Middle Persian documents and the making of the Islamic fiscal system: problems and prospects

Abstract

This article charts a new course for the study of the Middle Persian documents from early Islamic Iran, which takes their early Islamic context into account more fully than has hitherto been done. This approach and its potential fruits for the study of early Islamic history are illustrated through an in-depth treatment of four seventh-century documents from the Qom region (previously edited and discussed by Dieter Weber), each of which contains a fiscal term that is apparently otherwise unattested in the documentary corpus. I show that the existing interpretations of these documents anachronistically project the fiscal terminology and structures of a later time into early Islamic Iran, and that these documents, considered in aggregate, suggest a certain course of development for the Islamic fiscal system in the post-Sasanian territories in the decades following the initial conquests: from broad and relatively unspecific impositions to more targeted exactions, based on increasingly detailed assessments.

Introduction

The Middle Persian documents from early Islamic Iran offer the potential to transform our understanding of Sasanian and Islamic history alike, in many respects. In the realm of historical linguistics, they significantly expand our knowledge of the Middle Persian language and the developments that would eventually give rise to New Persian. In paleography and diplomatics, they furnish critical evidence for Sasanian documentary practices (insofar as these can be projected back into pre-Islamic times) and for the “Pahlavi substrate” that would decisively inform the evolution of Arabic documentary script.¹ They also offer a uniquely direct perspective on the administrative structures of early Islamic Iran, on the local, regional and perhaps even imperial levels, which cannot but expand our knowledge of the Sasanian state and society as well.

The Islamic Middle Persian documents largely fall into two groups – the Tabarestān corpus and what I will call the Qom corpus.² The Tabarestān corpus includes some 33 legal documents, which date from the mid-eighth century ce, and are in private hands. All but two of these have now been published.³ The main distinguishing feature of the Tabarestān corpus, aside from its legal orientation, is the striking continuity its documents exhibit with Sasanian customs and terminology, according to the investigations of Maria Macuch. As she writes, they furnish “direct evidence of legal practice in accordance with the age-old Sasanian juridical system, using its technical terminology and formulae and applying the norms known from pre-Islamic law in Iran”, as known primarily from the late Sasanian legal compendium Mādāyān ī Hazār Dādestān, or Book of a Thousand Judgments.⁴ This is in keeping with what is otherwise known of this region's history; Tabarestān only came under Islamic rule at a fairly late date, officially in 761, and the region's rulers’ titulature and coinage evince notable continuity with Sasanian models, throughout the mid-eighth century and beyond.⁵

The Qom corpus, the main focus of this article, includes over 300 documents, mostly financial in nature and dealing with estate management. Only a few of these are from Qom itself, but all are from the city's vicinity. They largely date from the mid-seventh to the early eighth centuries ce and, to date, some 190 of them have been published.⁶ By contrast with the situation in Tabarestān, this region came under Islamic rule relatively soon, a few years after the decisive Muslim victory at Nihāwand in 642.⁷ For some 60 years after the initial conquest, however, as noted by Andreas Drechsler, developments in Qom and its surroundings are touched on only very seldom in the Islamic historiographic sources, until a substantial group of Ashʿarī Arabs, who had been followers of Ibn al-Ashʿath, settled there in or after the failure of the latter's rebellion in 702/3. Although a modest Arab presence is recorded a decade or so earlier, for Drechsler this near-lacuna indicates that the initial conquest of Qom was decidedly limited in its local impact – the region remained by and large “untouched” by its new rulers over these 60 years, “in near-total isolation from political incidents”. Thus, in many respects life continued as it had under Sasanian rule, with taxation being perhaps the one notable exception, insofar as the region was administered, at least nominally, by Umayyad governors.⁸ Unbeknown to Drechsler, this 60-year span happens to coincide with the period in which the vast majority of the Qom documents were written; for Dieter Weber, this is far from a coincidence, but rather indicates that “the coming of these [Ashʿarī Arabs] put an end to the Zoroastrian life as is documented in the texts”.⁹ The extent to which Qom and its surroundings were integrated into the Islamic state in the decades immediately following their initial conquest, and what the Qom documents have to contribute on this score, remain open questions; this article will largely be concerned with addressing them.

Due in large part to the difficulties presented by the Pahlavi script, the decipherment of these documents is still very much a work in progress. Philippe Gignoux and Dieter Weber have done heroic work in producing editions of over 190 Islamic-era Middle Persian documents over the last 20-odd years, but much in their editions remains uncertain and, accordingly, subject to change. Moreover, a large portion of the Qom documents that are housed at the University of California, Berkeley (perhaps as many as 150) remains entirely unpublished, and it was only in 2022 that photographs of this critically important collection became available to scholars – an enormously promising development, albeit one that took place too late to substantially inform this article. For these reasons, at least on the philological level, if not on the interpretive and historical levels as well, any discussion of the Middle Persian documents must retain a somewhat tentative and speculative character.

These caveats and cautions notwithstanding, here I offer several substantive conclusions regarding the interpretation of these documents, the Middle Persian fiscal terminology encountered therein and the formation of the Islamic fiscal system. Examining four documents from the Qom corpus, each of which contains a fiscal term that is apparently otherwise unattested in the documentary corpus, I demonstrate that the existing interpretations of these documents are anachronistic, as they project the terminology and administrative structures of a substantially later period in Islamic history back to seventh-century Iran. Although we do have Middle Persian terms that are etymologically related to the Arabic words jizya and kharāj in Berk. 67's gazīdag and Berk. 27's harg (in the compound frašn-hargarīg), it should not be assumed that gazīdag and harg would have meant (respectively) “poll tax” and “land tax”, as their Arabic etymological relatives eventually would. Closer attention to the contexts in which these Middle Persian terms occur, and to what is otherwise known about the history of the Arabic terms jizya and kharāj, suggests that in these attestations gazīdag and harg simply meant “tax”.

