6.1 One Khwadāynāmag. Or Many?

For a long time, most scholars have spoken about many Khwadāynāmags, but usually without supporting this with sufficient evidence or defining more exactly what they mean by “many”. There seem to be five main reasons for assuming plurality of the Khwadāynāmags. The first is the mistaken belief that Firdawsī’s Shāhnāme, written some 400 years later and partly fictitious, can be read as documentary evidence for the Sasanian period: when Firdawsī mentions that a Book of Kings was read to some king, this is taken as evidence for the situation in the Sasanian period and, hence, it is concluded that there must have existed early versions of the book. The second is the equally mistaken belief, deriving ultimately from Nöldeke (1879a): xix, that if the same event in pre-Islamic Iran is described in two or more different ways in Arabic sources, each version must derive from the Khwadāynāmag and, hence, provides evidence for various redactions or versions of the book.¹ The third is the passage Ḥamza, Ta⁠ʾrīkh, pp. 9–10, which is taken to refer to a plurality of Middle Persian texts, while, in fact, it speaks about Arabic translations. The fourth is the reference in Ḥamza, Ta⁠ʾrīkh, p. 16, to a number of manuscripts of some texts, usually (mis)understood to refer only to the Khudāynāme, that were not found to be identical. The fifth and most weighty piece of evidence is Bahrām ibn Mardānshāh’s reference to twenty-some copies of the Khudāynāme in Ḥamza, Ta⁠ʾrīkh, p. 22.

The first reason can be easily dismissed. There is a lot of historical material in the Shāhnāme and it can be used for historical studies. Yet we cannot rely on its details, especially those that are used to create atmosphere or have other literary functions (writing of letters; entertainment of the kings and heroes; etc.). Unfortunately, the reading of ancient books clearly belongs to this category. While we may perhaps rely on the fact that a certain king waged a campaign against the Byzantines, we may hardly rely on the narrative that after the battle, the heroes drank wine, listened to music, or had books read to them for their edification or entertainment.

As we have already shown that the Khwadāynāmag was but one of Firdawsī’s ultimate sources and not necessarily even that (Chapter 4.2), it should be clear that his Shāhnāme cannot be taken as representing a carefully preserved official record of those times.

The second reason has already been amply answered (Chapters 2.2.1 and 4.6): there was a variety of historical sources in Pahlavi and many of these are known to have been translated into Arabic (and yet others, unknown to us, may well have been translated, too), so that there is no reason automatically to assume that all variant versions of the same incident must by necessity come from variant versions of one book only.

The remaining three points need some more discussion. First, we have to distinguish between the Khwadāynāmag and its Arabic translations. If one refers to the Arabic translations, then certainly there was variation in them. We have seen that extraneous material was added to these translations, but the “Arab” material shows that this will definitely have been done within Arabic tradition by tapping Arabic sources, thus having nothing to do with the Middle Persian Khwadāynāmag. The reasons the Arabic versions differed from each other are attributable to two factors: 1) different translations of the same original obviously differ from each other in wording and style, etc.; 2) as translators added new materials to their translations, the result is, of course, that they differ from each other.

How many of the Arabic translations were direct translations from the Middle Persian is not clear, but it is obvious that not all the authors discussed in Chapters 3.1–4 produced completely new translations and some may have merely revised the version of Ibn al-Muqaffaʿ by additions and corrections in the light of other transmitted materials (whether in Arabic or Middle Persian, whether the Khwadāynāmag itself or some other texts, or even oral transmission). This admitted, one should then point out the obvious: the multiplicity of translations does not permit us to speak about multiple originals. The multiplicity of the translations of the Qurʾān or Bible does not allow us to speak about multiple Qurʾāns or Bibles.

To come to the third reason, we may repeat what has already been pointed out in Chapter 3.1: Ḥamza, Ta⁠ʾrīkh, p. 9, unequivocally speaks of translations, not original texts. The crucial passages are marked below in bold:

Their chronologies are confused, rather than accurate, because they have been transmitted for 150 years from one language into another and from one script, in which the number signs are equivocal, into another, in which the “knotted” number signs (ʿuqūd) are also equivocal.

What is perhaps even more significant is that Ḥamza, whose main interest is chronological, refers to confusion in number signs. He does not claim that any incidents were told differently or that the contents of the translations would have been different. Immediately after this, Ḥamza goes on to speak about the various manuscripts he had collected (“In this chapter, I have had to take recourse to collecting variously transmitted manuscripts (nusakh), of which I have come across eight”), after which he lists eight Arabic translations of Pahlavi texts, not original manuscripts in Pahlavi. There is, moreover, no reason to assume that all the eight Arabic texts were translations of the Khwadāynāmag. On the contrary, it has been shown in Chapters 3.1–3 that some of them in all probability were not translations of the Khwadāynāmag.

