To beat the steppe, the Han became cavalry (after 200 BCE)

Before: the army the Han founded with, and the steppe it could not reach

When Liu Bang declared himself emperor of the Han in 202 BCE, the army that had carried him to the throne was an army of footsoldiers. It was built from peasant conscripts levied household by household under a registration system the Han had inherited intact from the Qin, men called up for a fixed term of service and armed from state arsenals with the halberd, the dagger-axe, the iron sword, and above all the crossbow ⁷. The crossbow was the weapon that defined this army. The Qin had already turned it into an instrument of mass infantry warfare, casting its bronze trigger mechanisms to interchangeable tolerances and issuing them in the tens of thousands, so that a conscript farmer with a few weeks' drill could, from behind a shield wall, kill an armoured aristocrat at two hundred paces ¹⁶⁷. Chariots survived in the order of battle as command platforms and as relics of an older warfare, but the decisive arm was the massed crossbow line, screened by spearmen, fighting on the flat alluvial plains of the Yellow River where Chinese armies had always fought.

This was a settled army built for settled war. It moved at the speed of a marching column and the slower speed of its grain carts; it was supplied from granaries along fixed roads; and it was designed to take and hold the things a farming empire valued — walled towns, river crossings, cultivated land. Against another such army it was formidable. Against an enemy that owned no towns, planted no fields, and could not be brought to battle unless it chose to be, every one of its strengths became irrelevant.

What the Han had, and what it lacked

The Han inherited from the Qin one of the most formidable infantry-and-logistics machines the ancient world produced. What it did not have was cavalry in any quantity that mattered. Horses were scarce and expensive in the agricultural interior, bred in small numbers on land that was worth more under the plough; the open grassland needed to raise war-horses by the hundred thousand lay north and west, in country the Han did not control ³. Mounted troops existed, but as a screening and scouting arm, not a strategic weapon. After the civil wars that founded the dynasty, the histories record, the empire was so short of horses that the emperor himself could not find four matched animals for his carriage and his ministers rode in ox-carts ³.

The categories the Han army possessed were the categories of a state that fought by standing its ground: the conscript register, the granary, the walled garrison, the crossbow bolt counted out by the crate and signed for by a clerk. The categories it lacked were the ones the steppe lived by — the remount herd that turned one rider into three, the short bow that could be loosed in any direction at the gallop, the saddle and trousers that freed both hands and both legs, and above all the man who had ridden since before he could walk and for whom war was only hunting with a different quarry. These were not things the Han could requisition. They had to be grown, and the Han had neither the grass nor the generations.

The lesson the Han had already been taught

The gap was not new, and the Chinese states had already seen one answer to it more than a century before the dynasty existed. In 307 BCE King Wuling of the state of Zhao, whose northern border ran against mounted Hu raiders, ordered his court and army to abandon the long robes of Zhou ceremonial dress for the trousers, short jacket, belt, and boots of the nomads, and to learn to shoot from horseback — the reform Chinese sources call hufu qishe, "barbarian dress and mounted archery" ⁸¹¹. The court resisted hard. Trousers were the clothing of peoples the Zhou world despised, and to put them on was to concede in public that the despised had something the civilized lacked. Wuling broke the resistance by wearing the dress himself in audience and arguing, against his own ministers, that a king governs for utility and not for the comfort of ritual. Within a few years Zhao fielded a genuine cavalry arm and pushed its frontier north into the steppe margin ¹¹.

The precedent was on the record, preserved and read. What the early Han lacked was therefore not the idea of mounted archery — it knew the idea — but the herds, the grassland, the breeding time, and the political will to pay for all three at the scale of an empire rather than a single frontier state. For sixty years it chose not to. The reasons were partly fiscal and partly ideological: the early Han ran deliberately light, taxing little and spending less, and its court intellectuals distrusted military adventure on principle. The bill for that restraint was a frontier it could not defend and a neighbour it had to bribe.

