The Origins of the Boxer Uprising - Softcover

Esherick, Joseph W. W.

Softcover

ISBN 10: 0520064593 ISBN 13: 9780520064591

Publisher: University of California Press, 1988

Synopsis

In the summer of 1900, bands of peasant youths from the villages of north China streamed into Beijing to besiege the foreign legations, attracting the attention of the entire world. Joseph Esherick reconstructs the early history of the Boxers, challenging the traditional view that they grew from earlier anti-dynastic sects, and stressing instead the impact of social ecology and popular culture.

About this title

About the Author

Joseph Esherick is Professor of History, University of Oregon, and author of Reform and Revolution in China: The 1911 Revolution in Hunan and Hubei (California, 1976).

Excerpt. © Reprinted by permission. All rights reserved.

The Origins of the Boxer Uprising

By Joseph W. Esherick

University of California Press

Copyright 1988 Joseph W. Esherick
All right reserved.
ISBN: 0520064593


Chapter One
Shandong: Where It All Began

The impact of the Boxer Uprising was truly global. The political context for the uprising was significantly influenced by national trends. But in the end, the Boxers were really a regional movement. With the exception of a few officially inspired Boxer "militia" in such places as Shanxi, Inner Mongolia, and the Northeastern provinces of Manchuria during the summer of 1900, the movement was essentially confined to the Shandong and Zhili portions of the north China plain. Before all else, it is essential to look at the geography, the political economy and the social formations of this regionand in particular at Shandong, where it all began.

The North China Plain

Late in the nineteenth century, an American medical missionary described a trip through the heartland of the Boxer movement:

We were now passing through a level farming country, which appeared, as usual, thickly inhabited. At an interval of about a mile we would pass through a village of adobe houses with straw or reed roofs. These villages were usually surrounded by a decrepit adobe wall, and some even had a gate-house and heavy wooden gate. Trees grew within these village enclosures, but the plain was entirely free from any tree or shrub, every available inch of ground being taken up by the cultivation of cereals. The dreary look of all these villages made me feel home-sick. Nothing but dirty, mud-dried brick houses falling to decay everywhere, with some remnants of the red paper mottoes pasted up last New Year's time, fading and filthy, still sticking to the miserable, rotten doorways.1

Behind the prejudice and disgust displayed by a Westerner newly arrived in China, we can see here some of the most important characteristics of the north China plain: flat land, cereal agriculture, dense population, and impoverished villages. This was the alluvial flood plain of the mighty Yellow River plus a number of lesser streams that flow from the Taihang Mountains along the Shanxi border into the Gulf of Bohai. To the naked eye, the ground lies utterly flat for hundreds of miles in every direction, and the eye is not much deceived. From the point at which the Yellow River leaves the mountains of Shanxi and flows onto the plain in northern Henan, it travels 550 kilometers before it reaches the seaand drops only 100 meters: less than 20 centimeters per kilometer. The muddiest great river in the world, its silt content reaches 25 percent. Millennia of such deposits have yielded the rich soil which makes this region one of China's most important agricultural areas. But the flatness of the plain makes drainage a major problem; and where water cannot easily be drained away, the soil grows saline and yields only the sparest of crops.2

The climate in this region is continental: temperatures average just above freezing during the winter months, with cold dry winds from the northwest, but the summers are hot and humid. Through most of the plain, rainfall averages less than 500 millimeters per year, and is heavily concentrated in the summer months of July and August. Along the coast and on the Shandong peninsula, rainfall is both more plentiful and more reliable, but on the plains, the variation from year to year can be considerable. Lacking any man-made irrigation network, the peasants lived wholly dependent on the heavens delivering summer rains in the right amount at the right time. All too often, the weather was not kind: too much rain would flood or waterlog the land; or too little precipitation would bring parching drought, leaving the soil hard and cracked and often covered with a thin white layer of salts risen to the surface.3

