Tang Empire

The second attempt

formationDrafted from Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009)

The Tang is the second dynasty in Chinese history to inherit a unified empire from a short-lived predecessor that had built it and then destroyed itself doing so. The first was the Han, taking over from the Qin. The parallel was obvious to everyone at the time and was discussed openly at court.

What the Sui had done

The Sui reunified China in 589 after three and a half centuries of division, and in thirty years built the physical apparatus the Tang then ran on: the Grand Canal linking the Yangtze rice country to the northern capitals, granaries on a scale that has to be excavated to be believed, a rebuilt Chang'an, the equal-field land system, and a revived examination.

Then Yang Guang invaded Korea three times, lost armies estimated in the hundreds of thousands, exhausted the corvée, and was murdered by his own guard in 618 while the empire came apart in revolt.

The advantage of being second

Li Yuan was a Sui general and a relative of the imperial house who took Chang'an in 617, and his family took the throne the following year. The Tang inherited the canal, the granaries, the land system, the law code and the capital, and inherited them intact — the Sui had been destroyed by its own exertions rather than by an invasion, so the machinery was still there.

They also inherited the lesson, and stated it. The verdict on the Sui was the verdict on the Qin: an empire lost through overreach and the exhaustion of the people. Taizong's court took that as a working constitution, and the Zhenguan zhengyao — the record of his policy discussions, compiled afterwards — is substantially a book about a ruler being contradicted by his officials and accepting it.

How much of that is genuine and how much is a later idealisation is argued about. The relevant point is that it became the model. Every Chinese emperor afterwards was measured against a portrait of a founder who listened.

Where Taizong came from

He had also killed two of his brothers at the Xuanwu Gate in 626 and forced his father to abdicate.

National Palace Museum, Taipei · Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009) · Unknown · Public domain

Both things are in the official record, which is itself informative: Taizong is known to have read and intervened in the court diaries covering the episode, which every historian since has noted, and the episode is still there. What he appears to have secured was not suppression but framing — that his brothers had been plotting first.

The dynasty that produced the most admired model of consultative rule in Chinese history was founded by fratricide in a palace gateway, and both facts were kept.

The shape of what they had

By the 640s the Tang held the north China plain, the Yangtze, the south, and a frontier that ran out into country nobody had governed from Chang'an since the Han.

National Museum of China, Beijing · Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009) · Gary Todd · CC0

The armies were fubing militia: farmer-soldiers holding land under the equal-field allotment, serving in rotation, providing their own kit and going home. It cost almost nothing and was designed so that no commander ever held a permanent army.

That system, and what happened when it stopped working, is the political history of the dynasty.

The monk who left illegally

expansionDrafted from Xuanzang, Si-yu-ki: Buddhist Records of the Western World (1884)

In 629 a Chinese monk left the empire without permission — the frontier was closed, and the prohibition on private travel west was in force — and came back sixteen years later with a description of everything between Chang'an and Bengal. The emperor, who had spent the intervening years conquering a good deal of it, received him personally.

Xuanzang went for scriptures. What he wrote is a gazetteer.

Metropolitan Museum of Art, New York · Xuanzang, Si-yu-ki: Buddhist Records of the Western World (1884) · Unknown · CC0

The road as he found it

The entries have a shape: distance, direction, terrain, the circuit of the capital in li, the soil, the climate, the character of the people, the number of monasteries and monks. It is the register of a man who has walked it.

Going north-west from Aksu, across a stony desert, comes the Ice Mountain, the northern plateau of the Pamirs.

Both hills and valleys are filled with snowpiles, and it freezes both in spring and summer; if it should thaw for a time, the ice soon forms again. The roads are steep and dangerous, the cold wind is extremely biting.

Travellers on that road, he adds, should not wear red garments or carry loud-sounding gourds: the least forgetfulness entails certain misfortune, a violent wind rises with flying sand and gravel, and those who meet it sink from exhaustion and are almost sure to die.

The advice is superstition. The wind is not.

The khan's deer

At the great lake beyond — Issyk-Kul — he records that the khan of the Türks comes every year to escape the heat, and that the place is full of deer wearing bells and collars, tame and unafraid of people, which the khan is fond of and has forbidden anyone to kill on pain of death without remission.

That is the western Türk khaganate at its height, described by a passing monk in one paragraph about a deer park.

Three hundred houses at Talas

The entry that repays the most reading is the shortest.

Talas is a town eight or nine li round where merchants from all parts live with the local people. Ten li to the south, Xuanzang writes, is a little deserted town that once had about three hundred houses, occupied by people of China. They had been carried off violently by the Türks; afterwards they gathered a number of their countrymen and occupied the place in common.

Their clothes being worn out, they adopted the Turkish mode of dress, but they have preserved their own native language and customs.

A colony of Chinese captives on the Talas river, keeping the language and losing the clothes, in about 630. A hundred and twenty years later a Tang army and an Abbasid army fought a battle at the same river.

What the empire did with him

He came back in 645, at the point when Tang armies had destroyed the eastern Türk khaganate, taken Turfan, and were pushing into the Tarim.