There cannot be the slightest doubt that the passage testifies to (limited) differences between various Arabic texts. Some, but only some of these, were translations of the Khwadāynāmag, and the differences between these would mainly concern confusion in the numerals, which complicated the use of these texts for chronological purpose, and this was Ḥamza’s main interest. What the passage does not say is that the differences between those texts that were translations of the Khwadāynāmag would necessarily have been significant outside of chronological matters or that all these eight texts would have been translations of the same text, or, finally, that any of them would have been in Pahlavi. What is more, if we accept the rather obvious conclusion that not all these texts were translations of the Khwadāynāmag, then the reference to differences can only refer to numbers and chronology: it would be superfluous to say that different books differed from each other in their content. It is the different chronologies, partly based on miswritten numerals, that were compared and found to differ from each other.

Next we come to the fourth argument, which is based on Ḥamza, Ta⁠ʾrīkh, p. 16:

Mūsā ibn ʿĪsā al-Kisrawī has said in his book: I looked into the book called Khudāynāme, which is the book that, when translated from Persian into Arabic, is called Ta⁠ʾrīkh mulūk al-Furs. I repeatedly looked into manuscripts (nusakh) of this book and perused them minutely, finding that they differ from each other. I was unable to find two identical copies. This is because the matter had been confused by the translators of this book when they translated it from one language into another.

At first sight, this would seem to refer to a Pahlavi text, al-Kisrawī first identifying Khudāynāme as the title of the original and then referring to the manuscripts of “this book”. However, we have seen that the same title, Khudāynāme, was also used for its Arabic translation (Chapter 1.1.1) and the end of the passage shows that al-Kisrawī is, after all, speaking about translations: the matter has become confused because of problems in translation. Had he used Pahlavi original(s), the sentence would make no sense.

We should also contextualize the passage: this is a prefatory note to the chronological list given on pp. 19–21. What it refers to are the regnal years of the kings, which, as Ḥamza had already noted (Ta⁠ʾrīkh, p. 9) get confused when texts are translated from one language into another. There is no indication that Mūsā ibn ʿĪsā al-Kisrawī was here speaking of various redactions or manuscripts which in broad lines differed from each other. What he says is that he perused the manuscripts minutely (baḥathtuhā baḥth istiqṣāʾ) and found that they differed from each other. What he seems to be speaking of are the usual scribal errors that take place during the transmission of a text and which are particularly problematic in chronology:² a small mistake may garble the chronology completely, whether the scribe uses letter or number signs or writes the numbers out in words. We cannot, however, completely rule out the possibility that there might have been other differences in the manuscripts used by Mūsā, but the passage of Ḥamza cannot be used as proof of this. Note, moreover, that Mūsā does not claim that the differences were great³ – on the contrary, he had to peruse the manuscripts minutely for comparison.

All in all, the first four arguments only show that there were, perhaps minor, differences in the Arabic translations of the Khwadāynāmag, as well as chronological differences between various Pahlavi texts in their Arabic translations. The main complaint against these Arabic translations concerns their numerals, not the texts as such.

Finally, we come to the fifth and most significant passage: Ḥamza, Ta⁠ʾrīkh, p. 22. The passage reads in its entirety:

(What follows) repeats what was mentioned in the first chapter of this History, with a commentary, which was brought by Bahrām ibn Mardānshāh, the mōbad of the district of Shābūr from the country (balad) of Fārs.

Bahrām al-Mōbadhānī said: I collected more than twenty manuscripts of the book titled Khudāynāme and corrected (aṣlaḥtu) from them (i.e., on their basis) the chronologies (tawārīkh) of the kings of Persia from Kayūmarth, the Father of Mankind until the end of their days and the transfer of kingship from them to the Arabs.

The same Bahrām appears on Ḥamza’s list (Ta⁠ʾrīkh, p. 10), where he is listed as the last authority and the title of his book is given erroneously as Kitāb Ta⁠ʾrīkh mulūk Banī Sāsān (see Chapter 3.2.6).

The language of this Khudāynāme is not indicated in the passage and we have just seen how al-Kisrawī (and following him, Ḥamza) refers to the Arabic translations under the same name. As a mōbad, Bahrām can be expected to have been familiar with Pahlavi, so he might well have used the original text, but this is not stated in the text.

However, there must have been Middle Persian copies of the text circulating at the time of Ibn al-Muqaffaʿ and available to those persons on Ḥamza’s list that really translated the text anew from the original language and did not merely elaborate on Ibn al-Muqaffaʿ’s translation. It is quite possible that Bahrām used copies of the book in both languages and there is no reason to deny that there might well have been several copies of the original Pahlavi text still circulating in the tenth century, and they may even have come up to a high number, twenty-odd, but there is no more reason to deny that some of these twenty-odd copies might as well have been Arabic translations, better and more certainly known to have circulated at the time.