Conversely, Berk. 34's drahm pad dēn and Berk. 154's bāǰ ī xwāstag, while etymologically distinct from jizya and kharāj, do capture some of these Arabic terms’ eventual meanings, while also revealing important differences between the fiscal system of early Islamic Iran and later regimes. Drahm pad dēn, “dirhams for the [Zoroastrian] religion”, does refer to a tax that was differentially assessed according to religion, but one that was apparently assessed on land according to its cultivation status and administrative region, and not as a poll tax. Meanwhile, Bāǰ ī xwāstag, literally a “money tax”, does seem to refer, at least in large part, to a land tax, but probably constitutes a heading under which such an imposition could fall, rather than a term meaning “land tax” in itself, as kharāj eventually would. Considered in aggregate, these four documents also suggest a certain course of development for the early Islamic fiscal system in the decades following the initial conquest of the Sasanian Empire: from broad and relatively unspecific impositions to more targeted exactions, based on increasingly detailed assessments.

Beyond the details of these conclusions, I demonstrate a method for interpreting these documents, dealing with their Islamic context in a more comprehensive manner than has been attempted hitherto.¹⁰ In particular, the papyri from early Islamic Egypt have not been considered sufficiently in connection with these Middle Persian documents, with which they are contemporary, nor have the medieval Islamic historiographical and legal traditions, and the relevant modern scholarship. Aside from the conclusions just surveyed, having to do with the fiscal system and the relevant terminology, a fuller consideration of this evidence leads to several additional conclusions and suggestions for further research, which I touch on as well, below and in the appendices: a more solid basis for the dating of the documents, based on a passage in the Tārīkh-e Qom whose significance has not been sufficiently appreciated (Appendix 1); the possibility that a Middle Persian term such as dar-handarzbed could refer to an official in the Islamic administration, as newly repurposed Greek terms do in the papyri from early Islamic Egypt; and the likelihood that the Middle Persian preposition ba- is a loan based on Arabic bi-, and, accordingly, an indication, heretofore overlooked, of substantial early contact between Qom's Persophone inhabitants and the Islamic administration (Appendix 2).

Berk. 67's “gazīdag”: what kind of tax?

I will begin by examining a document of just four lines – Berk. 67 – as well as Weber's corresponding interpretation and discussion. Editions and translations of Berk. 67 have been published three times: initially a partial edition and translation by Gignoux in 2010, and then fuller versions of each by Weber in 2013 and in 2019.¹¹ Although Weber's 2019 edition offers several intriguing possible new readings, on the whole this work does not supersede his 2013 edition; for an in-depth comparison of these two editions and their respective merits, see Appendix 3. Accordingly, the analysis that follows will be based on Weber's 2013 edition, which is reproduced here:

1. ZNE BYRH sp̅ndrmṭ QDM ŠNṬ XVI

2. MN gcyṭk' ml yzdʾnkrṭ XX-sl [?]

3. MN dlyk' OL lwdšn [?] BBA-hndlcpṭ'

4. YHYTYWNṭ' MN [?] [...]mṭ Y BBA-hndlcpṭ'

1. ēn māh Spandarmad abar sāl 16

2. az gazīdag mar Yazdāngird 20-s[atē]r [?]

3. az darīg ō rōyišn [?] dar-handarzbed

4. āwurd az [?] […]mad ī dar-handarzbed

[1] This month Spandarmad [12th month] of the year 16 [667/8 ce]. [2] From the account of the poll-tax Yazdāngird brought [line 4] 20 staters [?] [3] From the estate assistant to the prospering [?] tax collector. [4] From […] mad the tax collector.¹²

Much remains unclear here, to me and Weber alike; as Weber himself puts it at the beginning of the article in which his edition of 2013 appears, “it is clear from the difficulties of the script that, in the future, possibly new or better interpretations could be arrived at”.¹³ Hence, in lieu of a comprehensive interpretation of this document, I will offer a few comments on individual points, and trace their broader implications.

I will turn first to the terms that Weber has read as gazīdag and dar-handarzbed, which are of critical importance not only for the interpretation of Berk. 67 itself, but also for our understanding of the Sasanian and early Islamic fiscal systems. A deeper and more comprehensive engagement with the relevant Islamic sources than has been undertaken will be necessary to progress in our interpretation of these key terms.

Weber read gazīdag as “poll-tax” primarily based on its meaning in D.N. Mackenzie's dictionary of Middle Persian. It will be illuminating to venture beyond this dictionary entry, examining, in particular, the single other attestation of Middle Persian gazīdag that had been the basis for MacKenzie's definition, as well as the relevant evidence in other languages, particularly Arabic and Aramaic.¹⁴ Aside from yielding some modest progress in the interpretation of gazīdag and Berk. 67, attention to this evidence will clarify Berk. 67's importance for our understanding of the development of the early Islamic fiscal system.

In the single attestation of it that was known to MacKenzie, gazīdag does, unmistakably, refer to a poll tax: listed alongside the Arabs’ other resented impositions, in a poem of uncertain date lamenting the Islamic conquests, gazīdag is qualified as (a)bar sarān, “on the heads”.¹⁵ That such a qualification was required suggests, however, that gazīdag did not necessarily have such a narrow meaning; and attention to the early history of the Arabic word jizya, undoubtedly related to gazīdag in some way, only reinforces this impression, as we will see.

Weber did not take on the relationship of this Middle Persian word, gazīdag, to Arabic jizya.¹⁶ Aside from its importance for the interpretation of Berk. 67, this is an issue of direct relevance for the formation of the Islamic fiscal system, and even for the nature of its antecedents in pre-Islamic Iran. It could be an indicator of the extent to which the emergence of the Islamic jizya, in its “classical” sense of a poll tax specifically demanded of non-Muslims, was informed by a Sasanian precedent. Although the words and institutions are related, it will be helpful to consider each separately. As Kōsei Morimoto has noted in connection with early Islamic Egypt, “changes in terminology do not necessarily imply changes in the fiscal system”; nor, conversely, should terminological continuities necessarily imply institutional ones.¹⁷

As for the etymology of the Arabic word jizya, and its relationship to its relatives in Middle and New Persian, Aramaic and Bactrian, all of the available evidence indicates that this is a “specifically Arabic term”, in François de Blois’ words, and that it was only loaned into these other languages, ultimately from Arabic, in the wake of the Islamic conquests.¹⁸ The most plausible alternative account, originally expounded by Theodor Nöldeke, and still finding adherents to this day, holds that the word originally came from Aramaic into Arabic.¹⁹ But while a word gzytʾ (vocalized gzītā) does occur in Aramaic, it is, as de Blois points out, only attested in Islamic times, occurring nowhere among the “numerous pre-Islamic Aramaic texts and documents relating to taxation”.²⁰

