The 'U-Lag Problem And Attempted Solutions

As they did in other sparsely populated Inner Asian territory, Qing and Republican Chinese militaries and civilian bureaucracies in Kham used local corvée labor for government transport. This practice was not their invention. Chinese sources called the corvée “wula,” after the Inner Asian term that is Romanized as 'u-lag from Tibetan, ulaq from Turkic, and ulagh or ulag-a from Mongolian. Most successor states to the Mongol Empire had some form of tax that went by this name (Kumar Reference Kumar2007, 64–66). Broadly, it required owners of carrying animals to provide transport services to government officials and armies, either for free or for compensation at well below market transport rates.

This system was appropriate for the normal requirements of administrations like that in pre-twentieth-century Tibet, which delegated substantial authority to local monastic institutions and lay aristocratic families (Goldstein Reference Goldstein1989, 5). The central Tibetan government kept only a minimal military force, and did not maintain any kind of police force in rural areas. Even in these circumstances, the transportation corvée was still one of the “most difficult labour obligations” for taxpayers (Goldstein Reference Goldstein1989, 4). The much bulkier institutions and bureaucracy that went with the developmental state that Chinese officials attempted to establish in the twentieth century placed great pressure on the system and the livestock owners who performed the corvée. Overburdened households revolted or ran away, making the transportation of goods and military equipment impossible and undermining government efforts to increase the population and revenue base (H. Liu et al. [1938] Reference Liu, Zihe, Zhangfu, Shaocheng, Xinyu and Heping2004; Zhu [1914] Reference Zhu, Xinyu, Heping and Chuan2006).

Zhao Erfeng faced the same problem, but it was worse in the Republican era. This was partly because Republican-era officials were less able to offset the burden on 'u-lag providers by paying them, which had been the classic Qing solution since the Kangxi reign (Chen Reference Chen1984, 48–52). One investigator noted that “nobody can pay half a zangyang yuan,” the amount per animal stipulated by the regulations that Zhao Erfeng had created (Yu Reference Yu1937, 47). In addition, Republican-era governors were forced to collect more tax locally, and doing so required more official journeys. The late Qing administration in Kham had not bothered to collect tax from bands of miners numbering less than six, or households with fewer than around a dozen animals (X. Liu Reference Liu2007, 31; Sichuan Minzu Yanjiu Suo 1989, 3:836). As they sought to replace support from outside Kham, Republican-period military forces tried to collect as much tax from as many people as possible. In 1934, villagers near Dartsemdo complained via petition that:

Chinese governors routinely described the 'u-lag as one of the major burdens for the people of Kham, and its inefficiencies and limitations as one of the major obstacles to Chinese power in the region.

Liu Wenhui and his predecessors in Kham attempted two kinds of solutions. The first was to build on Qing attempts to ameliorate 'u-lag with regulation. To this end, most Han authorities in Kham announced new regulations to add to the already weighty code bequeathed to them by the Qing administration, and a whole new code was created in 1929 ( Sichuan Sheng Dang'an Guan and Sichuan Minzu Yanjiu Suo Reference Sichuan and Guan1990, 199, 200–202; Yu Reference Yu1937). Regulations mandated maximum journey lengths, maximum loads, payments for labor, and compensation for dead animals. This made 'u-lag—in theory—the most intensely managed aspect of the relationship between the regional Chinese government and local society. The second solution for the 'u-lag problem was the complete abolition of the system and the establishment of government-owned ranches to raise livestock for government and military transport. There were two attempts to do this. The State 'u-lag Agency (Guan yun wula shiwusuo) was established in 1932, and it was followed in 1938 by the Xikang Animal Transport Company (Mu yun gongsi). Both strategies had important implications for the development of Chinese colonial authority.