What seems to have remained unnoticed, though, is the content and extent of what “was brought by Bahrām ibn Mardānshāh, the mōbad of the district of Shābūr from the country (balad) of Fārs.” The excerpt from Bahrām covers only four pages of text in Ḥamza’s Ta⁠ʾrīkh (pp. 22–25): the following chapter (pp. 26–49) is not said to derive from Bahrām and, as we have seen (Chapter 3.6), some of its sources can be identified and these are known to have been books other than the Khwadāynāmag. Some individual pieces of this material (pp. 26–49) may come from Bahrām, but the chapter is given as an afterthought, or commentary, to the main chronological list, and it is this list that is the result of Bahrām’s collection of twenty-odd copies. Whether the next chapter is by Bahrām or not, the references to Kitāb al-Ṣuwar etc. make it impossible to claim that all information on pp. 26–49 would derive from the Khwadāynāmag and it is ultimately insignificant whether it was Bahrām, Ḥamza, or somebody else, who added this “commentary” section to the chronological list given on pp. 22–25, which is explicitly attributed to the Khudāynāme, in whichever language.

It is also worth noticing that in whichever language(s) the Khudāynāme copies were, they must have been short enough for a mōbad to collect twenty-odd copies and collate them with each other. We will return in Chapter 6.2 to the question of the size of the original Khwadāynāmag and its translations.

Bahrām does not say that there were any major differences between the copies of the Khudāynāme, and as his text contains only a chronological skeleton consisting of the names of the kings and their regnal years, with little elaboration, the differences between the copies must have related to this information.

At the end of the following chapter, p. 49, there is a centrally important passage which describes the contents of the books used for this chapter:

These short stories about the kings with which I fleshed this chapter (pp. 26–49) out are not found in the books of tawārīkh and siyar,⁴ except in small measure. The rest of them are in (i.e., come from) their other books.

Hence, the preceding chapter, pp. 22–25, comes from chronological and historical books, obviously the Khwadāynāmag being one of these, and the next chapter gives the commentary to this skeleton history of Persian kings, explicitly derived from a variety of other sources (“their other books”), not the Khwadāynāmag or any one book, for that matter.

Based on these five arguments and their discussion, we may now sum up our results for the time being. The evidence shows that the Arabic translations of the Khwadāynāmag had differences between each other, mainly in the use of numerals but admittedly probably going further than this, namely to the content of the texts. The Arabic Khudāynāme also circulated in a number of copies and versions, attributed to several translators, some/many of which were probably working on the basis of earlier translations, not the original text. In addition, we have reason to assume that the original Pahlavi text was also in circulation, but whether the number “twenty-odd” refers solely to Pahlavi copies of the text or whether it includes Arabic translations, is not clear. Nowhere are these Pahlavi texts described as different recensions or versions: they are only called different manuscripts (nuskha) of the Khwadāynāmag.⁵

There is only one piece of evidence that I am aware of that might be interpreted as speaking in favour of the plurality of Middle Persian Khwadāynāmags. This comes from Ḥamza, Ta⁠ʾrīkh, pp. 50–51:⁶

The fifth chapter of the first book, concerning narrating some passages (jumal) of what there is in the Khudāynāme, which neither Ibn al-Muqaffaʿ nor Ibn al-Jahm narrated. I have put them at the end of this chapter, so that the reader might consider them in the same way he considers the Arabs’ stories about Luqmān ibn ʿĀd and the Israelites’ stories about ʿŪj and Bulūqiyā.⁷ That should be understood.

I have read in a book that was translated⁸ from a book of theirs titled al-Ābistā (the Avesta) that … [there follows the story of Gayōmard, here Kahūmarth, and the twins Mashih and Mashyāna, in some seventeen lines].

I also read about this in a different form and with more commentaries to the narrative in another book: [there follows some ten lines with chronological and astronomical details that were lacking from the first narrative].

The passages come at the end of the chapter that discusses pre-Islamic Persia, as if an afterthought. There are three ways to explain the reference to passages in the Khudāynāme “which neither Ibn al-Muqaffaʿ nor Ibn al-Jahm narrated”. The first is that Ḥamza here uses the title Khudāynāme in a generic sense⁹ for various works on Persian national history. This would be our one and only such case.

The second possibility is that Ḥamza, who did not read Pahlavi, is here referring to the other versions of the Arabic Book of Kings (either under the title Khudāynāma or Kitāb al-Siyar), based on Ibn al-Muqaffaʿ’s translation (or new translations made from Middle Persian) that had been enlarged with other materials.

