THE MILITARY SYSTEM OF THE KARAKHITANS

The exile at Kedun (ca. 1124-1130 CE), the migration into Central Asia (ca. 1131-1134 CE), and the conquest of the Karakhanid city-states of Sogdia (ca. 1134-1143 CE) marked the final phase of development in the Khitan military system, resulting in the army of the Karakhitan Empire (1143-1218 CE). When Yelü Dashi made his strategic withdrawal to the Khitan outpost at Kedun (north-central Mongolia, near modern-day Ulaanbaatar), he initially had only 200 loyal men with him, but by the end of 1130 CE he was joined by as many as 100,000 exiles. Of these, only about 30,000 were combatants—20,000 regulars and 10,000 guards—and the remainder were divided between members of the two royal clans—the Yelü and the Xiao—and the families of the soldiers. Some of the outer tribes stayed loyal, but most did not, especially after the capture of emperor Tianzuo by the Jurchen (1125 CE). This meant that Yelü Dashi was left with very meager resources to use in any attempted reconquest of the empire, although the imperial horse herds had been grazed in central Mongolia, and at the time of his retreat to Kedun Yelü Dashi was able to seize the bulk of these herds. Nevertheless, the maximum size of the army Yelü Dashi could field at Kedun would have been less than half the size of an imperial field army. Although the Uyghur city-state of Qocho (modern Turpan) remained loyal to the Khitans, the Tanguts strove for neutrality, fearing a possible Song-Jurchen alliance against them. Uyghur support faltered after Yelü Dashi made a failed attempt to seize the eastern Karakhanid city-state of Kashgar, and his forces took heavy casualties (1128 CE), but he still had sufficient forces to repel a small Jurchen army (10-20,000 men) from Kedun (1131 CE)—although this victory was achieved with the help of the Shiwei Borjigin chief, Khabul Khan. However, with the Jurchen massing for another attack on Kedun, Yelü Dashi moved his main camp to the banks of the Emil River in eastern Kazakhstan, where he also received the submission of several tribes of Karluk/Qarluq and Kangli/Qangli Turks (this swelled his forces to around 40,000 combatants). It was at that time that he assumed the title of gürkhan (“universal king”), although at that time this rather grandiose and novel title (Dashi seems to have invented it) must be seen more as a declaration of intentions than as an expression of the status quo. Indeed, the material circumstances of the Khitan exiles by that time would have been desperate—the Karakhitans would have had no real source of income, being forced to barter with local tribes using whatever wealth they had managed to take with them to Kedun, and although the imperial horse herds were both an important source of wealth and a necessary tool of the cavalry, it also made the Karakhitans a tempting target for the raids of nomadic tribes (some Turks had submitted to his authority, but others did in fact raid the Karakhitans, whittling away at their material and manpower resources). Nevertheless, we are told by later Muslim Arab, Sogdian, and Persian sources that the Karakhitans went to great lengths to maintain their ancestral traditions, including the equipment and discipline of their armed forces—Dashi's army was likely a microcosm of the old imperial system, although the sources specifically say that Dashi abolished the guard battalions, folded them into the regular regiments, and placed the regular regiments under his direct command.

 

