Genghis Khan in Uzbekistan: What 1220 Left Standing
Genghis Khan in Uzbekistan, 1219–1221: Otrar, Bukhara, Samarkand and Gurganj as Juvaini told it, why the death tolls don't add up, and what survived.
Bukhara had one tall brick tower before Genghis Khan arrived in 1220, and it still has it. The Kalyan minaret, finished in 1127 and about 46 metres high, is the single piece of the city's pre-Mongol skyline that you can stand under today. The mosque beside it, the citadel walls, the wooden houses: all of that went. Here's the story of Genghis Khan in Uzbekistan between 1219 and 1221, city by city, with a warning attached to every number.
Otrar, 1218: 450 merchants and a governor's greed
The war started with a trade mission. According to the Persian historian Ata-Malik Juvaini, Genghis had his sons and commanders each fund two or three traders, and the caravan carried 450 Muslim merchants. In 1218 it reached Otrar, a Khwarazmian frontier town on the Syr Darya whose ruins are now in Kazakhstan.
The governor, Inalchuq, was a relative of the Khwarazmshah's mother. Juvaini's version is almost petty: an Indian merchant who'd known him years earlier greeted him by his plain name instead of his title. Offended, and eyeing the goods, Inalchuq arrested the lot, and Sultan Muhammad II approved their execution. One merchant escaped and carried the news east.
Other chroniclers add a second insult: an envoy sent by Genghis to demand Inalchuq's surrender was killed as well. Whether Genghis needed a pretext or was simply handed one, historians still argue; our history of Khorezm before Khiva sets out how the dynasty got itself into that position.
You'll often read that Inalchuq died with molten silver poured into his eyes and ears. It isn't in Juvaini, who says only that he was executed at the Kok-Sarai near Samarkand after Otrar held out for five months. Treat the silver as a later embellishment.
The shah's real blunder came after the massacre
Killing the merchants was reckless. What Muhammad did next was worse. Instead of meeting the Mongols with one field army, he spread his forces through the walled cities of Transoxiana and waited. Juvaini puts the Samarkand garrison alone at 110,000 men, 60,000 of them Turkic troops and 50,000 Tajiks. Otrar's governor had been given 50,000.
That's a lot of soldiers behind walls while the enemy picks the order of its visits. The Mongols took the towns one at a time, drove each one's captured men against the next, and never faced the empire's combined strength. The shah fled west and died on an island in the Caspian around the turn of 1220–21 (chroniclers differ on the date). Our view: the caravan killing gets the headlines, but the garrison strategy is what doomed the cities.
Bukhara, February or March 1220
Genghis didn't come down the Syr Darya with his sons. He took his own column across the Kyzylkum. On the way, Juvaini records the surrender of Nur, now Nurata, at the foot of the Nuratau hills. Mongol riders approached at walking pace, each holding a ladder cut from garden trees in front of his horse, and the watchmen took them for a caravan. Nur paid a ransom of 1,500 dinars, half of it, Juvaini says, in the women's earrings.
Then Bukhara, where the dates already wobble. Juvaini puts the Mongol camp outside the citadel at the start of Muharram 617, which his translator J.A. Boyle converts to March 1220, noting that Ibn al-Athir and Juzjani say February. The Encyclopaedia Iranica gives 10 February 1220 for the town and twelve days later for the citadel. Pick February if you need one date, but it's a reconstruction.
The scene everyone quotes happens in the Friday mosque. In Juvaini's telling, about 20,000 of the sultan's troops broke out and were destroyed, the town opened its gates next morning, and Genghis rode into the mosque on horseback. He asked whether this was the sultan's palace. Told it was the house of God, he climbed a few steps of the pulpit and said the countryside was empty of fodder. The Qur'an cases became mangers, wine went round, and the town's scholars were set to hold the horses.
The famous speech is often placed in the same mosque. Juvaini sets it afterwards at the festival prayer ground outside town, before 280 of the wealthiest residents (190 locals and 90 visiting merchants), and it's there that Genghis tells them he is "the punishment of God" sent because of their great sins.
The garrison held the citadel, so the Mongols set fire to the town. The houses were wooden and most of Bukhara burned within days. Juvaini makes an exception worth noticing: the Friday mosque and some palaces, built of baked brick, didn't burn. Townspeople were driven against the citadel as a screen, and when it fell more than 30,000 Qangli Turks were counted among the dead. Every surviving man fit to fight was marched off against Samarkand.
One refugee who reached Khorasan summed it up in a line Juvaini admired for its economy: they came, they sapped, they burnt, they slew, they plundered and they departed.
Reading Juvaini: a witness who wasn't there
Juvaini was probably born in 1226, six years after Bukhara fell. He began his History of the World-Conqueror at the Mongol court in Qara-Qorum in 1252 or 1253 and was still working on it around 1260, in the service of Genghis's grandson Hülegü. Later he governed Baghdad for the Mongols.
So what kind of source is that? A Persian Muslim official, writing three decades on, for Mongol masters, about the humiliation of Muslim cities. He had documents (he cites a Mongol decree listing the emirs killed at Samarkand) and survivors' memories, and a court style that loves a moral. That doesn't make him wrong. It does mean the mosque scene is a sermon on judgement as much as a report, and his numbers aren't a census.
If you want to read him yourself, Boyle's 1958 English translation is free to read on the Internet Archive (last checked: September 2026); the Bukhara chapter runs only about six pages.
Samarkand, spring 1220: the city that moved
Samarkand had that 110,000-man garrison and a newly strengthened wall. It lasted days. On the third day the Mongols entered, and the cadi and the shaikh al-Islam were allowed to keep more than 50,000 people under their protection inside. The citadel held out one more day. Its last defenders, about a thousand men, fought from the cathedral mosque until naphtha set it alight, and the mosque burned with them inside.