The third possibility is that Ḥamza, or the copyist, has simply been careless. A simple error of the copyist (or the author) may have set the expression fī Khudāynāme in a wrong position. If we change the place of this element, the passage conforms to our lack of any other evidence for a plurality of Khwadāynāmags. If, instead of “fī ḥikāyati jumal mā fī Khudāynāme lam yaḥkihā Ibn al-Muqaffaʿ wa-lā Ibn al-Jahm” we read “fī ḥikāyati jumal mā lam yaḥkihā Ibn al-Muqaffaʿ wa-lā Ibn al-Jahm fī Khudāynāme,” the passage becomes unproblematic: “concerning narrating some passages (jumal) which neither Ibn al-Muqaffaʿ nor Ibn al-Jahm narrated in (their translations of) the Khudāynāme.” This would also explain how Ḥamza suddenly first quotes the Avesta and then another book in Arabic (presumably through Abū Maʿshar al-Munajjim, who refers to the Avesta in a quotation on p. 11) after claiming to be quoting from the Khwadāynāmag.

A variant of this would be that we just admit that Ḥamza has here been rather lax with his formulations. In any case, I find it difficult to conjure a plurality of Khwadāynāmags on the sole basis of this one short and problematic passage (“supported” by the various misunderstandings discussed above) that can, moreover, easily be emended to conform with the rest of the evidence. If we were to claim that this passage has to be taken at face value, we would still have to explain how the first passage is, however, attributed to the Avesta, not the Khwadāynāmag, and the second passage comes from “another book”.

The existence of (at least) two different types of the Arabic Siyar, however, seems to be confirmed by Ḥamza, Ta⁠ʾrīkh, p. 20, who mentions the existence of a long and a brief version of the Siyar. The latter may well be taken to represent more closely the Khwadāynāmag, whereas the former could well refer to an Arabic version expanded by stories which were originally separate. It is improbable that at least those stories which have an Arab point of view would first have been written down in Middle Persian and only then translated in two versions into Arabic. More probably a succinct Middle Persian version was first translated into Arabic and then expanded by adding material relevant to the Arabs and, perhaps, tales from other sources, some of which may have been in Middle Persian.

It also seems to be a rather common idea that the Khwadāynāmag was a priestly text or that there were two separate Khwadāynāmags, one of which was priestly. It is difficult to trace the origin of this idea which is often repeated as self-evident. As far as I am able to see the idea again derives from a confusion between other sources for Persian national history and the specific book called the Khwadāynāmag: passages from Arabic, Persian, and Middle Persian texts contain material with a strong Zoroastrian interest or even bias, and there is no doubt that some of this is “priestly” in the sense that its authors were probably mōbads or hērbads and the texts also reflect their ideas in their attitudes towards the kings.¹⁰

However, no such attitudes can be found in the material that might be considered specifically to derive from the Khwadāynāmag. Obviously, the kings were Zoroastrian and the Empire had a “Zoroastrian bias” because that was the general, and at times official, religion of the dynasty, but no such bias in the sense of any antagonism between the kings and the Zoroastrian “Church” is to be seen in what we know about the Khwadāynāmag: kings are not assessed in this material according to their attitude towards the clergy or blamed for their lack of enthusiasm in religion. Even though there are various evaluations of them in the Arabic and Classical Persian material, this does not mean that they were differently evaluated in the Khwadāynāmag.

In short, “a priestly Khwadāynāmag” is a phantom based on inappropriate use of terminology, yet again an example of how the expression “the Khwadāynāmag tradition” has misled scholars.¹¹

Finally, while the present study shows that there is no need to speak of several Khwadāynāmags, this does not mean that there would not have been any variation between the manuscripts of the Khwadāynāmag. Most probably there was. As we know from any historical or epic work, manuscript variants, short additions, deletions (either conscious or not), and so forth tend to accrue over the years even to a text which has been carefully transmitted. The Khwadāynāmag need not have been an exception to this. But just as there are variants in the manuscripts of, e.g., al-Ṭabarī’s Ta⁠ʾrīkh, and we still do not speak of a plurality of his Ta⁠ʾrīkhs, the variants, if such there were, do not legitimate speaking about the book in the plural. The same goes for possible additions to the text. If the Khwadāynāmag was a brief chronological text (see the next Chapter 6.2), it is quite possible that the last few kings of Iran may have been added to the list by later scribes, either in the original Pahlavi text or, more probably, in its Arabic translation to keep it up to date.¹² Yet this does not give us reason to speak of several Khwadāynāmags.

6.2 The Contents, Size, Sources, and Date of the Khwadāynāmag

What, then, was the Middle Persian Khwadāynāmag like? To get an answer to this we must work backwards in time. The contents of Ibn al-Muqaffaʿ’s translation of the Khwadāynāmag may be deduced in rough outline from the extant references and quotations. As the second step, we may then speculate on the relationship between it and the original Khwadāynāmag, keeping in mind that the translation process for historical literature did not expect similar exactitude from the translator as with the translation of scientific and philosophical works. Even a complete reconstruction of Ibn al-Muqaffaʿ’s translation would still leave much to do to reconstruct the original Khwadāynāmag.