Baideng, 200 BCE: the founding humiliation

The bill came due almost at once, and in person. In 200 BCE Liu Bang — now the emperor Gaozu — led a great army north against the Xiongnu, who under their chanyu Modu had welded the steppe tribes into a single mounted power. Near Pingcheng, in the cold of the early year, the emperor outran his own infantry with a forward column and was surrounded on the heights of Baideng by a Xiongnu cavalry force that Han tradition numbers in the hundreds of thousands, drawn up by the colour of their horses ²⁹. For seven days the emperor of China sat trapped on a frozen hill, his men losing fingers to frostbite, his relief column unable to break through a moving wall of horse-archers that opened and closed around it. He escaped not by force but by bribery — a gift and a quiet argument carried to Modu's consort, and a gap left conveniently open in the encirclement ⁴.

What followed Baideng was not a rematch but an accommodation that lasted two generations. The Han adopted the policy called heqin, "peace through kinship," whose terms the court renewed and enlarged across the following decades:

• a Han princess sent north to marry the chanyu, making the two ruling houses formal in-laws;
• fixed annual shipments of silk, grain, wine, and silver carried to the Xiongnu court;
• recognition of the Great Wall as the agreed boundary between the two states;
• and a diplomatic protocol that addressed the chanyu and the emperor as equals.

The statesman Jia Yi, a generation later, called this arrangement an inversion of the natural order — the empire, which should be the head, feeding the feet — and proposed instead a scheme of "five baits" to corrupt the Xiongnu elite with Han luxuries ³. But for roughly sixty years, down to the accession of the emperor Wu in 141 BCE, the wealthiest agrarian empire on earth paid tribute to a confederation of herders because it could not defeat them in the field. That humiliation is the thing to hold onto, because almost everything the Han did over the following century was an attempt to undo it.

The transmission: how the steppe taught the Han to ride

The thing transmitted was not an object. No single artifact crossed the frontier the way an alphabet or a coin crosses it. What crossed was a military system — a way of raising, mounting, supplying, and fighting men on horseback — and it crossed because the Xiongnu demonstrated it, year after year, on the bodies of Han soldiers and the burned roofs of Han border towns, until the Han state concluded it had no choice but to learn it. Transmission here is closer to forced apprenticeship than to gift. The teacher charged tuition, and collected it whether or not the lesson was wanted.

Modu's machine

The system the Han faced was, above all, the creation of Modu, who seized the Xiongnu leadership in 209 BCE by a coup his own people remembered as a parable of discipline. He trained a corps of guards to loose their arrows at whatever his own whistling-arrow struck, executing those who hesitated; he made them shoot his favourite horse, then his favourite wife, killing the doubters each time; and when at last he turned the whistling-arrow on his father the old chanyu, every guard fired without pause ²⁵. The story is a fable about the thing that actually mattered: a steppe leader who could impose absolute, instantaneous obedience on dispersed horsemen had solved the problem that had always kept the steppe weak, and could now turn its mobility into strategy.

The army Modu built rested on a decimal organization — commanders of ten, of a hundred, of a thousand, and of ten thousand — that let a scattered pastoral population mobilize into coordinated columns and dissolve again into the grassland, a structure Sima Qian recorded with the precision of a man describing a rival state's constitution ²¹³⁵. Its weapon was the composite recurve bow, built up from horn, wood, and sinew, short enough to be drawn and loosed in any direction from a moving horse and powerful enough to kill at a gallop. Its mounts were the hardy steppe ponies that grazed and watered themselves. And its logistics were the herds: an army that rode and ate its own supply train needed no granaries, left no road to cut, and could not be starved out of a field it never had to hold. Modu turned this machine outward, destroying the Donghu to the east, driving the Yuezhi west out of Gansu, and assembling, in Sima Qian's account, the largest steppe dominion that had ever existed ²⁵.

Photograph by Gary Todd (WorldHistoryPics). Steppe-style gold belt plaque, Inner Mongolia Museum, Hohhot. CC0 via Wikimedia Commons. · CC0

Sima Qian, who knew the Xiongnu as a contemporary and not a memory, described how the system was grown into the body from childhood. "The little boys start out by learning to ride sheep and shoot birds and rats with a bow and arrow," he wrote, "and when they get a little older they shoot foxes and hares, which are used for food. Thus all the young men are able to use a bow and act as armed cavalry in time of war" ². There was no training establishment to build and no remount system to fund, because the entire society was the training establishment. This is what the Han confronted, and what Chinese strategists from Chao Cuo onward had to find an answer to: not an army that could be out-produced by a richer state, but a population that was already, in its ordinary life, an army.