Given, then, that Middle Persian gazīdag almost certainly stems from Arabic jizya, the early meaning of the latter should carry great weight in determining gazīdag's meaning in Berk. 67 – this document having been composed in 668 ce at the latest (on the dating of Berk. 67 and other Qom documents, see Appendix 1). As C.H. Becker has already shown, building on the work of Julius Wellhausen, and considering both historical and documentary sources, in the seventh century, Arabic jizya had a substantially different meaning from the “classical” sense it would eventually come to have: a poll tax, specifically collected from non-Muslims.²¹ In the seventh-century documentary evidence from Egypt, the term instead has the meaning “money tax”, encompassing taxes assessed in cash (but not in kind) on land, individuals and all other taxable entities; there is also no sign of any religiously based poll tax in this evidence, called jizya or otherwise.²² Only in the eighth century do these Egyptian papyri begin to refer to a “jizya of the head”, or otherwise, here and there, a jizya without such an explicit clarification, but distinctly in the sense of “poll tax”.²³ In its Quranic attestation, meanwhile, the term appears to mean nothing more specific than a “collective tribute” demanded from conquered peoples; the term's Syriac reflex gzytʾ, as attested in the Maronite Chronicle probably written in the 680s, likewise seems to have this meaning of a “tribute” submitted in the wake of a conquest.²⁴ Moreover, as Daniel Dennett notes, ninth-century authors such as al-Balādhurī and Ibn ʿAbd al-Ḥakam occasionally used jizya to refer to a tax on land: even at this point, following the emergence of the classical kharāj-jizya distinction in Islamic law, with the former referring to land tax and the latter to poll tax, jizya continued to have the potential to refer to a land tax in its own right.²⁵

Hence, it is not likely that Berk. 67's gazīdag should have referred specifically to a poll tax, collected from non-Muslims. It could mean a “money tax”, as opposed to a tax in kind, in line with jizya's meaning in the documents from seventh-century Egypt, or it could have had a broader meaning: simply “tax”.²⁶

Dar-handarzbed in Berk. 67 and elsewhere: tax collector, chamberlain or court adviser?

We now turn to another word that occurs in Berk. 67, dar-handarzbed. Weber initially translated this term as “tax collector” in 2013, before altering his translation to “chamberlain” in 2019.²⁷ In the later publication Weber continues to cite “tax collector” as a possible interpretation of the word, however, and is somewhat elliptical about the basis for his change of opinion; presumably it has something to do with his reinterpretation of gazīdag as something like “snake-bitten”, which, as discussed in Appendix 3, is probably not valid.²⁸ As we take a closer look at dar-handarzbed and the implications of its interpretation for Sasanian and early Islamic history, it will accordingly be necessary to consider both of the translations Weber has offered.

Dar-handarzbed's root meaning is “court adviser”, and while there is not much available evidence, every attestation of this term, whether in Middle Persian or Armenian historiography or late Sasanian seals, indicates that dar-handarzbed was a very high position in the Sasanian administrative hierarchy and suggests that we should take its root meaning literally. In Sasanian times, the dar-handarzbed almost certainly would have been an official who directly advised the ruler. It is probably safe to assume that this official would have had some more specific administrative functions, although these are essentially undefined in the sources and may well not have been fixed.²⁹ One could certainly do worse than Weber's “chamberlain” as an English equivalent for dar-handarzbed – its ambiguity and association with high administrative circles are helpful – but neither of the word's two specific meanings, whether the manager of a royal or aristocratic household or a treasurer, seems especially apt, on the face of it.³⁰ One would need to see more of a justification than Weber provides, at any rate, to accept his later translation without some reservations.

What of Weber's earlier translation, “tax collector”? It is tempting to restore this, to complement gazīdag's apparent fiscal meaning, but this will not do. For one thing, “tax collector” is a far humbler position than “court adviser”, dar-handarzbed's root sense; however certain we might be that gazīdag refers to some kind of tax, the mere fact that dar-handarzbed and gazīdag are both attested in Berk. 67 would not be enough to fix the former's meaning as “tax collector”. In both his 2013 and 2019 editions of Berk. 67, Weber adduced additional evidence for reading dar-handarzbed as tax collector, but this, too, ultimately does not amount to much.

If we set aside Berk. 67's association of dar-handarzbed with gazīdag, Weber's translation of the former term as “tax collector” is based on little more than an inaccurate citation, ultimately stemming from an etymological suggestion advanced by Shaul Shaked in 1991, in connection with the Aramaic word ʾdrgzr- as attested in the Book of Daniel. In that article, Shaked argued, pace W.B. Henning, that this term, while an Iranian loanword, probably has no etymological relationship to MP handarz, “advice”, and is rather to be connected with entirely distinct roots meaning, respectively, “debt” and “to injure, offend”; therefore, according to Shaked, this word has the sense of “one who oppresses or afflicts debtors” or more specifically “a collector of taxes”.³¹ In Shaked's account, then, this Aramaic word is unrelated to the Middle Persian term dar-handarzbed, and therefore, it is fair to say, all but irrelevant for determining the latter's meaning.

While Eduard Khurshudian's classic discussion of dar-handarzbed (and, more broadly, handarzbed, “adviser”), would touch on Shaked's discussion of ʾdrgzr-, in passing and representing it accurately, Gignoux's citation of Shaked via Khurshudian, although not inaccurate either, does not mention the fact that whereas Shaked did suggest “collector of taxes” as a translation for ʾdrgzr-, he also considered it to be unrelated to MP andarz, and therefore, by extension, unrelated to the term dar-handarzbed.³² Gignoux himself would translate MP dar-handarzbed as “court adviser” in the Qom documents and otherwise, but Guitty Azarpay, simply citing Shaked and Khurshudian, as “quoted by Gignoux”, opted to translate dar-handarzbed as “tax collector”. In this Weber has simply followed Azarpay, translating dar-handarzbed as “tax collector” and repeating her erroneous claim that Gignoux, following Khurshudian and ultimately Shaked, had also endorsed this interpretation.³³