Let us start with some negative remarks. Chapter 5.1 has shown that among Arabic scholars before al-Thaʿālibī very little was known about Rustam, which implies that he was at most a marginal character in the Siyar of Ibn al-Muqaffaʿ, and there is no reason to assume that Ibn al-Muqaffaʿ for some obscure reason purged him and the other Sistanians from his translation. Thus, one may surmise that there was little about Rustam in the Khwadāynāmag, too, and he may not even have been mentioned in the book. Neither do we have any evidence for the other heroes, Sistanians or others, in the Khwadāynāmag. The Khwadāynāmag clearly was what its title says, a book of kings, not of heroes.

Alexander and the Petty Kings are problematic. A Middle Persian Alexander Romance has been postulated, but in the light of the evidence its existence is seriously to be doubted (Chapter 2.3). Alexander was a problematic character for the Sasanids, who largely modified themselves on his enemies, the Achaemenids, about whom they knew little but were eager to imitate the little they did know. Hence, in Pahlavi literature Alexander is a negative character. The positive character in later, Islamic Persian tradition clearly comes from the Romance in one way or another, and there is no reason to assume that such features were included in the Khwadāynāmag. From a chronological point of view, Alexander was important, though, and one would presume that he was, in one way or another, mentioned in the Khwadāynāmag, presumably as a negative character whose evil deeds could be summed up in a few lines, perhaps mentioning the burning of Persian books and other acts of vandalism in the country or, perhaps, just giving the number of years of his interregnum. For the Sasanians, Alexander was hardly a legitimate Persian king, as he was for the later Persian tradition.

The Seleucids and the Parthians were little known in early Arabic historiography and were presumably only summarized in the Khwadāynāmag, if even that. We have to remain aware that this was a royal book of the Sasanids. The Seleucids would have been usurpers to them, and a tension must have remained between the Sasanids and the previous dynasty, the Parthians, even though, as Pourshariati (2008) has shown, we should not think that the Parthian element disappeared after the emergence of the Sasanian Empire. However, the lack of information on these two dynasties in Arabic sources shows that not much was told about them in the Khwadāynāmag, if anything. Whatever the extent of the earlier parts, it is clear that the Sasanids were the main focus of the book.

Firdawsī’s epic contains wonderful stories, both moving, entertaining, and full of suspense. But Firdawsī’s Shāhnāme is not the same as the Khwadāynāmag, as has often been emphasized throughout this book. How far the confusion has gone may be exemplified by an authoritative writer. Ehsan Yarshater writes about earlier Middle Persian sources in the Cambridge History of Iran 3/1 (1983): 393: “Some of these works must have been incorporated either in their entirety or in an abridged version in the later recensions of the Khwadāy-nāmag. We find, for instance, that a complete version of the Ayādgār ī Zarērān is reproduced by Firdausī and an abridged form by Ṭabarī and Thaʿālibī.”

What this actually proves is merely that the story had existed and been translated, or summarized, in Arabic by the time of al-Ṭabarī and had been incorporated into the common source of Firdawsī and al-Thaʿālibī, the Prose Shāhnāme (see Chapter 4.2). Nothing more. Zarēr may have been mentioned in the Khwadāynāmag, but there is no reason to assume that his story would ever have become part of the Khwadāynāmag.

We have already shown that the Sistanians did not play any role in the book and that there is reason to doubt whether they were even mentioned there. Likewise, there is also reason to believe that many of Firdawsī’s orphan stories, such as that of Bīzhan and Manīzhe, did not derive from Firdawsī’s main source, the Prose Shāhnāme, and, hence, have nothing to link them to the Khwadāynāmag, which Firdawsī used, if he used it at all, through the Prose Shāhnāme.

This already minimizes the potential narrative element in the Khwadāynāmag, and in Chapter 3.4, we have pointed out that many of the other stories, too, are unlikely to have been included in the Khwadāynāmag: stories about rebellions against the Sasanids and others centrally featuring Arab characters and full of Arab interest are difficult to assume to have been parts of the royal Sasanian chronicle, and there is no evidence that they would have been parts of the Khwadāynāmag or even its Arabic translation.¹³ Instead, these stories are known to have existed in separate Arabic translations (Chapter 2.2.1).

When the Khwadāynāmag was translated into Arabic, narrative elements may have been added to it, at least to some extent, but it is only in the tenth-century Classical Persian texts that we first encounter a fully-developed narrative history of Iran, in which the episodes, hitherto transmitted as separate texts, have been integrated into the chronological framework, possibly provided by the Khwadāynāmag, creating a powerful epic narrative of great literary merit, which culminated in Firdawsī’s Shāhnāme.

