Central Asian world cities?   


(XI – XIII century)


A discussion paper



 George Modelski




Central Asia is the vast landlocked region that comprises most of contemporary Iran, much of Afghanistan,as well as the five new post-Soviet republics  of Uzbekistan, Tajikistan, Turkmenistan, Kirgyzstan,, and Kazakhstan. Historically it consisted of the regions known as Khorasan, Khwarazm, Transoxania, and Bactria.


In the three centuries (XI-XII), roughly speaking between 950 and  1250, this area boasted not only

of prosperous cities  but also hosted vigorous intellectual and cultural life, and also was an active center of the Moslem world. “Central Asia” wrote A. G. Frank in 1992, has been known as ‘paradise on Earth’, ‘a land of a thousand cities’, and its capital Bactra as ‘the mother of cities’”   Bactria was part of Central Asia that in classical and modern eras was the cross-roads of Eurasia;  it was in an area “where routes converge from all quarters of the compass and from which routes radiate to quarters of the compass again”.  This was the keystone of the Silk Roads of the classical world, right into the modern age.


The Baghdad-centered (Sunni) Abbasid Caliphate gave definition to that space ever since about 750.   When its influence faltered, some two centuries later, it lost political power to its military commanders, the (Shia) Buyids - who occupied Baghdad in 946 – even while retaining its symbolic position in the Moslem world.   The Buyid period saw a loose confederation of Iranian principalities, such as the Samanids (875-999) in Bokhara and Samarcand; meanwhile the (Shia) Fatimids seized power in Egypt.


In 1055, the Seljuks displaced the Buyids in Baghdad.   They were (Sunni) Turks who had moved in from Central Asia and, adopting the title of ‘Sultans” set about restoring the central empire of Islam.  The Seljuks “instituted a new order in the Middle East, most of which was now united under a single authority” (Lewis 1995).  They sponsored a Sunni revival, and their political standing was greatly enhanced when Alp Arslan crushed the Byzantine army at Manzikert in 1071, and took over Anatolia.   Nishapur, and Merv ware the centers of Seljuk power.   Extremist Shia oppositionists retreated into the hills and came to be known as the Assassins; they found a stronghold at Alamut, (NW of Rayy) (1090) that remained impregnable until the Mongols destroyed it in 1256.


The Great Seljuks were severely weakened by the Karakitai, and were gradually displaced by the (Sunni Turk) Khwarazam- Shahs, their former tributaries.   By 1200 a Khwarazmian empire took shape, with a capital in Urgench on the lower reaches of the Amu Darya (Oxus) River.   For a short while it loomed large over Central Asia, assembling, under its authority,  all of its major cities,  Samarcand, Bokhara, Merv, Balkh, Herat, Nishapur, and Rayy, but its career was short.   Genghiz Khan attacked it in 1219-1220, and destroyed it in a swift campaign.   Resistance continued for a while but it ceased when the last ruler died in 1230.     When Helugu captured Baghded in 1258, all of Central Asia was controlled by the Mongols.     


The problem


The cities of Central Asia in the early modern era are not well known, and their population numbers are not well attested for.  This discussion paper brings together the available information, with one particular aim: to determine if this central region of the Moslem world might have accommodated, in that period, cities with a population of one million or more,  on the assumption that a high degree of urbanization might serve as a breeding ground of modernity.


The following Table 1 summarizes our information about the seven most important cities of Central Asia,

so that we might ask, which of them had risen in the early modern age, to millionaire-city status.   They are Merv, Nishapur, Herat, Samarkand, Bokhara, Rayy, Balkh, and Urgench.    This information will be compared with data for Zhongdu (Beijing), and Baghdad.  The important part of that information relates to the numbers reportedly killed by Mongols in the course of their campaign of conquest against the Khwarazm empire, These were then probably among the wealthiest, most literate and generally civilized areas of the world.  The question is the following:   do the extant historical accounts of Mongol massacres  over the period 1215-1258 allow us to draw conclusions about Central Asian city populations in 1200, and the preceding one-two centuries?         


