fragmentation842Beckwith, The Tibetan Empire in Central Asia (1987)
How it was governed
0 of 4 coded
Army raised byHow the fighting force was recruited, not how it fought.
No cited figure
Revenue fromWhat the state lived on. The values are unranked: the field cannot express which mattered most.
No cited figure
Succession byThe rule in force, not the outcome. A rule repeatedly violated is still the rule.
No cited figure
Right to rule fromThe ground publicly asserted — on coins, in the khutba, in titles.
No cited figure
Each line is drawn from a source that addresses the question directly. A line reads No cited figure where no source consulted does — never because the answer seemed obvious.
Turning points
No dated event is recorded as having changed this polity’s course. Not every history has one: a hinge is something some polities turn on and others simply do not.
Chapters
8 in order
The golden goose
Drafted from Bushell, The Early History of Tibet, from Chinese Sources (1880)formation
The Tibetan empire enters the written record in the seventh century as a
frontier problem in a Chinese archive, and the way the Tang historians tell it
is worth reading exactly, because they are describing a man they thought they
had absorbed.
How to obtain a princess
Songtsen Gampo — Lungtsan, in Bushell's transcription of the Tang text — asked
for a marriage alliance and was refused. So he invaded, carried off the herds
of a neighbouring people, and stayed in the field for several years without
going home; his chief ministers begged him to return, and when he would not
listen, eight of them killed themselves.
Then he sent a mission to apologise for his misdeeds and asked again, with five
thousand ounces of gold and several hundred precious articles. In 641 the
emperor gave him the Princess Wencheng.
That is a Chinese court history admitting, in its own words, that a marriage
alliance was extracted by force and then paid for.
What he did with her
The account of what followed is the part a Chinese historian is least likely to
have invented, because it is flattering to China and factually specific.
He met her escort in person and received the presiding prince with the rites
due from a son-in-law. Afterwards, the history says, he praised the costume of
the great empire, and the perfection of their manners, and was ashamed of the
barbarism of his own people. He told his relatives that no ancestor of his had
ever been allied to the sovereign empire, and that he would build a walled city
for the princess to proclaim his glory to later generations — and did, with a
palace for her inside the walls.
She disliked the Tibetan custom of painting the face red, so he ordered his
people to stop, and they stopped. He put off felt and skins and put on brocade
and silk. He sent the sons of his chiefs and his rich men to ask admission to
the Chinese national schools to study the classics, and invited scholars from
China to compose his official despatches to the emperor.
The goose
The despatches were worth having. When Taizong came back from campaigning in
Korea, Songtsen Gampo's minister was sent with congratulations and this:
The wild goose flies most swiftly, but not when compared with the speed of the
emperor. His slave and unworthy son-in-law, rejoicing a hundred-fold above
ordinary barbarians, in that the common goose is not unlike the wild goose,
has made a gold goose, and respectfully offers it.
The goose was moulded of pure gold, seven feet high, and held about ten gallons
of wine.
Read the letter and the object together and they are not the same document. The
letter is elaborately abject — his slave and unworthy son-in-law, above
ordinary barbarians — in the precise register a Chinese chancery expects, which
is exactly what you would produce if you had hired Chinese scholars to write
your despatches. The gift is a seven-foot wine vessel of solid gold from a
country the Tang had just refused a princess.
What he was actually building
Within seven years of the marriage, in 648, a Chinese envoy on his way to India
was robbed there, and it was Tibetan troops who marched into India with him,
defeated the offenders and sent word of the victory to the Chinese emperor. The
next emperor gave Songtsen Gampo the title Prince of the Western Sea and two
thousand pieces of silk. He replied offering to lead his warriors into China
personally to punish any officer who proved disloyal.
That is not a client. That is a power positioning itself as the guarantor of a
neighbour's order, in the vocabulary of a son-in-law, six years after being
told it was not good enough for a princess.
Under the same reign an alphabet was devised for Tibetan on an Indic model, and
the country acquired the ability to write itself down — which is the more
consequential of the two things Songtsen Gampo imported, and the one the Tang
historians do not mention.
Fifteen days in Chang'an
Drafted from Bushell, The Early History of Tibet, from Chinese Sources (1880)expansion
In the tenth month of 763 a Tibetan army walked into the capital of the Tang
empire, deposed the dynasty, appointed a new emperor with a new reign title and
a full slate of ministers, and went home a fortnight later.
