Liao Empire

The chief who would not stand for re-election

Drafted from Bretschneider, Mediaeval Researches from Eastern Asiatic Sources, volume I (1888)formation

The Khitan were a confederation of eight tribes on the Liao river, and their paramount chief was elected for a three-year term.

Abaoji was elected in 907. In 916 he was still in office, had killed most of the people who objected, had taken the title of emperor with a Chinese-style reign name, and had made the office hereditary in his family.

That transition — from a rotating tribal chieftaincy to a dynastic throne — is the founding act, and it is the one every steppe state has to perform and most of them fail at.

How he did it

By not letting the term expire, which produced the obvious result: his brothers revolted, three times, between 911 and 913. He put down all three and, according to the histories, executed several hundred of the participants.

Then he dealt with the other seven tribal chiefs. The account that survives says he invited them to a feast at a salt lake — pointing out that they had all been using his salt and had never thanked him for it — and killed them there.

Whether that happened as told is the usual question with a founding massacre. What is not in doubt is that after about 915 there were no other chiefs with a claim, and that the Khitan state thereafter had a single hereditary ruling house.

What he built it out of

Not a steppe empire on the old model. Abaoji's Khitan sat on the boundary between the Mongolian grasslands and the settled farming country of what is now Manchuria and northern Hebei, and he took from both.

He captured Chinese populations in raids and settled them in walled towns of their own — the histories describe towns built to replicate the ones the captives had come from, so that they would go on farming and paying taxes. Chinese advisers were used from the beginning; one of them, Han Yanhui, is credited with organising the agricultural settlements and much of the administrative structure.

He also commissioned a script. The Khitan large script was created in 920 on the model of Chinese characters, and a smaller script derived partly from Uyghur followed a few years later. Both were used, both are only partly deciphered, and the decision to have one at all is the point: a ruler who orders a writing system made for his own language is not planning to be a confederation.

926

The last thing he did was take Bohai.

Bohai was a settled kingdom in what is now eastern Manchuria and northern Korea, three centuries old, literate, organised on Tang lines with five capitals and a full bureaucracy. Abaoji destroyed it in about three months in 926, renamed it Dongdan — the Eastern Khitan state — and installed his eldest son as its king.

He died on the way home from that campaign.

What his successors inherited was therefore not a tribal following but a composite: a steppe confederation, a settled agricultural population in walled towns, and an entire conquered kingdom with a functioning Chinese-style administration already in it.

The problem of how to govern all three at once is what the next century of Liao history is about, and the answer they arrived at has no real parallel anywhere else.

The sixteen prefectures

Drafted from Bretschneider, Mediaeval Researches from Eastern Asiatic Sources, volume I (1888)expansion

In 938 the Liao acquired sixteen prefectures of settled Chinese territory south of the Great Wall, including the site of what is now Beijing, and they were not taken by conquest. They were paid over as a fee.

The transaction

North China in the 930s was in the middle of the Five Dynasties — a succession of short-lived military regimes in the old imperial heartland, each one overthrown by a general of the last.

Shi Jingtang was one such general, in revolt against the dynasty he served, and losing. He asked the Liao emperor for an army. The price was the sixteen prefectures, an annual payment of three hundred thousand bolts of silk, and Shi Jingtang's formal acknowledgement of the Liao emperor as his father — he was about ten years older than the man he was calling father, which contemporaries noticed.

He got the army, won, and founded the Later Jin dynasty as a Liao client.

Why it mattered for four hundred years

The sixteen prefectures are the mountain line. The passes through the ranges north of the plain are the reason a Chinese state can hold the north China plain against cavalry at all, and after 938 they were on the wrong side.

Every subsequent Chinese dynasty tried to get them back. The Song attempted it twice, in 979 and 986, and was beaten badly both times; the second failure led directly to a defensive posture that shaped Song military policy for the rest of the dynasty and to the treaty of 1005 that made the Song a tribute-payer. The territory did not return to a Chinese-ruled state until the Ming took it in 1368, four hundred and thirty years later.

What the Liao did with them

This is the part that makes the dynasty unusual. They did not try to make the prefectures Khitan, and they did not try to make the Khitan Chinese.