Each attempt at regulation of the corvée entailed efforts to bureaucratize the position of the Tibetan leaders who organized it because virtually all Chinese officials regarded the autonomy of such leaders as a key aspect of the problems with the system. Ren Zhuo (Reference Ren1940, 17), the author of an article in a journal produced for the Xikang provincial bureaucracy, Kham Guide Monthly (Kangdao yuekan) wrote:

There were cases in which local agents used their position in the 'u-lag system to derive private profit, developing positions of semi-independent authority in much the same way that the entrepreneurial brokers did in the northern Chinese villages studied by Duara (Reference Duara1988, 73–75). Liu Wenhui's garrison commander in Nyakchu complained about one Tibetan village leader who had exempted certain households from the corvée in exchange for cash payments, in effect appropriating some of the value of the corvée (Niansi jun jiyaochu you guan Kangqu jiaotong dang'an [1932] Reference Xinyu, Heping and Chuan2006, 473). The regime's Chinese agents did the same. A local strongman-merchant of Tau, “Barbarian King” Ding (Ding Manwang), arrived as a Qing-period settler and constructed a personal fortress that had enabled him to hold out against a 1911 lama-led uprising, which earned him enough authority to be made chief of the local militia (mintuan) when a Chinese army returned (N. Ren [1933] Reference Ren1996, 263). Ding also arranged the corvée duties, a role he profited from by exempting militia members from 'u-lag and charging a fee for the privilege of joining, which reportedly earned him more than 10,000 yuan.

People such as Ding contributed to the problems with the corvée system, but no more so than the officers who used 'u-lag for private business trips or neglected the welfare of the animals, causing them to die. Nevertheless, it was easier to blame the mostly Tibetan local agents, and Liu Wenhui's government attempted to step up its supervision of them and integrate them within the regime where possible. In 1932, Liu's military formalized the position of some of its Tibetan agents (whom they called tusi, “native officials,” even though the tusi title had been officially abolished by Zhao Erfeng) by giving them regular positions in the military bureaucracy.

In 1938 the regime attempted to increase its control over its agents further with the creation of an 'u-lag supervisory system (jiancha zhidu), according to which officers were posted at key locations around Kham. Zhang Zhen'guo, who filled this post at Tau, complained about how hard the job was (Zhang Reference Zhang1940, 132). It is difficult to determine how much impact the system had (if any) on the enforcement of the regulations, but it meant an extra official based in Tibetan communities with a mandate to supervise both Chinese and Tibetan leaders' use of corvée labor.

And despite “Barbarian King” Ding's prestige among the Chinese in Kham—one of Liu's employees commented that “there were no other Han merchants in the north of Kham with his ability and status”—he was increasingly isolated by Liu's government (N. Ren [1933] Reference Ren1996, 263). The Xikang Special Administrative Committee sided against him when he was sued by another merchant in 1928, and Liu's army did not help when he was attacked by Khampa near Darstemdo who wanted their 'u-lag money. Ding was forced to pay up, and his power and prestige faded in the 1930s. Ding was a member of the group of “Xikang residents” who complained about Liu in a petition to the Nanjing government in 1929, but to no avail (Sichuan Sheng Dang'an Guan and Sichuan Minzu Yanjiu Suo Reference Sichuan and Guan1990, 50).

Because of the corvée, the regime's interaction with Khampa leadership was very different from its interaction with Yi leadership in Ningshu, where there was no 'u-lag and no tax system in which Yi leaders had a comparable role. Chinese officials maintained relationships with Yi leaders, whom they contracted to provide security on the borders of the territory settled by Han (Chongqing Zhongguo Yinhang Diaocha Zuzhi 1935, 138). But this was a very different kind of role; it was less bureaucratized, and Liu's governors made no attempt to monitor Yi leaders' relationships with Yi society.

Another difference between Kham and Ningshu was that through a prejudiced interpretation of the 'u-lag system that blamed Tibetan agents for all of the system's problems, some officials in Kham became champions of state enterprise. Ren Zhuo (Reference Ren1940, 17) wrote:

In Ningshu there were plenty of officials with prejudices against the Yi, but this did not necessarily drive them to support state-owned industry or agriculture.