The Tang histories record all of it, because they had to.
How it happened
The empire was still broken from the An Lushan rebellion. Tibetan forces took
Jingzhou, whose governor Gao Hui surrendered to them; the next month they
invaded Binzhou. The great general Guo Ziyi was sent west to hold them, and had
to turn back when two hundred thousand Tuyuhun and Tangut troops came in from
another direction.
Then the emperor's carriage left for Shangzhou and the capital was abandoned.
The surrendered governor led the Tibetans in.
Inside the city, with the Tibetan commander-in-chief, they raised the Prince of
Guangwu to the throne, gave his reign the title Dashe, and appointed the various
officers of state.
What the empire looked like that autumn
Bushell's translation keeps the detail, and the detail is the humiliation.
When the emperor first went east, the officers and their families fled south or
went into the mountains to hide, and the soldiers of the imperial armies broke
up into armed bands blocking the roads. Guo Ziyi — the most celebrated soldier
of the dynasty — was in a mountain valley with a few hundred followers, his
wives, children and slaves, and some hundreds of camels, horses, carts and oxen,
not knowing which way to turn, until two of his officers came and told him he
ought to be ashamed of himself.
He agreed, and took a back road because the main one would be watched. His
column of a thousand men was strung out over a hundred li because the mountain
track was too narrow, and two of his officers left by a different route in case
of pursuit, since in a defile the van could not help the rear.
Meanwhile the general of the six armies had fled the capital with his own
standard and spent several days plundering the fleeing court officials and
robbing the townspeople of money, saddles and horses, until he too was talked
into behaving.
That is the strongest empire in the world, described by its own historians, in
the fortnight after a Tibetan army took its capital.
Then they left
The Tibetans occupied the city fifteen days and retired, and the imperial army
came back.
It is easy to read the withdrawal as a failure of nerve, and it was not. An army
of this kind could not garrison Chang'an: it was months from its own supply,
in a river valley in autumn, with a hostile countryside and no interest in
administering a Chinese province. What it could do was demonstrate — publicly,
in front of every polity in Inner Asia — that the Tang could be reached.
That demonstration is why the empire spent the following century holding the
Gansu corridor and the Tarim oases at Chinese expense, and why the Uyghurs
could charge the price they charged for cavalry.
Three powers, not two
The period between about 660 and 840 is usually told twice, once from Chang'an
and once from Baghdad, and each account has a frontier that stops where the
other begins. Between them was a third empire fighting both.
Tibetan armies took Tang oases in the Tarim, met the caliphate's forces in the
Pamirs and Transoxiana, and at different moments allied with each against the
other. The Karluks, Türgish and Uyghurs in between made and unmade arrangements
with all three.
Neither surviving imperial tradition has an obvious place for that, which is
why a fortnight in 763 tends to appear in accounts of the Tang as a raid, and
in accounts of Islam not at all.
The third power
Drafted from Beckwith, The Tibetan Empire in Central Asia (1987)peak
At its height the empire held the Tarim basin oases, the Gansu corridor, much
of what is now Qinghai and the approaches to the Pamirs, and fought both of the
great settled powers of the age at once.
The Abbasid frontier in Central Asia was not simply the far edge of the
caliphate; it was one face of a three-cornered contest, and the Karluk, Türgish
and Uyghur groups in between made and unmade alliances with all three. The
period is usually narrated twice — once from Chang'an, once from Baghdad — and
each version has a frontier that stops where the other begins.
The empire's other creations are a script and a religion. Songtsen Gampo's reign
saw an alphabet devised for Tibetan on an Indic model — a writing system made for
a language that had none. Trisong Detsen made Buddhism the religion of the court.
It broke apart after Langdarma's assassination in 842, and nobody conquered
it.
Which Buddhism
Drafted from Beckwith, The Tibetan Empire in Central Asia (1987)aside
Trisong Detsen made Buddhism the religion of the Tibetan court in the second half
of the eighth century and founded the monastery at Samye in the 770s. Then he had
to decide whose Buddhism it was going to be, and the choice was between his two
largest neighbours.
The question
Tibet sat between India and China and could have taken the religion from either.
