Hostname: page-component-586b7cd67f-gb8f7 Total loading time: 0 Render date: 2024-11-24T20:17:21.225Z Has data issue: false hasContentIssue false

THE REACH OF THE MILITARY: TANG

Published online by Cambridge University Press:  12 July 2017

David A. Graff*
Affiliation:
Kansas State University, email: [email protected]
Rights & Permissions [Opens in a new window]

Abstract

During the Tang period (618–907), war and the preparations for war significantly influenced the shape of the Chinese empire and the lives of its people. That influence can be found on at least four levels. The first of these is the state’s demand for military service, which imposed a variety of burdens upon a significant percentage of the population. The second is the state’s demand for cash, fabric, grain and other financial and material resources to meet the needs of the military establishment, a demand that affected the whole of the taxpaying population. Third is the loss and destruction caused directly by warfare, especially massive internal conflicts such as the rebellions of An Lushan and Huang Chao. Fourth, and most far-reaching and diffuse, is the range of cultural reactions to military affairs, from the emphatic embrace of martial attitudes and behaviors to their emphatic rejection.

Type
Research Article
Copyright
Copyright © Cambridge University Press 2017 

Although most Chinese dynasties were established through the application of armed force, it has been suggested that during the Tang dynasty (618–907) the role of the military loomed especially large.Footnote 1 In the early Tang, this martial aspect was on display in a bellicose expansionism that played out in military expeditions as far afield as Mongolia, Turkestan, the Korean Peninsula, and the Tibetan borderlands. In the dynasty's second half, it took the form of chronic warlordism in the interior of the empire as provincial military commanders asserted varying degrees of autonomy from the imperial court. Setting aside the probably hopeless task of defending the comparative claim that the Tang was somehow more martial than China's other major dynasties against the assaults of Song and Ming partisans, the purpose of the current essay is to assay the extent of military influences during the Tang period. It is divided into four sections, which will survey in turn 1) the state's demand for military service, 2) its extraction of financial and other resources needed for the support of the military, 3) the manifold damage done by military activities in times of peace as well as war, and 4) the influence of the military on Tang society and culture. The picture that emerges is, perhaps surprisingly, a mixed one. Although the impact of war and the preparations for war on the shape of the empire and the lives of its people was certainly substantial, it was nevertheless far from universal. There were long periods when most of the Tang territories were at peace and the burden of military service, taxation, and other forms of resource extraction was relatively light.

THE BURDEN OF MILITARY SERVICE

In any society, the single most obvious area of military influence over people's lives is the demand for military service. In different times and places, this demand has been met by means of a great variety of institutional forms. These range from temporary, ad hoc levies of untrained peasants to professional armies composed of full-time, long-service soldiers, with attested examples of every conceivable point in between—including warrior aristocracies, mercenary bands, tribal forces, and “national guard”–type organizations combining long-term but sporadic military obligations with participation in the civilian economy. During its 290-year reign, the Tang dynasty employed an impressive variety of institutional types. At the risk of oversimplification, the succession of Tang military institutions may be divided into three basic phases: 1) a period lasting for about sixty years after the dynasty's consolidation in the 620s, during which the fubing 府兵 system of part-time soldiers organized into territorially-based regiments served as the backbone of the Tang military; 2) a transitional period lasting from the 680s until the outbreak of the An Lushan Rebellion in 755, with the fubing gradually superseded by a long-service, professional soldiery filling the ranks of permanent frontier armies; and 3) a period of continuing dominance by full-time professional soldiers, now organized into provincial garrisons, that persevered from the An Lushan Rebellion until—and beyond—the end of the dynasty in 907.

In barest outline, the fubing system of early Tang was built upon several hundred regimental headquarters each controlling between 600 and 1,000 men, locally based in designated prefectures.Footnote 2 These headquarters were known as zhechongfu 折衝府 after 636. Each of them was affiliated with one of the imperial guard commands in the capital, where its men did one-month tours of guard duty according to a complex schedule of rotation. The more distant a unit was from the capital, the more shifts (fan 番) into which it was divided, and, as a result, the less often each man's turn came up. When the situation demanded, troops from these headquarters were assigned to ad hoc expeditionary armies led by generals dispatched from the guard commands. Once the emergency was over, the troops returned to their prefectures and the generals to the capital. When not on guard duty or campaign, the soldiers were expected to engage in military training and farming in their home communities under the supervision of their unit commanders. In exchange for full or partial exemption from taxation, they were supposed to furnish their own provisions and most of their own equipment. Men were enrolled as fubing around the age of twenty and continued to serve until sixty.Footnote 3 Tang regulations stipulated that new soldiers were to be selected from among the eligible males on the basis of the three criteria of wealth, strength, and number of males in the household.Footnote 4 It is thought that in the late seventh century, when the system was at its height, approximately 600,000 men were enrolled in the various regiments.Footnote 5 This would have amounted to more than one percent of the population of the empire, and an even higher percentage of the registered population over which the government enjoyed full administrative control.Footnote 6

Each fubing was expected to practice certain martial skills, such as archery, on a daily basis, and every winter, during the agricultural slack season, he would join his comrades for an intensive period of organized drill during which the regimental commander would organize his men into formations, hold mock combats, and teach cooperation and coordination by leading them on large-scale hunts.Footnote 7 Nevertheless, these arrangements were clearly predicated on the assumption that most of the time the soldiers were engaged in non-military activities that made them economically self-supporting, without any payroll to be met from the state's coffers. Most of them no doubt drew their living from agriculture. Since wealth was a criterion for selection, it seems unlikely that all the men would have worked the land with their own hands—though from the first half of the eighth century there is fragmentary evidence that fubing were most likely to be drawn from the poorest households.Footnote 8

More obvious than the inequitable distribution of the burden of military service by wealth and class is its skewed geographical distribution. Of the 638 Tang zhechongfu that can be located with confidence, no fewer than 419, or 65.7 percent of the total, were located in the two circuits of Guannei and Hedong (roughly corresponding to today's provinces of Shaanxi and Shanxi).Footnote 9 And within Guannei, as many as 131 regiments may have been concentrated in the metropolitan prefecture of Chang'an (the Jingzhaofu 京兆府).Footnote 10 The remaining 219 regiments were scattered across all the rest of the empire, with vast stretches of eastern, central, and southern China being almost entirely devoid of zhechongfu.Footnote 11 The reasons for this uneven distribution have been much debated, and are very likely connected to both the regime's origin in the northwest and its distrust of other regions of the empire.Footnote 12 Its implications with regard to the demand for military service are much clearer. The concentration of the great majority of the territorial regiments in the northwest meant that fubing accounted for a very high proportion of the population of the region. Kikuchi Hideo has estimated that in some of the prefectures of Guannei more than half the adult males were soldiers, and nowhere in that circuit were they less than 20 percent. In some prefectures of the Longyou circuit, west of the capital in today's Gansu, almost all the adult males in the registered population must have been required to serve as fubing.Footnote 13

The odds of being selected for military service were considerably less for inhabitants of the 240 or so prefectures that had no regimental headquarters, but they enjoyed no absolute exemption. The early Tang made extensive use of temporary levies of peasant conscripts from prefectures without fubing. Those assigned to frontier garrisons were known as fangding 防丁 (“defense conscripts”), who were selected by the local civil authorities from among the registered, corvée-eligible adult males in accordance with annual quotas imposed by the center. They were supposed to serve for a fixed period, probably one year, before returning home.Footnote 14 The ranks of the expeditionary armies, meanwhile, were filled out by the so-called “conscript-recruits” (bingmu 兵募).Footnote 15 Though they lacked the extensive training given the fubing and were therefore of less value on the battlefield, the bingmu represented a far more flexible form of military manpower. Men were chosen from among the general population to serve in an expeditionary force for the duration of its campaign, at the end of which they were demobilized and sent home. The center could assign troop quotas to any of the empire's prefectures, and the number of men demanded could be tailored to fit the anticipated requirements of the campaign. The conscript-recruit system enabled the government to extract manpower from the many prefectures in which territorial regiments had not been established, and troops could be drawn as needed from the areas nearest the scene of hostilities—or from districts whose inhabitants possessed useful skills, as when men from the Shandong coast and the watery Lower Yangzi region were sent on seaborne expeditions to the Korean peninsula. The numbers of bingmu raised for individual campaigns could be quite large; in 661, for example, more than 44,000 men were raised from 67 prefectures in Henan, Hebei, and Huainan to participate in operations against Koguryŏ.Footnote 16 In the early years of the Tang dynasty, and especially before the early 660s, it seems that bingmu quotas were at least partly filled through voluntary enlistment, with the local authorities then having recourse to conscription in order to make up the difference. As time passed, the voluntary element largely disappeared.Footnote 17

Major changes to the early Tang military system were set in motion in the late seventh century by the emergence of powerful new opponents, particularly the Tibetan empire and the revived Eastern Türk qaghanate. Expeditionary armies, even those of enormous size that were sent out in the last decades of the seventh century, were no longer able to inflict decisive defeats on the enemy and return home in a reasonable length of time with their missions accomplished. Unable to deliver a knockout blow, yet at the same time unable to withdraw when undefeated enemy forces were still capable of threatening exposed border prefectures, more and more expeditionary armies had to settle down in static encampments that gradually evolved into permanent garrisons. In the first half of the eighth century these garrisons coalesced into ten large frontier commands, nine of which were led by officials with the title of military governor (jiedushi 節度使). As of 742, each of the largest five frontier commands controlled between 55,000 and 91,000 soldiers.Footnote 18 In the late seventh century, recognizing that men with families and farms in the interior of the empire could not be retained on the frontier indefinitely, the government set up rotational systems to cycle replacements, both fubing and bingmu, in and out of the army garrisons.Footnote 19 As time passed, the authorities had more and more recourse to conscript-recruits, now almost entirely conscripts rather than willing recruits, to fill the ranks of the frontier armies.Footnote 20 They were a much more flexible form of manpower than the fubing, as the number of men in service could be adjusted from year to year and soldiers were not tied down by competing commitments such as rotational guard duty in the capital. By the early years of the Kaiyuan period (712–41) it was usual for conscript-recruits to serve two- or three-year tours of duty on the frontier, with an upper limit of four years established in 717 for the remote northwestern frontier, where long distances made the logistical cost of frequent rotation especially onerous.Footnote 21

The trend, however, was toward the gradual replacement of rotational service by long-term service in the frontier armies. Those who enlisted for longer periods of time became known as jian'er 健兒, or “sturdy lads.” Not only did the recruitment of such men reduce the cost of moving troops to and from the frontiers, but it also meant that the most experienced soldiers—those at the end of their tours of duty—could be retained for further service in the armies. The year 737 saw a decisive shift toward re-registration of soldiers and their dependents in the garrisons, and toward voluntary rather than coercive enlistment. Declaring that the relatively peaceful conditions prevailing on the frontier made it appropriate to reduce the heavy burden of military service that had been imposed on the populace, an edict issued on June 12, 737, called for military governors to calculate their manpower needs and establish quotas to be filled by the recruitment of currently serving bingmu and members of migrant families (kehu 客戶) who were expected to serve as long-term jian'er in the frontier armies. As inducements, the government offered material compensation “above the usual standard,” permanent exemption from taxes and corvée, and provision of land and houses for dependents who accompanied soldiers to the frontier; moreover, the military governors were instructed to treat these recruits with the utmost solicitude.Footnote 22 The offer proved attractive. In the first month of 738, with quotas already close to being filled, it was decreed that the remaining conscripts serving on the frontier would be allowed to return home, and that henceforth no more men were to be conscripted for service in the army garrisons.Footnote 23