As for dar-handarzbed's meaning in Berk. 67 and in the Qom documents more broadly, we must, at the very least, abandon the interpretation “tax collector” and return to square one. “Chamberlain” is somewhat more plausible, but not especially helpful in itself in advancing our understanding of what the dar-handarzbed did and their position in the administrative hierarchies of early Islamic Iran. Although I cannot offer an alternative interpretation of much substance here, given that five of the eight Qom documents known to mention a dar-handarzbed remain unpublished, I will briefly touch on some considerations that should inform further research in this area.³⁴

First, Gignoux's observation about the dates of the documents that mention dar-handarzbed is worth repeating: these dates tend to be very early. Of the four dated documents mentioning dar-handarzbed, one – Berk. 67 (as we have seen) – is dated to what is probably 667/8 ce, while two others – Berk. 101 and Berk. 217A – give the equivalent of 652/3 as their year.³⁵ Berk. 20, the other dated document with dar-handarzbed, does give a substantially later date, 691/2, although this may be an exception that proves the rule: as an “estate of the dar-handarzbeds” (dastgerd ī dar-handarzbedān) is referred to (here as also in Berk. 78), dar-handarzbed seems to be a family name in these later occurrences, rather than an administrative position.³⁶ As ever, a more thorough review of the evidence will allow for firmer conclusions, but we can at least form a hypothesis that dar-handarzbed only designated an administrative position for a short time following the Islamic conquests, and eventually transformed into a family name – somewhat in the manner of Middle Persian marzbān (“border-guard”), which seems to have been both a personal name and an official title.³⁷

This brings me to my second point, which is more general. Quite apart from their mischaracterizations of Shaked, Khurshudian and Gignoux's positions, it is significant that in grappling with an apparently unfamiliar meaning for Middle Persian dar-handarzbed in early Islamic documents, Azarpay and (in 2013) Weber resolved the dilemma by way of (what seemed to be) indirect evidence from the Achaemenid period.³⁸ They preferred an “Iranian” solution, however distant from early Islamic Qom the relevant evidence was. It might be better to approach the meaning of this term – which, indeed, even as an administrative position, does seem to differ somewhat from what it had been under Sasanian rule – within an Islamic frame of reference.

As we saw with gazīdag, a comparison with early Islamic Egypt, where the volume of documentary evidence is much greater, offers several useful insights that are directly relevant to the interpretation of dar-handarzbed in its occurrences in the documents, and any number of other, analogous cases. First and foremost, in comparing the pre-Islamic documents from Egypt in Greek and Coptic with their early Islamic analogues in the same languages “we are struck by the number of changes in Greek and Coptic administrative texts from Egypt starting immediately following the conquest”. Direct loans from Arabic are only the most obvious among these changes; entirely new Greek terms were also developed (to refer to the Arab caliph and governor, for example), and “existing Greek [and Latin] expressions were given new meanings [probably] to describe institutions or features of the new administration”, although it is also possible that these “reflect[ed] internal Greek (Byzantine) administrative developments”.³⁹

The implications for the interpretation of dar-handarzbed and, more broadly, the study of the early Islamic documents in Middle Persian, hardly need to be spelled out. Although there are differences between the respective Islamic conquests of Egypt and the Iranian plateau, and the administrative changes that ensued in each place, it should certainly be admitted as a distinct possibility, if not assumed outright, that if the Middle Persian documents from Sasanian-era Qom were fully brought to bear, analogous changes, with the same fundamental cause underlying them, could be discerned.⁴⁰ When, as in the case of dar-handarzbed's occurrences in the Qom documents, apparent disjunctions with the available Sasanian evidence do emerge, these should not simply be written off as otherwise-unknown Achaemenid survivals, any more than evidence from the Ptolemaic era should play a significant role in accounting for the differences between Egyptian documents in Greek and Coptic from just after the Islamic conquests and their immediate pre-Islamic predecessors. Rather, the Islamic conquests, and the incorporation of these regions into the new Islamic state, should be taken seriously as an impetus, if not the main impetus, behind these differences, which, accordingly, should really be viewed as historically conditioned changes.

To return to dar-handarzbed, there is a fairly direct analogy between its root meaning and those of Greek symboulos and protosymboulos, “adviser” and “foremost adviser”, which, though “unknown… in the papyri” prior to the Islamic conquests, come to mean, respectively, “governor” and “caliph” in the Greek documents from early Islamic Egypt. As ever, firm conclusions must await a comprehensive consideration of the evidence, but we should consider the possibility that something similar has happened with Middle Persian dar-handarzbed in early Islamic Iran.⁴¹

Berk. 34: “injurious [?] dirhams for the religion” in early Islamic Iran

Whatever the precise meaning of the word gazīdag in Berk. 67 – “tax” or something more specific – another Qom document, Berk. 34, though lacking a term such as gazīdag etymologically related to Arabic jizya, nonetheless very likely constitutes important evidence for the early background of differential taxation based upon religion. This document suggests that as late as the early 690s the poll tax and tax burden specifically falling on non-Muslims had not yet become entirely coextensive, as they would be in subsequent centuries (with only non-Muslims paying the poll tax, referred to as jizya, and all, Muslims and non-Muslims alike, paying the land tax, referred to as kharāj). Instead, non-Muslims seem to have paid the land tax at a higher rate. As we will see, this generally aligns with what we otherwise know of the early Islamic Empire, although the Middle Persian evidence may offer some uniquely early and direct corroboration for the notion that already in the late seventh century non-Muslims had recognized and even objected to religiously based impositions.

Berk. 34 is dated to the year 29, or (with the post-Yazdgerd era presumed) 680/1 ce. Weber has published two editions and translations of this document, one in 2013 and another in 2021, but, as with Berk. 67, in my view the later version has not entirely superseded the earlier one. Weber's two publications of Berk. 34 do not differ nearly as much as his publications of Berk. 67. Accordingly, rather than going to the trouble of reproducing both editions and translations of Berk. 34 in full, I will simply offer a single transcription and translation, largely based on Weber's edition and translation of 2021, but with several (noted) changes that are either my own or reversions to Weber's 2013 publication, as well as a few cosmetic alterations. As with Berk. 67, both of Weber's publications of Berk. 34 also include a transliteration, as well as an image of the manuscript itself.