Both transport organizations established by Liu's regime in Kham failed, probably because they were under-resourced. Their livestock died in large numbers, and the organizations were disbanded within a few years of their establishment. Nevertheless, advocacy for this position and the experiments that it generated made a significant contribution to making Liu's development policy different from Zhao Erfeng's. The post-1905 Qing administration had established a tannery at Batang, but displayed little real interest in livestock farming. While the tannery cost 24,000 taels, Zhao spent 60,000 taels on encouraging crop-growing in 1906 alone (He Reference He2001, 43; Ma Reference Ma2002, 161). His officials very rarely mentioned livestock farming in their writing (X. Wang Reference Wang2011, 232). In 1940, Liu Wenhui blamed the failure of the transport organizations on the fact that “Han people are several thousand years removed from nomadic life” and therefore did not know enough about livestock farming (W. Liu Reference Liu1940a, 5). Zhang Zhen'guo, the 'u-lag supervisor at Tau, also blamed ignorance and a lack of veterinary medicine, as well as poor management (Zhang Reference Zhang1940, 133). That year, Liu's government decided that the Xikang Agriculture Improvement Institute (nongye gaijin suo) should focus on livestock farming, and that a million yuan would be spent improving animal stock and veterinary medicine in the following year (Xikang sheng nongye gaijin suo gongzuo yuebao 1940).

Although the 1938 Animal Transport Company failed, its livestock ranch near the Gartar Jampa monastery in the region of Tau (at sites called Taining and Qianning in Chinese) continued to operate under the management of the Agriculture Improvement Institute. The Gartar ranch covered more than 1,000 mu (67 hectares) (Qianning kenmu chang cheng 1948). It was the best-funded government agricultural enterprise in Xikang Province, receiving 37 percent more investment than the second most highly funded enterprise in the province between 1939 and 1943 (Y. Liu Reference Liu1944, 5–6). When Peter Goullart visited in 1939, he found several hundred yaks, imported Holstein bulls, sheep, and pigs “everywhere” (Goullart Reference Goullart1959, 33–34, 49). In 1944 the Xikang Agriculture Improvement Institute and the central government Ministry of Agriculture and Forestry jointly established a Cultivation Management Office (kenwu guanliju) at the ranch, which employed a staff of fifty-two people, not including actual farmers (of whom there were few) (Nonglin bu Xikang Taining kenqu guanli ju sanshiwu nian qiansan renyuan gongliang fei baohe qingce 1946; Xikang Taining kenqu choubei banli 1944). After 1949 the farm underwent several changes to its name and character, but remained a part of the landscape of what is now called Bamei Township (C. Wang Reference Wang2005, 69).

A final effect of 'u-lag was that it amplified conflict between the regional Chinese government and Tibetan leaders who were outside its control. Tensions related to the corvée were important in the development of the conflict between Liu Wenhui and the Panchen Lama's field office in Kardzé. The office had gone to Kardzé in 1938 having run out of supplies at its temporary Qinghai base, where the lama's journey back to Tibet had stalled (Lin Reference Lin2006, 110–11). In late 1939, the field office disarmed Liu's garrison in Kardzé with the aid of Khampa allies from Kardzé and Nyakrong. Liu organized a counterattack and drove the field office out of Kham within two months, though he allowed 500 Khampa fighters and a leader from Nyakrong to return (S. Zhao Reference Zhao1941).

In most accounts, Liu's government is portrayed as a passive party in these events. Peng Wenbin summarizes the Kardzé conflict as a campaign “against Liu Wenhui, couched in the appeal of ‘Kham for the Khampas’” (Peng Reference Peng and Epstein2002, 72). However, Liu had certainly not been anxious to avoid the conflict; indeed he seems to have deliberately courted it. The office's attack on the Kardzé garrison was precipitated by Liu's arrest and imprisonment of Déchen Wangmo (Bde chen dbang mo), a daughter of the Khangsar family, one of the most powerful families in northern Kham, and the bride-to-be of a member of the field office's staff. There were political implications to the marriage, but there were many and more lucrative parts of Xikang Province that were partially or totally outside the control of Liu's forces, so it is important to ask why they singled out the field office. Granted, it was strongly connected to the Guomindang, but attacking the Guomindang's ally could also have backfired for Liu and cost him some of the support that he also received from the party.

Matters related to the transport corvée were an important factor in the buildup of tension between Liu Wenhui's regime and the lama's field office. The office levied a large amount of 'u-lag for the transport of goods and personnel (Li Reference Li1940, 11). Its Guomindang supporters noted that payments for corvée labor consumed a significant portion of its income, and the entourages of the office's visitors also made full use of the system (Guofang zuigao huiyi 1938). In a report of his 1937 visit, Dai Jitao wrote that the officials dispatched by the field office to accompany him “required more than 600 pack animals on the outward trip and about 400 for the return trip, which were all provided by local people and officers as per custom” (Dai Reference Dai1938, 1824). This was in addition to the animals conscripted by the unit that Liu Wenhui sent with Dai. Dai noted that the members of the field office “did their utmost to limit the number of animals they levied” (1824), which suggests that there was very little slack in the system.