They ran two governments. The Northern Administration governed the tribal population by Khitan custom, with Khitan officials and the ordo system of imperial camps. The Southern Administration governed the settled Chinese population by Tang institutions — prefectures, counties, a civil bureaucracy, examinations — with largely Chinese officials, in Chinese, on Chinese law.

The emperor was head of both and the two systems did not merge. That arrangement is the subject of a chapter of its own, and it is the single most important administrative invention on the northern frontier: the Jin copied it, and the Yuan and the Qing both governed their composite empires on some version of the same principle.

What being the senior power looked like

By the eleventh century the Liao held the steppe frontier from Manchuria to the Altai, took tribute from Korea and from the Tanguts, received Song silver and silk annually under a treaty its own armies had extracted, and had a court that foreign envoys attended.

They were also, by then, the state that everyone west of them meant by China. The word travelled: Khitan becomes Kitai in Russian, and Cathay in Latin Europe, and it stayed the western name for the country for six hundred years after the dynasty that carried it had ceased to exist.

Swine and deer

Drafted from Parker, A Thousand Years of the Tartars (1895)expansion

Before the Khitan were an empire they were one of a pair of tribes on the Sira Muren, and the Chinese record of them begins with a funeral.

The prayer

The Khitan and the Kumo Xi both laid their dead in the mountains, among the trees or hung upon them. Of the Khitan it is added that after three years the bones were collected and burned, a libation of liquor was poured, and this prayer was said:

In winter at noon may I eat towards the south, in summer towards the north, and always find plenty of swine and deer in my hunts.

Parker seizes on the pigs, because they place the people. The Tungusic forest peoples kept swine and the Turkic and Mongol steppe peoples did not; a tribe whose idea of a good afterlife includes hunting pig is a forest people, not a steppe one.

Everything else about the Liao follows from that. They were not nomads of the open grass like the Xiongnu or the Türk. They came out of the forest and the river valleys of what is now Manchuria, they had agriculture and pigs as well as horses, and they were therefore able to do a thing that pure steppe powers never managed: hold farmland and work it.

The record before the record

They appear in the Chinese histories as a nuisance and then as a client.

The Xianbei had pushed them and the Xi north of the Sira Muren; the Tuoba emperor punished the Xi for raiding in 388; from 440 the Khitan and the Xi sent regular tribute of horses; in 479 they moved south under a chief Parker renders Bagatur Murkan to the Loha river. In 493 the Xi raided and were beaten, forgiven, allowed to trade inside the Wall, and thereafter sent horses and sables regularly. In the sixth century the Khitan had to be punished again, separated from the Xi for twenty years, and went back north of the river; and the Northern Qi built three hundred miles of new Great Wall from the Nankou pass to Datong to keep them out.

Five hundred years of that, before anything that could be called a state.

Why it kept happening

Parker's explanation of the mechanism is the most useful thing in his book, and it applies to every polity in this part of the record.

The dynasties of North China that dealt with the Khitan — Tuoba, Yuwen — were themselves the more civilised tribes of the same race. Among the eastern Tungus the pattern repeats exactly: the earlier-civilised group founds a kingdom, the less-civilised founds an empire after it, and the most obscure tribe of all eventually rears its head over both. That sequence runs Bohai, then the Jurchen, then the Manchus — three states in eight centuries out of the same forest, each raised by a people the previous one had regarded as backward.

Horse-back forms the state.

It is always a petty tribe which, under some hero or some great provocation, takes the lead and gives a new name to a kingdom. Otherwise there is no idea of nationality, and things remain much the same.

The Manchus, when they had conquered China and Mongolia, decided that the Solon of the Amur were the descendants of the ancient Khitan — which is to say that the imperial dynasty of the seventeenth century looked at the imperial dynasty of the tenth and identified its survivors among the hunting bands of the far north.

What they were when Abaoji found them

Eight tribes, federated, electing a leader for a fixed term.

That is the condition out of which every steppe and forest empire in this history begins, and the thing that has to be broken before anything larger can be built. The Khitan had been in it, on and off, for five centuries — long enough to have a tributary relationship with four Chinese dynasties, a settled quarrel with their Xi cousins, and no state at all.