Though the vagaries of weather left this region prone to recurrent natural disasters, the plains had long been one of China's key agricultural regions. In fact, in the late nineteenth century agriculture was virtually the only source of livelihood there. Even in the 1930s, after the advent of the railway and some incipient commercialization, surveys showed roughly 90 percent (and often 95 percent) of the population classified as peasants. Though there were important areas where such commercial crops as cotton were grown, wheat, soy, millet, gaoliang (sorghum) and other subsistence grains dominated the

landscape.4 These crops supported an extraordinarily dense population. A 1950s economic geography puts the population density throughout the plain above 200 persons per square kilometer (over 500 per square mile), and in most regions over 300 persons per square kilometer.5 The population would have been somewhat less in the late nineteenth century, but extrapolating back from 1930s figures, I would estimate the average density in the critical western Shandong portion of the plain to have been about 250 persons per square kilometer.

The North China Macroregion

The vast Yellow River plain lies in the heart of what G. William Skinner calls the "North China macroregion." In analyzing China's cultural geography, no schema has been more influential in recent years than Skinner's regional systems theory. He divides China south of the Great Wall into eight macroregions: North China, Northwest China, the Lower, Middle and Upper Yangzi, the Southeast Coast, Lingnan (centerered on Canton) and the Southwestern region around Yunnan and Guizhou. The North China macroregion is bounded by the Great Wall on the north, the Taihang Mountains of Shanxi on the west, and the Huai River drainage on the south (see map 1). By Skinner's estimate, it was second only to the Lower Yangzi in population density (163 persons per square kilometer in 1893, as compared to an average of 100 for the eight macroregions), but lowest in level of rural commercialization. The unirrigated agriculture produced little marketable surplus: the relatively uniform ecology of the north China plain discouraged local specialization in commercial crops or subsidiary activities within the macroregion; and the northern winters gave peasants a long slack period in which all could engage in handicraft production (especially spinning and weaving) for household consumption. In fact, trade in general was relatively under-developed in North China. There were few navigable rivers and the cost of overland transport of grain equalled its production cost every 200 miles. As a consequence of this retarded commercialization, North China had a lower level of urbanization than all but the substantially frontier regions of the Upper Yangzi and the Southwesta rather remarkable fact given that North China was one of the oldest centers of Chinese civilization, and the location of the national capital since the tenth century. More than any other part of China, the North China

Map 1.
Macroregions of China. Based on maps in G. William Skinner, The
City in Late Imperial China (Stanford, 1977).

macroregion was a land of a few large cities and a densely populated, poor, and substantially self-sufficient rural hinterland.

China's macroregions, in Skinner's analysis, are defined by their fundamental economic independence: "The major commercial cities of the region had stronger economic links with one another than any had with cities outside the region." Within each macroregion there are densely populated lowland cores and more sparsely populated peripheriesusually lying in hilly regions that form watersheds between major river systems. The cores are not only more populated than the peripheries, they are also more fertile, better irrigated, more commercialized and served by better transport networks. In the regional peripheries, by contrast, "local society assumed its most heterodox and variegated guise," being characterized by heterodox sodalities from bandits to religious sects to secret societies and a "disproportionate number of smugglers, outcasts, political exiles, sorcerers, and other deviants."6

The analytic power and explanatory utility of Skinner's regional systems theory is beyond question. Perhaps especially in the Lower and Middle Yangzi, and the Lingnan region around Canton, the regional cores are substantially richer and more commercialized than the hill lands in their respective peripheries. Man-made irrigation works and waterborne transport are far better in these lowland cores. And it is unquestionably the case that the peripheries and regional frontiers have provided the base areas both for China's traditional rebellions and for her modern revolution.

But there are several respects in which North China fails to fit neatly into Skinner's model. The first problem is purely empirical and involves the southern boundary of the region. The basic definition of a macroregion requires that trade be primarily oriented to central places within the macroregionand in North China, especially towards its central metropolises, Beijing and (following its growth as a treaty port in the late nineteenth century) Tianjin (Tientsin). within North China, where even the Yellow River was barely navigable, the key carrier of such trade was the Grand Canal. This spectacular inland waterway was built to carry the tribute rice of the Yangzi valley, Henan, and Shandong to feed the capital in Beijing, but it also served as a major bearer of north-south commerce in all types of goods both on the tribute barges and on private merchant vessels. From the Yangzi port of Zhenjiang (Chinkiang), it proceeded north to the lakes on Shandong's southern border, and through Jining. Then, picking

up the water of the Wen River and following earlier natural watercourses, it meandered north to Linqing on the Shandong-Zhili border. From there the canal followed the course of the Wei River flowing northeast, until it eventually reached Tianjin and the final short stretch to Tongzhou outside of Beijing.