Dashi spent four years at the camp on the banks of the Emil, and after his failure to take Kashgar, the nearest Karakhanid rulers seem to have believed he was no longer a threat. Indeed, the Karakhanid ruler of Balasagun (the city-state that dominated the Chuy Valley in northern Kyrgyzstan), seems to have invited Dashi into his kingdom (ca. 1134 CE) to help him repel an invasion of Turkic Qarluq and Qangli nomads (apparently different tribes to those that had already submitted to Dashi). This Dashi did, but after defeating the Turks, Dashi demanded that they submit to him as gürkhan, and not to the Karakhanid king. Soon thereafter, the ruler of Balasagun was also forced to submit to Dashi. Dashi did not directly occupy Balasagun, instead establishing a royal camp (husi woerduo) on the plains north of the city (the Chuy valley is pretty much the only large, flat grassland in all Kyrgyzstan). Although Dashi steadily expanded his hegemony in Central Asia, he and his successors—Tabuyan (1144-1150 CE), Yilie (1150-1164 CE), Pusuwan (1164-1178 CE), and Zhilugu (1178-1211 CE)—contented themselves with occupying only the Chuy Valley, preferring to retain local dynasts as tributary allies in each region conquered by the Karakhitan Empire. This created a state that was basically parasitic, somewhat different from the integrated state of the Khitan-Liao. Following the usurpation of the Chuy Valley and Balasagun, Dashi received another boon to his growing army—there were 10,000 Khitan warriors that had taken up service with the Ilig Khan (i.e., overlord) of the western Karakhanid Kingdom (Ibrahim bin Ahmad, 1128-1158 CE) following the fall of the Khitan-Liao Empire, and these defected to Dashi. This would have raised the total number of core Khitan troops at Dashi's disposal to approximately 40,000, plus a roughly equal number of Turkic auxiliaries (Qarluqs and Qanglis). With these enlarged forces, Dashi was able to defeat Ibrahim bin Ahmad at the Battle of Khujand (1137 CE), and to thereafter seize the Ferghana Valley and the western regions of the Tarim Basin (including the important cities of Kashgar, Khotan, and Beshbalik). However, the Karakhanids were at that time vassals of the Seljuk Sultanate—a Turko-Persian empire that at that time controlled the Iranian Plateau, Mesopotamia, the southern Caucasus, the Levant, and most of Anatolia—and it was not until Dashi defeated the combined forces of the Seljuks and Karakhanids at the Battle of Qatwan (9 September 1141 CE) that he was able to make himself master of the Karakhanid city-states (spread throughout Uzbekistan, northern Afghanistan, Tajikistan, and Kyrgyzstan). As mentioned, the Qarluq and Qangli tribes of eastern Kazakhstan were already vassals, as were the city-states in the western half of the Tarim Basin, but following Dashi's victory at Qatwan, the Uyghur kingdom of Qocho/Idiqut, which dominated the city-states in the eastern half of the Tarim Basin, also submitted to the gürkhan, as did the Yaghma Turks, who were at that time inhabiting the region of Zungharia (a plain north of the Tarim Basin), and the Chigil Turks, who were at that time inhabiting the lands around Lake Issyk Kul (a large endorheic lake in the Tian Shan Mountains in eastern Kyrgyzstan). Among the vassals of the Seljuk emperor (sultan)—Ahmad Sanjar (1097-1157 CE)—that fought at the Battle of Qatwan was the king (shah) of the Kingdom of Khwarezmia (modern Turkmenistan), Ala ad-Din Atsiz (1127-1156 CE). After his defeat at Qatwan, the Khwarizmshah was also forced to become a vassal of the gürkhan, although this would prove portentous in a way that reveals the strategic weakness of the Karakhitan Empire. Ahmad Sanjar never recovered from the devastating loss at Qatwan, and when he died with no male heirs ten years later, the Seljuq Empire collapsed. With the support of their Karakhitan overlords, the Khwarizmians were able to expand (ca. 1194 CE) into Khurasan (northern Iran) and the Central Iranian Plateau, then into western Afghanistan (ca. 1204 CE). This placed the Karakhitans at the center of a vast web of vassal-states, but the growing power of their erstwhile Khwarizmian vassals led to the growing independence of the Khwarizmshahs, who also subverted the loyalty of the western Karakhanids. The Khwarizmian king, Ala ad-Din Muhammad (1200-1220 CE), and several of the western Karakhanid rulers rebelled in 1210 CE, and defeated the Karakhitan emperor, Zhilugu, at the Second Battle of Talas (the First Battle of Talas is another famous battle that occurred much earlier at this same place between the Abbasids and the Tang), and when Zhilugu was overthrown by a Naiman Mongol auxiliary commander (Kuchlug) in 1211 CE, Ala ad-Din Muhammad went on to conquer Transoxania (ca. 1212 CE) and Afghanistan (ca. 1215 CE). The Karakhitans were soon thereafter conquered by the Mongols under Temujin Chinggis Kha'an (ca. 1216-1218 CE).

 