What changed was one man deciding not to stand down at the end of his term.

Governing two things at once

Drafted from (eds), The Cambridge History of China, Volume 6: Alien Regimes and Border States, 907-1368 (1994)peak

The Khitan were a nomadic people of the Manchurian steppe whose ruler Abaoji declared himself emperor in 916 and whose successors took sixteen prefectures of northern China from a collapsing Tang successor state in 938. That left them governing pastoral nomads and Chinese farmers in the same polity, which is the recurring problem of every conquest dynasty on the Chinese frontier.

Their answer was to not integrate them. The Liao ran a Northern Administration for the tribal population on steppe principles and a Southern Administration for the Chinese population on Tang bureaucratic lines, with separate personnel, separate law and a mobile court that moved between five capitals on a seasonal circuit. It is a deliberate refusal of the assimilation that had dissolved previous steppe conquerors into Chinese officialdom.

They also commissioned two scripts of their own, a large one adapted from Chinese characters and a small one influenced by the Uyghur alphabet, neither of which is fully deciphered. Two writing systems designed to order, by a state, for a language that had never been written — which is a thing almost no dynasty has ever bothered to do, and which tells you what the Liao thought they were.

The Song paid them an annual tribute in silver and silk from 1005 to keep the peace, and described it in their own records as a gift.

How the two halves actually worked

The Northern Administration governed the tribes: the Khitan proper, the Xi, the Jurchen and the rest, by tribal units under their own chiefs, judged by their own customary law, liable for military service and for tribute in kind, and answerable to officials with Khitan titles.

The Southern Administration governed the Chinese and Bohai population of the sixteen prefectures: prefectures and counties, salt and iron, land tax and corvée, civil examinations, Tang statute law, and officials with Tang titles, many of them Chinese.

A man's law depended on what he was, not where he lived. The two systems had separate treasuries, separate archives and separate career ladders, and the emperor stood on top of both as khagan to one and huangdi to the other.

Why they refused to merge them

Because every previous conquest dynasty on that frontier had merged them and had ceased, within a few generations, to have an army.

The Tuoba Wei is the case everybody in North China knew. A Xianbei conquest state that took the Yellow River plain, moved its capital to Luoyang, adopted Chinese dress, Chinese surnames and Chinese court ritual by decree, forbade the Xianbei language at court — and within thirty years faced a mutiny of its own frontier garrisons, who had been left on the steppe and left behind, and which destroyed it.

The Liao answer was that a conquest dynasty must remain two things at once, permanently and by design, because the moment it becomes one thing it becomes the weaker one.

It held for two hundred and nine years, which is longer than any of the dynasties that took the other road, and the Jurchen who destroyed it and the Manchus who eventually took all of China both built systems that were recognisably the same idea.

The ordo

Drafted from (eds), The Cambridge History of China, Volume 6: Alien Regimes and Border States, 907-1368 (1994)peak

The Liao raised its armies from tribal levies, which is true of every steppe state and explains nothing. What was unusual is what happened to them afterwards.

The Liao emperors each raised an ordo: a standing household guard, drawn from tribal levies and from captured populations, with its own herds, its own territory and its own revenue. It went to war with him, and when he died it did not pass to his successor. It became the permanent establishment of his tomb.

So each emperor built an army from scratch, and the dead accumulated armies.

What that is for

It looks like waste and it is a solution to a specific problem. A steppe ruler whose guard passes intact to his heir has handed the heir's rivals a target and handed the guard commanders a vote in the succession — which is exactly what happened to the Yuan, whose guard commanders murdered an emperor at the summer camp in 1323 and did the same sort of thing repeatedly for thirty years.

An ordo that dissolves into a tomb establishment cannot be inherited, cannot be bargained with, and cannot make an emperor. The cost is that every reign begins with the ruler recruiting.

The succession itself was never fixed. There is a Chinese-style heir apparent in the arrangements and a steppe assembly behind it, and the empress dowagers of the Xiao consort clan — which supplied every Liao empress and much of the officer corps — settled several transitions by force. The ordo did not solve that. It made sure the army was not the thing that decided it.

And two revenue systems

The Song payments are the famous half of the Liao's income and they were never the larger half.