The Chinese route was the obvious one: the empire had been dealing with the Tang
for a century and a half, it had captured Chinese territory full of Buddhist
establishments, and Chan teachers were already in the country. The Indian route
meant bringing scholars over the Himalaya and working in a language nobody in
Tibet had grown up speaking.
The tradition records a debate at Samye, placed in the 790s, between an Indian
monk arguing the gradual path of study and accumulated merit and a Chinese Chan
master arguing sudden enlightenment. The Indian side is recorded as winning.
Whether it happened as a single staged event is doubted, and the sources for it
are Tibetan and later and have an interest in the outcome. That a choice was made
in that direction is not in doubt at all, because everything downstream of it is
visible.
What a king was choosing
It is worth being clear that this was not only a doctrinal question, because the
doctrinal question is the part that has survived and the political one is the
part that would have decided it.
Trisong Detsen was at war with the Tang for most of his reign; his armies took
Chang'an in 763 and he spent decades holding the Gansu corridor against Chinese
attempts to recover it. Adopting the religious establishment of the state he was
fighting, staffed by its clergy and reading its commentaries, would have put a
Chinese-trained institution at the centre of his court.
India was not a rival. There was no Indian empire on his frontier, no Indian army
he was going to have to fight, and no Indian court whose prestige his own
nobility might prefer to his. Importing from India meant importing scholars
without importing a patron.
What it committed the state to
Everything that followed was expensive.
Taking the religion in Sanskrit rather than in Chinese translation meant a
translation programme built from nothing: Indian panditas brought in and
maintained, Tibetan translators trained in a language with no prior tradition of
scholarship in it, and — the decision that made the whole thing work — a
standardised technical vocabulary decreed by the court, so that a given Sanskrit
term produced the same Tibetan word in every text by every translator.
A state deciding by edict what words mean, and enforcing it across a body of
translation running to thousands of volumes, is an administrative act on the
scale of a tax reform. It was paid for out of the same revenue as the armies in
the Tarim basin.
Nobody involved was buying an institution that would outlast the court. That is
simply what they got.
Like unto these victims
Drafted from Bushell, The Early History of Tibet, from Chinese Sources (1880)aside
The Tang histories carry a description of how the Tibetan state worked, compiled
by people who had to deal with it and who wrote down the parts a foreign ministry
needs to know. It is the only account of the empire's institutions written while
it existed by anybody who was not inside it.
The machinery
The sovereign is the tsanpo — from tsan, a famous hero, and pu, a man. His
consort is the momeng. Under him are ministers styled great lun and small
lun, and the histories set out the whole establishment: a chief minister and
his assistant, a commander-in-chief, a chief minister of the interior with an
assistant and a lesser minister, a chief consulting minister with the same two
grades under him, the whole body named collectively in one Tibetan phrase the
Chinese transcribe and do not translate.
Then the observation that a Chinese archivist would find worth making:
Although there are officers, they are not constantly employed, being only
appointed when there is stress of business.
An establishment that exists on paper and is staffed when required is a
description of a state run by an aristocracy rather than a bureaucracy, from the
point of view of the most bureaucratic government on earth.
The communications are the part they admired. A gold arrow seven inches long
served as a sign of office and was used to call warriors out. There was a post
station every hundred li; if the war was important the courier carried a silver
hawk on his breast, and if it was urgent, several hawks. When the country was
invaded, smoke-fires were lit from towers a hundred li apart.
No writing
The same passage states, flatly, that the Tibetans have no written characters,
and that notched pieces of wood and knotted strings are used in covenants.
That cannot be true of the empire the Tang were fighting. An alphabet was devised
for Tibetan on an Indic model under Songtsen Gampo; the treaty of 821 was cut in
Tibetan on a stone pillar; and the imperial chancery was producing documents in
Tibetan that survive.
What the sentence preserves is an earlier notice, copied forward into an account
of a later state — which is how these compilations were made, and a useful
warning about the rest of the description.
Punishment and hospitality
Two details sit next to each other and are worth keeping together.
Their punishments are most severe, and even for small crimes the eyes are
scooped out, and the nose cut off, or stripes inflicted with a leather whip.
They differ according to caprice, there being no fixed code. They imprison men
in holes several tens of feet under the ground, and release them only after two
or three years.
And:
When they entertain envoys from foreign countries, they always bring out a yak
for the guest himself to shoot, the flesh of which is afterwards served at the
banquet.