Despite the efforts of the field office to limit its use of the corvée, the Dartsemdo-based customs bureau and merchant community “complained vociferously” about the number of animals the office used (Li Reference Li1940, 11). Merchants argued that 70 percent of the goods the office transported from Sichuan to Kardzé—in trains of “hundreds of pack animals”—consisted of commercial goods for sale rather than vital supplies (Li Reference Li1940, 11). Whether or not this claim was true, merchants who believed it would have been angry for two reasons. Firstly, transport costs made up a large portion of the price of goods imported to Kham from outside. With 'u-lag transport at below-market rates, the field office would have been able to significantly undercut the merchants' prices. Secondly, because transport resources were severely limited, trains with hundreds of 'u-lag pack animals could very plausibly have driven up the prices that the merchants had to pay. War-related inflation began in 1939, so the rising transport prices were not necessarily due to the field office, but it is not surprising that the merchants drew such a connection. The Dartsemdo tax authorities predictably sided with the merchants. The latter were generally taxpayers, and they had a plausible case that they were losing business to an organization that did not pay tax. The 'u-lag levied by the Panchen Lama's field office was certainly not the only reason for the Kardzé conflict, but it meant that the office's presence in Kardzé had an impact on traders in other parts of Kham, and contributed to the antagonism between the office and Liu's officials.

In sum, the 'u-lag system had a highly significant impact on the development of Chinese warlord authority in Kham. To make use of the corvée, officers had to rely on Tibetan agents, but many blamed the agents for the system's problems. In response, warlord regimes attempted to increase their control over the middlemen. When this failed to fix the problems, and with dependence on distrusted Tibetan agents a continued source of anxiety among disgruntled officials, Liu Wenhui's regime attempted to forgo 'u-lag entirely and cater for government transport needs by establishing a state-owned ranch system. Without reliable outside support, this still had to be funded through taxes raised in Kham, though such taxes could have been designed to avoid some of the inevitable problems with 'u-lag: overtaxation of communities near major roads, and the lack of an incentive for users of 'u-lag to take proper care of animals. The attempts to build an alternative transport organization failed too, probably because they lacked sufficient funding. However, they did leave a substantial legacy in the form of a large government-managed livestock ranch that absorbed the bulk of the funding Liu provided for agricultural development in Kham.

Revenue And “Wasteland”

The unreliability of subsidies from Sichuan and the central government made it harder for Republican-era military governments to afford compensatory payments to 'u-lag providers than it had been for Zhao Erfeng's administration. It also forced them to attempt to collect more revenue locally. The Republican period was an age of fiscal innovation throughout China. Taxes that became important for Chinese authorities in Kham—the deed tax and the butchery tax—were also significant in other parts of the country. In Kham, they had important implications for government policy relating to land ownership.

To foster land cultivation, Zhao Erfeng ruled that all wasteland (huang) belonged to the state, and would be rented to anyone who wished to cultivate it (Sichuan Minzu Yanjiu Suo 1989, 1:97–98). Throughout the Republican period, regulations continued to proclaim that huang land belonged to the state (Sichuan Sheng Dang'an Guan and Sichuan Minzu Yanjiu Suo Reference Sichuan and Guan1990, 80; Tu Reference Tu1944, 63). However, there was a big difference between official regulation and actual practice. In 1936 an aggrieved member of a Bank of China research team wrote a tirade against the Dartsemdo local authorities in Sichuan Frontier Quarterly, a journal that aimed to promote investment in and political engagement with Kham and other non-Han territories west of the Sichuan basin. The county government, the writer alleged, was allowing local Tibetan households the right to charge Chinese settlers rent for use of huang land (Chongqing Zhongguo Yinhang Diaocha Zuzhi 1936, 162–3). The writer highlighted the case of a man who was forced to pay rent to bury his wife on a huang hillside. Many aspects of the case caused outrage. The rent was “ten times higher than in Shanghai” (163) (which is hard to believe). The land owners allegedly robbed and murdered people who did not pay up. But the core problem was the idea that this type of land could be privately owned at all. “Huang land like this, according to regulations in the president's Fundamentals of National Reconstruction (jianguo dagang) ‘belongs to local governments, and must be used for local enterprises for the public good’” (162). By rights, it was “national, publicly owned huang land” (162).

