What the Mongols did to Iran, and what Iran did back (1219-1335)

Before: who was actually running Iran in 1219

In the spring of 1219 the lands between the Syr Darya and the Persian Gulf answered to ʿAlāʾ al-Dīn Muḥammad, Khwārazmshāh, who had spent nineteen years assembling them and would lose them all in three. His army was Turkic — Qangli and Qipchaq cavalry loyal less to him than to his mother, Terken Khatun, who kept her own chancery and her own seal. His treasury, his land registers, his tax assessments, his diplomatic correspondence and his court poetry were Persian, and had been produced by the same kind of men, often literally the same families, for two hundred years ¹. The largest cities of his empire — Nishapur, Merv, Herat, Bukhara, Samarkand, Rayy — were among the largest anywhere west of China. Nishapur had a water-supply system of underground channels, a dozen congregational mosques and a book market. Merv had ten libraries, one of which the geographer Yāqūt used for three years and wrote about as the happiest period of his life, leaving in 1220 just ahead of the Mongol army ².

This was not a golden age and should not be narrated as one. It was a conquest society with an unusually stable internal arrangement. Since the Ghaznavids in the late tenth century and the Seljuqs after 1040, the Iranian plateau had been ruled by Turkic military elites who did not read Persian well, did not know how to assess a harvest, and did not intend to learn. They hired people who did. The arrangement had a name in the political literature — Niẓām al-Mulk's Siyāsatnāma, written for the Seljuq sultan Malikshāh around 1091, is an entire book explaining to a Turkish soldier why he needs a Persian secretary and how to avoid being cheated by one ³. By 1219 the arrangement was three centuries old and nobody involved in it thought it strange.

The dīvān was a technology, and it was portable

The Persian administrative apparatus was not an attitude or a tradition. It was a specific set of offices with specific procedures, and it could be handed over intact.

• The dīvān-i aʿlā, the central bureau, under a ṣāḥib-dīvān who was in practice the state's chief financial officer.
• The mustawfī's accounting office, which kept the registers — the daftar — in a specialised Persian accounting script and a specialised vocabulary, with rules for auditing that were taught as a trade.
• The munshī's chancery, which produced the state's outgoing documents in a formal Persian that had displaced Arabic for this purpose across Iran during the eleventh century.
• The kharāj assessment, a land tax calculated on cultivated area, crop type and water access, requiring a survey and a written record of every irrigated field in a district.
• The qanat: a gently sloping tunnel driven into an alluvial fan to bring groundwater to the surface by gravity, maintained by a guild of specialist diggers. Without qanats there was no agriculture worth assessing on most of the plateau, and the assessment system was built around them ⁴.

The people who operated all this were drawn from hereditary secretarial families concentrated in specific districts. One of those districts was Juvayn, in western Khurasan, whose leading family had supplied financial officers to one regime after another. In 1219 the head of that family, Bahāʾ al-Dīn Juvaynī, was serving the Khwārazmshāh as a senior accountant. He would later serve the Mongols in the same capacity. His two sons would run the Mongol state in Iran. None of this was treason as anyone then understood it; it was what the family did ⁵.

What the Mongols had instead

Chinggis Khan's empire in 1219 was thirteen years old as a unified entity and had no civil administration at all in the sense the word had in Iran.

What it did have was formidable. A decimally organised army in units of ten, hundred, thousand and ten thousand, cutting across tribal lines by design. A household guard, the keshig, that doubled as an officer school and an administrative staff. A body of rulings and precedents, the jasagh or yasa, that regulated the army, the succession, the hunt and the royal house. A script — Uyghur, adopted in 1204 from a captured Naiman scribe — and with it the beginnings of a written chancery. And a settled, unembarrassed policy of taking captured specialists into service: craftsmen, engineers, interpreters, scribes ⁶¹⁸.

What it did not have was any procedure for taxing settled agriculture. Steppe political economy ran on tribute, plunder and the herd; the yasa has nothing to say about how to assess a wheat field. The gap was real enough that in the 1230s, in north China, the Khitan administrator Yelü Chuzai had to argue to Ögödei against a serious proposal to depopulate the North China Plain and turn it into pasture — his counter-argument being arithmetic, that taxed peasants would yield more silver annually than the land would ever yield as grazing ⁵. He won. The same argument, in the same decades, was being won in Iran, and by the same method: someone showed a Mongol prince a ledger.