Taizong received him, and the sources agree on what he wanted: not the sutras. He wanted the countries — their rulers, their roads, their distances, their strength — and he offered Xuanzang a post in the government, which was declined. What he got instead was the Record of the Western Regions, presented to the throne in 646, compiled at imperial request.

Xuanzang, Si-yu-ki: Buddhist Records of the Western World (1884) · O.Mustafin · CC0

That is the transaction worth noticing. A monk broke the law to leave, and the state that had banned his journey commissioned its results. The most useful survey of Central Asia and India produced anywhere in the seventh century was made by a man on a religious errand and written up for a war ministry.

The Empress Wou

expansionDrafted from Boulger, The History of China, Volume I (1881)

For more than forty years, the person who actually ruled Tang China was a woman the dynasty's own historians could never quite forgive for it.

From convent to consort

Wu had been one of the previous emperor Taizong's own concubines, and after his death had retired, as custom required, into a Buddhist convent. His son and successor Gaozong brought her back to court and, in 655, deposed his own reigning empress to install her in the position instead - over the strenuous objection of ministers who saw exactly where the precedent led. It went further than they feared. Wu is recorded disposing of the two women she saw as rivals for Gaozong's favour by having their hands and feet cut off and their bodies thrown into a wine vat; whatever the literal truth of the detail, it is the story contemporaries told about how she cleared her own path.

She did not stop at the harem. Ministers hostile to her interests were removed from office and, in the accounts that survive, disposed of by the sword or the poisoned cup; her own son was named heir apparent; and Gaozong, whom the record credits with real capacity but no appetite for the daily labour of the throne, found her quick grasp of state business so useful that he let her handle more and more of it, until - in substance, years before it was ever formalised - she was ruling and he was reigning.

A dynasty renamed

Gaozong died in 683 after a nominal reign of thirty-three years. Wu had his eldest son proclaimed emperor and deposed him within days, installing a second son as a figurehead while she conducted the government openly in her own name - receiving petitions, filling offices, building ancestral temples to her own family, and eventually declaring a new dynasty altogether, breaking the fiction of a puppet Tang emperor she had kept up for years. No woman had ruled China in her own right before. She built temples, wore the regalia reserved for an emperor, and offered the sacrifices only an emperor was supposed to offer.

The opposition this provoked was real, and she met it without hesitation. Several risings and assassination plots against her all failed, and her own solution to the flood of accusations her spies generated was characteristically direct: an edict punishing false accusers with death, applied so rigorously that on one occasion eight hundred and fifty of a thousand charges brought before her were found to be fabricated - and their authors executed for it. Executions of the great nobles she distrusted were carried out publicly in the streets of the capital, occasions the chronicler compares, in their effect on the watching city, to the public scaffolds of Tower Hill.

Governing the case for her own rule

None of this was the whole of her reign. She fought Tibet to a standstill on the western frontier, recovering ground lost earlier in the dynasty; she made and then had to fight a Turkic khan she had tried to buy as an ally, after he turned the very arms she had sent him against her own borders; and she worked, deliberately, to build the positive case for her own legitimacy alongside the punitive one - commissioning books celebrating her administration, and instructing the empire's religious establishment to preach her own indispensability to the state's welfare. It is not the record of an idle usurper. Even the chronicler most inclined to condemn her, the record concedes, cannot show her failing any test ordinarily demanded of a competent sovereign.

Boulger, The History of China, Volume I (1881) · Gary Todd · CC0

The fall

In 704, in her eightieth year, illness kept her from her ministers for months at a stretch. Her enemies took the opening: a faction struck down her closest relatives and presented themselves at the palace in force. She had no army left to resist them with, and handed over the imperial seal with what the hostile chronicler himself concedes was real dignity. She died the following year. The son she had deposed decades earlier was restored to the throne, and the Tang dynasty she had displaced resumed its own name - though not, as the next chapters show, its own competence.

Boulger, The History of China, Volume I (1881) · Bairuilong · CC BY-SA 4.0

A second Taizong

peakDrafted from Boulger, The History of China, Volume I (1881)

Xuanzong took the throne in 712 meaning to be remembered as a second founder, and for its first three decades his reign gave every appearance of delivering on it.

The reforms

He cut the court's own expenses and led by his own example in the sumptuary laws he passed - a welcome change to a population that had just spent years under the tax-farmers Wu's own extended contest for power had left in place. He overhauled the army through a programme of reviews that doubled as military games, keeping the force that actually held the empire together engaged and watched. Astronomy and the systematic observation of natural phenomena, a science China had pursued since its earliest records, were placed on a new footing under his patronage. And in a gesture as much about restoring the dynasty's own intellectual prestige as about religion, he had Confucius formally proclaimed a prince and given the title King of Literature - an honor the sage himself, a near-contemporary chronicler notes dryly, would likely have valued above the secular rank.