It was not only Han settlers who found that notional state ownership of huang land was fictional in reality. With the exception of the ranch near the Gartar monastery created for the Animal Transport Company, other state-owned enterprises operated by the Xikang Agriculture Improvement Institute rented the land they used from private landowners, even when the land was registered as wasteland or forest. The huang land used by a farm at Simaqiao near Dartsemdo was all rented—from a Catholic mission, a Tibetan household called the Wasi in the Chinese-language documents, and other landlords of uncertain ethnicity (Kangding nongchang yewu gaikuang 1949; Simaqiao nongchang dang'an 1944). The Xikang provincial government did expropriate small amounts of cultivated and uncultivated land for infrastructure projects like the Dartsemdo airstrip, and in this case Liu ordered that the households that had lost land be given compensation (W. Liu Reference Liu1940b).

The decision to rent rather than expropriate huang land seems to indicate a surprising respect for private property from what Barnett called “one of the most oppressive (in my opinion, one of the worst) warlord regimes in China” (Barnett Reference Barnett1993, 448). It also had important consequences for the model of agricultural development employed by the government and for settlement in general. After the military campaigns had been paid for, it had cost the Qing administration nothing to appropriate huang land as state property and distribute it to garrisons or settlers. The Gartar ranch aside, Republican-period government enterprises rented the land they used and did not always find it easy to pay their landlords. The manager of the Simaqiao farm complained: “After paying the rent there is very little left. If the weather is dry, output drops significantly and self-sufficiency is very difficult” (Kangding nongchang yewu gaikuang 1949). Managers had good reason to think carefully about the value of acquiring more land and the use to which it could be put.

It is unlikely that the huang-owning Tibetans near Dartsemdo were simply too powerful for Liu's army. In 1933, Liu was forced out of the Sichuan basin by his first cousin Liu Xiang, but was still strong enough to suppress the Panchen Lama's field office in 1939. Several sources state that even after the retreat from the Sichuan basin, Liu's army forces had around 20,000 men (Cai Reference Cai1997, 46; Ni Reference Ni2006, 34). In the late 1940s, Barnett noted that they were well-armed and had plenty of ammunition (Barnett Reference Barnett1963, 220–21).

The regime's decision not to appropriate huang land was very likely related to the way that local officials in Chinese warlord governments in Kham drew on the territory's resources. In addition to the problems with the 'u-lag system, the increasing importance of taxes on livestock made the pastoral economy more important to Chinese regimes in the Republican period than it had been to their Qing counterparts. Consequently, a dichotomy could no longer be clearly posed between “productive” cropland and huang land that late Qing officials had seen as unproductive. Furthermore, taxes on the creation and trading of land title came to constitute an important source of local government revenue, which created an incentive to issue legal title to huang land.

Even though livestock farming was a more significant part of the Kham economy than crop-growing, almost all the revenue collected by Zhao Erfeng's colonial government from Kham resources had come in the form of taxes levied on cultivated land. Using the official price of 20 zangyang yuan per dan of highland barley, the Dartsemdo County land tax was worth 20,760 zangyang yuan in 1910— 99 percent of the tax collected by the county government. There were monetary taxes on livestock, but they only yielded 203 zangyang yuan (Sichuan Sheng Dang'an Guan and Sichuan Minzu Yanjiu Suo Reference Sichuan and Guan1990, 86; E. Zhao and Wu Reference Zhao and Fengpei1984, 81). The only other Kham resource that generated a substantial amount of revenue for the government was salt from Tsakalo (Ch. Yanjing) (X. Wang Reference Wang2011, 130, 198; Xikang sheng caizheng ting 1940, 37). Livestock owners faced the 'u-lag, which could be much more onerous than the land tax. The 'u-lag, however, did not yield cash income for governments. Assuming that local administrators in Zhao's regime believed there were sufficient livestock in their vicinity to meet the 'u-lag, the Qing tax system incentivized them to promote the conversion of grazing land to cropland (even though environmental conditions made increased cultivation impossible or unlikely in many places, as Zhao Erfeng came to realize).