As with the nature of the components of the Khitan-Liao military system, the change in the strategic posture of the Khitans during their exile at Kedun, their migration into Central Asia, and the period of their hegemony in Central Asia, as well as potential changes in the tactical stance of the constituent elements of the Karakhitan army, has been much debated. There can be no doubt that strategically, the period from the exile at Kedun to the establishment of the Karakhitans in the Chuy Valley was indeed a period of hardship during which the strategic resources of the Karakhitans were much reduced from those the Khitans had enjoyed during the period of the Liao empire. However, as I stated above, the basic lineaments of the system seem to have been retained and translated into the Central Asian context. All the Muslim authors that write about the Karakhitan state stress the fact that the Karakhitans were at great pains to maintain their ancestral traditions—so much so that many of the Central Asian peoples considered them to be Chinese. Although these authors only specifically mention the unique “battle dress” of the Karakhitans, and they do not refer to unique tactics (i.e., different from those prevalent in Central Asia, which were different from those in North/East Asia), many authors have suggested that the Karakhitan army had become at least partially Turkicized. Problem is, there is also no evidence for this. We know that Turkic auxiliaries made up a much larger portion of Karakhitan armies than they had in Khitan-Liao armies, and this continued into the period of the Karakhitan Empire, when there were substantial semi-independent vassal states. However, these same Muslim sources regularly refer to the Karakhitan component in these coalition armies as “a wall of steel,” and they are openly appreciative of the discipline of the Karakhitan units. “Wall of steel” could be pure hyperbole, but I think it probably encapsulates a kernel of truth—certainly a core force of disciplined and well-equipped Khitan regulars (troopers and foragers) would have been a formidable counter to the disciplined units of ghulam/ghilman slave-soldiers fielded by the Islamicized Turko-Persian states, and they would have outclassed every other troop type fielded by these same states. There is no reason to believe that the Karakhitans temporarily abandoned these practices while migrating from Kedun to the Chuy Valley—as I stated above, the army of the Karakhitans at that time was probably a model of a Khitan-Liao imperial army in microcosm. Certainly, the hardships of migration have led many peoples throughout history to cling to every last vestige of their cultural practices, and in the case of the Karakhitans, they ultimately proved to be conquerers in Central Asia that were able to settle in a relatively isolated region with sufficient resources for them to continue to maintain their own culture (i.e., there is no evidence of significant assimilation between them and the surrounding peoples). The Karakhitan emperor, Yilie (1150-1164 CE), conducted a census of the central territories of the Karakhitan Empire during what would prove to be its peak, and at that time his officials recorded that there were 84,500 Khitan “households” in the Chuy Valley—if indeed the Karakhitans followed Khitan-Liao precedent, where each household of the inner tribes was expected to support a regular and a forager, this would mean that at the time of the census the Karakhitans could have fielded as many as 169,000 cavalrymen (regulars and foragers). Unlike the Khitan-Liao state, these troops would not have been spread out over a large territory as garrisons because the Karakhitans ruled most of their empire through vassals. This could explain why Muslim sources regularly mention that Karakhitan armies numbered in the hundreds of thousands—it's still probably part hyperbole, but at its height the Karakhitan Empire likely was able to field armies as large as those of the earlier Khitan Empire. Such a large expansion of the Karakhitan populace and army, from 100,000 people at Kedun (and 30,000 soldiers) to a populace that could field 169,000 soldiers, within a span of twenty years, needs some sort of explanation, however. There certainly could have been further migrations of Khitans unrecorded by historians—the Jurchen are almost universally portrayed in Chinese sources as being far more barbarous than the Liao, and to have violently subjugated subject peoples, and it is certainly possible that many of the Khitans that initially supported the Jurchen eventually chose to flee. Yelü Dashi's successes after about 1134 CE could certainly have fueled a desire in the Khitans of Manchuria to again live in a place where they were the rulers, and they were a people with nomad roots for whom such a migration would not be as onerous as it would be for a more settled people. Those Khitans that remained behind certainly did not enjoy a favorable fate—Khitan auxiliaries were sorely used by the Jurchen against the Song, and when the Jurchen fell to the Mongols, a group of Khitans fled to Korea, where they essentially lived as brigands, and they are believed to be the ancestors of the baekjeong (“untouchable”) outcast group in modern Korea. Although there are many places in Manchuria named after the Liao, there are no other traces of them in China, and no groups that trace their lineage to them, despite the fact that there should still have been a population of 650,000 Khitans living in Manchuria at the time of the Jurchen invasion. My personal belief is that those that did not die or flee to Korea fled to Karakhitai. It has also been suggested that when Yelü Dashi declared himself gürkhan at the Emil camp, those Mongol and Turk tribesmen that had submitted to his authority at that time were folded into the regular forces of the Karakhitan army, and therefore were no longer auxiliaries. In this context, the title of “universal ruler” may have been initially intended to represent the unity of the Khitans, Mongols, and Turks thus forged. We do not have exact figures for the number of Mongols and Turks that joined Dashi at Emil, but for them to accept him as ruler it seems unlikely that they would have outnumbered the Khitans, but this could still have added a significant element to the core population of the Karakhitans.

 