The first is an accusation — a Chinese historian's standard complaint about a
government that has punishments without a written code, which is the thing his
own state most prided itself on having. The second is a piece of protocol
somebody experienced and reported, and it is the only glimpse anywhere of what
being received at a Tibetan court was like.
The oath
The central institution, and the one the Chinese describe at most length, is a
ceremony.
Once a year the officers were assembled for the lesser oath of fealty. Sheep,
dogs and monkeys were sacrificed — their legs broken first, then killed, the
intestines exposed and cut into pieces. The sorcerers were summoned and called
on the gods of heaven and earth, of the mountains and rivers, of the sun, moon,
stars and planets:
Should your hearts become changed, and your thoughts disloyal, the gods will
see clearly and make you like these sheep and dogs.
Every third year came the great ceremony. All assembled in the middle of the
night on a raised altar spread with savoury meats. The victims were men, horses,
oxen and asses, and the prayer ran:
Do you all with one heart and united strength cherish our native country. The
god of heaven, and the spirit of the earth, will both know your thoughts, and
if you break this oath they will cause your bodies to be cut into pieces like
unto these victims.
What that is a description of
Not barbarism, whatever the compiler thought. It is a constitutional mechanism.
An empire with no standing bureaucracy, whose officers are appointed when there
is business and whose army is summoned by sending out a gold arrow, is held
together by the personal obligation of a few hundred aristocrats to one man. The
Chinese state bound its servants with an examination system, a code and a
salary. The Tibetan state assembled them once a year and, every third year at
midnight, showed them what they would look like if they broke their word.
For two centuries it worked better than the alternative. When it stopped working
— when the great families took their own regions after 842 and stopped taking
orders — there was nothing underneath it, because the oath was the
administration.
A horse with no bit
Drafted from Bushell, The Early History of Tibet, from Chinese Sources (1880)contraction
The two empires swore treaties at each other for two centuries. The Tang
histories record the ceremonies in detail, including the negotiations about what
to kill and the occasion on which the ceremony was an ambush.
What to sacrifice
At the peace of 783 the two delegations agreed the shape of the thing in advance:
two thousand men each, half armed and drawn up two hundred paces from the altar,
half unarmed below it; seven men a side in court costume to go up and swear.
It had been agreed that the Chinese would sacrifice an ox and the Tibetans a
horse. Then the Chinese envoy, Zhang Yi — who, the history says, was ashamed of
his part in the ceremony and wanted to depreciate the rites — proposed a
substitution:
The Han cannot cultivate the ground without oxen, the Fan cannot travel without
horses, I propose therefore to substitute a sheep, pig and dog as the three
victims.
The Tibetan minister agreed. But there were no pigs to be had outside the
frontier, so he took a wild ram instead and the Chinese took a dog and a sheep.
They were killed on the north side of the altar, the blood mixed in two vessels
and smeared on the lips of the swearers.
An envoy embarrassed to be treating with barbarians downgrades the victims on a
pretext about agriculture, and the other side, being on the wrong side of a
frontier with no pigs, improvises. That is what two empires negotiating as equals
looked like from close up.
Pingliang
Four years later they arranged to swear again, and the Chinese chose the ground
carefully.
The Tibetan minister had proposed a valley called Tulishu. The general Ma Yulin
reported that the Tulishu country was full of difficult passes and that he feared
a secret ambush; Pingliang, by contrast, was a plain open on every side and close
to a Chinese garrison town. The court agreed and moved the ceremony there, and
had to send after the Tibetan envoy, who had already left, to tell him.
At Pingliang the two commanders agreed three thousand men a side drawn up east
and west of the altar, four hundred unarmed men below it, and mounted scouts on
both sides to watch each other.
Then the Tibetan minister massed several tens of thousands of picked horsemen
west of the altar. His scouts rode through and through the Chinese ranks. When
the Chinese general sent sixty horsemen into the middle of the Tibetan army they
were seized and bound on the spot.
Hsien had provided for nothing.
A message came across asking that the minister and his subordinates put on their
official robes and hats, swords and jewels, and wait for instructions — this was
to induce them to dismount from their horses so that they might be captured.
They went into the tent together, calm and unsuspecting. The Tibetan commander
had the drums beaten three times and his army came on shouting.