Seeing the Khwadāynāmag as a rather dry chronicle with little narrative also frees us from several problems. Had the Khwadāynāmag been a huge epic in prose, anywhere near the size of Firdawsī’s Shāhnāme, or even a quarter of it, it would have been the largest Pahlavi text of the time that we would be aware of. There are few long texts in Pahlavi literature, and the one that is the longest, the Dēnkard, is a text which was compiled much later and was based on shorter texts that have been excerpted and put together to form an encyclopaedia of knowledge, which despite this comes up in the modern facsimile edition to only 832 small pages.¹⁴

All other historical Pahlavi texts that are extant or that we know or presume to have existed are much shorter. The Kārnāmag ī Ardashīr covers less than 70 small pages in the modern edition and Ayādgār ī Zarērān even less (17 pages).¹⁵ Even religious texts are usually rather brief. The Bundahishn, again a later compilation, has 83 and 240 small pages in its two redactions.¹⁶ Of the Hazār Afsāne we know very little, and the book certainly did not contain one thousand stories, as the title would have it,¹⁷ and the existence of a Pahlavi Alexander Romance is extremely dubious (Chapter 2.3), but if it existed and was about the same size as the Syriac texts, it would clearly be the longest single non-religious Pahlavi text we can point out.

Similar conclusions may be reached by considering the translation of the Khwadāynāmag itself. If even some of the texts mentioned in Chapter 3.1 were new translations or even thorough reworkings of an earlier translation, one is hard put to claim that so many versions could have existed if the original was a voluminous book. Ibn Isḥāq’s long Life of the Prophet circulated in several recensions, but it was a centrally important text for the Muslim community, which the translation of the Khwadāynāmag certainly was not. Did twenty-something (Ḥamza, Ta⁠ʾrīkh, p. 22) scholars each copy and revise a book of, say, 250 pages for their private use? And would the result of the efforts of Bahrām ibn Mardānshāh be a disappointing four pages of text after having collated more than twenty copies of a voluminous work? Putting together all passages on pre-Islamic Iran in al-Ṭabarī’s Ta⁠ʾrīkh would hardly make up two hundred pages and it is clear that he used several sources to achieve this.

Ḥamza, Ta⁠ʾrīkh, p. 9, may again be used to argue for a chronicle-type content of the Khwadāynāmag:

Their (the Persians’) chronologies are all confused, rather than accurate, because they have been transmitted after 150 years from one language into another and from a script, in which the number signs are equivocal, into another language, in which the “knotted” number signs are also equivocal.

While not saying anything about what else the Khwadāynāmag might have contained, this passage refers to a text where numbers play a major role, which would tally with the contents of the Bahrām ibn Mardānshāh quotation in Ḥamza’s Ta⁠ʾrīkh: a dry list of kings and their regnal years.

These examples should suffice to show that even on the Arabic side, a voluminous Siyar al-mulūk would perhaps be out of line with what we might expect and with what the evidence would seem to point to. Let me here take a purely speculative step and spell out what I believe the Khwadāynāmag to have been, based on the studies presented in this book but going beyond them into the unprovable, but, I hope, never coming in collision with any of the available evidence.

For me, the Khwadāynāmag is a book of very small size, be it of 10, 20, or 30 pages. It contained a list of Persian kings and its main interest may well have been chronological – at least, Ḥamza, who himself is admittedly specifically interested in chronology, would give us this impression and Agathias’ evidence supports this. It clearly started with Gayōmard and continued until the time of its writing (cf. below), and individual copies may well have been expanded by adding a few lines on the last kings of Iran, to cover the whole story of pre-Islamic Iran until the conquest.

The Khwadāynāmag probably contained a rather short and dry account of each king, listing his regnal years, perhaps some throne speeches or maxims, mentions of the foundation of cities and Fires, and the main (positive) events during his reign, such as major victories. This image would equally well fit the evidence of Agathias and of Ḥamza, and it is hard to come by any tangible evidence to the contrary. In style, it would probably be comparable with the Pahlavi original of the Arabic Kitāb al-Ṣuwar.

Throne speeches and maxims are reported in several sources and they might well come from the Khwadāynāmag, although the genre of wisdom literature (andarz) was a favoured one in Middle Persian literature and there were certainly separate texts of that genre, many of which have even been preserved. But the tradition is rather unanimous in attributing a handful of maxims to many kings in contexts where we surmise the main source to have been the Khwadāynāmag, which shows that either the maxims were already there or some early author, be it Ibn al-Muqaffaʿ or someone else, had inserted them into the Arabic version.

This is further supported by a passage in Ḥamza, Ta⁠ʾrīkh, p. 49. After describing what was mostly lacking in the books of chronology/history, Ḥamza adds two sentences which show that these chronological texts (i.e., the Khwadāynāmag and other historical books) did contain pieces of wisdom literature:

fa-hādhā alladhī ḥashawtu bihi hādhā l-faṣl min qiṣār akhbār al-mulūk mā laysa fī kutub al-tawārīkh wa’l-siyar minhu illā qalīl wa-bāqīhi fī sāʾir kutubihim. fa-ammā rasāʾiluhum wa-waṣāyāhum wa-mā ashbaha dhālika mimmā huwa fī kutub al-ta⁠ʾrīkh fa-qad akhlaytu l-kitāb minhu.