Sources of local government income changed significantly in the Republican period. From the mid-1930s, opium was an important source of revenue for Liu Wenhui's regime, though most of this was grown in Liangshan and Ya'an rather than the parts of Kham that Liu Wenhui controlled (Guomin Canzheng Hui Chuan Kang Jianshe Shicha Tuan [1939] 1971, 406). The region around Zungchu (Ch. Songpan) produced a lot of opium but was not part of Liu's dominion. For the lowest-ranking officials in the Chinese warlord regimes in Kham, the key change was that revenue generated by livestock-related taxes and taxes on property deeds became much more significant than it had been in the Qing, which undermined the Qing legal distinction between cultivated and uncultivated land.

The new butchery tax (tu shui), which was an indirect tax on livestock, accounted for 20 percent of Dartsemdo County's revenue in 1932, and the livestock tax contributed a further 4 percent (Ge xian liang shui diaocha biao 1932). In 1939 butchery taxes (including the “supplementary butchery tax”) were equal to 36 percent of the value of the county's land taxes (including “supplementary” levies) (Xikang sheng caizheng ting 1940, n.p. following 112, 120, 246).

With the rising fiscal importance of taxes on animal products, Liu Wenhui's government began to distinguish between “huang” and “pastoral land” (mudi, muchang). A set of “land rights principles” (di quan yuanze) drafted in 1939 separated the two (but did not determine how they could be differentiated in practice): “Forested land and huang are without exception considered public property”; but “pastoral land shall be communally owned by groups of pastoralists (mu min)” (Tu Reference Tu1944, 63). Other institutions in Xikang, such as the Agriculture Improvement Institute, also adopted the term “pastoral land” but listed it as a subcategory of huang (Xikang sheng tudi hukou diaocha biao 1947). The term “huang” was therefore more ambiguous in the late Republican period than it had been in the Qing. Zhao Erfeng's colonial government posed a relatively simple dichotomy between cultivated land (taxed and subject to private ownership) and everything else, which was normally called huang, taxed very lightly if at all, and not considered subject to legitimate private ownership claims. By the late Republic, pastoral land was more heavily taxed and sometimes conceptually distinguished from huang, in which case government codes recognized the possibility of private ownership.

Even though a radical distinction between huang (inclusive of grazing land) and cultivated land now made less sense, the government could have decided to appropriate grazing land anyway as a part of an atavistic return to Qing policy. However, before 1911 almost nobody in the Kham had possessed title deeds to uncultivated land issued by a Chinese government, and this also changed dramatically in the Republican period, for two reasons. Firstly, cash-strapped county officials sold land appropriated by the Qing administration, though no documents used by this study indicate how much (Goré Reference Goré and Zhesheng1923, 52; Wissing Reference Wissing2004, 160). This appears to have happened more in the early Republican period, though in 1932 tax records from Dartsemdo also recorded income from sales and mortgaging of land (Ge xian liang shui diaocha biao 1932).Footnote ²

Secondly, in order to increase revenue, Chinese officials across Kham bullied as many Tibetans as they could into registering their land with them—for which they had to pay the deed tax. In 1914 the Sichuan Frontier Department of Finance established a Deed Tax Collection Office. It noted:

The same lines or variations of them appeared multiple times in government communications on the subject (Chuanbian zhenshou fu qishui chouban chu cheng 1914). Because only a very small number of landowners in Dartsemdo and Chakzam Counties already possessed land deeds, estimations of the potential income to be derived by forcing Tibetans to register their land were very large indeed. The magistrate in Gonjo (Ch. Gongjue) figured that about 35,000 zangyang yuan would be collected in deed tax from his small and thinly populated county (Sichuan Sheng Dang'an Guan and Sichuan Minzu Yanjiu Suo Reference Sichuan and Guan1990, 117). This must have been a great overestimate, but actual income could still be substantial. In 1916 the county authorities in Gyezur succeeded in raising 4,170 dayang yuanFootnote ³ by issuing deeds for property categorized as “fields,” “pasture,” “buildings,” or “temple estate” (120). In the past, officials had had an incentive to register land as huang so that migrants could be settled on it; now there was an incentive to register it in another category and permit private ownership of it so that more tax could be collected. The 4,170 dayang yuan may have been a disappointment, but it should have been enough to meet the grain requirements of a garrison of around 150 soldiers for about a year (in a county of around 12,000 people) (Qiu Reference Qiu1939, 42).