On the eve of the An Lushan Rebellion, the jian'er were the most common and characteristic troop type in the frontier armies. As with the fubing in the early years of the dynasty, however, they did not account for the whole of the military establishment. Non-Han contingents played an important role, especially in the cavalry.Footnote 24 Part-time militiamen known as tuanjie bing 團結兵, first recruited in response to the Qidan invasion of Hebei in the 690s, were grouped around the capital and in certain frontier prefectures. These militiamen in some ways resembled the old fubing, though there were also significant differences. The burdens placed on them were much less (there was, for example, no requirement that they perform guard duty in the capital) and their command structure was greatly simplified (in most cases, the tuanjie bing were controlled by the prefect of the prefecture in which they resided).Footnote 25 Nor had the conscript-recruits disappeared entirely. Despite the reported success of the 737 reforms, the army garrisons were not always able to fill their quotas with volunteers, and the jian'er system proved completely inadequate when it became necessary to raise large new forces in a very short time in order to mount an especially strong expedition or cope with a sudden emergency.Footnote 26 The disastrous campaign against the southwestern kingdom of Nanzhao in 754 cost the lives of tens of thousands of conscript-recruits, mostly to disease, and when the great rebellion of An Lushan erupted at the end of 755 bingmu were hastily raised in Hexi and Longyou and rushed to the defense of the capital.Footnote 27 One of Bai Juyi's 白居易 most famous poems tells of a man who crippled himself to avoid participating in the Nanzhao campaign.Footnote 28

The turbulent years of rebellion and civil war from 755 to 763 saw the number of military provinces increase from the original ten frontier commands to more than forty, most of which were now located in the interior of the empire. Some of these provinces, especially those in Hebei, were controlled by rebel generals who had come to terms with the imperial court, but the great majority of the new military commands had been created to meet the exigencies of the crisis and were assigned to loyalist commanders as bases from which to extract the human, material, and financial resources they needed to support their armies.Footnote 29 For the remainder of the Tang dynasty and on into the subsequent Five Dynasties period (907–60), the autonomous or potentially autonomous province remained the basic framework for military organization. With provinces of “loyalist” origin capable of being as troublesome and threatening as the ex-rebel commands, the imperial court found it necessary to place increasing reliance on the Imperial Palace armies. These had been rather modest guard forces before 755, but now they were greatly expanded, placed under the control of palace eunuchs who were the most trusted servitors of the throne, and sent into the field as a major component of imperial campaign armies.

Like the province-sized military command, the professionalized armies of long-service jian'er metastasized during the An Lushan Rebellion and continued to thrive in the centuries that followed. Though now more often referred to as guanjian 官健 (“government stalwarts,” or regulars) rather than jian'er, the majority of the soldiers in both the provincial forces and the Imperial Palace armies were long-service professional fighting men. The provincial soldiery in particular differed from their predecessors, the frontier jian'er of Xuanzong's time, in two important respects: They were completely separated from civilian life and productive economic activities, and they quickly developed a keen sense of corporate identity and collective self-interest that often led them to strike out violently when they felt their interests to be threatened. Although there were occasional exceptions, guanjian were usually obtained almost entirely through the voluntary enlistment of willing recruits.Footnote 30 As in the years before the rebellion, the greatest pool of potential recruits was provided by the landless, desperate, and destitute, and particularly by farmers who were no longer able to wrest a living from the land after the breakdown of the equal-field system and the formation of private estates; military commanders were actually discouraged from recruiting among the registered population for fear that it would weaken the tax base.Footnote 31 Once a man had enlisted he would most likely spend the rest of his life in the army, and he might remain on the roster even after he had become too old or infirm for active service. In contrast to the jian'er of the pre-rebellion period, many of whom had still been involved in rotational service and part-time farming, the late Tang regulars were full-time mercenary soldiers who were entirely dependent on the pay they received from their commanders to sustain both themselves and their families.Footnote 32 This pay included a grain ration for the soldier and enough silk or hemp cloth to provide him with outfits of spring and winter clothing, and there might also be allowances of salt, wine, soy sauce, and vinegar. In addition, soldiers expected to receive grain rations for their dependents.Footnote 33 Soldiers were no longer responsible for their own upkeep, nor were they bound to their commanders by traditional ties of dependence. In keeping with the growing importance of market forces in late Tang China, most soldiers were true mercenaries who were loyal to their commanders only to the extent that they found them good and reliable paymasters.

With loyalty now dependent on the cash nexus, both the central government and the provincial authorities made sure that soldiers were well compensated. In addition to their regular rations, soldiers might also receive many additional rewards of various sorts. There were bonuses of silk cloth for men setting out on campaign, and further rewards followed successes in the field. In some provinces, the soldiers received bonuses at major holidays (such as the Spring Festival) and on the occasion of the installation of a new military governor.Footnote 34 The men of elite headquarters guard units were especially well rewarded. Those of Xuzhou's “Silver Swords” who were on duty at the governor's yamen were treated to daily banquets, and one governor went so far as to ply them with drink, pat them familiarly on the back, and sing songs for their entertainment.Footnote 35 The soldiers came to think of above-quota rewards as a regular part of their compensation, and any reduction of customary rewards and bonuses could easily spark a violent mutiny. In the tenth month of 783, for example, the emperor Dezong was driven from his capital by the revolt of elements of the Jingyuan 經原 Army, about to set out for the fighting front in Henan, who discovered that their departure bonus of silk was less than they had expected.Footnote 36 A modern Chinese historian has counted ninety-nine instances of mutiny in the late Tang provincial armies, with financial matters the single most important source of the soldiers’ discontent.Footnote 37

Before the An Lushan Rebellion the Tang government had often encountered difficulty in finding and retaining men to serve in the military, and desertion had always been a serious problem. In the fragmented political landscape of the post-rebellion period, however, both central and provincial leaders found themselves dependent on soldiers who were determined to extract the best possible deal for themselves. As a result, the professional soldiery came to resemble a privileged and parasitic caste.Footnote 38 Some units took on a hereditary character as sons followed fathers in the ranks; when a man was killed in battle a son or brother might claim the privilege of inheriting his military status and emoluments. Soldiers were unwilling to give up their iron rice bowl when they became too old for combat, with the result that in some provincial armies many of the troops were not fit for duty.Footnote 39 All of this made for upward pressure on the size of the military establishment. In 805, at the beginning of Xianzong's reign, there were approximately 800,000 regular soldiers in the empire, and by the time of his successor, Muzong, the number had increased to nearly a million.Footnote 40 Nor did these regulars represent the full extent of the military. Short-term, ad hoc conscription was rarely seen after the An Lushan Rebellion, but militia forces of the tuanjie type (now more often tagged as tuanlian bing 團連兵) were maintained in many of the military provinces. Like the men of the tuanjie units in the pre-rebellion period, these militiamen supported themselves by farming, received periodic training during the winter slack season, and rendered part-time military service in exchange for exemption from taxes and corvée. When called to the colors they usually performed in a defensive role in their home areas, and were seen as a valuable supplement to the regular troops.Footnote 41 In one province, the mountainous and impoverished Ze-Lu 澤潞 command in southeastern Shanxi, militiamen were actually the mainstay of the army. Unable to support a substantial regular force from the meager local resources, the province's leaders in the 760s selected one out of three registered adult males to serve in the militia, and set them to practicing archery at home. Within a few years, Ze-Lu was able to field a crack infantry force of 20,000 men.Footnote 42

When considered in the aggregate, the demands for military service that Tang rulers imposed on their subjects do not appear to have been exorbitant. In the especially well-documented year of 742, when the Tang dynasty was at the height of its power, the registered population of the empire was recorded as 48,909,800. That same year the frontier armies numbered 476,900 men.Footnote 43 The ratio of soldiers to the overall population (or “military participation ratio”—M.P.R., to use the term coined by the sociologist Stanislav Andreski) was 1:103.Footnote 44 This was approximately the same ratio as in the late Roman Empire, and considerably less burdensome than the ratios for some late agrarian European states, such as France in the last years of Louis XIV (1:50) and Prussia in 1740 (1:27).Footnote 45 If we adopt the larger figure of 600,000 for the mid-eighth-century Tang military establishment suggested by some scholars, the ratio rises to 1:83.Footnote 46 But this may be more than offset by the consideration that the official figure of nearly 49 million represents the registered population of the empire, which was necessarily smaller than the actual population (since registration was a prerequisite for taxation, and the empire's administrative resources were not unlimited).Footnote 47 If we recalculate the mid-Tang M.P.R. on the basis of a larger military and the late eighth-century scholar-official Du You's 杜佑 estimate that the actual population before the An Lushan Rebellion was 13–14 million households (or about 70 million individuals), we are left with an even more favorable M.P.R. of 1:117.Footnote 48 It seems unlikely that the regular military establishment of the mid-to-late seventh century, also thought to have numbered about 600,000, could have yielded an M.P.R. of greater than 1:80 when measured against the actual population of the empire. The ratio may have risen to 1:70 by the early ninth century, when it was claimed, as we have seen, that there were nearly a million men enrolled as regular soldiers. Nevertheless, these figures strongly suggest that a great majority of the adult male subjects of the Tang dynasty were never sent into battle or required to perform military service, even when the ad hoc conscription of bingmu for campaign armies in the pre-rebellion period is taken into consideration.Footnote 49

One reason why this was so, especially during the first century of Tang rule, was the contribution made by the various nomadic peoples of the Inner Asian steppe who had accepted Tang authority. The defeat of the qaghan of the Eastern Türks in 630 was followed by the submission of more than 100,000 of his people who were then “settled” in the Ordos region and elsewhere along the northern periphery of Chinese habitation. Subsequent campaigns and diplomatic efforts induced more distant peoples, such as the Western Türks and the Tiele 鐵勒 tribes dwelling north of the Gobi, to become vassals of the Tang emperor. These outlying, non-Chinese groups were governed through the system of “bridle prefectures” (jimi zhou 羈縻州), with their leaders accepting nominal appointments as prefects or governors-general (dudu 都督) while continuing to exercise their traditional authority as tribal chieftains. As Tang vassals, they were often called upon to contribute soldiers for expeditionary armies. Though usually subject to the overall authority of an imperial general, the steppe warriors served under their own leaders and retained their own organization separate from that of the Chinese.Footnote 50 These tribal auxiliaries sometimes outnumbered the Chinese forces (as in 651, when 30,000 Han Chinese soldiers and 50,000 Uighur cavalry were mobilized to attack the Western Türk Shaboluo 沙缽羅 Qaghan), and on some occasions armies composed entirely of steppe warriors were ordered into the field (as in 646, when the Chinese general Li Shiji 李世勣 led 20,000 Tiele cavalry to attack the Xueyantuo 薛延陀).Footnote 51 The military contribution of the steppe became less pronounced during the course of the eighth century as the character of the Tang armies changed and the focus of conflict turned inward, though allied contingents of Uighurs played a major role in the defeat of the An Lushan Rebellion and the Shatuo 沙陀 Türks supported the Tang court against the rebellions of the late ninth century.Footnote 52

For Chinese, the likelihood of military service varied greatly depending on the time and place. During periods of internal conflict, most notably the Tang consolidation up to about 623 and the An Lushan Rebellion of 755–63, the demand for fighting men skyrocketed throughout the affected regions, and even those who were not formally enrolled as soldiers might find themselves pressed into service to defend a besieged town or driven to join a local protective militia, a rebel army, or even a predatory bandit gang as a matter of survival. And as we have seen, even in times of relative peace and order the burden of military service was unevenly distributed between regions. During the heyday of the fubing system, almost all the registered males of certain prefectures in the areas around the capital must have been enlisted in the regiments. Later, during the post-rebellion period, the militia forces raised by some of the military provinces embraced a very high percentage of the population. The 20,000 archers raised by Ze-Lu accounted for one-third of the province's registered adult males and yielded an M.P.R. of 1:27—the same as Prussia in 1740.Footnote 53 And when troops were raised for specific campaigns or for garrison service on the frontier, heavy reliance was usually placed on the prefectures nearest the war zone to reduce transport costs and ensure a higher level of locally appropriate skills (such as horsemanship or mounted archery). Many of those who were inducted into the military, whether as fubing in the seventh century or as jian'er in the eighth and ninth centuries, assumed obligations that were effectively lifelong. Untold thousands died of wounds or disease while serving in the field, often with dire consequences for their families back home. And even some of those who managed to avoid these risks, like Bai Juyi's “old man with the broken arm,” did so only at the cost of self-inflicted, debilitating injuries.Footnote 54 Viewed from this perspective, the Tang burden of military service cannot be considered to have been light.