The transmission: from Otrar to Tabriz

1219–1223: this was a campaign, not a government

The immediate cause is not in dispute. In 1218 a Mongol-sponsored trade caravan of some four hundred and fifty merchants was seized at Otrar on the Khwarazmian frontier and its members killed by the governor, Inalchuq; the envoys Chinggis Khan sent to demand redress were also killed, one executed and the others sent back shaved ⁷. The response began in 1219 and was not a raid. Bukhara fell in February 1220, Samarkand in March. Juvaynī preserves what a survivor of Bukhara is said to have reported afterwards, in six verbs: "They came, they sapped, they burnt, they slew, they plundered and they departed" ²¹. In 1221 Chinggis's youngest son Tolui took the Khurasani cities in sequence: Merv in February, Nishapur on 10 April, Herat in June ⁸.

What happened at those cities is the reason the rest of this record needs to be written carefully. It is also the least reliably documented thing in it. Our principal source for the Khurasan campaign is ʿAṭā-Malik Juvaynī's Tārīkh-i Jahāngushā, the History of the World-Conqueror, written between 1252 and 1260 — by a Persian who had lost his country to these armies, while employed by the men who commanded them. Juvaynī reports that a sayyid spent thirteen days counting the corpses at Merv and reached 1,300,000, and gives 1,747,000 for Nishapur ²¹. No modern historian accepts these figures. John Masson Smith's 1975 analysis of Mongol manpower and Persian population concluded that armies of the size required to kill and dispose of that many people in the time available did not exist, and that the cities in question could not have held such populations in the first place ¹². Current estimates for Nishapur run in the region of one to two hundred thousand, and even those are inferences.

Hold all three of these in view at once, because the record depends on it: the medieval millions are rhetoric, written by a man with motives; the modern ranges are estimates, built from carrying capacity and logistics; and the institutional facts — a city depopulated, a dynasty ended, a currency withdrawn — are certain. Conflating the three registers is how this period gets narrated badly in both directions.

For thirty years after Chinggis Khan's death in 1227, Iran was not governed so much as farmed. A succession of appointed overseers — Chin-Temür, then the Uyghur Körgüz, who was executed in 1242, then Arghun Aqa, who held the post from 1243 to 1255 — collected revenue for Karakorum through tax-farming contracts and through the ortogh, the state-merchant partnerships that advanced cash to the treasury and recovered it, with interest, from whoever could be squeezed ³. Arghun Aqa's registers show a bureaucracy being reassembled out of surviving Persian personnel, without any policy direction about what it was for.

1256: Hülegü arrives with a brief to stay

The quriltai of 1251 that elected Möngke as Great Khan also decided that the Islamic lands would become a Chinggisid appanage under Möngke's brother Hülegü, with a standing army and an explicit remit to govern rather than to raid ⁶. This is the hinge. Everything before it is conquest; everything after it is administration.

Hülegü crossed the Oxus in January 1256. The Nizārī Ismāʿīlī fortress of Alamūt surrendered in December of that year, and two men came out of it who would matter enormously. One was Naṣīr al-Dīn Ṭūsī, the polymath who had spent decades at Ismāʿīlī courts and who now entered Hülegü's service, where he stayed until his death in 1274. The other was a library. Juvaynī, who was travelling with the army, was given the run of Alamūt's collection and describes in his own history what he did with it: he selected the Qurʾāns and the scientific and historical manuscripts for preservation and had the Ismāʿīlī doctrinal works burned ²¹. It is an unusually honest sentence for a court historian, and it tells you exactly what kind of transmission this was — not preservation, not destruction, but sorting, by a man with a sorting authority he had just been handed by an invading army.

Baghdad, February 1258

Hülegü reached Baghdad in January 1258. The siege proper began around 29 January; the city surrendered on 10 February; the sack ran through the following days; the ʿAbbāsid caliph al-Mustaʿṣim was executed on 20 February, rolled in a carpet and trampled by horses so that no royal blood would touch the ground ⁷¹³.