Metropolitan Museum of Art, New York · Boulger, The History of China, Volume I (1881) · Han Gan · CC0

What the reforms did not fix

The empire Xuanzong inherited still had frontiers to defend, and here his own later record complicates the case for a second Taizong considerably. War with Tibet ran through most of his reign with no side gaining a lasting advantage: a Chinese army recovered lost ground in the 720s, a treaty in 730 bound both sides to stop raiding, Xuanzong broke it himself when Tibet moved against the small tributary state of Poulin, and the fighting that followed cost tens of thousands of lives without settling anything the earlier treaty hadn't already tried to settle. The siege of Chepouching in 749 alone is recorded as costing more than thirty thousand lives to take.

Elsewhere the record is starker still. An entire Chinese army of thirty thousand men was destroyed in the Gobi fighting the Khitan and allied Turkic tribes; a separate campaign against the southwestern kingdom of Nanzhao lost thirty-three towns in a single year and, three years later, a second army outright. The reign's total military losses are put, in total, at something close to two hundred thousand men - a toll that fell on the same frontier garrisons whose discontent would shortly produce the general who broke the dynasty from within.

A verdict earned rather than assumed

Later tradition remembers Xuanzong's early reign, the Kaiyuan era, as the high point of the whole dynasty, and by the measure of court culture and domestic reform that reputation has real substance behind it. But the same reign that produced the reforms also produced the frontier failures that fed directly into what followed: a court confident enough in its own greatness to overlook how thin the army actually protecting it had become, and a general named An Lushan who was about to demonstrate exactly how thin.

Boulger, The History of China, Volume I (1881) · Unknown · Public domain

Talas

peakDrafted from Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009)

In 751 an Abbasid army and a Tang army met on the Talas river in what is now Kyrgyzstan. The Tang force was defeated, partly because its Karluk allies changed sides during the battle.

Its importance is usually overstated and occasionally dismissed. It did not decide the fate of Central Asia — the An Lushan rebellion four years later did that, forcing the Tang to withdraw its western garrisons permanently to deal with a civil war that may have been the most destructive in world history to that point. Talas is the marker rather than the cause.

What it does mark is the only direct contact between the two largest states on earth. Tang China at Xuanzong's height and the caliphate under the early Abbasids were comparable in extent, in revenue and in the sophistication of their administration, and their frontiers touched exactly once, in a battle neither government treated as important. The tradition that Chinese papermakers captured at Talas brought paper to Samarkand and thence to Baghdad and Europe is a good story of uncertain support; paper does appear in the Islamic world at about this time, and Baghdad had a paper mill by the 790s.

The empire that fought at Talas was the most cosmopolitan in Chinese history — Sogdian merchants, Turkish generals, Persian refugees from the Sasanian collapse, Nestorian and Manichaean and Zoroastrian communities in Chang'an. An Lushan, who nearly destroyed it, was a general of Sogdian and Turkic parentage.

Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009) · Unknown · Public domain

What the Tang was actually doing out there

The battle is easier to place if the western position is described rather than assumed.

By the 740s the Tang held a line of garrison commands — the Four Garrisons of Anxi — at Kucha, Khotan, Kashgar and Karashahr, with a protectorate over the Sogdian city-states beyond the Pamirs and clients as far as the Oxus. This was not frontier defence. It was an imperial system a thousand miles beyond the Jade Gate, administered by Chinese officials, garrisoned by professional soldiers on long service, and supplied along a road of post stations.

It existed because of the Türk empires and then the Tibetans. Whoever held the Tarim held the flank of the route to the west and the approaches to Gansu, and the Tang had spent a century taking it precisely so that nobody else would.

The commander at Talas, Gao Xianzhi, was a Korean by descent — which is worth noting, because it is entirely characteristic of how that army was officered.

What the defeat cost, and when

Not the Tarim. The Tang went on holding the Four Garrisons after Talas.

What took them was An Lushan four years later, when the professional armies of the west were recalled to fight the civil war and never came back, and the Tibetan empire moved into the vacuum and took Chang'an itself in 763. The garrisons in the Tarim were left without relief or orders and held out, in some cases, for decades — a detail known because their coinage and their dated documents kept being issued long after the road home was closed.

Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009) · Jean-Michel Moullec · CC BY 2.0

That is why Talas is a marker rather than a cause. The Chinese position in Central Asia was not lost to the Arabs in a battle. It was withdrawn to deal with a general of Sogdian and Turkic parentage who had been given three army groups on the northern frontier, and it was never reinstated.

An Lushan

contractionDrafted from (ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979)

An Lushan was a frontier general of Sogdian and Turkic parentage who commanded three of the empire's northern military regions and perhaps a third of its army. He revolted in 755, took Luoyang and then Chang'an, and drove the emperor Xuanzong into flight to Sichuan. The rebellion took eight years to suppress and required Uyghur cavalry to do it.

Metropolitan Museum of Art, New York · (ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979) · Unknown · CC0

The census figures before and after show a fall of about two thirds. Almost nobody now reads that as the death toll — the registration system collapsed along with the state's ability to run it, and uncounted is not dead — but the disruption was real and prolonged, and the argument about how much of the drop was mortality is unsettled.