The ride
The Chinese commander Hun Jian got out of the back of the tent, caught a loose
horse and rode.
The horse had no bit in its mouth, and Hsien had to lean over its mane to guide
it with his hands, and it was only after he had ridden for several miles that
he got the bit into its mouth, and it was owing to this that the arrows of the
pursuing horsemen passed by without wounding him.
One of his generals collected a few hundred men on a hillock to the north and
fought until the whole army surrounded him. The treaty officials were taken.
Why the record kept it
A court history is not obliged to preserve an episode in which its own side
chooses the ground for safety, agrees the numbers, walks into the tent, and is
taken. The Tang histories preserve this one in circumstantial detail, down to the
horse's missing bit.
Part of the reason is that Hun Jian survived to report it and somebody had to be
blamed; the sentence about providing for nothing is a verdict. But the larger
part is that the treaty system itself mattered. The Tang had been making sworn
agreements with the Tibetans since the seventh century and would make another
thirty-five years later — the one cut on stone in Lhasa, with its two sovereigns
and its single border. A record of what happened when the ceremony was used as a
trap is exactly what a chancery preserves, and reads.
The stone outside the temple
Drafted from Bushell, The Early History of Tibet, from Chinese Sources (1880)end
In 821 and 822 the Tibetan empire and the Tang concluded the last of their
sworn treaties, and the text is the clearest statement anyone made of what the
two centuries of fighting had established.
The terms
The oath was sworn on an altar in the western suburb of the Chinese capital,
with a pit dug on the north side; the victims were sacrificed and buried with
the written text.
What the Chinese now rule shall have the Tang as the sovereign, and the
country of their western race shall have the great Fan as ruler. From this
time henceforward both shall put by weapons and armour, forget their
differences and old grievances, and respect the honoured kinship of their
sovereigns and the ancient bonds of mutual aid. The frontier guard-houses
shall be left ungarrisoned, and the watch-fires no longer lighted... the
barrier stations and fortifications shall be disused and invasion and plunder
shall cease.
And then the verse the ministers proclaimed:
The frontier mountains are lofty, the River flows on unceasingly: on a
propitious day and favourable season have we fixed the two boundaries, the
west to belong to the great Fan, the east to be ruled by the grand Tang: the
great ministers, holding up the sworn treaty, proclaim it afar to the autumn
country.
Two sovereigns, two territories, one border, no tribute either way. That is
not how the Tang concluded arrangements with frontier peoples. It is how two
empires conclude a peace.
The treaty was ratified in Tibet and cut on a stone pillar, which was set up in
Lhasa outside the Jokhang temple and is still standing.
Twenty years
The empire had about twenty years left.
What ended it is domestic and badly documented. Langdarma was assassinated in
842; the succession was disputed between two claimants and never resolved; the
great families that had supplied the empire's ministers took their own regions
and stopped taking orders; a revolt of the common people in the 860s finished
whatever central administration was left, and the imperial tombs were dug up.
There is no conquering power in it. Nobody took Tibet in 842 or afterwards. The
Tang, which had spent two hundred years failing to, was itself sixty years from
its end and in no condition to try. The Uyghur khaganate on the northern steppe
had collapsed two years earlier, in 840, for reasons of its own.
What went with it
The state religion went. Buddhism had been made the court religion by Trisong
Detsen and endowed on an imperial scale, and imperial endowment is what
monasteries live on. When the court stopped, the monastic establishment stopped
with it, for something over a century.
The empire's other creation did not stop. The translation programme had already
produced a library, and libraries do not need a government.
What came back in the eleventh century was built on the books the empire had
paid for, by men who no longer had any idea how to raise an army.
What the library did
Drafted from Beckwith, The Tibetan Empire in Central Asia (1987)afterlife
The empire fell apart in the 840s and Tibet did not become a state again for
four hundred years. What the empire had built that survived the empire was a
translation.
What the vocabulary did
The decreed vocabulary is the unglamorous part and it is the reason the whole
enterprise worked. It made the body of translation internally consistent across
thousands of volumes and hundreds of translators, and it made the Tibetan
versions reversible: a scholar can often reconstruct the Sanskrit from the
Tibetan word by word, because the same term was never rendered two ways.
Korea and Japan received their Buddhism through Chinese translations, which are
not reversible in that way, having been made by many hands over centuries with no
authority standardising them.