The casualty estimates run from Hülegü's own claimed two hundred thousand, in his letter to Louis IX, to eight hundred thousand and beyond in the Arabic sources; the lowest modern figures are around ninety thousand. As with Nishapur, the honest answer is that the number is unrecoverable and that its unrecoverability is part of what happened.

What is certain is institutional, and it is the largest single fact in this record. The ʿAbbāsid Caliphate had existed since 750. On 20 February 1258 it ended. Every Sunni polity from Morocco to Bengal had derived formal legitimacy from a caliph in Baghdad; after that date there was no caliph in Baghdad, and the successor states had to invent their legitimacy from other materials. The Mamluks in Cairo installed a refugee ʿAbbāsid as a figurehead within three years, which tells you how badly the position was missed ¹⁷.

The part of the story that does not survive scrutiny is the part everyone repeats. The claim that the Mongols destroyed Baghdad's libraries, threw the books in the Tigris until it ran black with ink, and thereby ended Islamic science, is not supported by the evidence. Michal Biran's reconstruction of the libraries of Ilkhanid Baghdad from biographical dictionaries and manuscript colophons finds a working scholarly economy in the city within a generation of 1258, with the Mustanṣiriyya madrasa functioning, teaching licences being issued, and books being copied, bought, endowed and moved ¹¹. Some libraries were destroyed. The Baghdad book trade was not. Ṭūsī, meanwhile, was carrying instruments and manuscripts north to a new observatory that the conquest was about to pay for.

Manuscript of Rashīd al-Dīn, Jāmiʿ al-tawārīkh, fourteenth century. Bibliothèque nationale de France, Supplément persan 1113, f. 239r. Public domain via Wikimedia Commons. · Public domain

The hiring of the Juvaynīs

In September 1260 a Mamluk army defeated a reduced Mongol force at ʿAyn Jālūt in Galilee. Hülegü had already withdrawn eastward on the news of Möngke's death the previous year, and the westward expansion stopped permanently at the Euphrates ¹⁷. At that point Hülegü's appanage became something new in Mongol experience: a Chinggisid state with a fixed frontier, no further conquests available, and a tax base it now had to keep alive rather than strip.

The men hired to do it were the family that had been doing it before. ʿAṭā-Malik Juvaynī — the historian, who had ridden with the army to Baghdad and watched it sacked — was appointed governor of Baghdad, lower Iraq and Khuzistan in 1259, and held the post until his death on 5 March 1283, a tenure of twenty-four years ²¹. His younger brother Shams al-Dīn Juvaynī became ṣāḥib-dīvān of the whole Ilkhanate in 1263 and ran its finances for twenty-one years. Persian became the language of the Ilkhanid chancery. Persian accounting practice became Ilkhanid accounting practice: the manual al-Murshid fī al-ḥisāb, written in Persian by Ḥasan ibn ʿAlī al-Qāshānī and analysed by Ryoko Watabe, shows the early Ilkhanid state assessing the qubchūr poll tax through a census procedure that is Persian bureaucratic technique applied to a Mongol levy ²³.

Nothing about this required the Mongols to admire Persian civilisation, and mostly they did not. It required only that they wanted revenue and that the revenue-collecting machine was standing there, undamaged in its essentials, staffed and willing.

The traffic ran the other way too

The second half of the transmission is the part the atrocity narrative has no room for. Because the Ilkhanate and Qubilai's Yuan in China were held by branches of the same family, personnel and texts moved between Tabriz and Dadu on a scale that nothing in Eurasian history had previously permitted. Thomas Allsen's study of this traffic describes the empire as a cultural "clearing house" and argues that the exchange was not a by-product of conquest but a deliberate policy: the courts consciously shared the specialist resources of their realms with each other ⁵.

The clearest single agent of it was Bolad Chingsang, a Mongol who had run Qubilai's agricultural and censorial bureaux in China, arrived in Iran on an embassy in 1285, and simply stayed — for twenty-eight years, until his death in 1313. He was Rashīd al-Dīn's principal informant on Mongol history, Chinese administration, Chinese agronomy and Chinese medicine, and much of what the Jāmiʿ al-tawārīkh says about the eastern half of Eurasia came out of his mouth ⁵⁷.