What is not disputed is the political consequence. To fight the rebellion the court granted regional commanders control over their own taxation and troops, and never took it back. The second half of the Tang is a central government negotiating with semi-autonomous provinces that remit what they choose. The western garrisons were withdrawn permanently and the Tarim basin passed to the Tibetan empire.

The comparison worth drawing is with the Abbasid civil war of 811 to 813, fifty years later: two halves of an empire fighting with their own provincial armies, after which the capital's garrison was never trusted again. Both empires survived their civil war and neither recovered the authority it had before.

Why a frontier general had that much army

Because of a reform that looked sensible at the time.

The Tang had begun with the fubing system — militia soldiers who held land in return for service, mustered in rotation, living at home between turns. It worked when campaigns were short and near. It could not garrison the Tarim basin, the Korean frontier and the Ordos simultaneously, because a farmer cannot be sent three thousand kilometres for three years and still be a farmer.

So from the 710s the militia was replaced on the frontiers by long-service professionals under permanent regional commands, the jiedushi, each with its own recruitment, its own supply and eventually its own revenues. By the 740s nine of these commands held the great majority of the empire's effective troops, and the interior of China was garrisoned by almost nobody.

An Lushan held three of the nine.

Museum of Far Eastern Antiquities, Stockholm · (ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979) · Daderot · CC0

This is the standing problem of every empire with a long land frontier, and the number of solutions is small. Rome ended with frontier armies that made emperors. The Abbasids imported slave soldiers and were governed by them within two generations. The Tang professionalised its frontier and lost control of it in forty years.

What eight years of it did

The court survived by borrowing. Uyghur cavalry was hired to retake the capitals, and was paid in silk, in the right to plunder Luoyang, and in a trade in horses at prices the Uyghurs set — an obligation that bled the treasury for a century and which the Tang could not get out of because it could not fight the people it owed.

Rebel commanders who surrendered were confirmed in their commands rather than punished, because there was no force available to punish them with. Those commands became hereditary. Three of the north-eastern provinces were in practice independent from 763 onward, sending nothing and obeying nothing, and the dynasty went on formally appointing the men who already held them for another hundred and fifty years.

The An Lushan rebellion did not destroy the Tang. It converted it into a different kind of state: a court with a rich southern tax base, a salt monopoly, a diplomatic relationship with its own provinces, and no army of its own worth the name.

The armies that changed

contractionDrafted from (ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979)

The Tang changed how it raised soldiers once, in the middle of its life, and the change is the political history of the dynasty.

It began with the fubing: militia units of farmer-soldiers who held land under the equal-field allotment system, served in rotation at the capital or on the frontier, provided their own kit, and went home. It cost the state almost nothing. It was also designed so that no commander ever had a permanent army.

(ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979) · Unknown · Public domain

By about 740 it had stopped working. The equal-field system underneath it had eroded the way such systems do — land accumulated, registers went stale, allotments were not there to be reallocated — and in any case rotating farmers to the far side of the Tarim basin for a season was not a way to hold a frontier against Tibet and the Türks.

What replaced it was ten regional commands on the frontier, each under a jiedushi, with long-service paid troops, its own supply, and eventually its own revenue.

755

An Lushan held three of those commands at once, and about a third of the empire's soldiers. In 755 he marched on the capitals.

The court survived the rebellion by granting the commanders who put it down the same autonomy the rebel had enjoyed, so the last hundred and fifty years of the Tang are a central government negotiating with provinces that remitted what they chose.

The same thing, seven hundred years earlier

The Han had done all of this before.

A militia resting on smallholding. The land system eroding underneath it. Professional regional armies raised to replace it. And then one of their commanders taking the capital.

Two dynasties, seven centuries apart, on the same ground, running the same sequence in the same order. Whatever else the Tang collapse was, it was not unforeseeable — the precedent was in the histories the court's own officials had been trained on.

Salt

contractionDrafted from (ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979)

The An Lushan rebellion destroyed the tax registers. A state that assessed households by head and by allotted land, and could no longer say who lived where or held what, had no way to collect.

What kept the Tang alive for a hundred and fifty years afterwards was salt.

From 758 the state bought salt at the production sites, resold it to licensed merchants at a large markup, and let them distribute it. It required no registers, no census and no control of the countryside — only control of a few hundred coastal pans and inland wells. By the 780s that one indirect tax was providing more than half of central government revenue.

(ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979) · Baomi · CC BY-SA 4.0

The Han had monopolised the same commodity in the same country nine hundred years earlier, and had held a court debate in 81 BC over whether a state should be in that business at all. The Tang did not reopen the argument. It reopened the monopoly, for the reason above: it was the only tax a government that had lost its records could still assess.

Two-tax

The other half of the answer is the reform of 780. Yang Yan abandoned assessment by head and by allotment and taxed households twice a year on what they actually held, in cash.

(ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979) · Gary Lee Todd · CC0

Very few fiscal reforms anywhere outlive the dynasty that makes them. This one outlived the Tang by seven hundred years: in outline it is how China was taxed until the sixteenth century, when the Ming Single Whip consolidated the descendants of Yang Yan's two taxes into one payment in silver.

And Huang Chao

The man who finished the dynasty was both of the things this fiscal system produced: a failed examination candidate and a salt smuggler.