1. ō xwadāyīg zād xrad-windād ōstāndār namāz Ōhrmazd-…

2. drōd was ○ ud čiyōn xwadāyīg ba-ābādānīh ī maydār

3. az xwāhišn ī man im sāl 29 pad maydārīh ī

4. Kōm rōstāg framūd gumārdan nar abāg xwadāyīg

5. paymān kerd kū hamē ka ba-ābādānīh drahm pad dēn

6. wizāyišnīg ēdōn čiyōn dād-ayār guftārīh [?] ō xwadāyīg

7. ud xwadāyīg ō man framūd kerdan [?] agar-iš wirāyišn ud pattān [?]

8. ī ān xīr rāy abāyēd burdan man pad xwēštan barēm

9. ud wizārēm ud ēn nāmag man pad gugāy-muhrīhā ī Xwadāgerd āwišt

10. ō xwadāyīg zād xrad-windād ōstāndār namāz Ōhrmazd-…⁴²

[1] To Sir ōstāndār, born with inherited wisdom, reverence! [From] Ōhrmazd-… [2] many greetings! And since Sir has designated⁴³ the wine-grower's [land] under cultivation⁴⁴ [3] according to my wish [?], this year 29 (680/1 ce), as a wine-growing [region] of [4] the Kōm district, men have made an agreement [line 5] with Sir,⁴⁵ that [5] as long as [it is] under cultivation, the injurious dirhams for the Religion [i.e. Zoroastrianism; dēn]⁴⁶ [6] will be according to the “law-helper's” [dād-ayār]'s statement to Sir [?]. [7] And Sir has ordered me to undertake [this]. If it is necessary to arrange it and bear the pattān [?] [8] on account of that matter, I myself will [9] bear and arrange it [i.e. the “injurious dirhams for the Religion”?]. And I have sealed this letter with Xwadāgerd's witnessing and sealing. [10] To Sir ōstāndār, born with inherited wisdom, reverence! Ōhrmazd-…⁴⁷

There is much to say about this document, and much here remains unclear. Thus for the moment I will concentrate on its implications for our understanding of the Sasanian and early Islamic fiscal systems, and in particular for the background of the jizya in its classical meaning – a poll tax uniquely exacted from non-Muslims.

Berk. 34 is a letter addressed to the ōstāndār from one of his underlings. As for ōstāndār, following Weber's usual practice I have left this frequently attested word untranslated. Although a comprehensive study of all its attestations in seals, documents and, especially, the literary sources, remains a desideratum, it apparently means “provincial administrator”, specifically the one in charge of the administrative unit that would have been known as an ōstān, and corresponds to a position that existed in Sasanian times as well. It is conceivable that this ōstān and ōstāndār referred more specifically to, respectively, “royal property” or a “crown province”, and an administrator in charge of the latter, as Michael Morony has claimed, largely based on the meaning of these terms in the Arabic sources. This is a point to which I will return below.⁴⁸

Although many details remain elusive, generally this document has to do with the reclassification of some vineyards as land under cultivation, in the “Kōm” (i.e. Qom) administrative region and the implications of this reclassification for taxation. Owing to the ōstāndār's (“Sir's”) classification of a certain parcel of land under cultivation as a wine-growing region in the district of Qom, a fiscal obligation has arisen, called “dirhams for the [Zoroastrian] religion”, which is possibly further qualified as “injurious”.

The key point here, as far as the classical jizya's background is concerned, is that by every indication the phrase “dirhams for the religion” – apparently (if the reading wizāyišnīg, “injurious”, is correct) a resented imposition, which fell specifically on Zoroastrians – refers to a land tax, and not a poll tax.⁴⁹ It seems that the specific mechanics of this tax were worked out between the ōstāndār (“Sir”, xwadāyīg) and a certain “law-helper” (dād-āyār), perhaps an official in the Islamic administration.⁵⁰ The author has apparently been ordered to undertake the collection of these taxes himself (“Sir has ordered me to undertake [this]”); presumably prompted by his superior, he also seems to offer to serve as a kind of guarantor for their delivery (“if it is necessary… I myself will bear and arrange [it]”). This personal guarantee may be reflected in Berk. 34's unusual sealing and witnessing procedure; while many documents end in a sealing formula, a third person, neither author nor addressee, is typically designated as the sealer.⁵¹ Here, by contrast, although a third person, Xwadāgerd, is mentioned as a participant in the procedure, the author seems to take the unusual step, otherwise unparalleled in the published documents, of sealing the letter himself: “I have sealed this letter with the witnessing and sealing of Xwadāgerd” (ēn nāmag man pad gugāy-muhrīhā ī Xwadāgerd āwišt).

As is the case for vineyards under cultivation being in a special fiscal category,⁵² and a provincial administrator's underling undertaking and serving as guarantor for tax collection,⁵³ the notion that a land tax should have been perceived as a religiously based imposition and, perhaps, an “injurious” one at that, likewise fits what we otherwise know of the Islamic Empire in the late seventh century. Although such securely early evidence as we apparently have in Berk. 34 has hitherto been lacking, the later writings of Muslim jurists strongly suggest that for the seventh century and beyond, land tax was largely assessed differentially according to religion, with Muslims paying a lower rate. Accordingly, paying the steeper rate was seen as a “humiliation”. It was only in the later years of the seventh century that “a new legal analysis was put forward”, which would eventually “[pave] the way for the imposition of kharāj on land belonging to [Muslims] and non-Muslims” alike, at a uniform rate.⁵⁴ In late seventh-century Iran, then, we can provisionally conclude that, in at least one important sense, the jizya, here as elsewhere in the early Islamic Empire, had not yet emerged in its classical form. At this point the poll tax was not the principal way in which the differential treatment of Muslims and non-Muslims was inscribed into the fiscal system. Moreover, Berk. 34 shows us that this differential treatment was perhaps not only perceived, but also resented, by the late seventh century. This resentment is certainly reflected in the literary sources, but, if Weber's 2013 reading of “injurious” is accepted, Berk. 34 would seem to constitute its earliest attestation in a document, produced by Iranian Zoroastrians or otherwise.⁵⁵

Berk. 27: harg and kharāj

We will now move on to Berk. 27, which apparently dates from 693/4 ce and has the only known attestations of the important compound frašn-hargarīg, which includes the likely Middle Persian etymological forebear of Arabic kharāj. Here, as with Berk. 34, I will largely reproduce Weber's transcription and translation, noting all significant divergences; those interested can consult Weber's original text for an image of the document and his transliteration.