The last two rows matter as much as the first five. This was not a one-way flow from a sophisticated east to a wrecked west. It was a circuit, and the Ilkhanate was one of its two poles.

What changed, and what it replaced

Iran becomes a word again

The single most durable consequence of Mongol rule in Iran is a name.

After the Sasanian collapse in 651 there had been no political entity called Iran. The plateau was a set of provinces of the caliphate — Jibāl, Fārs, Khurāsān, Ādharbāyjān — and then a set of successor sultanates. Īrān survived as a literary and legendary geography, the world of the Shāhnāma, with no administrative referent at all. Under the Ilkhans it comes back as a political term. Dorothea Krawulsky's study of Ilkhanid territorial vocabulary devotes its opening chapter to precisely this — Zur Wiederbelebung des Begriffes "Īrān" zur Īlḫānzeit, on the revival of the concept of Iran in the Ilkhanid period — and traces Īrānzamīn and the royal style pādishāh-i Īrān in Ilkhanid chancery and historiographical usage ⁸. Bert Fragner's world-historical reading of the same material puts the point bluntly: the Ilkhanate is where Iran becomes governable as a unit again, and the idea survives its governors ¹⁰.

It survives them by a long way. The Jalayirids, the Timurids, the Aq Qoyunlu and finally the Safavids all inherit a political geography in which Iran is a thing that can be ruled as a whole. So, eventually, does the modern state.

The irony is not subtle and the record should say it out loud. The conquerors who are remembered for destroying Iran are the ones who put its name back on the map.

Ghazan's reforms, 1298–1304

Ghazan converted to Islam in June 1295 and took the throne in November of the same year — the first Ilkhan to rule as a Muslim, at the head of a Mongol elite that converted with him or shortly after ¹³. What followed was the most serious attempt at fiscal reconstruction any Mongol successor state made.

The programme, as recorded by the man who largely designed and executed it, ran roughly:

1. Fixed, published assessment. Tax liabilities for each district set at a defined figure and inscribed on stone or metal tablets posted at the district mosque, so that collectors could not invent demands.
2. Abolition of repeat levies. The qubchūr had been collected two and three times in a single year by competing agents; Ghazan's decrees set a single annual collection with a stated calendar.
3. Reform of the yam. The relay post had been used by officials as a private transport subsidy, with horses requisitioned from villages along the route; the paiza tablets authorising its use were recalled and reissued on a restricted basis.
4. Currency standardisation. A dinar of fixed weight and fineness, struck to a uniform standard across the realm, replacing the local variations that had made cross-provincial accounting nearly impossible.
5. Land grants in place of cash pay. Army units were assigned iqṭāʿ holdings, registered in the daftar, both to relieve the treasury and to give the Mongol soldiery a stake in the land producing rather than in the land being plundered.
6. Reclamation of abandoned land. Decrees offering tax remission for a term of years to anyone who restored cultivation on land that had gone out of use — an admission, in the reforms' own language, of how much had ³.

Ann Lambton's institutional study is the standard treatment, and it carries a warning worth repeating: we possess the decrees, not the compliance. The source for Ghazan's reforms is overwhelmingly Rashīd al-Dīn's Tārīkh-i Mubārak-i Ghāzānī, written by the vizier who authored the reforms, in praise of the ruler who ordered them ³. It is a self-report from an interested party. What can be independently confirmed — the coinage standard, from surviving dies and hoards — checks out. Much of the rest cannot be tested.

The Jāmiʿ al-tawārīkh: world history as an administrative product

Rashīd al-Dīn Faḍlallāh Hamadānī was born around 1247 into a Jewish family of Hamadan, trained as a physician, converted to Islam around the age of thirty, entered Ilkhanid service as a court doctor, and became vizier in 1298. He held the office for twenty years. In roughly 1300 Ghazan commissioned him to write a history of the Mongols; Öljeitü, who succeeded in 1304, expanded the commission into something without precedent ⁷.