The wealth of the Karakhitan state, which came to control the middle portions of the Silk Road, meant these troops were probably also equipped to the same high standard of the Khitan-Liao period. Yelü Dashi had abolished the use of guard battalions, something his successors continued, but this would not have necessitated any major changes to the tactical functioning of the Karakhitan army (from a tactical standpoint, the guard battalions were analogous to the regular regiments), and from a strategic standpoint it probably made the Karakhitan emperors more secure on their throne due to the limited ability of any other noble to develop a regional powerbase far from the imperial court. One major innovation was the office of tayangu (“general”). This office did not exist in the Khitan-Liao army, where large field armies were usually commanded by the emperor himself, one of the regional viceroys, or a scion of one of the imperial clans. However, Karakhitan practice was to send an army in support of the forces of a given vassal state as needed (i.e., if their own troops proved insufficient to vanquish a given foe), usually in return for a supplementary payment of tribute and/or a portion of the spoils captured during the campaign. Sometimes, this meant there was need for Khitan support from more than one vassal simultaneously. In such instances, the emperor would lead the largest force where it was needed, while a tayangu would be appointed to lead a smaller army to wherever else the Khitans were needed (there does not seem to have ever been the need to appoint more than one tayangu). As with the Khitan-Liao system, however, this post was almost always reserved for a scion of one of the royal dynastic clans (still the Yelü and Xiao clans), and the command post was always limited to a specific campaign with limited objectives. Really, with the sole exception of the high degree of assimilation with the Chinese populace experienced during the Khitan-Liao period, the Karakhitan Empire looks a lot like the Khitan-Liao Empire. The Karakhanid, Sogdian, and Uyghur city-states along the Silk Road were not as large as the cities controlled by the Khitan-Liao in northern China (or the circuit capitals), but they filled much the same military role—masses of conscripts, some elite units (composite cavalry nobles, some horse archers, infantry spearmen and bowmen), and even the provision of artillery (mangonels, manjaniq, and ballistae, aradat). The Turkic auxiliaries available to the Karakhitans (Qarluq, Qangli, Yaghma, Chigil) would have filled much the same role as the outer tribes and Shatuo tributary states in the Khitan-Liao system, and would have supplied much the same types of troops. Even the Khwarizmians ultimately proved to be a nemesis of the Karakhitans reminiscent of the way the Jurchen had overthrown the Khitan-Liao. Thus, the differences between the Khitan-Liao and Karakhitan states can be summed up as a lack of assimilation between the Karakhitans and their subjects, the relatively small area inhabited and directly governed by the gürkhans, and the relative autonomy of the tributary states. Although Karakhitan armies were, during the period of exile and migration, much smaller and probably less well-equipped than those of the Khitan-Liao state, Yelü Dashi still had a sizeable force at his disposal (approximately 30,000 Khitans and perhaps another 15-20,000 Turks and Mongols), which only grew after the defection of the 10,000 Khitan warriors that had been in Karakhanid service, and which reached a maximum of 169,000 core troops by the mid-twelfth century CE. This was about the same overall size of the army of the Khitan-Liao state, but unlike the Khitan-Liao the Karakhitans did not spread these forces out over a large area to garrison territory, so the Karakhitans may actually have been able to concentrate larger numbers of troops for field armies. Unfortunately, we do not have anything like the regulations preserved in the Liao Shi to give us details about the tactical stance of the constituent units of the Karakhitan army—when I look at the evidence, I think it is probable that the Karakhitan system was broadly similar to the Khitan-Liao system, and in light of the fact that we do not have evidence to the contrary, it seems unwarranted to assume that the Khitans had reverted to the use of all light cavalry skirmisher formations, as I've read in the works of some authors. Muslim descriptions of the battles of the Karakhitans do indicate that the Khitans normally deployed for battle in a central division, and that their tributary allies were usually deployed on the wings, and this is certainly different from the three lines in succession described for the Khitan-Liao, but there is no reason to believe that the central division was not divided into a leading line of skirmishers (i.e., the foragers) and a supporting line of composite cavalry (i.e., the regular troopers), hence giving rise to the Muslim characterization of the Karakhitan forces as a “wall of steel.”

 

As a final addendum to the discussion of the Karakhitan military system, I would like to point out one significantly different feature of Central Asian warfare that seems to have found its way into the Karakhitan art of war—the war elephant. The Khwarezmians used war elephants, and these would undoubtedly have been fielded by Karakhitan-Khwarezmian coalition armies, but one such Karakhitan-Khwarizmian army defeated the army of the Ghurid Sultanate at the Battle of Andkhud in 1211 CE, which led to the acquisition of war elephants that became part of the central Karakhitan army. The Ghurids were ethnic Tajiks (probably the descendants of the ancient Bactrians) that inhabited Afghanistan and Tajikistan, but at the time of the Battle of Andkhud they controlled an empire that included Pakistan, eastern Iran, and much of northern India. War elephants were a major part of their military system, and when they were defeated at Andkhud, their surviving war elephants were taken back to Balasagun by the Karakhitans. Ironically, the citizenry of Balasagun chose that time to rebel, but the returning Karakhitan army used their newly acquired elephants to bash down the walls and gates of the city. We do not know what happened to these elephants during the final seven years of the Karakhitan Empire, or even how many elephants the Karakhitans captured from the Ghurids. Subsequently, when the Mongols conquered Karakhitai and Khwarezmia, they only encountered elephants when fighting the Khwarezmians—Temujin Chinggis Kha'an was initially impressed with the beasts, but when he discovered how much fodder they required to live (and that they wouldn't eat the same fodder as the Mongols' horses and camels), he ordered them set free on the steppes of Kazakhstan, where they promptly starved to death. It is extremely likely that the elephants captured from the Ghurids by the Karakhitans faced a similar fate, since the Karakhitans probably had no better understanding of how to care for them.