The initial deed tax bonanza ended after those landowners who could be coerced or otherwise persuaded into paying for an official title deed did so. Thereafter the tax was collected when the government issued new deeds to purchasers of real estate. In places where there was a significant amount of buying and selling of land, revenue could still be substantial. In 1939 the deed tax generated 7,650 yuan in Dartsemdo County (Xikang sheng caizheng ting 1945, 6:7). The same year, the land tax amounted to 956 dan of grain, which, according to the formula used by Liu's tax bureaucracy, was worth 14,334 yuan (Xikang sheng caizheng ting 1940, n.p. following 112). Although the land tax was still twice as valuable, the deed tax contributed a very significant proportion of county revenue. Liu's tax officials calculated that throughout Xikang Province, the deed tax income was equivalent to one-third of the land tax income in 1940 (Xikang sheng caizheng ting 1945, 5:16). Moreover, it was a tax that retained its importance through the era of hyperinflation because it was collected as a percentage of transactions' value, rather than as a fixed amount (like the livestock tax), and it was collected throughout the year.

Just as in early Republican Kham, county officials mindful of the importance of this source of revenue had an incentive to grant individuals title to uncultivated land. A Dartsemdo County government list of land transactions from 1940 gave a single-word description of the type of land involved for each transaction, and some stated that the land was, indeed, huang or forest (Kangding xian xianfu chengbao dian mai qishui zhengshou 1940). Most of the records simply described the land in question as “mountain land” (shan di) without noting what, if anything, it was used for. Buyers (or the tax office) may well have recorded uncultivated or forested land as “mountain land” as a precaution against problems that could have arisen had the government decided to make good its proclamations that huang land was public property. Although most of the names in the records are Han Chinese names, some transactions involved Tibetan purchasers (and more might have involved Tibetans using Han-sounding names).

Subjectively, the huang land that officials encountered in the late Republican period was different from the huang land that the Qing administration had dealt with. It produced almost no income for the Qing government, but a substantial amount of revenue for the Republican governments. Moreover, in the Qing it had not been covered with any form of legal title that Chinese governments recognized. By the late Republican period, large numbers of people, particularly those nearby Chinese administrative centers and including Tibetans, possessed titles to huang land that had been created by local Chinese authorities. The Qing notion that huang land belonged to the state continued to be repeated in policy statements. But a strict dichotomy between huang and cropland no longer existed in practice.

Colonialism in the Himalaya: Conclusions and Comparisons

James Scott (Reference Scott2009) describes a long-term struggle between communities in upland Southeast Asia that rejected the state as a form of political organization, and the lowland states that have sought to colonize their lands. It is difficult to situate the Republican-era Chinese warlords who operated in parts of Kham after the fall of the Qing within this interpretive framework. After the 1911 revolution, the bonds between different layers of authority in the Chinese state weakened and in some cases disappeared completely. The colonial regime the Qing had created in Kham was cut adrift after the Yunnanese occupation of Sichuan, and having established control over the Chinese garrisons in the region, Liu Wenhui sought to restrict outside influence. Chinese warlords in Republican-period Kham were not really agents of the Chinese state, but nor were they anarchists who rejected the idea of state institutions. Without strong connections to the Chinese central government, Liu and his post-1911 predecessors in Kham had to modify or elaborate on certain aspects of the policies left by the Qing colonial regime.

The eighteenth-century Qing body of practice for borderlands governance included both laissez-faire maintenance of indigenous custom and colonial settlements called tun that existed to provide food for frontier garrisons. These were state-controlled communities composed either of soldier-farmers or civilians who were subject to regimes of management reminiscent of those in the military. Scholars have been alert to the late nineteenth-century shift in governing principles that led to attempts to remove indigenous authority and establish Chinese administration and aspects of Chinese culture. But most of the policies that Zhao Erfeng created to integrate Kham more firmly within the empire drew on the more interventionist wing of earlier Qing practice. Migration was to be tightly controlled, and the agricultural colonies his regime founded were within the long tradition of tun settlement. Uncultivated huang land was to be owned by the state, and migrants who cultivated it would receive permanent tenancy rights but not full ownership (X. Wang Reference Wang2011, 193). Zhao stated repeatedly that such land was “state property (guojia gongchan). It is different from the property one inherits from ancestors (zuye)” (Sichuan Minzu Yanjiu Suo 1989, 1:97–98). Zhao also discouraged settlers from subletting their land by holding them legally responsible for any problems caused by the subtenant. Land policy like this differentiated his regime from administration in China.