The Jāmiʿ al-tawārīkh — the Compendium of Chronicles — is the first work of history in any language that seriously attempts to cover the whole known world on the assumption that all of its parts belong in one book. It has sections on the Mongols and Turks, on the pre-Islamic and Islamic dynasties of Iran, on the Arabs, on the Jews, on the Franks of Europe, on India, and on China ²². It was researched the way an intelligence bureau researches: from informants, each a specialist in the subject, assembled at the vizier's expense. Bolad Chingsang for the Mongols and China. A Kashmiri Buddhist monk, Kamalashrī, for India and for a summary of the Buddha's life that is the earliest account of Buddhism written by a Muslim scholar from a Buddhist source. A Frankish friar for Europe, which is why the book contains a papal king-list ⁵⁷²⁵.

Rashīd al-Dīn also solved the distribution problem, in the most characteristic way available to an Ilkhanid vizier: he endowed it. The waqf deed of the Rabʿ-i Rashīdī, his foundation in a suburb of Tabriz — a quarter with a mosque, a hospital, a madrasa, a paper mill and a scriptorium, studied in detail by Birgitt Hoffmann — obliged the foundation to produce two complete copies of his works every year, one in Arabic and one in Persian, checked against the master text and sent out to named cities ⁹. A surviving fragment of the Arabic copy of 1314 is in Edinburgh University Library as Or.MS 20; another is in the Khalili Collection ¹⁶. Those are the pages at the top of this record.

Stefan Kamola's study of the work's composition makes the political function explicit: the Jāmiʿ al-tawārīkh is a legitimation device, built to install a dynasty of foreign military occupiers into the Perso-Islamic historical record as rightful sovereigns ⁷. That is true, and it is not a debunking. The book is a piece of state propaganda that also happens to be the founding text of a genre, and both halves of that sentence are load-bearing.

Marāgha, and the arithmetic that outlived the empire

In 1259, before the Ilkhanate was formally anything, Hülegü authorised Ṭūsī to build an observatory at Marāgha in Azerbaijan and to fund it from waqf revenues — a financing decision that mattered more than the founding one, since endowment income kept the institution running for more than half a century, past the death of its patron and its founder both ¹⁹²⁰.

Marāgha was not one astronomer with a quadrant. It was a staffed research institution with a library, instrument workshops, and a roster that included Muʾayyad al-Dīn al-ʿUrḍī, Quṭb al-Dīn al-Shīrāzī, the Maghrebi Muḥyī al-Dīn al-Maghribī, and at least one Chinese astronomer, named in the Persian sources in a form usually rendered Fūmenjī, whose calendrical methods entered the discussion ⁵. Twelve years of observation produced the Zīj-i Īlkhānī, the Ilkhanid astronomical tables, completed around 1272 and dedicated to the khan who had paid for it.

The theoretical output outlived the tables. The mathematical devices developed at Marāgha and in its successor tradition to replace Ptolemy's equant — the Ṭūsī couple, ʿUrḍī's lemma — reappear, unattributed, in the work of Copernicus two and a half centuries later, a correspondence first documented in the 1950s and since accepted as requiring some route of transmission, even where the route remains unproven ¹⁹²⁰.

The chao: the one Chinese import that failed outright

By 1294 the treasury was empty. Gaykhātū's ṣāḥib-dīvān, Ṣadr al-Dīn Aḥmad Khālidī Zanjānī, proposed the solution that had worked for the Yuan: print money. The notes were copied closely from Chinese banknotes, down to a Chinese inscription, and the first issue circulated at Tabriz on 19 Shawwāl 693 — 12 September 1294 — with refusal to accept it made a capital offence ²⁴.

Prices rose more than tenfold within days. Merchants closed their stalls, much of the population left the city, and traffic in the bazaars stopped altogether. The edict was withdrawn within weeks. Gaykhātū was overthrown and strangled in March 1295 — his contemporaries Rashīd al-Dīn and Waṣṣāf both attribute his fall to a factional revolt among the Mongol nobility rather than to the currency, though the currency had not helped ²⁴.

It is the most instructive failure in the record. Chinese paper money worked because the Yuan state could enforce a monopoly, absorb notes back through taxation, and make them the only lawful medium. The Ilkhanate could do none of these things, and a Tabriz merchant with three hundred years of silver dirhams behind him had no reason to pretend otherwise. Transmission is not a matter of copying a good idea. It is a matter of whether the institutional substrate that made the idea work came with it.