Opinion varies about the value, and the reliability of the figures that have come down to us ,

and while some take seriously such quantitative data, others are obviously skeptical.   They regard them as ‘hyperbolic”, and a product of exaggeration, or a late result of Mongol propaganda.   Bernard Lewis, in his history of The Middle East (1995) argues that  the destructive effects of the Mongol conquests were neither as great, as lasting, nor even as extensive, as was once thought”. A recent student of Genghis Khan, Jack Weatherford (2004:118) calls them “not merely exaggerated or fanciful” but “preposterous”, and he gives a number of reasons.   He claims that his subject could be “more accurately described as a destroyer of cities than a slayer of people  - as though destroying cities, albeit for ‘strategic’ reasons,  were something to boast about.   But we know that accounts of his deeds also include accurate portrayals of the methods whereby the slaying of people reached a high degree of perfection in the Mongol armies, and that is additional evidence.



Table 1:   Basic Data 




No. inhabitants

T. Chandler  (TC), G. Modelski (GM)       

Juvaini  (Boyle tr.),

Boyle 1975

Encyclopedia Britannica (EB),  Encyclopedia of Islam (EI

Howorth (H))



1153  sacked by Ghuzz

1221 sacked by Mongols






1150   population 200,000 (double Nishapur), (largest city in the world)

over15 miles sq.(TC)

1100:  100,000

1200:    45,000

1221 “The Mongols (7000 men) ordered that … the whole population, including the women and children, be killed”

The sayyid Izz-ad-Din, with others, counted the people slain for 13 days and arrived “at a figure of more than 1.3 million” (163-4);

Each soldier allotted 3-400 persons to kill”

Ibn al Athir: 700,000 slain; (“allow for customary hyperbole”) (EI,VI)





828- overtakes Merv

1037-1140 Seljuk capital

1153 sacked by Ghuzz


1000:   125,000

1034    120,000

1100:    100,000

c.1150  ½ of Merv

1220    70,000 (TC)


Tole with “very large army” “drove all the survivors on the plain… not even cats and dogs left alive”(177)

“the whole population put to death” (Boyle 1975)

April 1221 

Genghiz Khan’s son-in-law Tokuchar killed, widow decrees death for all (Weatherford 117) 1,747,000 men lost their lives, acc. to Mirkhond (Erdman  420) (H I,88)




Timurid capital 1409-47






900:   250,000

1217:   44,000

1221:   60,000 (TC)

1221-2 captured by Elgigidei with 80,000 men;  “the entire population put to the sword” “seven days of slaughter”; Saifi: 1.6m slain;

Juzjani (contemp), (‘History of Herat’)  finds, in single quarter, 600,000 dead, hence total of 2.4m” (Boyle 1975)

Six months’ siege, by Noyan, with 80,000 men, “for a whole week, it is said, 1.6 m were killed”

(H I,91)






 968: c.790 ha.

1150:  30,000  (TC)

“surrender availed them not” “whole population driven out” “put to the sword” (130-1)

IX,X cent.geographers:

some 200,000,(7.7 km.sq.)

1200 great. mosques,

200 public baths (H I,80).











EB, W, EI, H



on Amu-Darya,


capital of Khwarazm




1200:   25,000 (TC)

Army of Chagatai and Ogodei, after7 mo. siege;

drove people out into the open”, “artisans consisting of more than 100,000 separated”.  Girls and boys in slavery, all the rest slain, “to each fighting man fell the execution of 24 persons” (127)

“artisans consisting of 100,000 families were set apart… the rest were divided among the soldiers, 24 to each, and all were then slaughtered”

(H II,33-4)



1035 laid waste by Seljuk


1000:  100,000

1200:    80,000

1220:    larger than Isfahan  (TC)

c.8000 ha (GM)


Sobotai: every male killed  (Boyle 1975)

Rivaled Damascus and Baghdad in VII cent.

c. 8000 ha (EI)

100,000 Mongols destroy it



1141 Kara-kitai capital;

1212 to Khwarazm;

May-June 1221 Mon-gol siege





1220:  106,000

1221:    26,000 (TC)

Defended by 110,000 men,  60,000 Turks & 50,000

Tajiks.  Nos of townspeople “beyond  computation”;  50,000 driven out of town;  30,000 chosen for their craftsmanship, and 30,000 youth for the levy.

30,000 Turks from the citadel brought into the open, “divided into groups of ten and hundred”, and slaughtered

Chinese traveler: 100,000 families before the Mongols, but only ¼ left afterward. (EI)




1000:  75,000 (TC)

earlier 25 km.sq.