There is no question that the Grand Canal made the boundary between the North China and Lower Yangzi macroregions extremely permeable. But these were clearly two separate economic regions and the key question is: at what point on the canal was the flow of goods primarily to and from Zhenjiang and the cities of the Lower Yangzi, and at what point was the orientation of trade primarily northward toward Tianjin and Beijing. Clearly this should be the boundary between the North China and Lower Yangzi macroregions. The line is fairly reliably established through the reports of the Imperial Maritime Customs, for goods imported through Zhenjiang moved on native boats under "transit passes" which exempted them from the collection of likin a tax on domestic commerce. These transit passes listed the destination of the goods, and the reports of the Zhenjiang customs commissioners indicate that southern Shandong, at least to the point where the northern (post-1855) course of the Yellow River intercepted the canal, lay within the Zhenjiang catchment.7 Since the transport costs would be the same for native and imported goods, we can be reasonably certain that this reflects the division of the commercial networks of the two macroregions.8

As we shall see below, this revision of the boundaries of the North China macroregion makes the analytical portion of Skinner's model better fit the facts. It recognizes much of southern Shandong as a periphery, appropriate for an area of widespread banditry. And since bandits, temporary laborers, and beggars from this area all tended to move southward in times of hardship,9 this revision has them moving more properly toward a regional core rather than further out toward a periphery.

But this is only a minor factual revision of the regional systems model. Far more serious is the fact that so much of the North China core is densely populated, but scarcely commercialized except in the immediate vicinity of the Grand Canal, and poorly served by any transport beyond the wheelbarrow along rutted dirt roads. It was, in fact, less commercialized than many areas of the Shandong peninsula which are actually in Skinner's peripheryareas that also produced an inordinate proportion of the province's gentry elite. Above all,

it was the plain that produced the "heterodox" practices of the Boxerspractices far more appropriate to a regional frontier. To explain these phenomena, we shall have to look for other sources of local variation beyond regional cores and peripheries and shift our unit of analysis from the North China macroregion to the province of Shandong.10

Socio-Economic Regions of Shandong

In the summer of 1900, when advancing foreign troops threatened the Boxers besieging the Legation Quarter in Beijing, these Boxers looked for the arrival of an "old corps" (lao-tuan ) from Shandong to revitalize their struggle against the alien aggressor. Shandong was the source of the Boxer magic; but the movement did not encompass all of the province. The Boxers emerged in the area north of the Yellow River, in what is usually termed "northwest Shandong" (Lu-xi-bei ). The predecessor of the Boxers, the Big Sword Society, was a product of the southwest, along the border with Jiangsu and Henan. Our task here is to define the features that distinguished these areas from the remainder of the province.11

We may begin our survey of Shandong on the peninsula in the east. (See map 2 for the boundaries of the six Shandong regions, and table 1 for a comparison of regions by key measures.) Though the end of the peninsula is characterized by bare, long-since deforested hills and fairly rocky soils, a broad low plain stretches from Jiaozhou Bay on the southern shore to the Guff of Bohai on the northa plain so level that the Chinese once linked two rivers to form a canal across the entire peninsula.12 In 1897, the Germans seized Jiaozhou Bay, on whose southeastern shore they would build the port city of Qingdao (Tsingtao), and soon began surveying for the railway which would link the port to the provincial capital of Jinan. But the railway and the growth of Qingdao were twentieth century developments, and beyond our period of concern. In the nineteenth century, the only treaty port for foreign trade was Yantai (Chefoo), opened in 1862, which developed a very modest trade largely limited to the peninsula and Shandong's northern shore.13