The Southern Administration taxed a settled Chinese and Bohai population on agricultural land, in the ordinary way, and that population was the majority of the empire's people. An empire whose whole subject is governing two things at once had two treasuries as well as two governments, and the tribute was the one that got written about because it was the one the Song had to explain to itself.

Calling tribute a gift

Drafted from (eds), The Cambridge History of China, Volume 6: Alien Regimes and Border States, 907-1368 (1994)aside?What aside means

The Treaty of Chanyuan in 1005 ended a war between the Liao and the Song with an agreement that the Song would pay the Liao an annual sum in silver and silk. The Song called it a gift. The Liao called it tribute. Both recorded the arrangement in their own terms and neither corrected the other, and it held for a hundred and twenty years.

The Chinese diplomatic tradition had a single template — the tribute system, in which foreign rulers acknowledged the emperor's superiority and received gifts in return — and Chanyuan does not fit it. The two courts addressed each other as formal equals, with a fictive kinship between the emperors, negotiated borders, and regular embassies. It is the first treaty of its kind in Chinese history and it was made because the Song could not win the war and the payments cost substantially less than continuing it.

That calculation is worth stating plainly, because the Song are conventionally remembered as weak for making it. The annual payment was a small fraction of the Song state budget and a large fraction of Liao revenue, the frontier stayed quiet, and both economies profited from the border markets it opened.

The arrangement's real cost was strategic and arrived a century later, when the Song allied with the Jurchen to destroy the Liao — as the Jin's own record describes — and discovered they had removed a neighbour that took payment in exchange for one that took Kaifeng.

A hundred thousand ounces

Drafted from Parker, A Thousand Years of the Tartars (1895)peak

In 1042 the Liao emperor massed troops in the Peking plain, demanded territory from the Song, and was talked out of it by an envoy who explained to him that his own ministers wanted a war and he did not.

The demand

Xingzong — Parker gives his Khitan household name as Mukpuku — saw an opportunity while the Song were occupied with the Tanguts of Xia, and wrote to his elder brother the emperor of China.

The letter reopened the question of the ten districts the Later Zhou had recovered, complained about the Song attempt on the Peking plain in 986, complained that two of the districts were still in Chinese hands, and complained about a frontier governor making strategic use of a river near Bazhou.

The Chinese saw through it at once. Parker's judgement is blunt: the Khitan envoys, while ostensibly claiming territory, were not skilful enough to disguise the fact that larger annual subsidies were really what they wanted.

The argument

The Song envoy Fu Bi found the emperor at his nabo — his seasonal hunting camp — living in a travelling tent, and made an argument in three moves.

First, gratitude: after the previous war the Chinese emperor, against his generals' advice, had let the Khitan army withdraw unmolested.

Second, arithmetic. When Cathay had taken a Chinese capital in the tenth century, what had it got? The monarch who did it lost more than half his best troops and horses and died himself; the only gain was that the survivors filled their pockets and stuffed their houses with plunder. China has since recovered its territory and has millions of seasoned troops, so victory is uncertain — whereas further loss of Cathayan life is a matter of complete certainty.

Third, and this is the one that landed: whose war is it?

In the event of peace being preserved, not only will the monarch not lose his soldiers' lives, but all the subsidy goes to himself alone, whilst his ministers get nothing but what they can make out of being occasional envoys. Thus it is plainly the interest of the counsellors to have war and plunder, the ruler losing his men and his subsidy.

A Chinese diplomat, in a tent on the steppe, explaining to a Khitan emperor that the men advising him to fight are the men who do not receive the tribute.

The offer

Then the settlement, and it is the most candid sentence in the whole exchange.

The emperor's position, Fu Bi said, was that the desire for the land was at bottom a desire only for the profits on the land, and that he would not spend lives on a question of profits. He therefore increased the subsidy by the amount of taxation leviable on the disputed territory.

Buy the revenue and keep the ground. If Cathay insisted on the land itself, that was evidently a pretext for overturning the treaty of 1005, and in that case his majesty could not refuse war.

The counter-offer

Xingzong was visibly impressed and proposed a marriage alliance instead.