Then the sorting. More than 30,000 Turkic soldiers were killed after being told they'd be spared. 30,000 craftsmen were shared among Genghis's sons, as many again taken as a levy, and the rest let back in for a ransom of 200,000 dinars.
The date? Juzjani says 19 March 1220. Juvaini's text gives a month that Boyle calls an error, and would put it in May–June.
For a visitor, the detail that matters is small. During the citadel assault, Juvaini says, the Mongols laid waste the Juy-i-Arziz, the "leaden watercourse". That channel carried water to the old city on its plateau, and the usual explanation is simple: without it, the plateau couldn't be lived on. Samarkand didn't die in 1220; it relocated. The old city, now called Afrasiyab, a field of brown mounds with a museum on top, was left empty. The town that regrew did so on lower ground to the south, and that's where Timur built his capital 150 years later. UNESCO's Samarkand listing treats Afrasiyab as an archaeological reserve because nobody built over it.
Gurganj, 1221: the siege Juvaini wouldn't total
Gurganj (Urganch to its own people) was the Khwarazmshahs' capital on the Amu Darya. Genghis sent his sons Chaghatai and Ögedei against it, with Jochi sending reinforcements. Its commander, Khumar Tegin, lost his nerve and gave himself up partway through.
Juvaini's details are oddly practical. There was no stone anywhere near Khorezm, so the Mongols cut catapult ammunition from mulberry trees. The river ran through the town under a bridge, and when 3,000 Mongols tried to seize it to cut off the water, none came back. The fighting then went street by street. At the end, more than 100,000 artisans were separated out and sent east; young women and children were enslaved; and the remaining men were divided among the soldiers, 24 to each, to be killed.
And then Juvaini does something unusual. He says he'd heard a figure for the dead so large he didn't believe it, and so he didn't write it down.
That hasn't stopped the internet. You'll find the claim that Juvaini recorded 50,000 Mongol soldiers each killing 24 people, making 1.2 million dead. We couldn't find the 50,000 in Boyle's translation of the Gurganj chapter; it looks like a modern multiplication. The 1.3-million figure often attached to Gurganj is Juvaini's thirteen-day count at Merv, a different city in a different chapter. The story that the Mongols broke the dams and flooded the ruins comes from other chronicles, and can't be checked on the ground.
The ruins, Kunya-Urgench, are in Turkmenistan, a separate UNESCO site needing a Turkmen visa; crossing into Turkmenistan from Uzbekistan covers the paperwork.
The numbers, and what they actually measure
Every figure above, with what it really measures:
| Place | Date as given | Figure in Juvaini | What it measures |
|---|---|---|---|
| Otrar | 1218; siege 1219–20 | 450 merchants; siege of 5 months | Caravan size; a duration |
| Nur (Nurata) | 1220 | 1,500 dinars | Ransom paid |
| Bukhara | Feb or Mar 1220 | 20,000 in sortie; 280 notables; 30,000+ Qangli killed | Troops, hostages, soldiers "counted among the slain" |
| Samarkand | 19 Mar 1220 (Juzjani) or May–Jun (Juvaini, per Boyle) | 110,000 garrison; 50,000 protected; 30,000 craftsmen; 200,000 dinars | Army size, survivors, deportees, ransom |
| Gurganj | 1221 | 100,000+ artisans; 24 per soldier; total refused | Deportees; a ratio, not a total |
Notice what's missing: a death toll for any of these cities. Chroniclers used big round numbers to signal scale and judgement, and modern writers have multiplied them into totals the sources never gave. The destruction was enormous. The figures aren't knowable.
What survived, and why
Survival in 1220 came down to brick, burial and luck. Wood burned. Fired brick, if nobody bothered to dismantle it, could stand.
- The Kalyan minaret, Bukhara (1127). Local legend says Genghis looked up at it, his hat fell off, and he spared it for making him bow. Juvaini never mentions the minaret at all. The boring explanation fits his text better: solid brick doesn't burn and there was nothing inside to loot. The Po-i-Kalyan ensemble around it came later.
- The Samanid mausoleum, Bukhara (early 10th century). It stood in a flood-prone cemetery where silt was slowly burying it, which may be why it passed unnoticed. UNESCO's Historic Centre of Bukhara listing calls it the best surviving 10th-century building in the Muslim world. More on the Ismail Samani Mausoleum and its brickwork.
- Magoki-Attori mosque, Bukhara. Its 12th-century facade survived because the ground rose around it; archaeologists dug it out in the 1930s.
What didn't survive: Bukhara's citadel walls (today's Ark is much later), Samarkand's cathedral mosque, Afrasiyab, and Gurganj. Even Bukhara's Friday mosque, which Juvaini says escaped the fire, is gone; the current Kalyan Mosque dates from 1514.
Bukhara did come back, faster than you'd expect. Juvaini credits the administrator Mahmud Yalavach with drawing people home, and by the 1250s calls it unmatched in the Islamic lands for scholars and endowments.
Seeing 1220 on the ground
Follow the Mongols' own order: Bukhara, then Samarkand, then Khorezm. In Bukhara the minaret and the Samanid mausoleum are about a 20-minute walk apart. In Samarkand, go up to Afrasiyab before the Registan, not after; the Registan makes more sense once you've seen the plateau the city abandoned.
And take the Boyle translation on your phone. Read the Bukhara chapter in the square below the minaret, knowing its author never once mentions the tower above you. It's the quickest lesson in how 1220 got told.