Chao Cuo's audit

The clearest contemporary statement of the gap came from inside the Han court itself. Around 169 BCE the official Chao Cuo submitted to the emperor Wen a memorial that reads like a comparative threat assessment, preserved in the Hanshu ⁹. Chao Cuo did not flatter his own side. On broken mountain ground and at the gallop, he conceded, the horses of China could not match the horses of the Xiongnu; in shooting from horseback over rough country, in riding while shooting to front and rear, Han riders could not match Xiongnu riders; in enduring wind, hunger, thirst, and cold, Han troops could not match steppe troops. These were not insults to be answered with bravado; they were operational facts.

But Chao Cuo named the Han advantages with equal precision. On level ground, Han chariots and disciplined infantry in formation could break a nomad charge; Han iron armour and edged weapons were superior to Xiongnu leather, bone, and bronze; Han crossbows, massed and volleyed, outranged and outpenetrated the composite bow; and in close fighting on foot, where the nomad was out of his element, the Han soldier prevailed ⁹¹⁶. From this audit Chao Cuo drew a programme: settle the frontier with armed colonists, use surrendered nomads and allied horsemen to fight in the steppe manner, and above all build a Han cavalry that could meet the Xiongnu in their own domain. The memorial's importance is that it framed the problem as solvable. The Xiongnu were not invincible; they were dominant in exactly one domain, mobility, and the Han had either to acquire that domain or to go on paying for the lack of it.

Zhang Qian and the road to the horses

Acquiring it meant, first of all, horses — better animals and more of them than the interior could breed. The search for them produced one of the most consequential journeys in Eurasian history. In 138 BCE the emperor Wu sent the envoy Zhang Qian west to find allies against the Xiongnu among the Yuezhi, the people Modu had driven out of Gansu a generation before, on the reasoning that an enemy's enemy on the far flank might open a second front ³⁸. The mission was a diplomatic failure and an intelligence triumph. Zhang Qian was captured by the Xiongnu almost at once and held for around a decade, married, fathered children, escaped, pressed on west to find the Yuezhi settled and unwilling to fight, was captured again on the way home, and finally returned to Chang'an in 126 BCE after thirteen years, one of two survivors of an embassy that had set out with a hundred ³.

He came back with something more durable than an alliance: the first detailed Han knowledge of the lands beyond the steppe — Ferghana (Dayuan), Bactria, the Wusun, Sogdiana, and the oasis kingdoms strung around the Tarim — together with word of the tall, fast, powerful horses bred in the Ferghana valley, animals of a size and speed the Chinese interior had never produced ³⁸. The road Zhang Qian mapped was the road the Han army would soon march, the road the Hexi commanderies would later protect, and the road that later generations would call the Silk Road. It was opened, at the start, in search of cavalry mounts.

The Heavenly Horses

The Ferghana horses became an obsession of the state. Han sources call them the "heavenly horses" and describe them as sweating blood — a detail modern writers have tied to a parasitic skin infection, but which in Han eyes marked them as supernatural mounts fit for an emperor and for the cavalry that would finally master the steppe ⁸. When a Han mission carrying a thousand pieces of gold and a golden model horse was rebuffed, and its envoys killed, by the kingdom of Dayuan, the emperor Wu turned a thwarted horse-trade into a war at the edge of the known world.

The first expedition, under the general Li Guangli in 104 BCE, was a disaster: it lost the great majority of its men to distance, hunger, and hostile oases that closed their gates and wells along the route, and it limped back without reaching Ferghana. The emperor's response was to double down. The second expedition, in 102 BCE, went out with tens of thousands of troops, a logistics train and reserve to match, and orders that allowed no second failure; it crossed the deserts, besieged the Dayuan capital and cut its water, installed a compliant king, and came home in 101 BCE with several thousand horses — of which, by the time the column reached Han territory, barely a thousand survived ⁸⁶. The empire had fought a multi-year war thousands of kilometres beyond its own frontier for breeding stock. The measure of how badly the Han wanted to escape Baideng is that it thought this a price worth paying — and the measure of the cost is in the bones of the men who did not come back.