And here Fu Bi, seeing his advantage, said the thing that shows exactly what kind of relationship this had become:

Marriage connections easily give rise to disputes, and are not so durable as subsidies. A hundred thousand ounces of silver is the dowry given with a first-class princess, which is immeasurably below the value of an annual subsidy.

Do not take a wife. Take the money; there is more of it.

The emperor said: go back, and when you return I will tell you which I will have — but bring your full powers with you.

What the negotiation shows

That by the eleventh century the Song and the Liao had a relationship neither of them could describe honestly and both of them understood perfectly.

To the Song court the payments were gifts to a younger brother, entered in the records in language that preserved the fiction. To the Liao they were tribute from a tributary. In practice they were a standing purchase of peace, priced against the tax yield of the disputed districts, negotiated by men who compared them openly with the cost of a princess's dowry.

It worked. The frontier held for a hundred and twenty years, the longest peace between a Chinese dynasty and a northern power in the whole of this history, and both states grew rich in it. What it could not survive was the arrival of somebody who had not been paid — which is what happened when the Jurchen of the forest, whom the Liao had been taxing in falcons and pearls, decided they would rather have the empire.

The forest people who supplied the falcons

Drafted from Bretschneider, Mediaeval Researches from Eastern Asiatic Sources, volume I (1888)contraction

The Liao was destroyed by a subject people it had been taxing in hunting birds, and the sequence took about ten years.

What the Jurchen were for

The Jurchen lived in the forests and river valleys of eastern Manchuria, north of the Bohai territory the Liao had annexed in 926. They were not one people; the Liao distinguished the "civilised" Jurchen who had been resettled inside the frontier from the "wild" Jurchen who had not, and dealt with the second group through their own chiefs.

What the Liao wanted from them was tribute in kind: furs, pearls from the Manchurian rivers, horses, ginseng, and above all gyrfalcons — the haidongqing, the white sea-eagle falcon of the far north-east, which the Liao court used for hunting swans and which had to be taken from nests on the coast and carried south alive.

The demands escalated. The accounts describe Liao commissioners going into Jurchen country each year for falcons and pearls, quartering themselves on the villages, and behaving in the way tribute collectors behave when there is no appeal. It is the ordinary grievance of an imperial periphery and it is usually survivable.

The dance

The story that everybody tells about the break is a scene at a fishing camp.

The Liao emperor Tianzuo held the annual first-fish feast on the Sungari, at which the Jurchen chiefs attended and were required in turn to rise and dance before him. Aguda, chief of the Wanyan, refused.

The emperor's advisers told him to have the man killed. He did not, on the grounds that it was a small matter.

That anecdote has the shape of a story constructed afterwards, and it is recorded in the histories of the dynasty Aguda founded, which is exactly the kind of source that would construct one. What it is doing is real enough: it identifies the moment at which the tribute relationship stopped being obeyed and marks that the centre noticed and did nothing.

Why the centre did nothing

Because it was busy, and because it had stopped being able to tell the difference between a frontier grievance and a threat.

Tianzuo had come to the throne in 1101 after a succession full of purges. He hunted, extensively — the accounts of his reign are substantially accounts of where he was hunting when news arrived. The Northern Administration that governed the tribal periphery had been drifting for a generation, and the court's attention was on the Song frontier and on court politics.

The structural problem underneath is the one the two-administration system never solved. It was designed to govern two populations, the steppe and the settled, and it worked. The Jurchen were neither: forest hunters and farmers on the eastern edge, administered as tributaries by a Northern Administration built for pastoralists, at a distance, through their own chiefs — which meant through a leader who could turn the whole arrangement into an army the moment he decided to.

1114

Aguda attacked in 1114 with, by the histories' count, about two and a half thousand men, and won. He took Ningjiang, then beat a much larger Liao force at the Chu river, and declared himself emperor of a dynasty called Jin — Gold — in 1115.

The Liao had eleven years left. What follows is the fastest collapse of a major state in the region's history, and the reason it is so fast is that a composite empire comes apart along the seams it was assembled on.