These short stories about the kings with which I fleshed this chapter out are not found in chronological and historical books, except in small measure. The rest of them are in (i.e., come from) their other books. I have, however, omitted from this book their letters and testaments and such material that is found in chronological books.

The foundation of cities is also very often mentioned in the texts belonging to this tradition, and Shahrestānīhā ī Ērānshahr might well have been compiled by taking these parts of the Khwadāynāmag aside to form a book with geographical orientation, although it has to be emphasized that this remains wholly speculative.

In addition, there will have been bits of information that are not listed here. Ḥamza himself obviously abbreviated the material, but this does not say how much was left off. Did he cut half of the text away? Or a third or two thirds? We have no way of knowing, but it remains clear that the arguments presented in this chapter have to be taken into account in estimating this.

Reading the Nihāya, one gets a similar picture of the situation.¹⁸ Excluding the long stories, as should be done (Chapter 3.4), the Sasanian biographies are usually built on only three or four elements. To take a typical example, the short biography of Bahrām ibn Sābūr ibn Sābūr Dhī l-Aktāf (Nihāya, pp. 247–248) consists of four elements:

1. words spoken by him on ascending the throne;

2. a throne speech;

3. the sending of an encyclica (this element is missing in many short biographies);

4. a short report of his death and the number of his regnal years, to which the towns founded by the king are sometimes added.

The mention of an encyclica is especially interesting in the light of what Ḥamza, Ta⁠ʾrīkh, p. 49, says about the kings’ letters having been quoted in Pahlavi chronological works (kutub al-ta⁠ʾrīkh).

When it comes to the sources of the Khwadāynāmag, we are on even more speculative ground. Shahbazi (1990): 209–213, lists what he thinks were the sources of the Khwadāynāmag: old sagas (not further elaborated by Shahbazi), archival texts; narrations of contemporary events; and the “Ctesian” method, wherewith he means “anachronism whereby old history was enriched and its lacunae filled in by the projection of recent events or their reflections into remoter time” (p. 211).

While there were certainly some kinds of archives in the Sasanian Empire and while these may well have been fleshed out by knowledge of contemporary events, Shahbazi’s claim to take them as sources for the Khwadāynāmag is purely speculative. Shahbazi also uses “documentation” from Firdawsī’s Shāhnāme, a text written four centuries after the Khwadāynāmag, where literacy is not only attributed to the Sasanians – whose Empire certainly was literate – but even to the mythical kings, whose kingdom certainly would not have been literate, had they any standing in real history in the first place.

Narrations of contemporary events could well have contributed to the Khwadāynāmag, but first we should be able to show that the book did contain extensive narratives, which does not seem to have been the case. Shahbazi also falls victim to the confusion between one specific book, the Khwadāynāmag, and Persian national history in general, claiming without the slightest evidence that, e.g., the great campaign of Kay Khusraw against Afrāsiyāb “ultimately derived from” the Khwadāynāmag (p. 211), which we have very good reason to claim was not narrated, at least not extensively, in the Khwadāynāmag but in other Middle Persian texts (see Chapter 2.2.1).

Jackson Bonner (2015): 142, concludes his study by asking himself “whether there was any real historiography of Sasanian Iran at all.”¹⁹ While I would not go so far, this study would also imply that scholars tend to have an exaggerated idea of Sasanian historiography: the Khwadāynāmag is easily seen as a huge compilation living on in a great number of versions or recensions and, it would seem, mounting to hundreds of pages. About all this we have no tangible evidence.

The use of Firdawsī’s Shāhnāme as a historical document pops up again in Shahbazi’s speculation on the date of the Khwadāynāmag’s composition (1990: 213–215). Shahbazi claims that the book existed at the time of Bahrām Gūr because the Shāhnāme tells that Bahrām asked the book of kings (nāme-ye khusrawān) to be read in his presence. Again, this is a topos which should not be taken as historical truth, no more than the several letters written by various legendary heroes, such as Zāl and Sām, which cannot be used as evidence for the literacy of these legendary heroes.

That the Khwadāynāmag was first written down at the time of Khusraw Anūshirwān is a legend based on the Bāysunqurī Preface.²⁰ It is, though, quite possible, although Nöldeke’s note (1879a: xv) that until Khusraw Parwīz (r. 590–628) the information on Persian national history in Islamic sources is usually uniform would instead favour the dating of the book to his reign.