1. ēn māh Hordad ī

2. sāl 42 ud rōz Ādur

3. padīrēd frašn-hargarīg

4. ī Paywēšagestān

5. abāg farroxtar ōstāndār

6. pad rāh burdan

7. rāy az Dēn-abzānēd

8. kas az Xwadāgerd

9. padīrēd kē xwāstag

10. ī az wizārišn ī kār

11. andar kerd az dar ī

12. Dēn-abzūd pēlag [?]

13. 2 ud padīrāy

14. frašn-hargarīg pad

15. gūgāy-muhrān ī ān ēwēnag⁵⁶ ī

16. miyānǰīgān āwišt.

[1] This month Hordad [3^rd month], of the [2] year 42 [693/4 ce] and the day Ādur [3] the frašn-hargarīg ⁵⁷ of Paywēšagestān⁵⁸ receives, [from 7] for [6] taking a journey [5] with the most fortunate ōstāndār, [7] from Dēn-abzānēd [8] (anyone receives from Xwadāgerd, [9] who has obtained money [10] [for] the remuneration of work), [11] from the dar of [12] Dēn-abzūd 2 pēlag. [13] And [14] the frašn-hargarīg [from 16] has sealed [from 13] the receipt, with [15] witnesses’ seals in the manner of the [16] miyānǰīgs.^{59, 60}

As with the other documents we have examined, many aspects of the reading and interpretation here remain uncertain, but a review of Weber's treatment of it, and what seem to be its likely meaning and function, will be instructive. Generally speaking, this document seems to record the payment of a certain official, the frašn-hargarīg, for doing his job, although it is unclear who exactly has paid him and in what. Weber suggests, plausibly, that there are three layers to the payment: Dēn-abzānēd is recorded as the one who paid the frašn-hargarīg, but a certain Xwadāgerd, being “in charge of distributing payments”, is the one who gave the frašn-hargarīg the money, which actually ultimately stems from the “fund” (a “chapter” of which is referred to by dar) of yet another person, Dēn-abzūd.⁶¹ Weber further suggests that the payment may be in land, the otherwise-unattested pēlag being potentially related to New Persian pēlah, which refers to a kind of countryside.⁶² This too is plausible, although in this case it would be somewhat curious that land should count as some kind of “money” (as xwāstag in the heterographic spelling <NKSYA> typically means).⁶³

The key question here, as far as the implications for Sasanian and Islamic fiscal history are concerned, have to do with the function of the frašn-hargarīg. As Weber has pointed out, frašn-hargarīg is a compound, composed of the elements frašn, “question” or “enquiry”; harg, a kind of tax; gar, whose initial /g/ has been lost owing to the /g/ that precedes it, and which, in compounds like this, means “doer”; and a suffix -īg, typically an adjectival suffix, but here clearly occurring in a noun.⁶⁴ Hence, the meaning of the whole is something like “the one who makes enquiries in connection with the harg-tax”.

What, then, is the nature of this harg-tax? Weber assumes that it would have meant much the same thing as its etymological relative kharāj would come to mean, in most cases, in medieval Arabic and, eventually, in New Persian as well: a land tax. Hence he has translated frašn-hargarīg as “the measurer of taxable land”.⁶⁵ However, just as jizya did not have its usual medieval Arabic sense in the seventh century, nor, by all indications, did kharāj (or its early synonym kharj) or, by extension, its relatives (if not direct antecedents), Middle Persian harg and xarg.⁶⁶ None of Middle Persian harg/xarg's attestations, whether in Manichaean or Zoroastrian texts, refer distinctly to a land tax. Rather, the word has a sense of a generic “tribute” or “obligation” whose precise meaning varied according to context. It was not necessarily based on land, or even reckoned in cash or kind; for at least one of its attestations, it has been suggested that the term could refer to corvée labour.⁶⁷ Similarly, this sense is absent in those etymological relatives of harg that passed into the non-Middle Persian languages of Sasanian subject populations: krgʾ in the Babylonian Talmud, clearly related to Middle Persian harg, distinctly refers to a “poll tax” assessed by the Sasanian authorities;⁶⁸ and classical Armenian hark is a “general term for tribute and taxes” and does not refer specifically to a poll tax.⁶⁹ Nor is there any hint that kharāj or kharj referred specifically to a land tax in their respective Quranic attestations; there these terms seem to refer to a generic “tribute” or “reward”.⁷⁰

It is not clear when Arabic kharāj came to refer largely or exclusively to “land tax”. This shift seems to have happened substantially later than is typically assumed. As Gladys Frantz-Murphy shows, in the Egyptian papyri the term replaces jizya in the sense of “tax assessed in money” in the wake of the Abbasid takeover and, as noted by Marie Legendre, continues to have this meaning “until the end of the Abbasid period”. This broader sense, Legendre relates, is also discernible in Abū Yūsuf's Kitāb al-kharāj (which discusses “the whole fiscal system”, not just land taxes) as well as, possibly, the Arabic documents from early Abbasid Afghanistan in the Khalili collection, whose attestations of kharāj have yet to be conclusively demonstrated to refer distinctly to a land tax.⁷¹

In the ninth- and tenth-century historical accounts in Arabic that stem from the Middle Persian Xwadāy-nāmag traditions (probably in some way via Ibn al-Muqaffaʿ's influential translation of the mid-eighth century) the Sasanian fiscal system has largely been described in the Arabic terms current at that later time (i.e. jizya meaning “poll tax”; kharāj as “land tax”), but even here we get glimpses of an earlier terminology. Al-Dīnawarī, writing in the later ninth century, makes it clear that Persian speakers, not only in Sasanian times, but in his day as well, referred to the land tax not by the word kharāj, as al-Dīnawarī himself does, but rather by a term meaning “calculation” or “number”, which, although vocalized slightly differently, must be the Persian word shumāre.⁷² Al-Ṭabarī, meanwhile, although he usually uses jizya to refer to a poll tax, sometimes uses kharāj in this capacity as well, telling of “the kharāj of Kisrā [presumably Khusrō I], which was on the heads of men”.⁷³

The available evidence, then, indicates that harg had a broader meaning than assumed by Weber: harg is simply a “tax”, perhaps with further nuances varying according to context. There is accordingly no reason to confine the responsibilities of the frašn-hargarīg, “the one who makes inquiries regarding the harg”, to land-surveying, as Weber does. Census-taking, along with gathering information about other taxable entities, seems at least as likely to have been part of this official's responsibilities.