THE EXTRACTION OF RESOURCES

In the states and empires of the pre-industrial age it was usual for a larger portion of government revenues to be dedicated to the military than to any other purpose,Footnote 55 and Tang China was no exception to this general rule. Denis Twitchett found “the upkeep of the armies” to be “the largest item of state expenditure” during the reigns of Wu Zetian, Zhongzong, and Ruizong, exceeding even official salaries.Footnote 56 This situation continued and even intensified under Xuanzong. Around the end of the eighth century, Du You reported that before the beginning of the Kaiyuan period (713) frontier defense had cost no more than two million strings of cash each year, as opposed to ten million at the beginning of the Tianbao period (742) and fourteen or fifteen million on the eve of the An Lushan Rebellion in 755.Footnote 57 After painstaking analysis of the more detailed figures for imperial finances during the Tianbao period provided by Du in Chapter 6 of the Tong dian, the modern historian Li Jinxiu has estimated that out of a total annual expenditure of 33,064,300 strings of cash, 9,042,500 strings (or 27.35 percent of the total) went to support the military.Footnote 58 The increasing cost of the armies is usually attributed to the replacement of short-term conscripts and part-time, self-supporting fubing with the much better remunerated jian'er, though Li considers increases in the emoluments and other rewards given to officers to be responsible for the growth of military spending by an additional 1,440,000 strings of cash before the end of the Xuanzong's reign.Footnote 59 Overall military expenditure in the post-rebellion period, taking both central and provincial armies into account, was even greater. The governors of the autonomous military provinces taxed their populations heavily, with most of this revenue going to support the armies.Footnote 60 Considering the empire as a whole, with the upkeep of a soldier costing twenty strings of cash or more per annum, the entire military establishment required nearly 20 million strings of cash—almost two-thirds of the entire government revenue for a single year.Footnote 61 And these were only the peacetime costs. When wars broke out, especially large-scale civil wars, expenses ballooned as military governors demanded subsidies and their soldiers expected above-quota rewards. The campaign against the renegade military governor Li Xilie 李希烈 in Henan in 783 required more than 1,300,000 strings of cash each month, at a time when the Tang court had many other security concerns and other armies in the field; imperial operations against Wang Chengzong 王承宗 in Hebei in 809–10 cost a total of 7,000,000 strings of cash.Footnote 62

The Tang government regarded the upkeep of the military as a matter of primary importance. During times of fiscal constraint military expenditures were given priority over other disbursements, and prior to the An Lushan Rebellion the funds were drawn from the court's largest and most regular sources of revenue, the zu 租 tax (collected in grain), the diao 調 tax (collected in silk or cotton cloth), and the yong 庸 corvée exemption tax (usually also paid in textiles).Footnote 63 The grain was used to feed the soldiers while the cloth was put to several uses. The largest part was intended to be made into clothing, but since cloth also served as a form of currency it could be allocated for the local purchase of grain on the frontier (hedi 和糴) or earmarked for “special payments” which included rewards and bonuses for the soldiers.Footnote 64 Thus, the burden of supporting the Tang military spread far beyond the men who were expected to perform military service to include all the taxpaying households of the empire, virtually the whole of the registered population. During times of relative peace and prosperity, such as the first half of the eighth century, there is no reason to think that the tax burden relating to the military was unbearable, but periods of intense and widespread internal conflict such as the An Lushan Rebellion and the revolts of the military governors in the early 780s were a different story. To finance his war against the Hebei military governors, Dezong not only imposed extraordinary taxes on business transactions and homes and buildings, but also resorted to extracting forced loans from merchants.Footnote 65 Those who were caught trying to evade the taxes by concealing their assets were punished with sixty blows of the heavy stick. These taxes, which fell hardest on people with fixed assets but little cash, were deeply resented and may help to explain the ease with which the soldiers of the Jingyuan Army were able to get control of Chang'an after they mutinied in the tenth month of 783.Footnote 66

In addition to taxation, the Tang government had various other means of extracting resources needed by the military. During the early Tang the fubing were required to furnish or purchase their own clothing and many items of equipment, including the pack horses (or donkeys) for each ten-man squad and perhaps even basic weapons such as bows and sabers.Footnote 67 Short-term conscripts (bingmu and fangding) received more from the state, but when funds were insufficient neighbors were expected to chip in to provide or help purchase necessary items—a custom that was later applied to the tuanjie militias as well.Footnote 68 As a sort of informal levy, local governments could also ask people to contribute toward the purchase of military horses, with ten men together providing either a horse or funds sufficient for the purchase of one.Footnote 69 And during military crises the government was capable of requisitioning privately owned horses, with or without compensation.Footnote 70

Human labor was also subject to requisition, with porters conscripted from local communities to provide logistical support for military campaigns.Footnote 71 This was a particularly onerous burden during the civil wars of the late Tang and Five Dynasties, though Emperor Yang of the Sui dynasty had reportedly called up more than two million men to haul supplies for his 612 expedition against Koguryŏ.Footnote 72 More routine and circumscribed tasks in support of the military were performed by men chosen for miscellaneous labor service (zayao 雜徭) or special corvée duties (seyi 色役, fanyi 番役) in their home areas. Men in these categories were, for example, used to staff the network of beacon stations that provided an important element of the empire's strategic communications and a sort of early warning system on the frontiers. According to a modern estimate, the beacons were served by a total of at least 31,000 and possibly as many as 68,000 men.Footnote 73 The prefectural workshops that produced weapons, armor, and other items of military equipment for government use were also in part staffed by ordinary commoners who were assigned to them as a form of miscellaneous labor service.Footnote 74 Although several prefectures in the Yangzi valley were renowned for their production of high-quality weaponry, in the mid-eighth century the largest center of arms making was almost certainly the Tang northern capital of Taiyuan, the site of the Directorate for Armaments (junqi jian 軍器監) from 728 onward. Tasked with the fabrication of “heavy” weapons such as crossbows, this agency is thought to have had approximately 20,000 workers.Footnote 75

The Tang military did not simply rely on commodities and funds, deriving from tax revenues, that were supplied to it by the government, but also made a very large contribution to its own sustenance. Military agricultural colonies (tuntian 屯田) were set up near military garrisons, especially in the northwestern regional commands of Hexi, Longyou, and Shuofang where transported grain had to come by overland routes from or through Guannei, itself a grain-deficit region. By about 737 there were no fewer than 932 military colonies distributed through the border provinces, and in 749 a total production of 1,913,960 shi of grain was recorded—enough to feed more than 265,000 soldiers for a year.Footnote 76 The tuntian were cultivated mainly by soldiers, supplemented with tunding 屯丁 (ordinary commoners assigned to a form of special corvée service) and even exiled convicts.Footnote 77 According to Li Jinxiu's estimates, during the Tianbao period the military agricultural colonies generated about 21.5 percent of all the resources needed for the upkeep of the empire's armies.Footnote 78

Another area where the Tang military was to a significant extent self-supporting was the supply of cavalry mounts and other horses for riding. Given the important tactical role of cavalry in the warfare of the time, an adequate supply of good horses was essential for military effectiveness. Prior to the An Lushan Rebellion, the Tang military drew most of its saddle horses from the extensive government-run pastures in Longyou and Hexi, to the west and northwest of Chang'an. During the 660s the pastures had at least 400,000 and perhaps as many as 706,000 horses; in 754, on the eve of the An Lushan Rebellion, the number was 325,700.Footnote 79 According to Li Jinxiu, the annual cost of operating the pastures in the mid-eighth century was 1,744,400 strings of cash, of which only 700,000 were provided by the government out of tax revenues, with the remaining 60 percent being generated by the pastures themselves.Footnote 80 From 756 onward, Tibetan encroachments on the horse-breeding areas dictated that the Tang military would have to rely mainly on horses purchased from the Uighurs and other steppe peoples, financed from tax revenues.

This is just one instance of a more general pattern. During the early Tang the military was to a significant degree self-supporting, and the tax burden imposed on the populace to maintain the armies correspondingly light. This was no longer true after 755 as military expenses increased at the same time that the contribution from the tuntian was sharply diminished. The later Tang was nevertheless a period of marked economic growth. The relationship, positive or negative, between the financial demands of the military and the overall health of the economy is a problem that deserves further investigation.

THE DEVASTATION OF WAR

The Tang military had an important role in protecting public safety and social order. During the early Tang the territorial regiments in the provinces were responsible for guarding imperial tombs, detached palaces, canals, granaries, passes, and fords, while the fubing doing their rotational service in the Left and Right Jinwu 金吾 Guards acted as a sort of metropolitan gendarmerie, patrolling the broad avenues of the capital at night and manning police posts at the major intersections.Footnote 81 Inevitably, however, not all soldiers were well behaved and law abiding. When an expeditionary army was on campaign, it was required to make its camp at least three miles away from the nearest city or town and was also supposed to stay clear of crops growing in the fields. “Should there be a need to enter the town to make purchases,” wrote the early Tang general Li Jing 李靖, “an administrative assistant from among the encampment functionaries gives permission and dispatches a man to provide control and leadership; it is not permitted simply to enter the walled town.” This was done, he explained, in order to prevent “drunkenness, brawling, theft, and sexual improprieties.”Footnote 82 When troops were operating in hostile territory or had just defeated the enemy in battle, their behavior was likely to be even worse and might include wanton or excessive killing, widespread looting, digging up graves (probably also for purposes of looting), setting fire to houses and huts, trampling crops, and cutting down trees. All of these activities had to be prohibited on pain of death.Footnote 83

By widening the gap between soldier and civilian, the shift to a full-time, long-service soldiery in the eighth century does not appear to have produced any improvement in behavior. Politically privileged corps d’élite in both the capital and the military provinces were especially problematic in this regard.Footnote 84 At the center, the eunuch-controlled Shence Army was a sort of state within the state whose members were not subject to investigation and prosecution by the Censorate and other judicial authorities. Since men who enjoyed the status of soldiers of the Shence could get away with murder and a great many other crimes, the force attracted many recruits who were not really interested in soldiering. And the eunuchs, for their part, were happy to enroll them in order to collect the commutation fees that these nominal “soldiers” paid to avoid having to show up for duty.Footnote 85

