exact form.⁹⁶ Had he wanted to, he could have copied al-Ṭabarī’s text, letter by letter. He simply did not want to do so, yet he explicitly claimed to be quoting from al-Ṭabarī. “Quoting” obviously meant to him something else than it does to us, and the same goes for translating. While “quoting” al-Ṭabarī, al-Thaʿālibī simply aims at making the text as readable and as relevant for the reader as possible.⁹⁷

Translations of historical and literary texts differ from each other, but in most cases one might from a modern point of view speak of adaptations, re-creations, or redactions, rather than translations proper. For our purposes, a translation may be defined as any new text in the target language that reproduces, partly or completely, a text in the source language, with or without enlargements and embellishments, abbreviations and changes.

Texts may be abbreviated or expanded, and often cases of both may be found in the same text showing that length itself was not – at least, not always – an issue, but the primary reason for changing the text was to maintain the interest of a new audience. From historical texts, information that is no longer relevant to contemporary readers may be excised and replaced by new material. This may be seen, e.g., in the insertion of pieces of Islamic sacred history into Persian national history in texts translated from Middle Persian into Arabic or Classical Persian and directed at an Islamic readership.⁹⁸

Usually such changes are made without comment, but sometimes they may be made explicit, as is done in the Preface to Narshakhī’s Tārīkh-e Bukhārā. The book was written in Arabic in 332/943 and translated into Persian some two centuries later, in 522/1128. The translator states in his preface (Tārīkh, p. 2):

This book was written in Arabic in an elegant style during the months of the year 332/943. Since most people do not show a desire to read an Arabic book, friends of mine requested me to translate the book into Persian. (…) Whenever unimportant items were mentioned in the Arabic manuscript, by the reading of which the temper became more fatigued, an account of such things was not made.

Translation by Frye 2007: 2

In other words, the translator abbreviated the text without scruple when he thought his audience might otherwise lose interest.

In metrical texts, freedoms taken by the translator are usually even greater. Let us take one example from the Book of Kings tradition. Even though both Arabic and Classical Persian prose use a lot of hendiadys and parallelism, the number of repetitions in Firdawsī’s epic was radically diminished in its Arabic prose translation, al-Bundārī’s al-Shāhnāma. Likewise, battle scenes are often abbreviated in the text, yet in some cases the translator adds passages, which are usually discernible by their use of rhymed prose (sajʿ ) and strong parallelism. In the following excerpt of Ḍaḥḥāk explaining to his courtiers why he was paralyzed with fear when Kāwe spoke to him (from al-Bundārī, Shāhnāma i: 34), clear similarities with Firdawsī’s Shāhnāme i: 49, have been marked in boldface:

lammā dakhala ʿalayya hādhā l-mutaẓallim ra⁠ʾaytu ka-anna jabalan min al-ḥadīd ḥāla baynī wa-baynahu. wa-qad awjastu fī nafsī minhu khīfatan qalqalat aḥshāʾī wa-shaghalat khāṭirī. wa-mā arā dhālika illā min ʿalāmāt zawāl mulkī wa-nqilāb ḥālī. wa-laʿalla shams dawlatī ādhanat bi’l-ghurūb wa-wajh ḥaẓẓī ʿalat’hu yad al-shuḥūb.⁹⁹

Translation, however, should not be the end of the story, but a beginning. If it is the end, then the text will have had little influence on the receiving culture and its transmission is in a certain sense a dead end or a miscarriage. Successfully received texts are retranslated into further languages, back-translated into the original and transmitted and modified in the target language. The Book of Kings tradition supplies good examples of the processes.

In some cases, its very prestige may, somewhat paradoxically, have been the reason why a translated text had no afterlife, being neither copied nor circulated. Such texts were sometimes buried in (usually royal) libraries, where they were limited to their owners’ pleasure and became tokens of wealth and power instead of being accessible to those who could have used them.¹⁰⁰ Several scientific translations seem to have suffered this fate, as well as at least one version of the Book of Kings, namely the manuscript in the treasury of the Caliph al-Ma⁠ʾmūn (r. 198–218/813–833), which left no identifiable traces of its existence to later literature.¹⁰¹

In order to be successful a translation has to have an influence on later literature and culture in general. This it may do in various ways:

1. through circulation: the translation is read,¹⁰² copied, and circulated.