Wadād al-Qāḍī has convincingly argued that the late 680s and early 690s ce marked a turning point in Islamic fiscal and administrative history. In 691–2, as Abū Yūsuf's (d. 798 ce) Kitāb al-Kharāj and the anonymous Syriac Chronicle of Zuqnīn (wr. c. 775 ce) agree, the caliph ʿAbd al-Malik undertook a comprehensive census and land survey in Mesopotamia, on a scale unprecedented in Islamic history.⁷⁴ This was not only “compulsory, comprehensive, organized, and meticulously recorded”, but also “supra-provincial”, as its scope, by Abū Yūsuf's account, included both Mosul and Syria.⁷⁵ Al-Qāḍī brings further evidence from Syrian and Egyptian documents and historical narrative sources to bear, further bolstering her case that the early years of ʿAbd al-Malik's (r. 685–705 ce) reign saw a pronounced intensification of census-taking and land surveying, which extended across provincial boundaries.⁷⁶ The scope of al-Qāḍī's study only extends as far east as Iraq. Further clarity on this point will only come with a far more comprehensive treatment of the Middle Persian documents than is possible at present, but Berk. 27, dating from 693–4 ce and, possibly, pointing to some kind of centrally directed census and/or land survey in early Islamic Iran, may indicate that the effects of this intensification under ʿAbd al-Malik were felt as far east as the Qom region.

Berk. 154: evidence for the persistence of Khusrō I's “misāḥa” system?

I will now briefly examine one last Middle Persian document edited by Weber, which likewise has something to say about taxation. Berk. 154, dated to the year 31 or (assuming the post-Yazdgerd era) 682/3 ce, is a kind of cheque, entitling its bearer to a certain collection of agricultural products, with a certain cash value, which has been collected to pay a certain tax.⁷⁷ The document apparently has to do with the collection and payment, “at the ōstāndār's demand”, of the bāǰ ī xwāstag, a phrase that probably means “money tax”, although it is difficult to ascertain how precisely the transaction ordered in Berk. 154 fits into the broader fiscal administrative process. In its few other attestations in Middle Persian, bāǰ refers to a “tax” or “tribute”, and this is the meaning of its New Persian reflex bāj as well; as discussed above, xwāstag, when spelled <NKSYA> as here, means “money” and not “property”.⁷⁸ As we will see, there is also no basis, whether in the word's other attestations, or the context in which it occurs, to further specify its meaning as “sales tax”, as Weber does, apparently to set it apart from what he assumes (or had assumed) to be the meanings of harg and gazīdag, “land tax” and (in 2013) “poll tax”, respectively.⁷⁹

As with the documents discussed above, what I give here is based on Weber's text, but with some modifications. These are mostly cosmetic, but I have made two more significant adjustments to the translations offered by Weber in 2013 and 2022: in my rendering of bāǰ ī xwāstag, just discussed, as well as in my leaving pahrist untranslated.

1. ēn māh Spandarmad ī sāl 31

2. ud rōz Day pad Ādur ^wizārd^⁸⁰ wīr-ud-gīr ī

3. pad tis ī yazdānbāng zād xrad-windād ōstāndār

4. pad kulān [?] az framān ī ōstāndār

5. pad bāǰ ī xwāstag az wahāg ī

6. gandum ī pad pahrist ī g[rīw] 1 pad drahm 1

7. s[tēr] 38 drahm 3 ud az wahāg ī

8. wēnōg ī pad pahrist ī g[rīw] 1

9. pad drahm 1 s[tēr] 5 az wahāg ī

10. aspast ī pad pahrist ī 30 [gerd] [?]

11. pad drahm 1 s[tēr] 1 drahm 1 hāmist s[tēr] 45

12. ud čak ī pad čehel ud panj stēr⁸¹ ba-aband [?]⁸²

13. āwišt

[1] This month Spandarmad [12^th month] of the year 31 [682/3 ce] [2] and the day Day pad Ādur [8^th day]: to be paid in good faith [3] regarding the affairs of the ōstāndār [?], protected by the gods and having innate wisdom, [4] in detail [?] at the ōstāndār's demand, [5] for the money tax, from the value of [6] wheat in the pahrist, 1 grīw [worth] 1 dirham, [7] 38 stēr 3 dirhams. And from the value of [8] lentils in the pahrist 1 grīw [worth] 1 [9] dirham, 5 stēr. From the value of [10] alfalfa in the pahrist, 30 [bundles?] [11] [worth] 1 dirham, 1 stēr 1 dirham, altogether 45 stēr. [12] And the čak which is for forty and five stēr is [13] sealed [from 12] without guarantee [?].⁸³

Although apparently nominally assessed in cash, this tax seems to have been actually collected in kind: 155 grīw of wheat are accepted as the equivalent of 38 stēr (i.e. tetradrachms) and 3 dirhams; 32 grīw of lentils for 5 stēr; and 150 of a certain other unit of alfalfa for 1 stēr and 1 dirham.⁸⁴ Altogether, the crops to be submitted are worth 45 stēr or 180 dirhams. Weber has opted to interpret pahrist as the “store” where these lentils, wheat and alfalfa are held, but the word, in line with the meaning of its Arabic derivative fihris(t), could also conceivably refer to a text – perhaps the sort of “register” or “catalogue” giving the official cash values of various crops we see in the closely parallel Berk. 46 and Berk. 97, which both record price lists under the heading pahrist and include an opening formula that may refer to public announcements.⁸⁵

It is significant that various amounts of three different crops have been cobbled together to arrive at this round figure of 45 stēr, to pay what is literally called the “money tax”. For one thing, this suggests a distinction between this “money tax” and a hypothetical “tax in kind” or “wheat tax”, along the lines of the fundamental distinction between jizya (money tax) and ḍarība (wheat tax) which can be discerned in the contemporary Arabic administrative papyri from Egypt.⁸⁶ One of course would like more data on which to base these kinds of suppositions, but it is at least suggestive that, some 15 years after Berk. 67's gazīdag of 667/8 ce, an apparently undifferentiated “tax”, here, in Berk. 154 dated to 682/3, we may have evidence for a somewhat more complex and differentiated fiscal terminology.