Over 100,000  (Frye)

300,000  (Man)

March 1220   first city to be attacked.  Fell after a few days’ siege (Boyle)

“Most of the city was burned, and the slaughter …was enormous, although not as complete as elsewhere” (Frye 184)).

“Noble Bukhara

Caravan city, consisting of citadel, shahristan, and rabid (3 walls).

 ZHONGDU (Chung-tu)


Jin Capital 1153-1214)

DADU (1272-

(on new site)

1200:  130,000 (TC)




1300: 1m (GM)


1215   Mongols burn it to the ground, population is massacred (Grousset 224);

1265:   160,389

1280:   401,000

50 km.sq. within walls (EB)


seat of caliphate


932:   1.1m  (TC)

1000: 1.2 m

1100: 1.2m

1200: 1m (GM)


X cent. pop. 1 ½ m;

1258   estimates of the dead range from 800,000 to 2 million (EI)




Weatherford writes (ib.): “…conservative scholars place the number of dead from Genghis Khan’s invasion

 of central Asia at 15 million within five years.   Even this…total would require that each Mongol kill more than a hundred people;…    To set against this argument, we have two specific accounts from the cases of Urgench, and Merv, where each Mongol soldier was required to kill 24 , or 3-400 people respectively.


In Urgench, after separating the artisans, and the young women and men that were to be reduced to slavery,

 each Mongol warrior – in an army group that might have consisted of two tumens (units of 10,000) (one for each son of Genghis Khan) but also included allied contingents – was required to execute 24 people (Juvaini 127, Howorth II,33-4)).    That one deed, if confirmed,  alone would account for some 480,000!   In Merv (1221), by contrast – we are told – “people were driven out of town for four days and each soldier was allotted 3-400 persons to kill” (Boyle 1975, Juvaini 163-4).   Maybe the higher quota was due to the small size of the Mongol force – 7,000 men – but the total computes, at the lower level, to an astounding 2.1m.   Clearly, Genghis was not just a destroyer of cities, he was also a slayer of people.  For one major city, he ordered his commander: ”You must execute the whole population of Herat”. (Boyle)   Zhongdu, a predecessor city of Beijing, was destroyed in 1215, and its population massacred.   A similar fate befell Baghdad half a century later.


The Mongol method of dealing with conquered cities had four interrelated components:


1.   The Mongol army was numerate, and was organized on a decimal system, with the basic unit of 10 men, becoming part of one of 100 men, which in turn formed a regiment of 1,000, and ten of which formed a ‘tumen’ - ‘division’-  of 10,000.  The core army consisted of 100,000 men, to which allied forces and conscripted levies were added.   The conscription of local labor was an extension of the decimal organization, as each Mongol fighter was required to round up ten local men to work under his command (Weatherford 92).   Victims of massacres were similarly allocated “proportionately among the soldiers in accordance with their usual custom” (Juivani 139).


2.  Every city or town that refused surrender and resisted the Mongols was subject to destruction.   That was widely known, and broadcast, and was an essential element of the terror inspired by this army.   The Table shows that process for the most important cities.   But the city would also be destroyed, e.g. when Genghis Khan’s son in law was killed by an arrow at Nishapur, and his widow demanded revenge: “death for all” (Weatherford 117).   Weatherford cites “strategic reasons” for destroying cities: to facilitate control of trade, but cites no source for that statement, or cites specific cases.


3.    Genghis Khan did not as a rule enter the cities he conquered;  “when victory was assured, he withdrew with his court to a distant and more pleasant camp while his warriors completed their tasks” These tasks consisted of emptying the city, and driving the population into the open spaces outside it.  (Weatherford 3).    This rendered the people defenseless, made possible a count of the population, and what is more, facilitated looting.   Mongol leaders treated looting as a serious matter of state, and from the early campaigns onward the emptying of cities was developed as a means of facilitating systematic looting,   But it soon also became a way of disposing of large numbers of people.   The accounts we have describe masses of people, effectively managed, conducted into the countryside, allotted to the army units, and efficiently massacred.  As already mentioned. in Tirmiz, on the Oxus: “all the people, both men and women, were driven out onto the plain, and divided in accordance with their usual custom, then they were all slain” (Juvaini 139)  The bodies apparently remained unburied, hence subject to rapid decay even in drier climes.