What changed, and what was displaced

Between the accession of Wudi in 141 BCE and the great northern campaigns of the 120s and 110s, the Han army stopped being an infantry army with a cavalry screen and became a cavalry army with an infantry base. The change was deliberate, expensive, and total, and it remade not only how the Han fought but how the Han state was shaped, where its borders ran, who its generals were, and what it did with its revenue.

The cavalry state

Under Wudi the Han built state horse-pastures on a scale the early dynasty would not have contemplated. Administrators counted the government's remount herds in the hundreds of thousands, raised and held in the northern and western frontier zones where grassland allowed, and a dedicated bureaucracy of pasturages and stud-farms grew up to manage them ³⁶. Private horse-raising was encouraged by tax incentive; horse-theft and the export of horses and iron to the Xiongnu were punished as crimes against the state. Cavalry ceased to be auxiliary and became the point of the spear.

The campaign armies that went north from 127 BCE onward were built around tens of thousands of mounted troops, capable for the first time of doing to the Xiongnu what the Xiongnu had always done to the Han: striking fast, deep, and where they were not expected. The chronology of the offensive is the chronology of a state cashing in its new instrument:

• 127 BCE — Wei Qing drives the Xiongnu from the Ordos, the great loop of steppe inside the northern bend of the Yellow River, and the Han plant commanderies and colonists on it.
• 121 BCE — Huo Qubing's deep raids into the Gansu corridor shatter the local Xiongnu kings and bring the Hunxie king over to the Han with tens of thousands of his people.
• 119 BCE — at Mobei, "north of the desert," two great cavalry armies under Wei Qing and Huo Qubing cross the Gobi, break the chanyu's main force, and drive the Xiongnu court north beyond the desert — a thing no Han army could have attempted one generation earlier ⁶⁸¹.

Wei Qing and Huo Qubing

The new kind of war produced a new kind of general. Wei Qing, a former household slave raised by the favour of an imperial consort, and his nephew Huo Qubing, who led deep cavalry raids before he was twenty-five, were not the chariot-borne aristocrats of the old order. They were commanders whose reputation rested on speed, range, and the willingness to cut loose from the supply line and live off captured herds and grass in the enemy's own manner ⁶⁸. Huo Qubing in particular fought in the Xiongnu way — fast columns, no baggage, deep strikes — and was rewarded with honours that the older military aristocracy could only watch. The careers of these men mark the transformation as clearly as any single battle. The Han had not merely adopted the steppe's weapon; it had adopted the steppe's kind of soldier, promoted him over the heads of the well-born, and built its proudest victories on his methods.

The Hexi commanderies and the soldier-farmers

Victory in the field was converted into territory and held by colonization. Across the Gansu, or Hexi, corridor — the long grassland-and-oasis passage between the Tibetan plateau and the Gobi that linked the Han heartland to the Tarim and the west — the Han planted a chain of commanderies through the late second century BCE:

• Wuwei, anchoring the eastern entry to the corridor;
• Zhangye, holding the centre;
• Jiuquan, guarding the route toward the desert;
• Dunhuang, the western gate onto the Tarim oases and the roads beyond.

These were not merely garrisons. They were held by the tuntian system of military-agricultural colonies, in which soldier-settlers and their families were moved to the frontier to farm the land, feed the garrisons from their own fields, and stand as the first line of defence — turning conquest into self-supporting occupation and easing, in theory, the fiscal strain of the campaigns ³⁸. Han records describe resettlement on an enormous scale, soldier-colonists and relocated households counted in the hundreds of thousands, driven north and west to make the conquered corridor permanently Han ³. The commanderies cut the Xiongnu off from their subjects and allies in the Tarim and from the Qiang of the Tibetan margin, and they opened the protected road west; the Protectorate of the Western Regions, established in 60 BCE, formalized Han authority over the oasis kingdoms ⁸³. The arms-race against mounted archery had become an engine of imperial expansion, and the route that would carry silk west and ideas east ran through the ground it conquered.