Berk. 154 does not tell us what kind of assessment its figure of 45 stēr is based on, but it is most likely that, largely or all in all, it deals with the payment of a tax on land, here treated under the broader heading of “money tax”. While not impossible, it is somewhat difficult to imagine that Berk. 154's 45 stēr worth of produce should have gone mostly or entirely to pay a poll tax, for instance. At any rate, there is nothing at all here to indicate that bāǰ ī xwāstag simply meant “sales tax”, as Weber proposes.⁸⁷ Berk. 154 accordingly suggests that in Iran in the early 680s ce, the land tax was demanded not as a proportion of crop yields, but rather as a fixed monetary amount.

Whether due to continuity or mere coincidence, in this respect the fiscal system of early Islamic Iran would have functioned similarly to the “misāḥa” system supposedly introduced under the Sasanian rulers Kawād (r. 488–96, 498–531) and Khusrō I (r. 531–79), and not the proportional system (Ar. muqāsama) their reforms supposedly replaced. In this new system, the land tax would have been demanded as a fixed monetary amount, as Berk. 154's bāj ī xwāstag is arrived at on the basis of a detailed and wide-ranging cadastral survey (Ar. misāḥa), rather than as a proportion of total yields.⁸⁸

As touched on above, Morony concluded that the ōstāndār, who orders the collection of Berk. 154's bāǰ ī xwāstag, and is otherwise ubiquitous in the early Islamic Middle Persian documents from the vicinity of Qom, was an official specifically in charge of Sasanian “crown land”.⁸⁹ While this still may be true, and requires further investigation, Berk. 154 raises problems for his further claim that, even after Kawād and Khusrō I's reforms, the muqāsama tax regime remained in place in these crown lands, called ōstān and governed by ōstāndārs, into the Islamic period.⁹⁰ In at least one case, we have, in an administrative region governed by an ōstāndār, and presumably itself called an ōstān, what is probably, at least in large part, a land tax, but one that is distinctly not assessed as a proportion of yields. Either ōstān and ōstāndār do not invariably refer to crown lands and the official in charge there, or a muqāsama tax-assessment regime did not invariably prevail in such places.⁹¹

Conclusion

One of my primary aims here has been to illustrate the value of a complementary approach to these Middle Persian documents, bringing in not only Iranian-language evidence, and the insights of Iranian Studies, but also a thorough examination of their Islamic context and the relevant sources and scholarship. Beyond the more specific conclusions advanced here, I have thus demonstrated a method that, I hope, will usefully inform the further study of these important Middle Persian documents from early Islamic Iran.

It can already be said that these documents add substantially to our understanding of early Islamic Iranian history, however. The most basic point is, again, simply that the terminology and structures of the later Islamic fiscal system cannot be projected back into seventh-century Iran. Where we have a probable reference to a land tax, it is not called kharāj or even harg, but rather falls under what is most likely the broader heading of bāǰ ī xwāstag, “money tax”. Where a fiscal imposition specific to Zoroastrians is discussed, meanwhile, this is not called jizya, or gazīdag, nor is this even a poll tax. Instead, it seems that non-Muslims paid special land taxes, rather than having a distinct poll tax imposed upon them. And while Middle Persian terms etymologically related to both jizya and kharāj do surface in the Qom documents, in gazīdag and (in a compound) harg, there is nothing to indicate that their meanings corresponded to what would become these Arabic terms’ ordinary senses at this early stage. A closer examination of their context, as well as what is otherwise known about the early history of jizya and kharāj, indicates rather that gazīdag and harg probably had a broader meaning, each simply meaning “tax”.

We may also be able to find some significance in these documents’ relative and absolute chronology; the Islamic authorities’ demands and knowledge seem to grow more detailed and differentiated over time. While Berk. 67's (667/8 ce) gazīdag apparently refers to a “tax” which was more or less undifferentiated, the fact that the imposition discussed in Berk. 154 (682/3) is specified as the bāǰ ī xwāstag, “money tax”, may well imply a distinction from another kind of tax – perhaps one, unlike this bāǰ ī xwāstag, officially to be paid in kind.

The two documents from the early 690s, meanwhile, each attest to a somewhat higher level of administrative knowledge and concern, which may be connected to an increasing Arab presence in the Qom region, as well as the administrative reforms undertaken by ʿAbd al-Malik, touched on above.⁹² Berk. 34 (690/1) seems to indicate that taxes were now paid on land, according to its cultivation status and the administrative region in which it was located, as well as its proprietor's religious status – all information of which the administration obviously took careful note. And although, as discussed above, the harg that occurs in Berk. 27's (693/4) frašn-hargarīg (“maker of inquiries regarding the harg”) most likely refers to a generic “tax”, it is highly significant, in its own right, that at this point the state seems to have seen fit to send someone to make enquiries regarding taxation – whether such enquiries constituted land-surveying, census-taking or, what is most likely, both, along with gathering other information.

More comprehensive and detailed analyses of these documents promise to reveal much more. First and foremost, we can use them to assess how integrated into the early Islamic state the Qom region and the western Iranian plateau were and, by extension, how centralized and powerful the early Islamic state was. Matthieu Tiller and Annelise Nef have influentially argued that the Umayyad state was essentially “polycentric”, marked by administrative practices that differed substantially by region: does the documentary evidence from early Islamic Iran bear this out?⁹³ A full treatment of the preposition ba- and its relationship with Arabic bi-, for instance (as briefly sketched in Appendix 2), considering the chronology and contexts of its attestations, is not only helpful for filling out the history of the Middle Persian language: as a proxy for interaction between the Qom area's Persophone inhabitants and speakers of Arabic, such an analysis will also be an important indication of the extent to which this region was integrated into the early Islamic state. Studies of the documentary occurrences of administrative titles such as dar-handarzbed, dād-ayār and miyānǰīg, which establish their functions, and whether these refer to figures in the Islamic administration with close contemporary parallels in other regions, or holdovers from the Sasanian era, will provide important guidance on these questions of centralization and authority as well.