Though the hilly end of the peninsula made this the most sparsely populated region (135 per square kilometer), the well-watered plain, where yields were among the highest in the province, was densely settled and the region as a whole accounted for 24 percent of the

Map 2.
Six Regions of Shandong

TABLE 1 COMPARISON OF SIX REGIONS OF SHANDONG
Region (number of counties)	Population per km2	Yield Indexa	Land Rented	Non-agricultural households (percent)b	Disaster Indexc	Ju-ren per 50,000 population (1851-1900)
Peninsula (17)	135	396	27.5	11.1	1.5	2.95
North Slope (15)	311	472	17.8	16.0	(190.9)	4.40
South Hills (12)	191	381	26.3	12.4	11.0	1.34
Jining (10)	291	461	20.1	11.4	106.7	2.91
Southwest (9)	312	378	23.1	(9.9)	225.1	0.81
Northwest (44)	252	367	10.6	7.1	196.7	1.85
SOURCE : See Note 11 for a discussion of the various sources from which these data have been developed.
a Yield index = wheat yield + soy yield + average of gaoliang and millet yields (all in catties per mu ). This figure, while most useful for comparative purposes, in very rough terms represents the number of catties of grain provided on one mu of land over two years in normal crop rotation. A catty (jin ) was about one-half kilogram, and a mu one-sixth of an acre.
b In the Southwest, Shah county was dropped from the average as the 40,000 peasant households recorded for it (clearly somebody's wild guess) yielded an unrealistic figure of 32% non-agricultural households in the county.
c See note 11 on the composition of this index. The North Slope figure is inordinately high because a few areas northeast of Jinan were either covered or repeatedly flooded by the new course of the Yellow River. The rest of the region was quite free of disaster.

province's population. An early missionary visitor found "the land well-watered by rivulets, the soil rich and productive, abounding in grain, fruits and vegetables."14 The economy was perhaps the most diverse in the province. In addition to fruit trees, the hills grew mulberry and oak treesthe latter providing the leaves for the famous pongee silk, just as the former fed the ordinary silk worm in this ancient silk-raising region. Some streams carried gold dust, which was panned by the local inhabitants, tobacco was grown as a commercial crop, straw braiding and bamboo working were important handicraft industries, and fishing was common along the coast.15

Though the peninsula's relative immunity to natural disasters meant that few were reduced to utter destitution, and both begging and banditry were rare,16 there are a number of indications that society was relatively stratified here. Landlordism (27.5 percent of the cultivated area was rented) was higher than any other region, as was the production of sweet potatoesnormally the diet of the very poor.17 Wealthier folks and town-dwellers ate and drank significantly better (often including wine with their meals), lived in tiled houses and wore fine-spun cloth.18 A traveller through Jiaozhou was particularly impressed by the many fine houses with "high poles before their front door, indicating that some member of the family had been a mandarin."19

Fifteen counties along the northern slope of the central Shandong mountains, from Changyi and Wei in the east to Licheng (Jinan) and Changqing in the west, form the second region in Shandong. In many respects it closely resembled the peninsula, especially in its diversified economy. Silk production was widespread. There were large coal mines and iron founderies in Wei county, and more coal mines plus pottery and glass works in Boshan. But with much less hilly ground than the peninsula, population density was much higher (311 per square kilometer), and agriculture resembled the best that the peninsula had to offer. Streams flowing down from the hills and relatively accessible underground water made this the best irrigated region of the province; on the average, grain yields in this "exceedingly fertile" area were the highest in Shandong.