Huang Chao's rising crossed the empire twice, sacked Guangzhou, and took both capitals. A monopoly high enough to fund a state is high enough to make smuggling worth an army; and an examination system that had become the only respectable road to office produced, every three years, a cohort of literate men who had been told the empire was open to talent and had then been failed.

The dynasty survived him by twenty-three years and was destroyed by the general who put him down.

The wine of immortality

contractionDrafted from Boulger, The History of China, Volume I (1881)

By 805, when Xianzong came to the throne, the shape of the second half of the dynasty was already set: a court negotiating with provinces that answered to their own military governors rather than to Chang'an. Xianzong's fifteen-year reign is remembered as the one real attempt to reverse that arrangement rather than simply manage it.

Boulger, The History of China, Volume I (1881) · Unknown · Public domain

Putting down two governors

Within his first year, the governor of Sichuan, Liu Pi, tested how far the new emperor's patience would stretch, expanding his own authority well past what the court had tolerated from his predecessors. Xianzong tried accommodation first, overlooking conduct he might reasonably have punished - and found, as rulers usually do with that approach, that overlooking it only encouraged more of it. He sent an army instead. Liu Pi's rebellion collapsed, and the province that had been drifting out of the court's reach was reattached to it.

A larger test followed in Shandong, where the governor Li Shidao held out with something close to open independence for years. When the imperial army finally moved against him, his own following judged the cause lost before a battle was even fought and refused to defend him; Li Shidao and his two sons were taken from their own palace and executed in the public square. It was the clearest demonstration in a generation that a rebellious jiedushi could still be broken by force rather than merely tolerated - and for a court that had spent thirty years negotiating with exactly this kind of governor since An Lushan, that was no small thing to prove.

The emperor's own weaknesses

Xianzong is remembered, fairly, as well-meaning rather than brilliant, and his own court showed the same faults that had weakened every Tang emperor since Xuanzong's decline: an excessive trust in the palace eunuchs, whom he protected even when his own advisers warned him what that trust would cost his successors, and a genuine credulity about Buddhist and Daoist claims to prolong life through alchemical preparation. The second weakness killed him. In 820, after an overdose of what the court called the "wine of immortality" left him in a fury that cost several eunuchs their lives, the eunuch Chen Hongzhi finished what the elixir itself had started, poisoning the same cup Xianzong kept returning to.

What the restoration actually restored

Xianzong's reign did not undo the settlement An Lushan's rebellion had forced on the dynasty - the great provincial commands, largely self-taxing and self-arming, remained the basic shape of Tang government for the rest of its existence. What it proved was narrower and, in its own way, more useful: that a court willing to commit force decisively, rather than negotiate indefinitely, could still bring an individual rebellious governor back under its own authority. It was a lesson his own less capable successors would prove no better able to apply than his own eunuchs were to be trusted.

Drones of society

contractionDrafted from Boulger, The History of China, Volume I (1881)

Wuzong took the throne in 840 by the same method the eunuchs had used to make and unmake emperors for a generation - they passed over his predecessor's own sons and installed him instead, expecting a pliable client. They got, in this one instance, something closer to the opposite.

Boulger, The History of China, Volume I (1881) · Wang Qi · Public domain

An able ruler by the standard of his own age

Wuzong protected the frontiers with an energy the court had not shown in years, defeated a persistent Tangut tribal challenge in Shaanxi through his own lieutenants, and granted asylum within his dominions to a displaced people from Central Asia rather than turning them away. He was, by his contemporaries' own account, a genuine huntsman and a man given to military exercise rather than to court ceremony - closer in temperament to the dynasty's early rulers than to the figures who had occupied the throne since Xianzong's death twenty years before.

The edict

It was this same practical, unsentimental cast of mind that produced the reign's defining act. Chinese Buddhism by 845 supported an establishment on a scale the state had never systematically measured or taxed - close to four thousand monasteries and something like a quarter of a million monks and nuns, by the count later chroniclers settled on, all of it standing outside the ordinary registers of land and labour the rest of the population was taxed and conscripted through. Wuzong's own reasoning, as the record presents it, was not theological. A monk was, in his own blunt phrase, an able-bodied man living in idleness on the credulity of his neighbours, and it took no particular flaw in the faith itself to explain why so many had found their way into a monastery's comfortable retirement. In 845 he ordered the monks and nuns of the empire's religious houses to leave their orders and return to their own families - and, with them, to the tax rolls and the corvée labour their monastic status had shielded them from.

What it did and did not undo

The suppression was a fiscal and administrative act dressed in religious language, and it worked on its own narrow terms: land and labour that had been outside the state's reach for generations came back within it in a single year, at exactly the moment a court starved of revenue by decades of provincial autonomy needed every registered household it could find. It did nothing to the faith itself outside China's own borders - Buddhism had already spread through Japan and Korea by way of Chinese teaching and scripture decades earlier, and no domestic edict in Chang'an could touch what had already propagated abroad.