4.   Depending upon the size of the army unit, and the numbers of victims rounded up – they were apparently herded, decimally,  in tens, and hundreds etc. – each soldier was required to execute a number of persons that varied according to circumstances.   We have reports of 24 per warrior for Urgench, and 3-400 in Merv, as noted.


All this does not mean that we must accept without qualification the figures that we are offered.   The city figures of victims were likely inflated by significant numbers of refugees driven by fear, and by Mongols.   The figures also need to be cross-checked against archeological findings on the extent of the ruins, stripped of the element of exaggeration, and put in context of political and economic developments.   We must also remember that the population we are estimating is that of 1200, and not of the year the Mongols attacked.   But we cannot altogether ignore them.


The sources


The first element in regard to these sources must be caution.   Comparison with Chandler’s (generally reliable) census shows him to be a skeptic, opting mostly for low estimates.   But his Central Asian estimates seem to be unusually cautious, and likely too low.   For example his figures for Baghdad, 1000 to 1200, seem to be drastically flawed.    How are we to take the 1200 figure for Urgench, at that time the seat of the Sultan of Khwarazm, the leader of the great, if only recently assembled, empire in these parts of the world, as  rating only 25,000.   And how are we to regard his assertion that Merv – whose population he put at 200,000, was, around the year 1150, the “largest city in the world”.   For we know that between 1100 and 1200, at least three world cities, Baghdad, Kaifeng, and Hangzhou, had a population in the million range:


We might also remark on the reliability of our main source “The History of the World Conqueror” whose author, Juvaini, a Persian, was a high official in the service of the Mongol empire and, after 1258, the administrator of Baghdad.  Most generally, his account of the fall of the empire of the Khwarazm Shah, that is the center piece of his work, carries conviction.   In his judgments, he exercises caution.   For instance, in closing his chapter on the siege of Urgench he writes :   “As for the fighting and killing,… I have heard such a quantity of slain that I did not believe the report and so have not recorded it” (Boyle tr.128),


Overall Jovaini’s is probably the best source on the human cost of the Khwarazm campaign, even though he gives an overall figure only for Merv (while another might perhaps be inferred for Urgench).   But he does note several cases where all were “put to the sword”.   In fact, while he describes the sad fate of the three other main cities of Khorasan, he fails to report the disaster that struck Herat - that other writers have rated as even bloodier than that of Merv.   Nor does he extend his narrative to the fall of Baghdad, another catastrophe.   In fact his entire work is so structured as to contrast the relatively benign events prior to the Mongols with the horrific destructions that followed it.


What then do the extant sources tell us about the population of the system of major cities found in Central Asia

 in the XI-XII centuries?




First, we note the thoroughness of the destruction.   Of the eight Cental Asian cities surveyed, all but two were totally destroyed.  Only ruins have remained of Merv, Urgench, Rayy (on the outskirts of Tehran), and Balkh (close to Mashar al-Sharif).   Heart, thoroughly devastated,  was only rebuilt under the Timurids.  Samarkand “barely survived”, with only one quarter of its former population.  Nishapur sunk into oblivion.   Only Bohara, the only city which is recorded as having been entered by Genghis Khan, survived seemingly  intact



Let us review the record for individual cities.


Merv   Of the seven cities reviewed, this is the only one for which Juvaini provides actual numbers, albeit those of a witness, for 1.3 million, to which he adds another one hundred thousand or more later on.   Cited in the Encyclopedia of Islam is another writer, for 700,00.  Either of these are very big numbers .   Are we to credit these as justifying the claim that Merv, known as “very rich and populous”, might have had a population in the one-million class?


The website of the International Merv Project (at University College London) accounts for an urban (Abbasid-Seljuk) site approaching 1000 ha (in an oasis of some 1900 miles sq.), one that would justify a population in the quarter-million range.  Explaining the eye-witnesses larger numbers might be (1) exaggeration, and (2) refugees, driven into the city by the Mongols.  On balance, these figures seem too hih.