Although the military presence could be disruptive even in times of peace, the disturbance caused by war was much greater. At various times, frontier regions in the northern and western parts of the empire were troubled by invasions of external forces. During the 620s the pressure of the Eastern Türks was felt all along the northern frontier zone, from Hebei and Hedong through Guannei (and the environs of the capital) to Longyou. The years 696 and 697 saw a brief but devastating invasion of Hebei by another nomadic people, the Qidan, followed by an incursion of the Türks into the same area in 698. In the west, meanwhile, the Tibetans were a near constant menace, real or potential, from the 630s down to the collapse of their empire in 840. Their position enabled them to threaten everything from the Sichuan basin northward to Hexi, Longyou, and the remote garrisons in what is now Xinjiang. Tibetan inroads in areas immediately west of Chang'an became especially serious in the aftermath of the An Lushan Rebellion, and the capital itself fell under their control for several weeks near the end of 763.Footnote 86

By far the worst damage, however, was done by China's internal conflicts and civil wars. The decade-long, multi-cornered struggle that extended from the terminal crisis of the Sui dynasty through the consolidation of the Tang devastated many of the core regions of the empire. The heavily populated North China Plain was repeatedly ravaged by the contending armies, many of which subsisted by plundering the districts through which they passed. When the city of Luoyang was under siege in the spring of 621, it was reported that, “The millet in the granaries was nearly gone, and people in the city were eating one another. Some took dirt and put it in vessels, using water to separate out the impurities. The sand and pebbles having settled to the bottom, they skimmed off the mud that was left floating on the surface, threw scraps of rice into it, formed it into cakes and ate them.”Footnote 87 According to another source, “Everyone was sick, with swollen bodies and weak feet. Corpses lay on top of each other in the streets.”Footnote 88 In 632, when the empire was once again at peace, one of Tang Taizong's ministers, with pardonable exaggeration, described what had once been one of China's most populous regions as a vast wasteland stretching from Luoyang eastward to the seacoast.Footnote 89

After more than a century of recovery and peace in the interior of the empire, military disaster struck again at the end of 755 with the rebellion of the frontier general An Lushan. For the next eight years, both rebel and loyalist armies rampaged across the same populous core regions of northern China that had suffered most in the early seventh century. Hebei and Henan were especially hard hit, along with Hedong and Guannei, and several smaller revolts broke out in the Yangzi valley. The two capitals, Chang'an and Luoyang, both changed hands more than once, with Luoyang being subjected to an especially savage sack by the Uighurs, fighting as allies of the Tang court, in 762.Footnote 90 Another great city of the empire, the commercial center of Yangzhou in the lower Yangzi region, was pillaged by a nominally loyalist army in 760, with several thousand foreign merchants (shang hu 商胡) slaughtered.Footnote 91 After less then two decades of relative peace, Hebei, Henan, and Guannei were fought over again in the civil wars of the early 780s. And yet another round of internal conflict occurred in the early ninth century when Xianzong sought to subdue the autonomous provincial military commands.

Worse was to still to come in the second half of the ninth century with the devastating rebellion of Wang Xianzhi 王仙芝 and Huang Chao 黃巢, which had its roots in widespread social distress and banditry in the areas between the Yellow River and the Yangzi. After Huang Chao emerged as the main rebel leader in 878, he led his army to the far south of China and sacked the prosperous trading port of Guangzhou before turning north again to capture the secondary capital of Luoyang and then Chang'an itself at the beginning of 881. The nearly two and a half years that the rebels occupied the capital were especially bloody and brutal. Their initial entry into the city was marked by robbery, murder, and rape on a very large scale, an episode famously memorialized in Wei Zhuang's 韋莊 poem “Lament of the Lady of Qin.”Footnote 92 Tang officials and other members of the elite were particular targets, both of reprisals ordered by Huang Chao in response to their unwillingness to serve his regime and of random violence fueled by the social resentments of Huang's followers. But great suffering was also inflicted on the urban population more generally. When Huang recovered Chang'an after having been briefly evicted from the city by an opposing army, he ordered the massacre of tens of thousands of inhabitants for their perceived disloyalty. And as is often the case, war gave rise to famine. The prolonged confrontation between Huang's forces and ostensibly loyalist armies in Guannei led to a near cessation of farming in the region, with the result that deaths by disease and starvation probably exceeded those caused directly by violence.Footnote 93 During the summer of 882, “the people of the capital region built stockades in the mountain valleys to protect themselves, and were unable to farm. Grain sold for thirty strings of cash for a single peck, and people crumbled tree bark into powder and ate it. There were those who seized people from the stockades and sold them to the rebels as provender; a person brought in several hundred thousand cash.”Footnote 94 Although Huang Chao's army was forced to evacuate Chang'an in the spring of 883 and Huang himself was pursued to his death in the summer of 884, civil war, banditry, and warlordism continued to plague northern China for more than two decades.

As should be evident from this brief sketch of Tang China's internal conflicts, some parts of the empire did not suffer as much as others and there were two long, multigenerational periods when the interior was almost entirely at peace. Very few of those who were born after 624 and died before 755 experienced the horrors of war in their home regions and, setting aside the relatively limited campaigns of Xianzong's reign, the same could also be said of the great majority of the Tang subjects whose lives were lived within the span of 785–870. Nevertheless, over the long run there were very few places that were completely spared. Suzhou, for example, was a second-tier provincial city in the lower Yangzi region, far from usual arenas of conflict in the North China Plain and Guannei. Although spared from invasion during the Tang consolidation due to the peaceful submission of the regional warlord, Du Fuwei 杜伏威, Suzhou was captured in 760 by forces of the rebellious general Li Zhan 劉展, who held it briefly until they were defeated early in the next year by the same government army that had recently sacked Yangzhou. Suzhou was later affected by the rebellion of the Zhexi 浙西military governor Li Qi 李錡 in 807 and then by the rebellion of Wang Ying 王郢 in 876.Footnote 95 Then, as Tang authority unraveled in the last years of the ninth century, the region became “a battleground… and every record describes appalling suffering and destruction as the city of Suzhou was repeatedly sacked by local warlords and bandits.”Footnote 96 The experience of civil war was thus a living memory in the community (if only, at some times, among its eldest members) from 760 onward to the end of the dynasty and beyond. Elsewhere in the empire, especially in Guannei and the North China Plain, the same was true from 755 or 756 onward—and also for the whole of the early Tang period up through the 670s.

THE MILITARY AND TANG CULTURE

The first three Tang reigns were a time when military elites played a leading role in the imperial government and martial values enjoyed widespread acceptance. In this respect, the early Tang can be seen a continuation of the period of the Northern Dynasties, when North China was dominated by non-Chinese elites who had imposed their rule by force of arms. The tribal invaders gradually mingled and intermarried with their Chinese subjects, but the ruling class of the north retained many of the values and attitudes of the steppe peoples. In the early years of the Tang dynasty this steppe heritage was still very much in evidence. The Tang founders, Li Yuan 李淵 and his sons, were themselves of mixed blood, as were a great many of their followers. The Li forebears had held high office under the Xianbei regimes of Western Wei and Northern Zhou, and belonged to the group of powerful, prominent northwestern clans that had intermarried with the ruling houses of Northern Zhou and Sui. Like their nomadic ancestors, this northern aristocracy attached great importance to the martial skills of horsemanship and archery and had a passion for the hunt.Footnote 97 By the seventh century, to be sure, men of this class were usually literate in Chinese and some were accomplished writers and gifted poets, but a boy who devoted himself entirely to book learning and neglected the study of the fighting arts might find himself held up to ridicule by members of his own family.Footnote 98

The early Tang aristocratic ideal, endlessly repeated in the biographies and funerary inscriptions of the period, was to possess both the martial skills of the warrior and the literary ability of the civil official, and, on the basis of those skills, to be able to serve with equal competence as either a general in the field or a minister at court (chu jiang ru xiang 出將入相).Footnote 99 Though not all of the northern aristocrats managed to live up to this ideal, the biographies of early Tang officials afford numerous examples of men who moved back and forth between civil and military posts without difficulty. The preeminent example of this career pattern was Li Jing, conqueror of the Eastern Türks and the Tuyuhun, who held important posts at court—including Minister of Justice (xingbu shangshu 刑部尚書) and Right Vice-president of the Department of State Affairs (shangshu you puye 尚書右僕射)—when he was not leading an army on campaign.Footnote 100 In the early years of the Tang dynasty civil and military career paths were much less distinct than they later became; men with military skills and combat experience held some of the highest civil offices, and the prestige of the military was relatively high.

Military officers, even those who were not of aristocratic origin, were men of some prominence in their communities and enjoyed fairly high official rank and status. The colonel (zhechong duwei 折衝都尉) of a territorial regiment would in most cases have ranked only slightly below the local prefect, a civil official responsible for administering a population numbering in the tens (or even hundreds) of thousands. Like the prefect, he belonged to the relatively small group of officials—those of the fifth rank and above—who were appointed to their positions by the emperor himself rather than through the routine selection procedures of the Board of Civil Office and the Board of War.Footnote 101 A major (xiaowei 校尉) of a territorial regiment held roughly the same rank as a county magistrate, and even the lowly dui zheng 隊正, commander of a unit of only fifty men, might outrank half of the civil officials on the magistrate's staff. Beyond the testimony of rank alone, there is evidence that Taizong, himself a battle-hardened warrior and victorious general, felt a special bond of closeness with the commanders of the fubing regiments. When he heard that a county official had taken the very unusual action of having a zhechong duwei put to death for criminal misbehavior, the emperor exclaimed, “What sort of county marshal (xian wei 縣尉) is this, to kill one of my colonels!”Footnote 102

The history of the early development of the fubing institution under Western Wei, Northern Zhou, and Sui would suggest that command positions were most often given to local elites (haojia 豪家 or haoyou 豪右), members of influential landholding families prominent at the prefectural or county levels, rather than to scions of the higher-ranking aristocratic clans that frequented the imperial court, and that these men commanded units that were based in their home communities.Footnote 103 By Tang times, regimental offices were no longer necessarily identified with local power.Footnote 104 However, there were still good reasons for local elites to seek command positions in the fubing system. They offered the prestige of government office, the possibility of upward mobility through promotion to, say, a generalship in the Guards,Footnote 105 and distinct advantages for wealthy, landholding families under the equal-field system of land distribution (since officers were entitled to up to 600 mu of “office land” [zhifen tian 職分田] and might be able to claim additional landholdings on the basis of honorific rank [xun guan 勳官] won in battle).Footnote 106

Military service may well have appealed to men of less exalted status who served as ordinary soldiers in the territorial regiments for much the same reasons. They were not entitled to grants of office land, but they could win honorific rank, and with it the possibility of a larger landholding, through their exploits on the battlefield. An outstanding performance in combat might also lead to rapid advancement, as in the case of Xue Rengui 薛仁貴, a volunteer who was made lieutenant-colonel of a territorial regiment after his heroics caught the eye of the emperor at the battle of Anshicheng 安市城 in 645.Footnote 107 It seems that in the first half of the seventh century, there was no shortage of willing recruits for the empire's major campaigns. At the end of 644, Taizong boasted that when he set about recruiting men for his expedition against Koguryŏ, “we took only those who were willing to go; when we wanted to recruit ten, we obtained one hundred, and when we wanted one hundred we got a thousand. Those who were not able to join the army were bitterly disappointed.”Footnote 108

In the early Tang empire, political power was held by an aristocracy with strong martial traditions. The military establishment loomed very large in the state structure and martial enterprises attracted the willing participation of a large segment of the population. However, this is not to say that the entire society subscribed to a warlike ethos. Medieval China was a very complex society with several distinct elites.Footnote 109 There were many families, especially in the south and the northeast, that maintained strong traditions of scholarship and coveted the opportunity to serve the state as civil officials. Men from such cultured, literate backgrounds usually lacked the martial skills characteristic of the northwestern aristocrats,Footnote 110 but they brought administrative and communications skills and a knowledge of ritual and institutional precedent that was essential for the conduct of government. As heirs to Confucian ideals of statecraft, they were committed to a belief in the transforming effect of benevolent and virtuous civil administration and were inclined to regard martial exploits with distaste, if not outright hostility. One of the most prominent civil officials of Taizong's court, Wei Zheng 魏徵, ostentatiously signaled his disdain for the “smashing the formation music” (pozhen yue 破陣樂), a performance by dancers with armor and halberds commemorating Taizong's military victories, by looking down at the ground and refusing to watch.Footnote 111 This attitude had serious policy implications; for example, literati-officials strongly opposed Taizong's military adventures in Korea in the 640s.Footnote 112 The more warlike aristocrats reciprocated from time to time by belittling the literati as “petty clerks.”Footnote 113 These incipient civil-military tensions would become much more pronounced in the eighth and ninth centuries, when the power of military men was on the rise at the same time that their prestige and social status were declining in comparison with the civil bureaucracy.