Commerce was equally flourishing. The most remarkable town was Zhoucun. Though not even a county seat, it was described as "un doubtedly the most important market place in the Province, excepting Chi-ning-chou [Jining]."20 Zhoucun collected and distributed the silk from throughout the central mountain region.21 The city of Wei,

with 100,000 people was a "commercial and manufacturing city of considerable importance" as early as 1866,22 and was the major entrept for goods from the peninsula and the treaty port of Chefoo. Between there and the capital city of Jinan, a traveller described traffic as so great as often to impede progress.23 The contrast to the plains to the northwest was marked. Another road cut off from Wei to cross those plains in the direction of Tianjin. That road was described as "destitute of any place of importance."24

The peninsula and north slope were not only the most developed parts of the province, they also represented its political center of gravity. In the fifty years prior to 1900, 58 percent of Shandong's provincial degree-holders (ju-ren ) came from these two regionswith each providing about half of that total. The northern slope's graduates, coming especially from the provincial capital and the fertile spring-watered counties to its east, were particularly out of proportion to its population. There the fifty-year total represented 4.4 for every 50,000 populationmore than twice the figure for the northwest plains, and four times that of the southwest. But the peninsula was more successful in securing official positions for its gentry: fully 38 percent of the province's metropolitan officials came from the peninsulagiving that region by far the greatest political influence in the capital.

The hills of southern Shandong contrasted sharply with the north. Here was a true regional periphery. In percentage of land under cultivation (27 percent) this region was the lowest in the province. Its population density of 190 per square kilometer was lower than any area but the peninsula. With most of its land hilly, crop yields were low, though the hills also caught the summer storms, providing more dependable rainfall and making the region much less subject to natural disasters (and thus much more stable) than the western plains.25 Still, the peasants were poor here, and there was little commerce to provide alternative employment. Silk moved across the low hills to the towns on the north slope; and in the 1890s, the region began shipping substantial quantities of peanuts south to the Grand Canal and export via Zhenjiang.26 Some bean oil, silk, and straw braid also moved south to the Canal and eventually to Shanghaiindicating the primary economic orientation of southern Shandong toward the Lower Yangzi macroregion.27 On the whole, however, the region was so poorly served by transport that mercantile development was slight. The same traveller of the 1860s who remarked on the bustling

commercial activity north of the hills, was above all struck by the smallness of the towns and the lack of trade in the southcalling Yishui "the smallest city I have ever seen in China."28

The relatively retarded development of this region went together with fairly high rates of landlordism (26 percent of cultivated landsecond only to the peninsula). This no doubt reflected the curvilinear relationship between landlordism and commercializationwith more patrimonial forms of landlordism in more "backward" areas, and a more purely cash-nexus rentier landlordism prevailing in highly commercialized regions. The south Shandong landlords appear to have been resident village landlords, no doubt often of a very rude and uncultured typeand thus relatively high landlordism did not go together with any gentry strength in this area. Though it included 15.6 percent of the province's population, the region furnished only 8.4 percent of the provincial degree holdersa mere 1.3 per 50,000 population over the last fifty years of the nineteenth century, one-third the ratio north of the hills.

Continuing west we come to the Jining region around the Grand Canal. With most of its land sloping gradually toward the canal, and water available from the streams flowing from the hills, this region was second only to the northern slope in agricultural productivity. Commercial crops were fairly common, as tobacco had been cultivated in the area since the early Qing; and a wide variety of fruitpeaches, persimmons, pears, apricots and dateswere sent down the canal to Jiangnan.29 It was not, however, uniformly prosperous. The southernmost areas, where the canal skirted the Shandong-Jiangsu border, were prone to periodic flooding in the late summer and fall as the canal was unable to handle all of the run-off from the hills. When the Amherst Mission passed down the canal in 1816, they found that "whole villages, with extensive tracts of cultivated land, must have been submerged."30 At 291 persons per square kilometer, population was extremely dense. Jiningthe commercial center of the regionwas the largest city in the province, with a population estimated at 150,000. A 1907 source describes Jining as "formerly a large trading center, now an industrial city, where copper, iron and bamboo articles are manufactured"to which list we could add processed tobacco, leather goods, and pickled vegetables.31

In many respects, this region reflects an affinity for the Lower Yangzi quite unlike anywhere else in Shandong. Certainly most of the region's trade was with Jiangnan and the southern end of the Canal. The Jining residents regarded their city as a "little Suzhou"; with the

Grand Canal running through its center, and several scenic spots commemorating the wine-loving poet Li Bo's sojourn in the city during the Tang dynasty, the claim seems not entirely unwarranted. This was a cultured areaencompassing the ancient state of Lu, homeland of Confucius. The lineal descendants of the great philosopher continued to live in nearby Qufu, where the head of the lineage held noble rank as lord of the Confucian estate, and where an academy for the descendants of Confucius and three of his disciples produced an inordinate number of successful examination candidates. Not surprisingly, the region produced a relatively high 2.91 ju-ren per 50,000 populationroughly the level of the peninsula.

