Western Xia

A family of Toba extraction

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

Parker calls it, in passing, "a well-organized state" — a description he does not extend lightly to most of the polities on China's northern and western frontier. It was founded in the Ordos region by a family descended from the Toba branch of the Xianbei, the same steppe lineage that had ruled northern China centuries earlier as the Northern Wei, and it governed there "almost independently" from the year 890 onward, long before anyone in it claimed the title of emperor.

Two dynasties, two surnames

The family's Chinese surname changed with whichever dynasty happened to be granting it favour. The Tang court gave it the name Li, in the way a struggling imperial centre rewarded a frontier family for military service it could not otherwise compel. When the Song reunified China in 960, the compliment was renewed in the currency of the new court: the family took the surname Zhao — the House of Song's own name — "presented as a compliment in turn to the early 'Emperors' of Hia." Parker's scare quotes around "Emperors" mark exactly the ambiguity the family lived inside for a century and a half: nominally a Chinese governor's house holding a Chinese-granted name, in practice a hereditary power in the Ordos answering to nobody's day-to-day command.

What the histories agree on

Parker is explicit that this was not an obscure or marginal state, whatever its position between larger neighbours might suggest. "The histories of the Cathayan, Niichen, and Sung dynasties all contain lengthy descriptions of this country" — Liao, Jin and Song each found it worth a sustained chapter of their own record, which is a different thing from being a footnote to any of the three. It had, by the time its rulers stopped being satisfied with a princely patent, an administration capable of fielding a drilled army and inscriptions in a script of its own — the same script Chinese and Western scholarship would still find largely undeciphered eight centuries later.

The patent nobody in the family found sufficient

For most of this first century and a half the family's rulers accepted the patent of Prince of Hia from whichever Chinese court was issuing one, paid the outward respect the arrangement required, and used the interval to build the state that patent nominally subordinated. It was not, on the evidence of what came next, an arrangement any of them regarded as a settled description of what they actually were.

A higher style

Drafted from Boulger, The History of China (1898)peak

Li Yuanhao inherited a state his grandfather had spent a reign consolidating and his father had spent a reign defending, and he did not intend to spend his own reign accepting the terms either of them had settled for. He drilled his soldiers daily "after the fashion of the Chinese manual," built up stores of arms, and watched the Song army opposite him grow soft on the same long peace that was letting him prepare. In 1034 he turned those preparations into open war, using a border governor's incursion into his own territory as pretext for an offensive that needed very little pretext to begin with.

Encouraged by his success

The early fighting went his way, and Boulger records what that success did to his ambitions directly: "encouraged by his success, Chao Yuan resolved to assume a higher style than that of Prince of Hia" — the patent his family had held from Song and, before Song, from Tang. His justification rested on descent rather than conquest alone: he came, he argued, of a family that had in the past borne the imperial dignity itself, and a prince descended from emperors owed no throne the deference a patent implied. He took the title of Emperor, returned the letters patent Song rulers had sent his predecessors over the years, and answered the resulting insult with an army already in the field.

Sanchuan and Yang Moulong

The war that followed went badly for Song almost throughout. At Sanchuan the two armies fought for three days before the advantage settled with Xia. The decisive engagement came near the town of Yang Moulong, where Li Yuanhao commanded in person and out-manoeuvred Song's field commanders badly enough to surround their army entirely. The fighting that followed was, by Boulger's account, never in doubt once the encirclement closed: Song's generals died one after another rather than abandon the field, and by nightfall the flower of the Song army lay dead on the plain. One commander, urged by his own soldier to retreat from a lost battle, refused outright — "I withdraw! the general of this army. The battle is indeed lost, but I can and ought to die" — and was found afterward among the thickest of the slain.

What the war bought

Song did not recover quickly from the defeat, and further trouble followed almost immediately: Liao, watching a weakened neighbour, seized the moment to occupy ten Song border cities under the terms of an entirely separate grievance. Negotiations eventually brought Xia and Song to peace in 1042 on terms that left Li Yuanhao's imperial title untouched in practice, whatever face-saving language the treaty used for Chinese consumption — an annual payment in silver, silk and tea flowing north in exchange for the return of formal deference that had cost Song several armies to fail at collecting by force.

Elder brother

Drafted from Parker, A Thousand Years of the Tartars (1895)aside?What aside means

Xia's rulers spent the eleventh century playing Song and Liao against each other with a skill neither larger neighbour ever quite matched in return. Parker's narrative of the period reads less like the history of a client state than like the history of a small power that had worked out, well before anyone had a word for it, that two rival patrons will always outbid each other for a border ally's loyalty — provided the ally never lets either one feel secure in it.

The 1028 opening

In 1028 the Xia ruler captured the Uyghur city of Ganzhou, a conquest that extended his territory west but also, more usefully to him, gave Liao a fresh grievance to raise against Song at exactly the moment Liao wanted leverage of its own. The Liao court massed troops in the Beijing plain, reopened a decades-old dispute over ten border districts, and wrote to the Song emperor — addressed, with deliberate condescension, as "elder brother" — complaining as much about frontier river-works as about Xia's expansion. Song's envoys saw through the manoeuvre immediately: the ostensible territorial claim was cover for what Liao actually wanted, which was a larger subsidy. Song's own position on Xia was blunt and confident — "China is quite competent to deal with that power satisfactorily" — a confidence the following decade would not vindicate.

The war Song lost and the war Liao won

That confidence broke first at Sanchuan and Yang Moulong, where Li Yuanhao handled Song's armies as badly as Song had claimed no border state could. Liao read the same defeat as an opportunity and used it to extract an increased annual payment from Song without having to fight for it at all — diplomacy riding on the back of a war it had contributed nothing to.