Boulger, The History of China, Volume I (1881) · Zhangzhugang · CC BY-SA 3.0

Wuzong did not live to manage what he had unsettled. He died the following year, his reign judged by the record itself to have been too brief for the good it might otherwise have done, and the eunuchs who had expected a compliant nominee in 840 went back to the same kingmaking they had practiced before him - proof, if any more were needed, that one capable reign was not enough to break a pattern that had already outlasted three of them.

British Library, London · Boulger, The History of China, Volume I (1881) · Unknown · CC0

A certain herb

aside?What aside meansDrafted from al-Sirafi, Ancient Accounts of India and China, by Two Mohammedan Travellers, Who Went to Those Parts in the Ninth Century (1733)

An Arab merchant wrote an account of China around 851, and it is the earliest description of the Tang state by somebody from outside it. It is also the first description of tea in any language other than Chinese.

The herb

The emperor, he says, reserves to himself the revenues of the salt mines, and of

a certain Herb which they drink with hot Water, and of which great Quantities are sold in all the Cities, to the amount of great Sums. They call it Sah, and it is a Shrub more bushy than the Pomegranate-tree, and of a more taking Smell, but it has a kind of Bitterness with it. Their way is to boil Water, which they pour upon this Leaf, and this Drink cures all sorts of Diseases.

That is tea, in 851, described by a man who has clearly drunk it and has no word for it. Everything lodged in the treasury, he adds, comes from the poll tax and the duties on salt and on this leaf — which is an accurate statement of late Tang public finance, arrived at by a foreigner from what he could see of the customs houses.

Famen Temple Museum · al-Sirafi, Ancient Accounts of India and China, by Two Mohammedan Travellers, Who Went to Those Parts in the Ninth Century (1733) · G41rn8 · CC BY-SA 4.0

What he noticed about the taxes

There is no tax on land. There is a tax on men, levied according to their condition and capacity. Arabs and other foreigners in the country are taxed in proportion to their substance.

Famen Temple Museum · al-Sirafi, Ancient Accounts of India and China, by Two Mohammedan Travellers, Who Went to Those Parts in the Ninth Century (1733) · Xoesan · CC BY-SA 4.0

He puts fifty thousand dinars a day into the treasury from the poll tax at Canton alone — although this City be none of the largest in China, which is the detail that shows he understood the scale of what he was looking at.

The granaries

When any Dearth makes Necessaries very dear, then does the King open his Storehouses, and sell all Sorts of Provisions much cheaper than they are to be had at Market; and hence no Dearth is of any long Continuance among the Chinese.

An ever-normal granary system, working, described by a merchant from a part of the world that had nothing of the kind. He does not editorialise; he states it as a fact about how prices behave in China.

The bell on the string

The most remarkable thing he records is a machine for petition.

In each city a small bell hangs on the wall above the governor's head. A string runs from it for about three miles, crossing the highway, so that anybody can reach it. When the string is pulled the bell strikes over the governor's head, and he orders the person who has demanded justice to be brought in, and the complainant states his case in person and the wrongs he suffers.

And the same Practice is in use throughout all the other Provinces.

Whether every city in Tang China really had a three-mile bell-rope is not the point. What the merchant is recording is that he was told this, everywhere, as the normal arrangement — a state that advertised to foreigners that any subject could compel the attention of the governor.

The passes

And against that, the other half of the same system.

To travel from one place to another a man must carry two passes: one from the governor, which permits him to set out and records his name; and one from the eunuch, the emperor's own officer in the province. Both are inspected on the road.

A state that will hear any man's complaint and will not let any man move without papers. The merchant reports both without finding them contradictory, and they are not: they are two expressions of the same thing, which is a government that intends to know where everybody is.

Why the account matters

Because it is outside evidence for what the Tang was like at ground level, from a man with no interest in flattering it and no access to its official record.

The dynastic histories describe the salt monopoly, the granaries, the household registration and the travel permits, and a sceptical reader may reasonably wonder how much of that was real and how much was the paperwork of an ambitious bureaucracy. Here is a Persian Gulf trader, a century before the dynasty ended, who had gone there to buy and sell, reporting that the granaries opened, that the passes were demanded, that the tax was on heads rather than land, and that the government was making a fortune out of a bitter leaf infused in hot water.

What was copied, and what is not counted

aside?What aside meansDrafted from Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009)

The Tang is the state whose institutions the rest of East Asia took, and by the usual measures of influence it scores nothing. Both halves of that are worth setting out, because the reasons are different.

The script

Chinese characters reached Japan, Korea and Vietnam, and all three adapted them: kana derived from cursive character forms, hangul devised later in a deliberately different direction, chữ Nôm built out of characters for a language of another family entirely.

None of that is a Tang creation. The writing system those countries adapted is the clerical script the Han standardised, which is the ancestor of everything written in Chinese characters since. The Tang transmitted it; it did not make it.

The claim nobody made

Harder, and more interesting.

The Tang state model was copied wholesale. Japan's Taihō codes are Tang codes. The capital plans of Nara and Heian are Chang'an's grid. Korea's Silla and Goryeo took the ministries; Vietnam took the examinations and was still running them in the nineteenth century. Very few states anywhere have been imitated at that scale by that many independent governments.