Further west we arrive at the southwestern corner of Shandongbasically the portion of Caozhou prefecture south of the Yellow River. In terms of its agriculture, there was little to distinguish it from the Jining region.32 Crops were similar: wheat, soy and sorghum predominating, with smaller amounts of millet and a little cotton. Yields were a little lowerespecially toward the west where the high banks of the relocated Yellow River blocked the natural drainage of the plain. With almost no hills at all, the proportion of land under cultivation and the population density were higherin fact the highest in the province.

But in three crucial respects the Southwest differed from the Jining region. First, it was twice as prone to natural disastersespecially flooding along the Yellow River and the border with Jiangsu to the south. And in part because of this unstable environment, the Southwest, with a population roughly comparable to the Jining region, produced only one quarter as many provincial degree-holders, and three times as many bandits! With only 0.81 degree-holders per 50,000 population, the Southwest had the sparsest gentry presence of any region in the province; and Caozhou was notorious as a bandit stronghold. In fact, the low-lying marshy regions near the current course of the Yellow River provided the lair for the Song dynasty bandit gang immortalized in the novel Water Margin (Shui-hu zhuan ). It was surely no accident that the Big Sword Societywhose initial function was defense against banditryshould find its origins in southwest Shandong.

Home of the Boxers

Finally, we arrive in the northwest plains, the Boxer heartland. Stretching across the entire region north of the Yellow River, and

including a few counties of coastal delta south and east of the river, this was both the largest and the most populous of our regions, with 26 percent of the province's land area and the same proportion of population. This made population density about average for Shan-dong, over 250 persons per square kilometer (or 650 per square mile). With 93 percent of its inhabitants peasants, it was the most purely agricultural of Shandong's regions. But the area was anything but prosperous. Average yields were the lowest of any region in the province, reflecting the persistent problems of waterlogging and salinity of the soil. Even those averages probably understated the precariousness of peasant livelihood here, for the Northwest was also particularly prone to natural disasters.

When the Yellow River shifted to a northern course between 1852 and 1855, it followed the old bed of the Daqing River across western Shandong. As the dynasty fought for its life against the Taiping and other rebels, it was too preoccupied and short of funds to strengthen the embankments and protect the surrounding countryside. Consequently, especially in its lower reaches, the Yellow frequently flooded over the lower ground to its north.33 But the problem did not become severe until about 1880, by which time gradual silting had raised the bed of the river above ground level through most of Shandong, and floods of terrible destructiveness became almost an annual occurrence. In 1886-87 it briefly appeared that the river gods had come to Shandong's aid as the Yellow again broke its banks in Henan and returned to a southern course. Shandong peasants were said to have resisted government requisitions of millet stalks to repair the break, while Shandong officials lobbied to let the river resume its former course. But the province's political weight was no match for that of Jiangnan and its powerful governor-general, Zeng Guo-quan. After a year's respite the breach was repaired and the river returned to continue its devastation of northwest Shandong.34

The shift in the Yellow River not only brought floods. It also delivered a mortal blow to interregional commerce on the Grand Canal. Crossing the Yellow River had always been the most difficult hydrologic problem on the canal. Even in its southern course, Yellow River silt had been causing major problems for travel on the canal since 1785.35 With the shift in the Yellow River, the clear waters of the Wen, which had fed the canal between Jining and Linqing, could not pass beyond Dong'e. The only recourse was to divert the waters of the Yellow into the section of the canal between Dong'e and Linqing. This

could only be done during the summer rise in the Yellow, and only for a short time as the muddy Yellow deposited enormous amounts of silt in the slow-moving canal.