What was displaced

The transmission was not additive. Things went.

• The caliphate as a legitimating institution. After 1258 Iranian kingship could no longer be authorised from Baghdad. The Ilkhans authorised themselves from Chinggisid descent and from the Persian theory of the just ruler that Ṭūsī and Rashīd al-Dīn supplied, and that combination — steppe genealogy plus Persian kingship — became the licence every later Iranian and Central Asian dynasty used ¹³.
• Arabic's remaining share of Iranian high culture. Arabic had held the fields of theology, law and much of science. Under the Ilkhans, Persian took a decisive share of historiography, administration and eventually science; the Zīj-i Īlkhānī was composed in Persian.
• Baghdad's primacy. A thousand-year centre of gravity shifted to the Azerbaijani northwest. Tabriz, Marāgha and Sulṭāniyya became the axis of the Persianate world, and stayed there through the Safavids.
• Sunni institutional dominance on the plateau. With ʿAbbāsid authority gone and the Ilkhans indifferent for forty years, Shīʿī and Sufi institutions expanded in a way that would have been much harder before. Öljeitü converted to Twelver Shīʿism in 1310 and struck the names of the twelve imams on his coinage ¹³¹⁴.
• A Buddhist presence in Iran that lasted about forty years and was then erased. Hülegü and Arghun patronised Buddhist monks and built temples in Azerbaijan; after Ghazan's conversion in 1295 the temples were destroyed or converted, and almost nothing above ground survives ¹³¹⁵.

Mardetanha (2011). Mausoleum of Öljeitü (Gonbad-e Soltaniyeh), Soltaniyeh, Iran. CC BY-SA 3.0 via Wikimedia Commons. · CC BY-SA 3.0

What it cost, and who paid

The three registers, again

The cost of this transmission is enormous, unusually well documented at the institutional level, and unusually badly documented at the demographic level. The reader is owed the distinction.

Rhetoric. Juvaynī's 1,300,000 at Merv and 1,747,000 at Nishapur; the Arabic sources' 800,000 and up at Baghdad. These are figures written to convey enormity, by authors who had reasons — Juvaynī to demonstrate the terrible justice of the masters he served, the Arabic chroniclers to memorialise a catastrophe. They are evidence of how the events were experienced and understood. They are not counts.

Estimate. Smith's manpower analysis, the carrying capacity of the excavated urban areas, and the survival of tax registers put Nishapur's dead in the range of one to two hundred thousand and Baghdad's somewhere between ninety thousand and two hundred thousand ¹². These are inferences with wide bands.

Certain. A caliphate of five hundred and eight years ended on 20 February 1258. Merv, one of the great cities of the Islamic world, never recovered its position and its site is today an archaeological park. A currency was issued at Tabriz on 12 September 1294 and withdrawn within weeks. Six of the Ilkhanate's viziers were executed by the state they administered. These are not estimates.

Thirty-five years of fiscal wreckage

The part of the cost that the atrocity narrative misses entirely is what happened between the conquest and the reforms — and it did more cumulative damage to Iranian agriculture than any single sack.

Petrushevsky's assessment in the Cambridge History of Iran is that the seven decades from the 1220s to the 1290s constitute a period of colossal economic decline, caused as much by the administrative practices of the conquerors as by the destruction of the conquest itself ². The mechanisms are documented and mundane: revenue farmed to the highest bidder, who recovered his bid with a margin; the qubchūr levied repeatedly within a single year by agents of different princes, each with a valid warrant; the ortogh merchant-financiers advancing cash against future collections at rates that guaranteed default; requisition of transport animals and fodder along the yam routes without compensation; and, at the end of the chain, cultivators abandoning land because the assessment exceeded what the land could produce. Ghazan's reform decrees are themselves the best evidence for the scale of it — a state does not offer multi-year tax remission for restoring abandoned fields unless a great many fields have been abandoned ³.

The irrigation, and an honest uncertainty

Petrushevsky's further claim is that the damage was structurally irreversible in the northeast because of what was destroyed: qanats. A qanat represents decades of accumulated tunnelling, needs continuous maintenance, and cannot be rebuilt by a reduced population in a season or a generation. On this account the destruction of the qanat systems of Khurasan converted a landscape of relatively continuous settlement into isolated oasis towns, and the pattern persisted for centuries ²⁴.