The conversion of expeditionary armies into static garrisons from the late seventh century had a depressing effect on the morale and combat effectiveness of troops in the field and greatly reduced the desirability of military service in the eyes of many, both elite and commoners, who might once have considered it an attractive option. This was already apparent in the early 660s, when bingmu who went with expeditionary armies to the Korean peninsula were retained as garrison forces for two years or more. In a memorial submitted in 664, their commander, Liu Rengui 劉仁軌, reported that the troops were deeply dispirited. Their supplies of food and clothing were inadequate, but more important, there was a keen sense that the rewards offered by the state were no longer commensurate with what it was demanding of them. The result was that the eager volunteers who had filled the ranks of the expeditionary armies in the early years of Gaozong's reign had disappeared. Members of wealthy families made use of their connections with local officials to avoid being called for service, as a consequence of which the burden fell all the more heavily on the poorest members of the community who were least able to afford it. For them, the only ways to avoid service were desertion or self-mutilation.Footnote 114 Though Liu's report dealt with the bingmu in Korea in the 660s, the same dynamic must also have been at work when fubing were held over in the new army garrisons in the 670s and 680s and helps to explain the absconding of guardsmen and their families that was reported in Guannei in the 690s.

By the early decades of the eighth century, Chinese elites lacked military experience. More specifically, far fewer of them had any experience of leadership at the lower command echelons than had been the case in Taizong's day. Scions of the dynasty's founding military elite had begun to opt out of military careers, a pattern that was already evident as early as the 660s when Du Huaigong 杜懷恭, a son-in-law of the great general Li Shiji 李世勣, refused to serve with the Tang forces fighting in the Korean peninsula.Footnote 115 In the second half of the seventh century it became the usual pattern for the sons and grandsons of the political elite to pursue careers in the civil bureaucracy, especially after the system of literary examinations became an important means of fast-track recruitment during the ascendancy of Empress Wu. Military leadership—especially at the middle and lower levels—was increasingly left to the illiterate and the marginally literate. In 702, Empress Wu saw fit to introduce a new system of regular annual military qualification examinations (wuju 武舉) in order to attract a better class of military officer. Examinees were scored on their skill with bow and arrow and with the cavalry lance, and were also tested for physical strength and command “presence.” Those who passed with relatively high scores might be appointed to posts in the Imperial Guards.Footnote 116 In fact, the military examinations had no discernable influence on the composition of the Tang officer corps. The only graduate who name is known to us is Guo Ziyi 郭子儀, a prefect's son who rose to command the Shuofang 朔方 Army during the An Lushan Rebellion.Footnote 117 It seems that by the 750s the great majority of frontier officers at all levels were either Chinese from very humble backgrounds or non-Chinese, usually of steppe origin. Such men were often promoted from the ranks on the basis of military examinations administered at their garrison by the army commander or military governor.Footnote 118

The late Tang was marked by a much sharper distinction between civil and military elites than had been the case in the early years of the dynasty, when northwestern aristocrats and omnicompetent general service officers such as Li Jing had bridged the civil-military divide. The differentiation of military and civil-bureaucratic leadership roles and career tracks was already far advanced by the middle of the eighth century and provided one of the essential preconditions for the rebellion of An Lushan. That episode in turn pushed the gap even wider as the court, its bureaucracy, and educated men in general came to regard the suddenly dangerous military men with an attitude of suspicion and hostility that had not been seen before the cataclysm of 755. The reassertion of civilian control over the military—and of the scholar's superiority over the man of violence—became one of the central policy goals of the Tang government. During the post-rebellion reigns of Suzong, Daizong, and Dezong the court repeatedly signaled its distrust of its own generals, and relations with Tang field commanders were often strained—or worse.Footnote 119 Guo Ziyi, a figure of exemplary loyalty and one of the government's most successful generals, was not entrusted with a major field command during the last three years of the rebellion. Guo's colleague and sometime rival Li Guangbi 李光弼 fell under suspicion shortly before his death in 764.Footnote 120 Lai Tian 來瑱, who had won the sobriquet “Bite-iron” during his stubborn defense of Yingchuan 穎川 against An Lushan's men, was denounced by eunuchs and forced to commit suicide at the beginning of 763.Footnote 121 And it was during this same period in 763–64 that Pugu Huaien 僕固懷恩, who had succeeded Guo and Li as commander of the Shuofang Army, was badgered into rebellion by a cabal of military rivals and hostile eunuchs.Footnote 122

With political reliability now the paramount consideration, important military commands might be entrusted to civil officials or eunuchs rather than experienced generals. An army commanded by a senior scholar-official, Fang Guan 房琯, was defeated near Xianyang 咸陽 in the tenth month of 756, and another disaster occurred in the spring of 759 when the powerful eunuch army supervisor Yu Chao'en 魚朝恩 had de facto operational control over nine Tang contingents that were routed by Shi Siming 史思明 in front of Xiangzhou 相州. Undeterred by these failures, the court continued to name ranking civil bureaucrats as well as soldiers to the empire's top military commands. Pei Du 裴度, who directed the successful military operations against the rebellious province of Huaixi 淮西 in 814–15, was a jinshi of 789 who had served as Investigating Censor, Court Diarist, President of the Censorate, and Vice-president of the Chancellery before he was given overall command of the field armies in Henan.Footnote 123 Whenever local circumstances permitted, the court preferred to appoint civil bureaucrats as military governors, and the eunuch army supervisors were employed much more systematically than in pre-rebellion days.

In the seventh century, the anti-military sentiments of the literati had found expression primarily in the form of opposition to the aggressive, expansionist foreign policies of Taizong and Gaozong. In the post-rebellion period, however, we find scholar-officials expressing strong antipathy toward military men in a variety of ways. In 821, when the Tang court had briefly recovered control of Youzhou 幽州 from the local military rulers, civil officials sent from the capital taunted the soldiers: “Now that the empire is at peace, fellows like you who can pull a strong bow are not worth as much as someone who can recognize the simplest written character!”Footnote 124 Another sign of literati hostility toward the military was the attack on military ritual. In 731 temples honoring Qi Taigong 齊太公, the semi-legendary strategic advisor to the founders of the Zhou dynasty, were established in the two capitals and in every prefecture of the empire; these “military temples” (wumiao 武廟) and their associated rites were intended as a counterpart to the civil officials’ cult of Confucius, and their establishment, as Gao Mingshi has pointed out, was a symptom of the increasing differentiation of civil and military elites.Footnote 125 Significantly, the first recorded objections to the official cult of Qi Taigong came in 758; stronger criticisms were voiced by civil officials in 786, and a full-blown debate over the status of the military cult was held at court in 788. As a consequence of this controversy, observances in the wumiao were downgraded in such a way as to indicate that they were no longer considered to be on a par with the cult of Confucius.Footnote 126

The scholar-officials of the late Tang did not, however, turn their backs on military affairs. On the contrary, there were strong new assertions of the fitness of scholars and civil officials to exercise military command, amounting to a denial of the exclusive competence of military men within the military sphere. This was surely a defensive response to the enhanced power and autonomy of the empire's military elites. As David McMullen has pointed out, civil officials “were deeply opposed to the growth of professionalization or specialization at the highest levels of military command. Essentially, this development meant loss of civil bureaucratic control of the armies and therefore directly challenged the sectional interest of the civilian element in the polity.”Footnote 127 Although the cultural ideal of the rujiang 儒將—the scholar-general whose profound learning, strategic acumen, and understanding of human relations provided the basis for his success as a leader of armies—would not reach maturity until the Song period,Footnote 128 the historical writing of the late Tang and the succeeding Five Dynasties (907–60) offers many examples of men dismissed as mere scholars (shusheng 書生, ruzhe 儒者) who nevertheless succeeded in teaching the soldiers a thing or two. At the beginning of Xianzong's reign, for example, the prefect of Huzhou 湖州, Xin Mi 亲秘, was given command of the forces of five prefectures to put down the rebellion of Li Qi 李錡. The scholarly prefect was greatly underestimated by his opponents, until he sent several hundred men to make a successful surprise attack on them. After that, we are told, everyone recognized that Xin had the talent to be a general.Footnote 129 The ability to master one's own military underlings was also emphasized. Wang Jin 王縉, who held several military governorships during the reign of Daizong, brought the unruly Hedong garrison to heel by executing several insubordinate troop commanders who had scorned him as a ruzhe.Footnote 130 In the early years of the dynasty, military command had been identified with leadership on the battlefield. From the late Tang, however, the stress was on the ability of the tough-minded, strong-willed, and ingenious scholar-official to succeed as a strategist and military executive.