Shahbazi claims to have found geographical evidence for the dating of the Khwadāynāmag to Khusraw Parwīz’s time from the Preface of the Prose Shāhnāme (1990: 214–215). There Iran is defined in geographical terms, extending from the Oxus to the Nile and from Rome (Byzantium) to the Land of the Berbers (North Africa?). While it is interesting that this defines with some exactitude the limits of Iran at the time, and only at the time, of Khusraw Parwīz, there are three points that make one hesitant to accept this. First, there is no indication that this definition would come from the Khwadāynāmag – it may well be, and probably is, a definition given in the tenth century, possibly in remembrance of the maximal area the Sasanians ruled at the height of their power. Second, the larger definition of Ērānshahr is also attested in late Pahlavi works (especially Shahrestānīhā ī Ērānshahr),²¹ and as Daryaee (2002): 6–7, notes this became the late Sasanian (and post-Sasanian, one may add) concept of the Ērānshahr. Third, nāme literature (Chapter 4.7) extended the adventures of Iranian and Sistanian heroes far and wide, from Spain and North Africa to India and deep into Central Asia and China. This literature was well known in the mid-tenth century and may well have influenced the common idea of the area that, in some sense, belonged to the by then legendary Ērānshahr.

The date is tempting, though. If we accept it, then the next question would be, do we find any evidence for dating the Khwadāynāmag to the time of Khusraw Anūshirwān in the first place? The answer has to be in the negative. Khusraw Anūshirwān’s fame as a patron of literature was in the Islamic period,²² as it had already been in Agathias’ time, great and it was only natural to ascribe any important book routinely to his reign. Outside of the Islamic tradition, only Agathias might be taken as evidence for the Khwadāynāmag’s existence in Khusraw Anūshirwān’s time. Agathias (d. 582) died merely three years after Khusraw (d. 579). Now, if the Royal Annals that he was using indeed refers to the Khwadāynāmag, then the work must date back to the time of Khusraw Anūshirwān.

Some support for either dating might be got from Ḥamza al-Iṣfahānī’s Ta⁠ʾrīkh, p. 9 (“Their (the Persians’) chronologies are all confused, not sound because they have been transmitted after 150 years from one language into another”), if we read this as a reference to the translation of the Khwadāynāmag from Middle Persian into Arabic. Ibn al-Muqaffaʿ translated the work around 750, so the number could refer to the original having been written around 600 – but I am afraid we cannot demand great exactitude from the numbers to exclude Khusraw Anūshirwān’s reign, and the fact remains that we cannot even be quite certain whether Ḥamza is here referring to the time between the original and the translation or something else.²³ In any case, if we date the Khwadāynāmag to Khusraw Parwīz’s time, then there is no reason to attribute the text to Khusraw Anūshirwān’s times and postulate a new redaction of the book,²⁴ as the connection between Agathias’ Royal Annals and the Khwadāynāmag is not certain.²⁵

A further possibility would be to date the text to the time of the last Sasanian king, Yazdagird iii. The Preface to the Bāysunqurī Shāhnāme²⁶ mentions a historical compilation based on earlier works and compiled by Dānishwar dihqān at the beginning of Yazdagird’s reign, reaching up to the end of Khusraw Parwīz’s reign.²⁷ The source is late, and the whole story very dubious, contradicting all the other sources and, moreover, often anachronistic. With good reason, Ṣafā (1374): 80–81, presents doubts as to the name of Dānishwar dihqān, which does not fit the Sasanian naming tradition and sounds like a name invented in conformity with other such expressions in Firdawsī’s Shāhnāme (dihqān-e sakhungūy, dānā, etc.). Together with the name, we have good reason to doubt the whole story.²⁸

In Ḥamza, Ta⁠ʾrīkh, p. 22, Bahrām is quoted as saying that he compared his twenty-odd manuscripts and “corrected from them (on their basis) the chronologies of the kings of Persia from Kayūmarth, the Father of Mankind, until the end of their days and the transfer of kingship from them to the Arabs.” Taken literally, this would mean that at least some of his copies, whether in the original Pahlavi or in Arabic translation, took the story up to the end of the Sasanian dynasty. In the case of Ibn al-Muqaffaʿ’s translation we have reason to believe that his version did so (Chapter 3.7) and we might expect the same to be true also in the case of the other Arabic translations.

Whether the Pahlavi texts did the same is not clear. Bahrām may himself well have added the final kings (and queens) from other sources, but it is also quite possible that an individual scribe copying the Khwadāynāmag updated the manuscript by adding a few lines on the last rulers. It does not matter whether it was Bahrām or some scribe of the Khwadāynāmag who added the last rulers, the one to an Arabic compilation or the other to the Pahlavi original, but what does matter is that even if the addition came from one or several of the original Pahlavi manuscripts, the addition of a few lines at the end of the manuscript hardly allows us to call such a manuscript a new version of the Khwadāynāmag. If we did so, then almost any work written in Pahlavi, Arabic, or Persian should be said to exist in several versions.

Ultimately, dating the Khwadāynāmag is full of problems, and the best we can say is that in all probability it stems from the reign of either Khusraw Anūshirwān or, more probably, Khusraw Parwīz.