The need to raise more cash locally forced the warlord governments of the Republican period to innovate and move away from Qing precedent. They expanded taxes on livestock and land title, which meant allowing private ownership of huang land in practice if not in rhetoric. Less support from the central government also made problems with the 'u-lag corvée labor system worse because officials needed to levy more 'u-lag to collect more tax locally and had less money to pay those who performed the corvée. Twice Liu's regime attempted to deal with the problem by establishing ranches that would raise pack animals for official transport. Neither was successful, but the efforts meant that a large state-owned ranch was the most significant enterprise managed by the Xikang Agriculture Improvement Institute. The other solution to the 'u-lag problem consisted of expanded attempts to regulate the corvée, and efforts to supervise the actions of local agents, integrate them into the military bureaucracy, and, on at least one occasion, undermine their authority.

Other colonial regimes in the Himalaya and foothills faced the same problems as the Qing officials and their Chinese warlord successors in Kham, and interacted in similar ways with local people and resources. Zhao ruled that huang land in Kham belonged to the Qing state, and in the several laws promulgated from 1865 onwards the British Indian state granted itself control over forested land (though it allowed people to make certain claims to use-rights). The British also adapted indigenous systems of corvée labor (called begar and utar) for transport in mountainous territory without readily available or cheap commercial transport. At precisely the same time that Zhao Erfeng declared that there would be no corvée without payment in Kham, British officials made similar pronouncements in Uttarakhand and Kashmir (Pathak Reference Pathak1991, 265). As in Kham, however, repeated attempts at regulation failed to prevent officials' continued conscription of labor for transport at levels that caused significant hardship in mountain communities.

The 1910s and 1920s saw a divergence between patterns of colonial rule in the southern and eastern Himalaya. Opposition to the British use of corvée was well-organized, galvanized by hill soldiers returning from service in World War I, and coincided with Gandhi's non-cooperation movement (Pathak Reference Pathak1991, 266). There were more than 146 anti-begar meetings in Uttarakhand in the first four months of 1921 (Guha Reference Guha2000, 112). British and Chinese authority also followed vastly different trajectories. The commissioner for Kumaun (one of the two subregions of Uttarakhand) saw the territory that he was in charge of as (in Pathak's paraphrasing) “a small, hot part of a vast country” (Pathak Reference Pathak1991, 275). The British repression of protest against begar was harsh, but in 1921 provincial officials calculated that it was not worth continuing the fight for their right to corvée labor for transport in a territory they themselves viewed as not particularly important. In that year, begar was abolished and in the following year the government of India spent more than 160,000 rupees on hiring porters for officials travelling in the hills, an elaboration on the mid- and late-Qing governors' approach that was not available to the cash-strapped, semi-independent Chinese warlords in the Republican period (Guha Reference Guha2000, 112).

Post-1911 China is not the only case in which an imperial state has fragmented and left dependent colonial administrations with major financial problems; Tajikistan in particular faced enormous budget shortfalls after the Soviet collapse (Lezhnev Reference Lezhnev2005, 54). Accounts of places affected by a sudden withdrawal of central state support for a colonial government often tell a story of a descent into rule by bandits and warlords (Lezhnev Reference Lezhnev2005). Contemporary observers tended to characterize the Chinese regimes in Kham similarly. Whether Liu Wenhui ran one of the worst warlord governments in China, as Barnett thought, or an administration that was fairly similar to other military regimes, the taxation policies and practices of warlords merit analysis just as much as those of unified states. There were clear patterns in the way that the Republican warlords in Kham attempted to respond to the challenges of having less support than their imperial predecessors. However mixed the results, these strategies affected how officials dealt with uncultivated land and Tibetan leadership. It remains to be seen whether the fragmentation of an imperial state met with similar responses from other orphaned colonial regimes in Inner Asia.