Yet it would be wrong to assume that all of the scholar-officials of the late Tang had lost the vigorous, warlike attributes of the northwestern aristocrats and general service officers of Taizong's day. In contrast to the image of the refined, sedate literatus familiar from the Song and later dynasties, many scholars and civil officials of the late Tang were keen participants in martial sports and activities. There are reports of late Tang civil officials who were avid hunters,Footnote 131 and the game of polo, probably introduced from Tibet or Central Asia about the middle of the seventh century, enjoyed widespread popularity not only among military officers, who regarded it as a form of training for war, but also among civil officials and educated elites more generally.Footnote 132 For every scholar (such as Han Yu 韓愈) who criticized indulgence in this violent and dangerous sport, there were probably several other bureaucrats (such as Han Yu's sometime superior, the Xuzhou military governor Zhang Jianfeng 張建封) who were enthusiastic players. At least one emperor (Muzong) died of injuries received in a polo match, and in the early 870s polo was actually included among the entertainments and activities provided at the capital for each year's new crop of jinshi graduates.Footnote 133

There was never a complete separation of civilian elites from military command and warlike activities. During periods of internal disorder and rebellion, civil officials (and especially those with territorial responsibilities, such as prefects and county magistrates) might find military command thrust upon them by circumstances; during the An Lushan rebellion for example, Yan Gaoqing 顏杲卿, the civilian governor of Changshan 常山commandery, perished while leading a force of loyalist partisans against the rebels’ line of communications in Hebei, and Zhang Xun 張巡, the magistrate of Zhenyuan 真源 county in Henan, gave his life defending the strategically important city of Suiyang 睢陽 on the Bian 汴 Canal.Footnote 134 During the much more prolonged and widespread breakdown of civil authority that accompanied the Huang Chao rebellion in the 870s and 880s, local elites were driven to organize private militias (styled “righteous armies,” yijun 義軍) to maintain order and safeguard their homes and property.Footnote 135 This phenomenon was not unique to the late Tang, representing but one instance of a recurring motif in Chinese history. Landowners and local strongmen had mobilized their retainers, kinsmen, and dependents to resist non-Han invaders during the fourth century; the same pattern of local militarization reemerged after the breakup of the Northern Wei empire in the first half of the sixth century and again, albeit more briefly, during the chaotic transition from Sui to Tang rule. As late as the middle of the nineteenth century, local elites would respond in much the same manner to the challenge posed by the Taiping Rebellion.Footnote 136

CONCLUSION

This brief survey of the various ways in which war and the preparations for war affected the people of Tang times does not necessarily validate the proposition, once floated by the eminent Sinologist Edwin G. Pulleyblank, that wu 武, the military side of China's civil-military dichotomy, loomed larger in this period than under any of the other “major native dynasties.”Footnote 137 Nor does it demonstrate that each and every Tang person was profoundly influenced by war. Between 618 and 907, untold millions avoided both military service and taxes and lived entirely peaceful lives in the interior of the empire. Yet there were also many millions who served as soldiers, belonged to soldiers’ families, paid taxes, or engaged in other activities that supported the military, and there were vast areas where entire generations were devastated by the impact of war, especially in the early years of the dynasty and from the An Lushan Rebellion onward. The lesson to be drawn from the Tang military experience may not be that the Tang was somehow unique in this regard, but rather that it is generally consistent with the experiences of other dynasties and with new interpretations of the history of imperial China that assign much greater weight to war and the military than previously recognized.Footnote 138

References

1 Pulleyblank, Edwin G., “The An Lu-shan Rebellion and the Origins of Chronic Militarism in Late T'ang China,” in Essays on T'ang Society: The Interplay of Social, Political and Economic Forces, edited by Perry, John Curtis and Smith, Bardwell L. (Leiden: E.J. Brill, 1976), 35 Google Scholar.

2 Regiments varied in size. In the early years of the Tang there were three categories of zhechongfu: superior regiments of 1,000 men, intermediate regiments of 800, and inferior regiments of 600. During the Chuigong period (685–88), Empress Wu raised these quotas by 200 men. See Shigekuni, Hamaguchi 濱口重國, “Fuhei seido yori shin heisei e” 府兵制度より新兵制ヘ, Shigaku zasshi 41 (1930), 1266–70Google Scholar.

3 See Xiu, Ouyang 歐陽修 and Qi, Song 宋祁, Xin Tangshu 新唐書 (Beijing: Zhonghua shuju, 1975)Google Scholar [hereafter Xin Tangshu], 50.1324–26, and Swee Fo Lai, “The Military and Defense System under the T'ang Dynasty” (PhD diss., Princeton University, 1986), ch. 1.

4 Junwen, Liu 劉俊文, Tang lü shuyi jianjie 唐律疏議箋解 (Beijing: Zhonghua shuju, 1996)Google Scholar, 16.1173 (commentary to Article 227); Johnson, Wallace, The Tang Code, vol. 2: Specific Articles (Princeton: Princeton University Press, 1997), 215–16Google Scholar.

5 One of the sources supporting this is You, Du 杜佑, Tong dian 通典 (Beijing: Zhonghua shuju, 1988)Google Scholar 29.810.

6 A registered population of 48,909,800 was recorded for 742; see Xu, Liu 劉昫, Jiu Tangshu 舊唐書 [hereafter Jiu Tangshu] (Beijing: Zhonghua shuju, 1975)Google Scholar, 9.216. Late seventh–century figures are not available but were almost certainly less.

7 Xin Tangshu 50.1325–26; Jiu Tangshu 43.1834.

8 Hideo, Kikuchi 菊池英夫, “Tō setsushōfu no bunpu mondai ni kansuru ichi kaishaku” 唐折衝府の分布問題に關する一解釋, Tōyōshi kenkyū 27.2 (September 1968), 89 Google Scholar; Jiguang, Gu 谷霽光, Fubing zhidu kaoshi 府兵制度考釋 (Shanghai: Shanghai renmin chubanshe, 1962), 205 Google Scholar.

9 張沛, Zhang Pei, Tang zhechongfu huikao 唐折衝府匯考 (Xi'an: San Qin chubanshe, 2003), 1112 Google Scholar. Note that Zhang counts a total of 692 zhechongfu, but for 54 of them the location is uncertain. An earlier study found that 452 (68.8%) of the regiments were concentrated in Guannei and Hedong: see Gu, Fubing zhidu kaoshi, 154.

10 Gu, Fubing zhidu kaoshi, 154; Xin Tangshu 37.961. But Zhang Pei is able to identify only a hundred of these by name; see Tang zhechongfu huikao, 24–58.

11 Zhang, Tang zhechongfu huikao, 11–12.

12 Zhongmian, Cen 岑仲勉, Fubing zhidu yanjiu 府兵制度研究 (Shanghai: Renmin chubanshe, 1957), 59 Google Scholar; Pulleyblank, Edwin G., The Background of the Rebellion of An Lu-shan (London: Oxford University Press, 1955)Google Scholar, ch. 6; Gu, Fubing zhidu kaoshi, 155–57; Guogang, Zhang 張國, Tangdai zhengzhi zhidu yanjiu lunji 唐代政治制度研究論集 (Taipei: Wenjin chubanshe, 1994), 2324 Google Scholar.

13 Kikuchi, “Tō setsushōfu no bunpu mondai,” 8–9.

14 For more detailed treatment of the fangding system, see Tamai Zehaku 玉井是博, “Tō dai bōtei kō”  唐代防丁考, in idem, Shina shakai keizai shi kenkyū 支那社會經濟史研究 (Tokyo: Iwanami shoten, 1942), 231–44Google Scholar, and Zhang, Tangdai zhengzhi zhidu yanjiu lunji, 77–92.

15 The most important studies of the bingmu system are: Hideo, Kikuchi, “Tōdai heibo no seikaku to meishū to ni tsuite” 唐代兵募の性格と名稱とについて, Shien 68 (May 1956), 7598 Google Scholar; Tang Geng'ou 唐耕耦, “Tangdai qianqi de bingmu” 唐代前期的兵募, Lishi yanjiu 1981.4, 159–72; and Zhang, Tangdai zhengzhi zhidu yanjiu lunji, 29–53.

16 Guang, Sima 司馬光, Zizhi tongjian 資治通鑒 (Beijing: Guji chubanshe, 1956)Google Scholar, 200.6323.

17 Kikuchi, “Tōdai heibo no seikaku to meishū to ni tsuite,” 86, 90–92.

18 A list of the frontier garrison armies believed to date from 742 can be found in Jiu Tangshu 38.1385–89, Sima, Zizhi tongjian 215.6847–51, Du, Tong dian 172.4479–83, and Jifu, Li 李吉甫, Yuanhe jun xian tuzhi 元和郡縣圖志 (Beijing: Zhonghua shuju, 1983)Google Scholar, 4.92, 13.361–62, 34.886, 39. 991–92, and 40.1018.

19 Hideo, Kikuchi, “Fuhei seido no tenkai” 府兵制度の展開, in Iwanami kōza: sekai rekishi 岩波講座:世界歷史 (Tokyo: Iwanami shoten, 1969–71)Google Scholar, vol. 5, 426; Tang, “Tangdai qianqi de bingmu,” 167; Jimin, Sun 孫繼民, Tangdai xingjun zhidu yanjiu 唐代行軍制度研究 (Taipei: Wenjin chubanshe, 1995), 103 Google Scholar.

20 Tamai, “Tō dai bōtei kō,” 244; Le, Kang 康樂, Tangdai qianqi de bianfang 唐代前期的邊防 (Taipei: National Taiwan University, 1979)Google Scholar, 157, 177; Tang, “Tangdai qianqi de bingmu,” 160, 165; Zhang, Tangdai zhengzhi zhidu yanjiu lunji, 36.

21 The text of the 717 edict is in Minqiu, Song 宋敏求, Tang da zhaolingji 唐大詔令集 (Beijing: Shangwu yinshuguan, 1959)Google Scholar, 107.553.

22 Qinruo, Wang 王欽若, Cefu yuangui 冊府元龜 (Beijing: Zhonghua shuju, 1960)Google Scholar, 124.21b; supplemented with information from Li Linfu 李林甫, Tang liu dian 唐六典, rpt. as Dai Tō rikuten 大唐六典 (Tokyo: Hiroike gakuen jigyōbu, 1973)Google Scholar, 5.18b. Also see Zhang Guogang, “Tangdai de jian'er zhi” 唐代的健兒制, Zhongguo shi yanjiu 1990.4, 106.

23 Sima, Zizhi tongjian 214.6832.

24 Skaff, Jonathan Karam, “Barbarians at the Gates? The Tang Frontier Military and the An Lushan Rebellion,” War & Society 18.2 (October 2000), 2930 Google Scholar.

25 The most thorough study of the tuanjie bing is Kaisaburō, Hino 日野開三郎, “Dai Tō fuheisei jidai no danketsu hei ni tsuite” 大唐府兵制時代の團結兵について, Hōseishi kenkyū 5 (1954), 79133 Google Scholar.

26 Zhang, “Tangdai de jian'er zhi,” 102; and idem, Tangdai zhengzhi zhidu yanjiu lunji, 48.

27 Jiu Tangshu 107.3262; Kikuchi, “Tōdai heibo no seikaku to meishū to ni tsuite,”  80, 91–92. For the Nanzhao expedition, see Sima, Zizhi tongjian 216.6906–7 and 217.6926–27.

28 Quan Tang shi 全唐詩 (Beijing: Zhonghua shuju, 1960)Google Scholar, 426.4693–94, translated by Waley, Arthur as “The Old Man with the Broken Arm” in Chinese Poems (London: George Allen and Unwin, 1946), 129–31Google Scholar.

29 This process is described in Pulleyblank, “The An Lu-shan Rebellion,” 53–55; also see Sima, Zizhi tongjian 218.6983–84.

30 Masuo, Kurihara 栗原益男, “An Shi no ran to hanchin taisei no tenkai” 安史の亂と藩鎮體系の展開, in Iwanami kōza: sekai rekishi (Tokyo: Iwanami shoten, 1969–71)Google Scholar, vol. 6, 165–66.

31 Kaisaburō, Hino, Shina chūsei no gunbatsu 支那中世の軍閥 (Tokyo: Sanseido, 1942), 59, 209Google Scholar.

32 Zhang, “Tangdai de jian'er zhi,” 100; idem, Tangdai fanzhen yanjiu, 111–12.

33 Zhang, “Tangdai de jian'er zhi,” 106–7; Sima, Zizhi tongjian 225.7245.

34 Fang Jiliu 方積六, “Guanyu Tangdai mubing zhidu de tantao” 關於唐代募兵制度的探討, Zhongguo shi yanjiu 1988.3, 114–15.

35 Jiu Tangshu 19A.653; Toshikazu, Hori 堀敏一, “Hanchin shineigun no kenryoku kōzō” 藩鎮親衛軍の權力構造, Tōyō bunka kenkyūjō kiyō, no. 20 (March 1960), 9193 Google Scholar.