Herat   That city gives rise to claims even larger than Merv.   Of the two writers cited by Boyle (1975), Juzjami (less friendly to the Mongols than Juvaini) puts forward a figure of 2.4 million (based, we are told, on an actual count of 600,000 in one quarter, extrapolated to the whole city), and Saifi, for 1.6 billion.   Such figures are hard to grasp, But it is worth bearing in mind that Heart was attacked with as many as 80,000 men, and the killing reportedly went on for an entire week..   But unfortunately the city was so thoroughly destroyed that not much remains of the city briefly controlled by Khwarazm, and the extent of its built-up are is hard to judge.   Even allowing a large discount for exaggeration, this might still leave us with a substantial figure, possibly with a million-range.


Urgench   This capital of the Khwarazm Sultan is little known archeologically or textually, and is little studied, but is now claimed to have had its ‘golden age” in1150-1220. The figures reported by Juvaini and Howorth reach into the one-million range if the group of artisans who were spared were thought to consist of their entire families.   But that might not be the right way to do it.


Nishapur   Described as the most important city of Khorasan, this was a Seljuk capital, and at that time was overtaking Merv, with a built-up area reaching 17 km.sq., suggesting a population in the 300-400,000 range.   Reports (from less reliable sources) of some 1.7 m slain seem to be far-fetched, and hard to credit.


Rayy   Said at an earlier point to have rivaled Damascus and Baghdad this city presents an interesting problem.   Its urban site, as reported in the Encyclopeadia of Islam, representing some 8,000 ha,  that was attacked by a Mongol army of 100,000, clearly places in the million-people range.   But  while definitely destroyed by the Mongols, with “every male killed: (Boyle 1975), we do not have information about the aggregate toll taken here.   Let us reckon it as a possibility.


Samarkand   In 1220, the city had a large garrison, of 110,000 men.   We also have a report of a Chinese traveler, mentioning 100,000 families (undoubtedly a round figure), that is some 500,000 inhabitants prior to the Mongol attack.   The data do not seem to rise to a one-million level.


Bokhara    The city’s modern biographer, Richard Fry(1996:94) regards population estimates as “pure guesswork…(but)  one might hazard a guess that the population was one hundred thousand, but estimates reaching from one half to one million are surely exaggerated”.  Frye’s numbers might be on the lower side, but Man’s figure of 300,000 could be nearer to the mark.   .


In closing


This review raises a number of interesting questions.   The broadest  might be this:  why did this network of cities, at the

core of the Moslem world, fail to assume a leading role in the building of the modern world system?   Maybe its commitment to the ways of the traditional (overland) Silk Roads made it incapable of reshaping its priorities.   Just as importantly, its land-locked position equipped but poorly for a role in the new, oceanic world system.  Yet what was perhaps the land of the world’s wealthiest cities in 1200 is among the poorest portions of the globe to-day. 


More narrowly, this has been a discussion paper raising questions about the population of an important group of early

 modern, and potentially world, cities.   Our discussion suggests that the city populations might have been larger than is now estimated.  On evidence so far, we might consider Rayy as a potential candidate for that category but would need additional evidence for Urgench, Rayy, and Herat before reaching any firm conclusions.   Comments on any of these points, or suggestions for additional data are welcome.



September 29, 2007





Boyle, J.A.   (1977)   The Mongol World Empire   Variorum Reprint, London


Chandler, Tertius, (1987)   Four Thousand Years of Urban Growth,  Lewiston:  St. Gavid’s. 


Frank, A. G.,   (1992)   The Centrality of Central Asia,   Amsterdam: VU University Press.


Frye, Richard N.   (1996) Bokhara: The Medieval Achievement,  Costa Mesa, CA: Mazda


Grousset, R.   (1953)   The Rise and Splendor of the Chinese Empire,  Berkeley:  UC Press,


Howorth, H.H. (1880)   History of the Mongols,   New York:  Burt Franklin, 4 vols.


Juvaini   (1997)   The History of the World Conqueror,  J.A. Boyle tr, Seattle: UW Press.


Lewis, Bernard    (1995)   The Middle East,   London:   Oxford University Press.


Modelski, G.   (2003)   World Cities: -3000 to 2000,   Washington:  Faros2000


Man, J.   (1999)   Atlas of the Year 1000;   Cambridge:  Harvard U.P.


Weatherford, J.  (2004)   Genghis Khan and the Making of the Modern World,  New York:  Three Rivers Press.





Dr Ute Franke, of the German Archeological Institute, Eurasian Division, Berlin wrote on August 2, 2007:


“…I think that Chandler’s estimate is more correct, at least for Herat city proper;

1.6 million is out of the question.”