Liao's turn against Xia came a few years later. In 1044 the Liao court declared war on Xia over its habit of harbouring Liao deserters, announced the fact formally to the Song court as though seeking its endorsement, and won. In 1049 Liao announced a second victory. Xia, which had just spent a decade proving itself more than a match for Song's armies, discovered in the same decade that it was not remotely a match for Liao's.

What the arrangement was actually for

None of the three powers trusted the other two, and all three kept dealing with each other anyway, because the alternative — a war on two fronts against allied neighbours — was worse for each of them than an uneasy triangle in which loyalties shifted with the balance of recent battles. Song insisted, in its official correspondence, on being addressed as "Great Song" rather than merely "the Court of the South," a formality that cost Liao nothing to grant and Song a great deal to feel it had preserved. Xia took no side permanently and every side temporarily, and outlasted Liao by a full century on exactly those terms — still standing, more or less on its own, when the neighbour it had fought, allied with and outmanoeuvred by turns finally fell to a conqueror neither of them had yet had reason to fear.

No asylum

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

Liao's collapse in the 1120s put every one of its neighbours in the position of having to decide, quickly and with incomplete information, whether the power replacing it deserved the loyalty the old one had commanded. Xia had a century of practice at exactly this kind of calculation, and it made the choice with a speed that left no room for sentiment about an ally it had fought as often as it had supported.

The fugitive emperor turned away

When Yeliu Yanxi, the last Liao emperor, was finally run to ground by Jin forces after months as a fugitive with a shrinking escort, he had already tried and failed to find refuge with the Ordos kingdom that his own dynasty had fought wars against and made treaties with in roughly equal measure. Parker records the refusal without elaboration: "the Tanguts were afraid to give him asylum." Xia had read the direction the war was running clearly enough to know that sheltering a fugitive Liao emperor was a liability with no plausible upside, and declined to be the last friend of a dynasty that was, by that point, a corpse still technically breathing.

A new patron, chosen fast

The refusal was not neutrality. Once Liao's fall was no longer in doubt, Xia moved to secure standing with its replacement rather than wait to be conquered or ignored: it "hastened," in Parker's word, "to declare herself a vassal of the Jin" — language that credits the decision to Xia's own initiative rather than to any Jin demand. It was the same manoeuvre the dynasty had run for a century against Song and Liao in turn, adapted now to a patron with no track record yet established and therefore, if anything, more reason to value an early and voluntary submission.

What a century of triangulation had actually bought

The habit that had let Xia survive between two roughly matched rivals for a hundred years worked exactly as well against a single power with no rival left to play it against. Jin accepted the vassalage, and Xia's border, army and internal administration continued much as they had under Song and Liao's alternating attention — a state that had never depended on the balance between its neighbours surviving intact, only on being useful enough to whichever of them currently held the advantage.

The last campaign

Drafted from Boulger, The History of China (1898)end

Genghis Khan attacked Xia three times before he finished with it, and each campaign taught him something the next one used. The first, in 1207, was a raid rather than a conquest — he captured one fortress, collected the tribute the victory brought him from a section of the Kirghiz, and withdrew. Two years later he came back to finish the job and found it harder than expected: Xia's king put his full strength into the field, the campaign dragged into a river crossing and a failed attempt to flood the country, and it was only when Genghis threatened complete overthrow that the Xia king sued for peace. Genghis accepted, married the king's daughter, and folded Xia's considerable military resources into his own — depriving Jin of a potential ally on his own western flank in the same stroke. The war with Jin that had been building for years broke out the following year.

An alliance that did not survive Genghis's absence

Xia's forces fought alongside the Mongols against Jin more than once in the years that followed, an alliance neither side entered from affection. When Genghis left for four years to break the Muslim states of Central Asia, from Kashgar to the Jaxartes and the Indus, the Xia court concluded that an absent conqueror and a dead general — Muqali, his lieutenant in China, died the same year as the old Xia king — added up to an opportunity. The young ruler who inherited the Xia throne judged, correctly as far as it went, that his father's deference to the Mongols had cost more than it had bought, and refused to continue it. What he had not judged correctly was how fast Genghis could return from four years of campaigning eight thousand miles away.

The campaign that ended two lives

Genghis came back in 1223, resumed personal command against Jin, and spent 1224 in preparation for a reckoning with Xia specifically — the largest army he had yet assembled for any of his Chinese campaigns. The Xia king, who boasted he could field half a million men, discovered in the event that the boast did not matter. The Mongols took the principal cities and fortresses in succession, and in a battle fought over the frozen Yellow River — Boulger draws the comparison to Napoleon's artillery on the ice at Austerlitz — the Xia army was all but destroyed. The king who had broken with the Mongols died, Boulger writes, "of the grief caused by his numerous misfortunes" before the war was even finished; his successor gave formal submission in 1227, the year the conquest was completed.

Whose last campaign it actually was

Genghis did not live to see Jin finished, though he and Muqali between them had already broken its military power beyond real recovery. He is recorded to have refused, as one of his last acts, a peace offer from the Jin ruler whose earlier protestations of friendship he had never trusted. He died in his camp on the Xia frontier on the twenty-seventh of August, 1227 — by the most probable account a natural death, though several less probable ones were told afterward. Boulger states it plainly: the conquest of Xia was the last military undertaking of Genghis Khan's life, closing out fifty years of campaigns with the one that had, in the end, taken the two years he did not have left to give it.