Why that turned out to matter to everybody else
Buddhism disappeared from India. Over the centuries after the empire fell, the
monastic universities were destroyed, patronage went elsewhere, and a very
large part of Indian Buddhist literature was lost in its original language.
The Tibetan canon has it. For a great many Indian Buddhist works — including
central texts of Madhyamaka and Yogacara philosophy, and much of the tantric
literature — the Tibetan translation is the only complete version that exists,
and it is exact enough to be used as evidence for what the Sanskrit said.
So a decision taken by a Tibetan king in the eighth century about which of two
neighbours to import a religion from is the reason a substantial part of Indian
intellectual history is recoverable at all. It is not a Tibetan legacy in the
ordinary sense. It is an Indian legacy that Tibet happens to be holding.
The four hundred years
The intervening period is usually called the era of fragmentation and it is
badly served by the name, which suggests nothing happened.
The imperial line did not end; a branch of it went west and founded the kingdom
of Guge in Ngari, which in the eleventh century paid for the second wave of
translation and invited Atisha from Bengal. The monastic orders that dominate
Tibetan history afterwards — Kadam, Sakya, Kagyu — all form in that period, out
of the second transmission, working from and adding to the books the empire had
paid for.
What did not come back was the state. When Tibet next appears as a political
unit under a single authority, in the thirteenth century, it is as a
territory administered for the Mongols by the abbots of Sakya — a monastery
governing a country on behalf of a steppe empire.
That is the empire's real afterlife and its strangest one. It had made Buddhism
the instrument of the Tibetan state; four centuries later the Buddhist
establishment was the Tibetan state, and the arrangement lasted, in one form or
another, until 1959.
Succession
What led here
Tibetan Empireabsorbed remnants ofZhangzhung
The kingdom of the upper Sutlej and Ngari in western Tibet, with its own language, its own script and the Bon tradition attached to it. Songtsen Gampo married its ruler's sister, and the kingdom was absorbed into the empire during or shortly after his reign; its language survives in a handful of Dunhuang documents and in Bon texts.
What led away
the kingdom of Guge, and the second transmission
No power took Tibet in 842 and none replaced the empire for four hundred years. A branch of the imperial line went west and founded Guge in Ngari, which paid for the second wave of Buddhist translation and invited Atisha from Bengal in the eleventh century; the monastic orders that dominate Tibetan history afterwards all form in that period, out of the books the empire had commissioned. When Tibet is next governed as a unit, in the thirteenth century, it is by the abbots of Sakya on behalf of the Mongols.
Derived from cited spans, not from a separate source. Where both spans are ranges and they meet only on the widest reading of each, the pair is listed as possible rather than resolved to one answer. Unlinked entries are here for scale and are not written up.
Extent
An illustration, not a measurement
Snapshot year
800
This is the nearest available snapshot to the cited peak of 800. It is not the peak.
Cited extent
4,600,000 km²
The area of the shape on the left will not match this figure. That is expected. The cited figure is the figure; the map is an illustration, and the two are never reconciled.
Tibetan Empire, as the snapshot draws it
Everything else on the same snapshot
Edges are blurred from the dataset’s own border-precision field. Every feature in this period is marked 1, approximate, so every border here dissolves.
The dataset author’s caveat, which belongs here rather than in the footnotes: territorial boundary as a concept is meaningful in Europe only after Westphalia, ancient polities overlap, and old vector borders drawn on modern coastlines mislead because rivers and shorelines move. Boundaries from historical-basemaps, CC-BY-4.0.
Rating
Four axes, all computed from cited numbers and never from editorial judgement. Percentiles are against every polity written up here — 280 of them — plus 50 more included for scale. Absolute figures. A missing axis is excluded from the total rather than counted as zero.
Reach
4,600,000 km²
66th percentile
Longevity
224 years
54th percentile
Demographic weight
No cited figure
Influence
Three separate counts. This site never publishes a single influence number of its own; the sliders on the rankings view fuse them only in your view.
Descendant scripts
1
Tibetan script
Religions carried
1
Tibetan Buddhism
Successor claims
No cited figure
Weighted total
69th percentile
computed from 3 of 4 axes, renormalised across the axes that have figures. A polity with two documented axes is not penalised against one with four.