36 Sima, Zizhi tongjian 228.7351–54.

37 Guogang, Zhang, Tangdai fanzhen yanjiu 唐代藩鎮研究 (Changsha: Hunan jiaoyu chubanshe, 1987), 106 Google Scholar; also see Shounan, Wang 王壽南, Tangdai fanzhen yu zhongyang guanxi zhi yanjiu 唐代藩鎮與中央關係之研究 (Taiwan: Guoli zhengzhi daxue zhengzhi yanjiusuo, 1969), 228 Google Scholar.

38 Zhang, “Tangdai de jian'er zhi,” 108.

39 Fang, “Guanyu Tangdai mubing zhidu de tantao,” 115–16.

40 Hino, Shina chūsei no gunbatsu, 209–10.

41 Hori, “Hanchin shineigun no kenryoku kōzō,” 82–83.

42 Jiu Tangshu 132.3647; Sima, Zizhi tongjian 223.7172. There is some reason to believe that the Ze-Lu militiamen were eventually converted to regulars; see Fang, “Guanyu Tangdai tuanjie bing de tantao,” 103.

43 Jiu Tangshu 9.216; Sima, Zizhi tongjian 215.6847–50.

44 For the definition of M.P.R., see Andreski, Stanislav, Military Organization and Society (Berkeley and Los Angeles: University of California Press, 1971), 33 Google Scholar.

45 Contamine, Philippe, War in the Middle Ages (Oxford: Basil Blackwell, 1984), 12, 307Google Scholar. For France, see Weigley, Russell F., The Age of Battles (Bloomington and Indianapolis: Indiana University Press, 1991), 52 Google Scholar.

46 The figure of 600,000 is from Hamaguchi, “Fuhei seido yori shin heisei e,” 118.

47 Households sought to avoid registration in order to escape taxation, and local officials were unlikely to report inflated figures because they would be expected to come up with higher tax revenues if they did so. See Pulleyblank, Edwin G., “Registration of Population in China in the Sui and T'ang Periods,” Journal of the Economic and Social History of the Orient 4 (1961), 292–93, 299CrossRefGoogle Scholar.

48 For Du You's estimate, see Pulleyblank, “Registration of Population,” 300–301.

49 The army sizes reported for major Tang campaigns are listed in David A. Graff, “Early T'ang Generalship and the Textual Tradition” (PhD diss., Princeton University, 1995), 49–53. Before 755, no campaign army is reported as numbering more than 200,000, and many of these soldiers would of course have been members of the regular military establishment rather than ad hoc “conscript-recruits” (bingmu).

50 Sun Jimin, Tangdai xingjun zhidu yanjiu, 114–15; Zhang Guogang, Tangdai zhengzhi zhidu yanjiu lunji, 97–98, 100–101; Qun, Zhang 章群, Tangdai fanjiang yanjiu 唐代蕃將研究 (Taipei: Lianjing, 1986), 96 Google Scholar. For bridle prefectures, see Skaff, Jonathan Karam, Sui-Tang China and Its Turko-Mongol Neighbors: Culture, Power, and Connections, 580–800 (New York: Oxford University Press, 2012), 6162 CrossRefGoogle Scholar.

51 Sima, Zizhi tongjian 199.6274–75; Jiu Tangshu 199B.5348, and Sima, Zizhi tongjian 198.6237.

52 For the Shatuo contribution, see Yinke, Chen 陳寅恪, Tangdai zhengzhi shi shulun gao 唐代政治史述論稿 (Beijing: Sanlian, 1956), 158 Google Scholar, and Gungwu, Wang, The Structure of Power in North China during the Five Dynasties (Stanford: Stanford University Press, 1967), 16, 19, 56Google Scholar.

53 Xin Tangshu 39.1008; Sima, Zizhi tongjian 223.7172.

54 Self-mutilation to avoid service was apparently a serious problem, prohibited by the Tang Code and mentioned in documents such as the memorial that Liu Rengui 劉仁軌, one of the expeditionary army commanders in Korea, submitted to the throne in 664. See Johnson, Wallace, The T'ang Code, vol. 2: Specific Articles (Princeton,: Princeton University Press, 1997), 225–26Google Scholar (Article 236), and Jiu Tangshu 84.2793.

55 See, for example, Webber, Carolyn and Wildavsky, Aaron, A History of Taxation in the Western World (New York: Simon & Schuster, 1986)Google Scholar, 122, 125, 180–81, 238–39, and 289 (Table 6); also see Showalter, Dennis E., “The Prussian Military State,” in Early Modern Military History, 1450–1815, edited by Mortimer, Geoff (Basingstoke, Hampshire: Palgrave Macmillan, 2004), 118 CrossRefGoogle Scholar.

56 Twitchett, D.C., Financial Administration Under the T'ang Dynasty, 2nd edition (Cambridge: Cambridge University Press, 1970), 86 Google Scholar.

57 Du, Tong dian 148.3780.

58 Jinxiu, Li 李錦綉, Tangdai caizheng shigao 唐代財政史稿 (Beijing: Beijing daxue chubanshe, 1995), 1277 Google Scholar and the table on 1272–73. Also see Du, Tong dian 6.110–111, translated in Twitchett, Financial Administration, 153–56. Li's figure includes the expenses of the frontier armies, the capital guards, and the imperial pastures that provided horses for the military. Since the Tong dian passage uses all three of the currencies/commodities of the Tang fisc (strings of copper cash, standard lengths of silk or hemp cloth, and shi of grain) and their exchange rates varied by time and place, Li's conversion of everything to strings of cash is a fraught exercise. The resulting values are best taken as indicative rather than precise.

59 Li, Tangdai caizheng shigao, 1270.

60 Zhang, Tangdai fanzhen yanjiu, 67–69, 71–73.

61 Fang, “Guanyu Tangdai mubing zhidu de tantao,” 118.

62 Hino, Shina chūsei no gunbatsu, 160.

63 Li, Tangdai caizheng shigao, 1271. For more about these taxes, see Twitchett, Financial Administration, 25.

64 Du, Tong dian 6.110–111, translated in Twitchett, Financial Administration, 153–56.

65 Hino, Shina chūsei no gunbatsu, 121.

66 Juxtaposing passages from Sima, Zizhi tongjian 228.7346–7 and 7352–53, this would appear to be the conclusion suggested by Sima Guang.

67 Jimin, Sun 孫繼民, Dunhuang Tulufan suo chu Tangdai junshi wenshu chutan 敦煌吐魯番所出唐代軍事文書初探 (Beijing: Zhongguo shehui kexue chubanshe, 2000)Google Scholar, 5, 16, 26.

68 Yiqing, Gu 古怡青, Tangdai fubing zhidu xingshuai yanjiu 唐代府兵制度興衰研究 (Taipei: Xin wenfeng chuban gongsi, 2002)Google Scholar, 154, 239–43, 460; Sun, Dunhuang Tulufan suo chu Tangdai junshi wenshu chutan, 53, 64–65.

69 Yongxing, Wang 王永興, Tangdai qianqi xibei junshi yanjiu 唐代前期西北軍事研究 (Beijing: Zhongguo shehui kexue chubanshe, 1994)Google Scholar, 411–12.

70 Junmin, Ma 馬俊民 and Shiping, Wang 王世平, Tangdai mazheng 唐代馬政 (Taipei: Wunan tushu chuban gongsi, 1995), 119–20Google Scholar.

71 Gaohua, Chen 陳高華 and Haihao, Qian 錢海皓, Zhongguo junshi zhidu shi 中國軍事制度史 (Zhengzhou: Daxiang chubanshe, 1997)Google Scholar, vol. 4, 166

72 Zheng, Wei 魏徵, Sui shu 隋書 (Beijing: Zhonghua shuju, 1973)Google Scholar, 4.81.

73 Xilin, Cheng 程喜霖, Han Tang fenghou zhidu yanjiu 漢唐烽堠制度研究 (Taipei: Lianjing, 1991), 248–49, 252Google Scholar; for the various forms of labor service, see Twitchett, Financial Administration, 30–31.

74 Li, Tangdai caizheng shigao, 788.

75 Chen and Qian, Zhongguo junshi zhidu shi, vol. 4, 168–69.

76 Twitchett, “Lands under State Cultivation,” 175–76; Yinglin, Wang 王應麟, Yu hai 玉海 (Zhejiang shuju, 1883)Google Scholar, 177.23a–24a.

77 Li, Tangdai caizheng shigao, 685.

78 Li, Tangdai caizheng shigao, 1271.

79 Xin Tangshu 50.1337.38; Jiu Tangshu 148.3861.

80 Li, Tangdai caizheng shigao, 1183.

81 Gu, Tangdai fubing zhidu xingshuai yanjiu, 33. For the Jinwu Guards, see Jiu Tangshu 44.1901, Xin Tangshu 49.1284–86. Before 662, the Left and Right Jinwu Guards were called the Left and Right Wuhou 武候 fu (Du, Tong dian 28.789).

82 Du, Tong dian 157.4030–31.

83 Du, Tong dian 149.3824.

84 For example, see Jiu Tangshu 181.4692 regarding the Headquarters Army of the autonomous military province of Wei-Bo 魏博.

85 Huang Liangming 黃良銘, “Tangdai Shence jun hujun zhongwei ji Shence jun zhi yanjiu” 唐代神策軍護軍中尉及神策軍之研究, Taidong shi zhuan xuebao 1976.4, 191; Yongcheng, He 何永成, Tangdai Shence jun yanjiu 唐代神策軍研究 (Taipei: Taiwan Shangwu yinshuguan, 1990)Google Scholar, 93, n. 15; Liu Yat–wing, “The Shen-ts'e Armies and the Palace Commissions in China, 755–875 AD” (PhD diss., University of London, 1970), 394–95.

86 Skaff, Sui-Tang China and Its Turko-Mongol Neighbors, 302–12, provides a table listing recorded attacks on North China from 599 to 755. Zhenping, Wang, Tang China in Multi-Polar Asia: A History of Diplomacy and War (Honolulu: University of Hawai‘i Press, 2013)Google Scholar, devotes considerable attention to Tang China's foreign conflicts. For tensions with Tibet in particular, see Twitchett, Denis, “Tibet in Tang's Grand Strategy,” in Warfare in Chinese History, edited by van de Ven, Hans (Leiden: Brill, 2000), 106–79Google Scholar.

87 Jiu Tangshu 54.2233.

88 Sima, Zizhi tongjian 189.5908.

89 Jing, Wu 吳競, Zhenguan zhengyao 貞觀政要 (Shanghai: Shanghai Guji chubanshe, 1978), 70 Google Scholar.

90 For more on these events, see Pulleyblank, “The Rebellion of An Lu-shan,” 41–48.

91 Sima, Zizhi tongjian 221.7102.

92 Yates, Robin D.S., Washing Silk: The Life and Selected Poetry of Wei Chuang (834?–910) (Cambridge, Mass.: Council on East Asian Studies, Harvard University, 1988), 108122 CrossRefGoogle Scholar.

93 The horrific damage done by Huang Chao in and around Chang'an is recounted in Tackett, Nicolas, The Destruction of the Medieval Chinese Aristocracy (Cambridge, Mass.: Harvard University Asia Center, 2014), 191200 CrossRefGoogle Scholar.

94 Xin Tangshu 225C.6460. A parallel passage in Jiu Tangshu 200B.5394 makes it clear that those doing the seizing were government soldiers.

95 Suzhou's experience of war is conveniently summarized in Milburn, Olivia, Urbanization in Early and Medieval China: Gazetteers for the City of Suzhou (Seattle and London: University of Washington Press, 2015), 245–47Google Scholar.