2. through dissemination of the information it contains: this mainly takes place in microunits, which are quoted in other books. The quotations are often not acknowledged. In some cases, substantial parts of the text, or even its entirety, may be given as a quotation, or several quotations, within a larger text.¹⁰³

3. through transmission in redactions and rewritings: the translated text is modified by a later author, who needs not know the original source language. It should be emphasized that very often the redactor may, on purpose or not, give the impression that he is giving a new translation of the source text, while in fact he is merely elaborating on an existing one. Differentiating between the two in Arabic sources is made even more difficult by the fact that the verb naqala refers to both translating and transmitting.

4. through retranslations and back-translations: a target text may become a source text for another translator into a new language. A special case is when the new target language is the original source language or its descendant (e.g., Middle Persian → Arabic → Classical Persian). Sometimes material of the Book of Kings tradition has made interesting roundtrips between the three languages. To take but two examples:

   1. MPers. Khwadāynāmag (sixth c.) → Ar. Ibn al-Muqaffaʿ (eighth c.), Siyar →(?) Ar. al-Ṭabarī (tenth c.), Ta⁠ʾrīkh → CPers. Balʿamī (eleventh c.), Tārīkhnāme → Ar. trans. of Balʿamī (eleventh c.)

   2. MPers. Khwadāynāmag → Ar. Ibn al-Muqaffaʿ, Siyar →(?) CPers. Prose Shāhnāme (mid-tenth c.) → CPers. Firdawsī, Shāhnāme (early eleventh c.) → Ar. al-Bundārī (thirteenth c.), al-Shāhnāma¹⁰⁴

In order to understand the dissemination process (2) fully, one should keep in mind that in a manuscript culture the copying of bulky works was both time consuming and expensive and the writing materials were far from cheap. Hence, an essential part of transmission is when a text starts circulating in fragments, i.e., quotations in other works (microunits). This we can very often see in the case of the Arabic Book of Kings tradition: individual items are mined from the original translation(s) and set into other books, where they start a new life and continue their circulation as parts of a new book.

In the case of redactions (3), a text may be translated only once, but then freely modified in new redactions; or it may be translated several times, if it is prestigious enough and retains this prestige for a longer period, as the works of Aristotle did.¹⁰⁵ A mixed case is when an existing translation is used as a basis for a new version which is done by correcting the translation against the original and partly translating it anew. Both cases are amply documented in, e.g., the case of Ḥunayn ibn Isḥāq.

Thirdly, texts may be re-translated (4) into other languages – some scientific texts were first translated from Greek into Syriac and then from Syriac into Arabic, sometimes also from Greek through Arabic into Syriac – or even back-translated into a later form of the original language. Much of Middle Persian literature went this way, being first translated into Arabic and then from Arabic back into Classical Persian. It is rare that the text is translated back into the original language, but, e.g., the Arabic translation of Balʿamī’s rather free Persian translation and adaptation of al-Ṭabarī’s Arabic Ta⁠ʾrīkh is a good case of such back-translation.¹⁰⁶

To sum up, the Persian Book of Kings tradition presents a good case study for textual transmission and translation. It contains a suitable selection of languages (mainly Middle Persian, Arabic, and Classical Persian) with a sufficiently complex transmission history. It also exhibits all the four cases of what happens to a successful translation, as delineated above. In addition, there is reason to assume that it also benefited from oral transmission. Finally, the Book of Kings tradition cuts across the genre boundaries of historical and literary texts, partaking in some measure of both. This, added to the large number of texts involved and their long transmission period (roughly, 500–1200 ad and onward) makes it an excellent case study which illuminates the features discussed in this chapter.