Photograph by Gary Todd (WorldHistoryPics). Eastern Han bronze crossbow trigger mechanism, Anhui Provincial Museum, Hefei. CC0 via Wikimedia Commons. · CC0

The crossbow answer, and the doctrine that lasted

The Han did not simply imitate the steppe; they fused what they learned with what they already had. The crossbow remained the Han signature, now carried also by mounted troops and massed against cavalry charges, its standardized bronze triggers turned out under bureaucratic quality control with workshop, date, and inspector's name cast into the metal ¹⁶⁷. The mature Han answer was a combined system: disciplined crossbow infantry to hold ground and break a charge; mobile horse-archers to screen, pursue, and envelop; fortified colonies to occupy and feed; and a state remount and supply apparatus to keep the whole machine in the field far from home.

That synthesis — settled firepower welded to steppe mobility — became the template of Chinese imperial military organization and the frame of its strategic thought. The central problem the Han worked out against the Xiongnu, how a settled agrarian empire defends and projects power across an open steppe frontier against an enemy faster than itself, defined Chinese statecraft through the Tang confrontation with the Türks, the Song failure against the Khitan, Jurchen, and Mongol, and the Ming and Qing management of the northern frontier — a continuous strategic conversation across two thousand years, opened by the defeat at Baideng ¹⁴¹³.

What was displaced

The transformation buried as much as it built. The war-chariot, already obsolescent, vanished as a fighting vehicle, and with it the last institutional trace of the Zhou aristocratic warrior whose rank had been bound to his chariot team ¹¹. The defensive, tribute-paying posture of the early Han — the heqin settlement that had treated the chanyu as an equal and bought quiet with silk — was repudiated as a humiliation to be avenged, and a foreign policy of accommodation gave way to one of expansion, garrison, and subjugation ⁴³. And the light-tax, low-spending statecraft of Wudi's predecessors, the deliberate frugality that had let the early Han recover from civil war, was displaced by an interventionist war-economy of monopolies, confiscations, and direct state commerce — the change that carried the heaviest cost, and to which the reckoning now turns.

What the cost was

The transmission's bill was not paid by the steppe alone, and not in a single currency. It was paid in horses, in silver, in conscripted and resettled human beings, in the fiscal and social fabric of the Han state, and — in one exact and documented case — in the body of the man who wrote the history we have been quoting. The cavalry transformation worked: it ended the tribute, broke the Xiongnu's dominance, and opened the west. It also came close to breaking the dynasty that achieved it, and it fell hardest on people who had no say in any of it.

The bill in horses and silver

War at steppe range devoured horses. The single Mobei campaign of 119 BCE that broke the Xiongnu is recorded as having cost the Han on the order of a hundred thousand horses, lost to combat, distance, cold, and exhaustion — a loss so heavy that it limited the empire's ability to mount further deep offensives for years afterward, because the herds could not be replaced as fast as a campaign consumed them ⁶⁸. The Ferghana wars spent whole expeditionary armies to bring home a few thousand breeding animals. And the standing cavalry establishment was not a one-time purchase but a permanent charge: the pastures, the remounts, the fodder, the saddlery, and the garrisons strung across thousands of kilometres of frontier had to be funded every year, in war and in peace alike ³⁶. An empire that had once paid tribute to avoid war now paid far more to wage it, and kept paying after the fighting paused.

The bill in people

Behind the campaigns stood a vast mobilization of ordinary lives. The conscript armies were levied from the farming population, and the deep campaigns and long garrisons spent those men — in battle, on the march, and in the disease and hunger of distant frontiers — at rates that contemporaries noticed and resented ³. The tuntian colonies that held the Hexi corridor were settled by households moved hundreds of kilometres from their homes by state order, to break ground and stand guard on an exposed frontier under arms, far from the graves of their ancestors. Late in Wudi's reign the combined weight of conscription, taxation, and corvée pushed parts of the countryside into flight and banditry, and the histories record uprisings of the desperate that the state put down by force ³⁷.