Museum of Fine Arts, Boston · Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009) · Unknown · Public domain

And not one of them claimed to be the Tang. Adopting somebody's civil service is not claiming to be them: Japan in particular took the apparatus while explicitly refusing the subordination, and its rulers used a title for themselves that the Chinese court regarded as an impertinence.

That is the shape of Chinese influence generally, and it is why it is so easy to undercount. It travelled as a design that other people made their own, rather than as a title somebody inherited — and a design, unlike a title, leaves no claimant behind to be counted.

Huang Chao

endDrafted from (ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979)

The Tang was finished by a failed examination candidate with a salt licence.

Who he was

Huang Chao came from a family of salt merchants — which after the reforms of the 760s meant a family in the smuggling business, because the state monopoly had made salt the most heavily taxed commodity in the empire and smuggling the most profitable trade in it. He sat the examinations, repeatedly, and failed.

That combination is the dynasty's two great structural problems in one biography. The salt monopoly had become the state's main revenue and had created an armed criminal economy underneath it. The examinations had become the only respectable route to office and were producing far more qualified candidates than there were posts, in a system where the great families still took most of what there was.

What he did

He joined an existing rising in 875 and within a few years was leading it. What followed was not a regional revolt. He marched the length of the empire twice — south to Guangzhou, where his army sacked the port and, on the report of an Arabic source, massacred a very large foreign merchant community; then north again, taking Luoyang and, in 880, Chang'an.

The emperor fled to Sichuan, as an emperor had fled in 756 and another would in 881. Huang Chao proclaimed his own dynasty and held the capital for two years.

He was driven out by an alliance of provincial commanders and a Shatuo Turkic cavalry force under Li Keyong, and was killed in 884.

What was destroyed

The capital. Chang'an had been sacked in 763 by the Tibetans, in 881 by Huang Chao, and in 883 by the armies retaking it; the poet Wei Zhuang's Lament of the Lady of Qin describes the last of these, and the city never recovered its position. It was formally abandoned as a capital in 904 and has never been one since.

(ed.), The Cambridge History of China, Volume 3: Sui and T'ang China, 589-906, Part 1 (1979) · Unknown · Public domain

The southern trade. Guangzhou's foreign merchant community was the eastern terminus of the Indian Ocean network, and the sack of 879 is followed in the Arabic sources by a redirection of that trade to the Malay ports, where merchants met Chinese junks instead of sailing to China themselves.

And the aristocracy. The great clans of the north-west that had supplied Tang officialdom for two centuries — the families whose genealogies the dynasty had compiled and ranked — were destroyed in the fighting and the massacres of the 880s and 890s as a class, not as individuals. They do not reappear.

907

The rest is bookkeeping. Zhu Wen was one of Huang Chao's generals who defected to the imperial side, was rewarded with a governorship, and spent the 890s eliminating his rivals. In 903 he massacred the eunuchs. In 904 he moved the emperor to Luoyang and had him murdered. In 907 he deposed the boy who had replaced him and took the throne.

The dynasty had been governed by whoever held the emperor for about twenty years by then, which is the same condition the Han was in for the thirty years before 220 — and it ended the same way, with an abdication ceremony, for the same reason. The name was worth something to the man taking it.

What followed was fifty-three years of five dynasties in the north and ten kingdoms in the south, and then a reunification under the Song that put the military governors permanently under civil officials, because everyone had just watched what happened when it was the other way round.

The mulberry trees

endDrafted from al-Sirafi, Ancient Accounts of India and China, by Two Mohammedan Travellers, Who Went to Those Parts in the Ninth Century (1733)

The fall of the Tang has a foreign witness, because the rebellion that broke it destroyed the largest foreign merchant community in Asia, and one of the survivors' countrymen wrote down why the silk had stopped coming.

Abu Zayd

Around 916 Abu Zayd al-Sirafi added a continuation to the merchant's account of 851, explaining what had happened to the China trade in the meantime.

His answer is that it had been destroyed, and he begins by saying so in terms of navigation: the troubles which broiled the affairs of that empire, and put a stop to the justice and righteousness formerly practised there, have, in fine, interrupted the ordinary Navigation from Siraf to China.

Then he names the cause. An officer of no royal family, considerable in his employment, who revolted, marched about the country to the great loss of the inhabitants, won a party over by his liberalities, and gathered a multitude of vagabonds and abandoned people into a considerable body of troops.

He calls him Baichu. He is Huang Chao.

Khanfu

The rebel marched on Canfu — Guangzhou — one of the most noted Cities in China, and, at that Time, the Port for all the Arabian Merchants. The citizens shut the gates. The siege lasted a long while. At last he took the city and put the inhabitants to the sword.

al-Sirafi, Ancient Accounts of India and China, by Two Mohammedan Travellers, Who Went to Those Parts in the Ninth Century (1733) · Sekino Tadashi · Public domain

Abu Zayd then gives a figure and, more unusually, gives his reason for trusting it.