This thesis has been pressed hard since. Critics point out that the evidence is largely literary, that regional archaeology is uneven, and that fourteenth-century Iran also absorbed plague, civil war and Timur's campaigns, any of which could carry part of the blame. The atlas's position is that the mechanism is real and well understood, the attribution to 1220–22 specifically is contested, and the record says so rather than choosing the more dramatic version.

The viziers

Here is the cost almost nobody narrates, and it is the one this record was written for.

The Persian administrative class took over the Mongol state in Iran. The Mongol state then killed them, one after another, on charges that were usually poisoning and usually invented, at intervals of roughly a decade, for as long as the Ilkhanate lasted.

Read down the right-hand column. Six of the most capable administrators of the fourteenth-century Islamic world, holding the office that made the Ilkhanate solvent, and five of them died by execution.

Rashīd al-Dīn's is the case that carries the whole thing. He had served two khans for twenty years, built the Rabʿ-i Rashīdī, written the first world history, reformed the coinage and endowed a hospital. In 1318 his rival Tāj al-Dīn ʿAlī Shāh brought a charge that he had poisoned Öljeitü four years earlier. He was executed on 13 July along with his son, and his head was carried through the streets of Tabriz while a crowd chanted the words the sources preserve — "This is the head of the Jew who abused God's name, may God's curse be upon him" — a man who had been a Muslim for forty years, killed with his origin in the charge sheet ⁷⁹. His estates were confiscated, his foundation was handed to the soldiery, and its library — the scriptorium producing two copies of his work a year, in perpetuity, by the terms of his own endowment — was looted. His son rebuilt it, held the vizierate for eight years, and was executed in 1336, whereupon it was looted again ⁹.

The converts' bill

Ghazan's conversion in 1295 is usually written as the happy ending of this story, the point at which the conquerors join the conquered. It was that. It was also a licence.

The Buddhist temples of Azerbaijan were destroyed or converted, the monks given a choice between Islam and departure. Churches and synagogues were attacked in the first months. The Christian communities of Iraq and Azerbaijan — Nestorians who had enjoyed Ilkhanid favour for forty years because the khans' mothers and wives were often Christian — lost that protection permanently, and the violence against them in the 1290s and again in 1310 was severe enough to be a hinge in the history of Eastern Christianity in the region ¹³. Saʿd al-Dawla's fall in 1291 had already been followed by a campaign against Jews under Ilkhanid rule, and 1318 supplied the coda.

None of this is a footnote to the transmission. The settlement between Mongol power and Persian administration was made in an Islamic idiom, and the people who did not fit that idiom paid for the fit.

The reckoning

So what was the bill, and what was bought?

Bought: a state. Iran as a governable political unit with a name, for the first time in six hundred years. A Persian administrative tradition carried across a catastrophe intact and handed forward to every successor regime to the present. A research observatory whose mathematics reached Copernicus. The first attempted history of the world. A manuscript-painting tradition that fused Chinese landscape convention with Persian book art and produced, within two generations, the Great Mongol Shāhnāma. A Eurasian circuit of specialists that ran in both directions for eighty years.

Paid: the cities of Khurasan, with a death toll that cannot be counted and was certainly in the hundreds of thousands. The ʿAbbāsid Caliphate. Seventy years of fiscal predation that emptied villages the armies had never reached. The irrigation of the northeast, in part and arguably. The Buddhist and much of the Christian presence in Iran. And the individual men — Shams al-Dīn, Saʿd al-Dawla, Ṣadr al-Dīn, Rashīd al-Dīn, Ghiyāth al-Dīn — who did the actual work of the integration and were executed for it by the beneficiaries.

The popular story says the Mongols ended Islamic civilisation in Iran. They did not. They were absorbed by it, converted to it, paid for its most ambitious century of scholarship, and restored its political name. The popular story is wrong in a way that matters, because the correction is not consoling. A civilisation that survives its conquest is not a civilisation that was spared. It is one that kept working through the wreckage, on the conquerors' payroll, and buried its administrators as it went.