96 Milburn, Urbanization in Early and Medieval China, 247.

97 To take only one of its members as an example, the second Tang ruler, Li Shimin 李世民 (Taizong) claimed to have killed more than a thousand men by his own hand in battle. See Quan Tang wen 全唐文 (Taipei: Jingwei shuju, 1965)Google Scholar, 4.25a. He was also an avid hunter; see Wu, Zhenguan zhengyao, 283–87.

98 See the biography of Dou Wei 竇威, Jiu Tangshu 61.2364.

99 See, for example, the epitaphs of Tong, Qutu 屈突通 and Shentong, Li 李神通, in Hanguang, Mao 毛漢光, Tangdai muzhiming huibian fukao 唐代墓誌銘彙編附考, vol. 1 (Taipei: Institute of History and Philology of Academia Sinica, 1985), 85 and 144–45Google Scholar. For the chu jiang ru xiang formulation, see Taiping guangji 太平廣記 (Beijing: Renmin wenxue chubanshe, 1959)Google Scholar, 169.1233.

100 Biographies in Jiu Tangshu 67.2475 and Xin Tangshu 93.3811.

101 Ranks of county, prefectural, and regimental offices (as of the early eighth century) are given in Jiu Tangshu 42.1905–6 and 1918–21, Xin Tangshu 49A.1287–88, and Xin Tangshu 49B.1317–19. Hamaguchi holds that the ranks of the colonels had been slightly lower before 685; see “Fuhei seido yori shin heisei e,” 1269.

102 Jiu Tangshu 84.2789 (biography of Liu Rengui); also see Michio, Tanigawa 谷川道雄, “Sei E Hoku Shū Zui Tō seiken to fuheisei” 西魏北周隋唐政權と府兵制 in Chūgoku ritsuryōsei no tenkai to sono kokka shakai to no kankei 中國律令制の展開とその國家社會との關係 (Tokyo: Tōsui Shobō, 1984), 149 Google Scholar.

103 See Michio, Tanigawa, “Fuheisei kokka ron” 府兵制國家論, Ryūkoku daigaku ronshū, no. 443 (December 1993), 126 Google Scholar.

104 Su Dingfang 蘇定方, one of the few zhechong duwei who eventually rose high enough to merit biographies in the Tang histories, came from a Hebei family that was able to raise a local defense force of several thousand men during the Sui-Tang interregnum, but at the beginning of Taizong's reign he commanded a territorial regiment that was based in Chang'an. Su's biographies are in Jiu Tangshu 83.2777 and Xin Tangshu 111.4136; the location of his regiment, the Kuangdao fu 匡道府, is given in Zhang, Tang zhechongfu huikao, 25. For a similar case see the biography of Li Junqiu 李君球 (Jiu Tangshu 185A.4789), a man of Qizhou 齊州 whose regimental command was in Huazhou 華州.

105 See the biography of Guo Zhiyun 郭知運, Jiu Tangshu 103.3189–90.

106 Cen, Fubing zhidu yanjiu, 49; Li, Tang liu dian 3.34a–b.

107 Jiu Tangshu  83.2780.

108 Sima, Zizhi tongjian 197.6216.

109 See Twitchett, Denis, “The Composition of the T'ang Ruling Class: New Evidence from Tunhuang,” in Perspectives on the T'ang (New Haven: Yale University Press, 1973), 5051 Google Scholar.

110 For one example (Xiao Yu 蕭瑀), see Taiping guangji 254.1974.

111 Taiping guangji 203.1534; the performance is described in Pu, Wang 王溥, Tang hui yao 唐會要 (Beijing: Zhonghua shuju, 1955)Google Scholar, 33.612.

112 Rogers, Michael C., The Chronicle of Fu Chien: A Case of Exemplar History (Berkeley and Los Angeles: University of California Press, 1968), 34, 46Google Scholar; Yanhong, Meng 孟彦弘, “Tang qianqi de bianfang” 唐前期的邊防, Tang yanjiu, no. 1 (1995), 269n77Google Scholar.

113 Jiu Tangshu 60.2341.

114 The text of Liu's memorial is in Jiu Tangshu 84.2792–94.

115 Tang, “Tangdai qianqi de bingmu,” 167; Sima, Zizhi tongjian 201.6351.

116 For more detailed treatment of the military qualification examinations, see Xin Tangshu 44. 1170 (translated in Robert des Rotours, Traité des examens, 209–12); Mingshi, Gao 高明士, “Tangdai de wuju yu wumiao” 唐代的武舉與武廟, in Di yi jie guoji Tangdai xueshu huiyi lunwenji 第一屆國際唐代學術會議論文集 (Taipei: Zhonghua minguo Tangdai xuezhe lianyihui, 1989)Google Scholar, 1017, 1020–24; and Terrence Douglas O'Byrne, “Civil-Military Relations During the Middle T'ang: The Career of Kuo Tzu-i” (PhD diss., University of Illinois at Urbana-Champaign, 1982), 20–27, 288. The edict establishing the examinations is mentioned in Sima, Zizhi tongjian 207.6558, and Wang, Tang hui yao 59.1030.

117 Gao, “Tangdai de wuju yu wumiao,” 1034, 1037; Jiu Tangshu 120.3449.

118 Gao, “Tangdai de wuju yu wumiao,” 1032. O'Byrne (292–94) calculates that 21% of the men who served as military governors between 755 and 779 had risen from the ranks. Most would have begun their military careers before 755.

119 Tensions between the Tang court and its generals are examined in Pulleyblank, “The An Lu-shan Rebellion and the Origins of Chronic Militarism in Late T'ang China,” 54–58; Peterson, Charles, “P'u-ku Huai-en and the T'ang Court: The Limits of Loyalty,” Monumenta Serica 29 (1970–71), 423–55CrossRefGoogle Scholar; Wang Jilin 王吉林, “Tangdai de Shuofang jun yu Shence jun” 唐代的朔方軍與神策軍, in Di yi jie guoji Tangdai xueshu huiyi lunwenji, 914–21; and Qun, Zhang 章群, “Pugu Huai'en yu Li Huaiguang de fanpan 僕固懷恩與李懷光的反叛,” in Cho–yun, Hsü et al. , Zhongguo lishi lunwenji 中國歷史論文集 (Taipei: Taiwan Shangwu yinshuguan, 1986)Google Scholar, especially 112–15.

120 Sima, Zizhi tongjian 223.7166; Jiu Tangshu 110.3310–11; Xin Tangshu 136.4590.

121 Sima, Zizhi tongjian 217.6960; Jiu Tangshu 114.3366–68; Xin Tangshu 144.4700–4702.

122 Peterson, “P'u-ku Huai-en and the T'ang Court”; Pugu's biographies are in Jiu Tangshu 121.3477 and Xin Tangshu 224A.6365.

123 Pei's biographies are in Jiu Tangshu 170.4413 and Xin Tangshu 173.5209. For more on his role in the Huaixi campaign, see Peterson, Charles A., “Regional Defense Against the Central Power: The Huai-hsi Campaign, 815–817,” in Chinese Ways in Warfare, edited by Kierman, Frank A. Jr., and Fairbank, John K. (Cambridge, Mass.: Harvard University Press, 1974), 123–50Google Scholar.

124 Jiu Tangshu 129.3611.

125 Gao, “Tangdai de wuju yu wumiao,” 1046–56, 1060; McMullen, D.L., “The Cult of Ch'i T'ai-kung and T'ang Attitudes to the Military,” T'ang Studies 7 (1989), 9193 Google Scholar.

126 McMullen, “The Cult of Ch'i T'ai-kung,” 94–101.

127 McMullen, “The Cult of Ch'i T'ai-kung,” 75.

128 See Zijian, Liu 劉子健 [James T.C. Liu], “Cong rujiang de gainian shuo dao lishi shang dui Nan Song chu Zhang Jun de pinglun” 從儒將的概念説到歷史上對南宋初張浚的評論, in Guo shi shi lun: Tao Xisheng xiansheng jiu zhi rong qing zhushou lunwenji  國史釋論:陶希聖先生九秩榮慶祝壽論文集 (Taipei: Shihuo chubanshe, 1988), 481–85Google Scholar.

129 Jiu Tangshu 157.4150–51. For other examples of successful scholar–generals, see Jiu Tangshu 178.4634, Jiu Tangshu 140.3821–22, and Jiu Tangshu 128.3590.

130 Jiu Tangshu 118.3417.

131 See, for example, Jiu Tangshu 122.3506 and Jiu Tangshu 138.3783.

132 See Xianglin, Luo 羅香林, “Tangdai boluo qiu xi kao” 唐代波羅毬戲考, in idem, Tangdai wenhua shi yanjiu 唐代文化史研究 (1944; rpt. Taipei: Taiwan Shangwu yinshuguan, 1996), 165202 Google Scholar, and Liu, James T.C., “Polo and Cultural Change from T'ang to Sung China,” Harvard Journal of Asiatic Studies 45 (1985), 203–24CrossRefGoogle Scholar.

133 Luo, “Tangdai boluo qiu xi kao,” 186–87.

134 Yan Gaoqing's biographies are in Jiu Tangshu 187B.4896 and Xin Tangshu 192.5529. For Zhang Xun, see Jiu Tangshu 187B.4899, Xin Tangshu 192.5534, and Graff, David A., “Meritorious Cannibal: Chang Hsün's Defense of Sui–yang and the Exaltation of Loyalty in an Age of Rebellion,” Asia Major (Third Series) 8.1 (1995), 116 Google Scholar.

135 The most thorough work on the late Tang militias is Kaisaburō, Hino, Tōmatsu Godaisho jiei gigun kō 唐末五代初自衛義軍考 (Fukuoka: Hino Kaisaburō, 1984)Google Scholar.

136 See Kuhn, Philip A., Rebellion and Its Enemies in Late Imperial China: Militarization and Social Structure, 1796–1864 (Cambridge, Mass.: Harvard University Press, 1970)Google Scholar.

137 Pulleyblank, “The An Lu-shan Rebellion,” 35.

138 This is not the place for a full-blown historiographical essay, but a few of the more important works are Graff, David A., Medieval Chinese Warfare, 300–900 (London and New York: Routledge, 2002)CrossRefGoogle Scholar; Lorge, Peter, The Reunification of China: Peace through War under the Song Dynasty (Cambridge: Cambridge University Press, 2015)CrossRefGoogle Scholar; Swope, Kenneth, A Dragon's Head and a Serpent's Tail: Ming China and the First Great East Asian War, 1592–1598 (Norman: University of Oklahoma Press, 2009)Google Scholar; Perdue, Peter C., China Marches West: The Qing Conquest of Central Eurasia (Cambridge, Mass.: The Belknap Press of Harvard University Press, 2005)Google Scholar; Waley–Cohen, Joanna, The Culture of War in China: Empire and the Military under the Qing Dynasty (London: I.B. Tauris, 2006)Google Scholar; and Dai, Yingcong, The Sichuan Frontier and Tibet: Imperial Strategy in the Early Qing (Seattle: University of Washington Press, 2009)Google Scholar. Peter Lorge in particular has devoted most of his career to overturning the longstanding conventional wisdom that the Song was the most un-military and even anti-military of all China's dynasties.

There have also been several edited volumes of papers, among which Military Culture in Imperial China, edited by Di Cosmo, Nicola (Cambridge, MA and London: Harvard University Press, 2009)Google Scholar stands out as especially worthy of note.