There are Persons fully acquainted with the Affairs of China, who assure us, that besides the Chinese who were massacred upon this Occasion, there perished one Hundred and twenty Thousand Mohammedans, Jews, Christians, and Parsees, who were there on Account of Traffic. The Number of the Professors of these four Religions, who thus perished, is exactly known; because the Chinese are extremely nice in the Accounts they keep of them.

That last clause is worth pausing on. The number is credited because the Tang registered its resident foreigners by religion, and the register was available. A merchant in the Persian Gulf is citing Chinese household registration as the warrant for a casualty figure — which is both the best argument for the number and an unintended confirmation of everything the earlier account said about passes and counting.

The trees

Then the detail that explains why a Baghdad reader would care.

He also cut down the Mulberry Trees, and almost all the Trees of other Kinds; but we speak of the Mulberry in particular, because the Chinese carefully cultivate it for the sake of its Leaf, wherewith they subsist and propagate their Silk-worms. This Devastation is the Cause why Silk has failed, and that the Trade which used to be driven with it, in the Countries under the Arabs, is quite stagnated.

A commodity shortage in the caliphate, correctly traced across five thousand miles of ocean to an orchard.

The rest of it

Having sacked Canfu he took other cities one after another, the emperor being unable to stop him; he advanced on the capital, Cumdan — Chang'an — and the emperor made a confused retreat to a city on the Tibetan frontier. The rebel then went on demolishing cities and killing most of the inhabitants, intending in the general butchery to catch every branch of the royal blood, so that nobody should survive to dispute the empire with him.

That is the Huang Chao rebellion as it reached the Persian Gulf: accurate in outline, accurate about the flight to Sichuan, and accurate about the ambition.

What the account is evidence of

Two things the Chinese sources cannot give.

First, the scale of the foreign presence. A hundred and twenty thousand Muslims, Jews, Christians and Zoroastrians resident in one southern Chinese port for the purposes of trade is a figure that reframes what the Tang was. That is not a trading post. It is a quarter of a great city, registered, taxed, and killed in a single sack.

Second, the interdependence. The Tang collapse is normally told as a Chinese catastrophe with Chinese causes, which it was. It was also the event that stopped the silk reaching Baghdad, emptied the Gulf shipping lanes, and ended the direct sea route between China and the Islamic world for a century and a half.

When an empire that size comes apart, the people who notice first are the ones who were buying from it.

The model

afterlifeDrafted from Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009)

The Tang left something unusual: not a claimant, not a language, not a religion, but a template that three other countries adopted deliberately, in writing, while it was still running.

Copied on purpose

Japan sent embassies for two centuries specifically to study the arrangements and bring them home. What came back was the ritsuryō system: the Tang legal codes, the ministries, the provincial structure, the land allotment, the court ranks, the calendar. Nara and then Heian were laid out on Chang'an's grid — a walled rectangle with a palace at the north end and a chequerboard of wards, built at a scale the Japanese state could not fill and never did.

Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009) · SY · CC BY-SA 4.0

Korea's Silla and then Goryeo took the ministries and the ranks. Vietnam took the examinations and was still running them in the nineteenth century, in Chinese, on the same canon.

None of that was conquest. Japan was never part of the Tang empire, and Silla fought a war against it. The transmission is a set of foreign governments deciding that this was how a state should be built, and sending people to copy the plans.

Lewis, China's Cosmopolitan Empire: The Tang Dynasty (2009) · Tokumeigakarinoaoshima · CC BY-SA 4.0

What travelled with it

The characters, which all three adapted — kana out of cursive forms, later hangul in a deliberately different direction, chữ Nôm built out of Chinese graphs for a language of another family entirely.

Buddhism in its Chinese forms — Chan became Zen and Seon and Thiền, and the schools that shaped Japanese Buddhism for a millennium were imported from Tang China by monks who had gone to study there.

And the poetry. Tang verse became the classical standard for literate people across the whole region, and stayed the standard longer than any political arrangement it travelled with.

Why none of it is a claim

Nobody afterwards claimed to be the Tang. There is no equivalent of the Ottoman claim on Rome or the German claim on the Roman empire — no later state asserting that it was the Tang continuing.

That is a real difference and it is worth naming rather than treating as an absence. The mandate doctrine made succession exclusive: at any moment exactly one dynasty holds the mandate, and the previous one has lost it. There is no room in the theory for a state that is the old empire persisting, because if it persisted the current one would be illegitimate. Every Chinese dynasty succeeds the last; none continues it.

So what the Tang exported went out as a design rather than as a title, and the countries that took it took it as their own rather than as an inheritance. Japan did not become a Tang successor. It became a state built on the Tang plan, with its own emperor, and it kept the plan for eleven hundred years after the dynasty that drew it was gone.

The one thing that came back

In 1868, when Japan rebuilt its government on European lines, one of the words it coined for the new institutions was assembled from Chinese characters — and a good many of those coinages were then borrowed into Chinese, in the other direction, to name things the classical vocabulary had no word for.

The traffic ran one way for a thousand years and then reversed inside a generation, using the same script the Tang had exported, which is the only kind of afterlife an institution ever really gets.