1114

Drafted from (eds), The Cambridge History of China, Volume 6: Alien Regimes and Border States, 907-1368 (1994)end

The Jurchen were a subject people of the north-eastern forests, organised in tribes, paying the Liao in furs, horses, pearls and gyrfalcons. In 1114 they revolted. Eleven years later the Liao empire did not exist.

What makes the sequence worth a chapter is that it then happened again, twice, on schedule.

• 1114: the Jurchen, a forest people the Liao extracted tribute from, destroy the Liao in eleven years.
• 1211: the Mongols, a steppe people the Jin extracted from and manipulated, destroy the Jin in twenty-three.
• 1616: a Jurchen leader in the same forests declares a state, names it after this second dynasty, and his son's dynasty takes China in 1644.

Three times in five centuries, a conquest dynasty holding north China was ended by a population on its own frontier that it had been managing as a resource.

Why the same shape

The Liao and the Jin were both governing two things at once: a settled Chinese population under a Chinese administration, and a steppe or forest constituency under its own. The arrangement worked, and it required that the outer constituency stay outer — tributary, divided, and supplying men and goods.

Managing a frontier people takes attention, and the machinery that does it is also the machinery that teaches them how the empire works. The Jurchen who revolted in 1114 had been trading with, campaigning for and being administered by the Liao for two centuries.

Calling it a conquest is true and it misses the mechanism. The conqueror was a component of the thing conquered — a people the empire had built into its own administrative structure as a supplier, who had spent two hundred years learning how the machine worked from inside it.

How fast it went

The speed is the part that needs stating, because it is barely credible.

In 1114 Aguda, chief of a Jurchen tribe that had been supplying the Liao court with falcons and river pearls, refused to perform the dance at the emperor's fish feast and went to war. He had, on the Jurchen account, a few thousand men.

In 1115 he declared himself emperor of a dynasty he called Jin, gold — chosen, the story goes, because iron rusts and gold does not, and Liao was said to mean iron.

By 1120 the Song had agreed to attack the Liao from the south in concert with him. By 1122 the Liao capitals were falling. In 1125 the last emperor was captured in the west, and the dynasty that had taken tribute from China for a hundred and twenty years had ceased to exist.

Eleven years, from a refusal to dance to the end of an empire.

Why the Song joined in

Because they wanted the sixteen prefectures back, and this looked like the moment.

The alliance of 1120 is one of the worst decisions in Chinese diplomatic history and it was taken for an entirely intelligible reason: the territory had been lost in 938, every Song emperor had been expected to want it back, and here was a chance to get it without facing the Liao army alone.

What the Song got was the destruction of the buffer state that had been keeping a forest people out of north China, and a demonstration — conducted in front of the Jurchen — that its own armies could not take walled cities the Jurchen took in weeks. Within two years of the Liao collapse the Jin turned south, and in 1127 they took the Song capital and carried the emperor, the retired emperor and three thousand of the court into captivity in Manchuria.

The Liao had cost the Song two hundred thousand ounces of silver and two hundred thousand bolts of silk a year. The peace that replaced it cost them the north of China.

The half that went west

Drafted from (eds), The Cambridge History of China, Volume 6: Alien Regimes and Border States, 907-1368 (1994)afterlife

When the Jin took the Liao capitals, a member of the imperial clan named Yelü Dashi took what he could gather, went west across the steppe, and founded a state in Central Asia that outlasted the one he had come from: the Qara Khitai.

It lasted about ninety years, held Transoxiana and the Tarim basin, beat a Seljuq army at the Qatwan steppe in 1141, and was ruled by Buddhists over a largely Muslim population, which almost never happens in that direction.

They also kept the dynastic name. Two thousand miles from the nearest Chinese-speaking administration, ruling Turks and Persians and Muslims who had no reason to be impressed, the rulers used Chinese reign titles, struck coin in Chinese form, ran a chancery in Chinese, and went on calling themselves Liao.

And the other thing that went west

The name Khitai travelled with them, and travelled further than they did. It is how Central Asia, Persia and eventually Russia referred to China — Cathay in Marco Polo, Kitai in Russian today.

A dynasty that ruled the northern part of China for two centuries and never held the whole of it gave its name, in half the languages of Eurasia, to all of it, and kept it for six hundred years after the last man who could have objected was dead.