On the receiving end of the expansion were the peoples the Han now fought, displaced, and absorbed:

• the Xiongnu driven from the Ordos and the Gansu corridor, their kings killed or forced into surrender, their subjects scattered;
• the Yuezhi pushed west out of Gansu a generation earlier, an exile that rippled across Central Asia;
• the Wusun drawn into Han alliance and sealed with another princess sent to a steppe marriage;
• and the oasis populations of the Tarim, brought under Han garrisons, taxes, and hostage-taking as the protected road was driven through their territory ¹³¹³.

The opening of the Silk Road, narrated elsewhere as a story of connection and exchange, was on this end of it a story of conquest, garrison, resettlement, and the subjugation of small peoples between two large ones.

The fiscal reckoning

The deepest internal cost was structural. To pay for the cavalry empire, Wudi's government overturned the fiscal order of the early Han. It imposed state monopolies on salt (from 119 BCE) and iron (from 117 BCE), took direct control of coinage after a series of currency manipulations, taxed merchants' property and carts, sold offices and ranks and pardons for cash, and built the "equable transport" (junshu) and "price stabilization" (pingzhun) systems that let the treasury buy cheap, move goods, and sell dear on its own account ³¹⁰. This was a war-finance apparatus, run by officials like the merchant's son Sang Hongyang, and it was resented exactly in proportion to its reach.

After Wudi's death the resentment surfaced in the open court debate of 81 BCE, recorded in the text known as the Discourses on Salt and Iron ¹⁰. There Confucian critics arraigned the whole monopoly system as the ruinous and corrupting legacy of the Xiongnu wars — a state turned merchant, competing with its own people and grinding them for revenue — while the officials defended the monopolies as the only thing that had paid for victory and still paid for the frontier. The monopolies largely survived the debate, because the state could not do without the revenue and the border could not be left unfunded. That survival is itself the measure of the cost: the cavalry transformation had so deformed the empire's finances that, a full generation on, it could not be undone without admitting the frontier could not be held.

Li Ling, and the historian's body

The cost has a face, and a name we have already been leaning on. In 99 BCE the general Li Ling led a force of five thousand Han infantry deep into Xiongnu country, was surrounded by a cavalry army many times his size, fought a running battle until his arrows were spent, and surrendered rather than see the last of his men killed ². At the Han court, where the emperor wanted the defeat condemned and the defector damned, one official spoke for Li Ling: the court astrologer and historian Sima Qian, who argued that a man who had inflicted such losses against such odds before being overwhelmed was no simple traitor and had surrendered to fight another day.

For that defence the emperor had him condemned to death. Sima Qian, to live long enough to finish the history his father had charged him to write, accepted in its place the punishment of castration — a disgrace a man of his class was expected to refuse by suicide ². He chose mutilation and shame over an unfinished book. The Records of the Grand Historian, and within it the very account of the Xiongnu that gives us our fullest picture of Modu's machine and the steppe boys who learned the bow before they could walk, was completed by a man the Xiongnu war had personally maimed. The transmission's bill, itemized honestly, includes the historian who recorded it.

The longer reckoning

By the end of his reign even Wudi seems to have registered the exhaustion. In 89 BCE, in the document later known as the penitential Edict of Luntai, he refused a proposal for further military colonization in the far west and acknowledged the burden his wars had laid upon the people — an extraordinary public admission for a Chinese emperor, and one that later historians read as the moment the dynasty turned back from the edge of fiscal and demographic collapse ⁸³. The decades after his death were spent recovering.

The Xiongnu were not destroyed by any of this. They were broken northward and stripped of the Western Regions, then fractured by their own succession wars, until in 51 BCE the chanyu Huhanye submitted to the Han as a tributary — the relationship of Baideng exactly reversed, the steppe now the supplicant ⁴¹³. But the cavalry doctrine the Han had paid so much to build outlived the dynasty and every dynasty after it, and so did the strategic lesson buried in the cost: that a settled empire could indeed master the steppe's way of war, but only by remaking its army, its borders, its finances, and its idea of itself — and that the mastery, once bought, had to be paid for again in horses, silver, and men every year it was kept.