The shift in the Yellow River came at a time when the Taiping rebels were occupying the lower portions of the canal, and the Yangzi valley tribute rice was already being carried to Tianjin by sea. The success of this sea route (especially when steamships were used from the 1870s), meant that the canal would never be restored to its former grandeur. By 1896, a foreign observer would write (with some exaggeration): "As a means of communication between north and south, the Canal need not be considered, as not a single vessel except the junks carrying the tribute rice ever passes."36 Gradually the cities along its route went into decline. The prefectural capital of Dongchang was a prime example. When the Amherst mission passed through in 1816, they found it "well-built, extensive, and populous."37 For a time after the river shifted, the city maintained its position. An 1860s traveller still found it a "very important city" with "warehouses rivalling many in Tientsin [Tianjin] and Shanghai."38 But by 1897, it was "going to decay, its commerical prosperity rapidly on the wane."39 Linqing's fall was even more dramatic, in part because it was also devasted by the nineteenth-century rebellions. From a city of 100,000 during much of the Ming and Qing, it fell to an estimated 40,000 by the late nineteenth century. Said one traveller: "The most notable feature inside the city is its emptiness."40

Though it was certainly the most dramatic, the Yellow River was not the only source of disasters in northwest Shandong. Lesser streams frequently met natural or human impediments to quick runoff on this flat plain, and local flooding of low-lying pockets was a recurrent problem throughout the region.41 If there was not too much water, there was likely to be too littlefor the crops here depended completely on timely rains, and north of the Shandong hills the weather was at its most unreliable.42 A terrible drought was said to have carried off nearly two million people in 1876, and full relief did not come until several years later.43 Ten years later, famine struck again, and "husks, chaff and leaves became part of the daily diet."44

Historically, the Northwest was Shandong's disaster areabut the calamities were human as well as natural. West of Jinan, an imperial highway ran north and south, entering the province at Dezhou, and passing through Gaotang and Chiping before crossing the Yellow River's northern course at Dong'ea path which took it directly

through the heartland of the Boxer movement. From ancient times, this road had carried invading armies both north and southwith the local populace always the greatest victim. The Mongol invasion passed through herespreading a devastation shared by so much of northern China.45 Early in the Ming (1368-1644), much of the area was devasted by the battles surrounding the Yong-le's usurpation of the Ming throne. The Gaotang gazetteer describes the destruction as so great that all the major lineages of the department date their arrival only from the mid-Ming.46

The Manchu invasion probably matched the Mongols in destructiveness, especially if one adds the distress brought by numerous petty rebels during the course of the Ming-Qing transition. Most cities in the area fell repeatedly to rebel forces, before the might of the Manchu armies finally restored a harsh order on what was left of the populationthose few who were not killed or driven off by the rebels, the northern invaders, or the famines and epidemics which accompanied their devastation.47 If the tax records are to be believed, some counties lost 80 percent of their population, and two-thirds of the land lay uncultivated.48 In many areas, the government had to bring in settlers to repopulate the countryside.49 The Grand Canal city of Linqing was probably the greatest metropolis in this region, and the Manchu invasion seems to have delivered a blow from which, in some respects, it never fully recovered. Further southand into the Southwestern regionthe Elm Forest Army (Yu-lin jun) mobilized thousands to resist the Qing, many of them famine victims who joined the army out of desperation. Their effort devastated much of the surrounding countryside, and came to an end only when a flood in the Yellow River left them with no place to flee.50

Though Wang Lun's sectarian uprising of 1774 (discussed in chapter 2) brought physical destruction to much of Linqing, it was a limited affair, and a minor interruption in the long peace that the Qing had brought. But the nineteenth century once again introduced Shan-dong to a time of troublesand again it was west Shandong which suffered the most. On top of the Yellow River floods and the disruption of the Grand Canal came the great mid-century rebellions.

First the Taiping Northern Expedition passed through the area in 1854 after being driven back from the approaches to Beijing. Passing down the great north-south highway, the Taiping invested the city of Gaotang for much of the winter. The Mongol commander Senggerin-chin finally eliminated the rebels in Chiping by surrounding them and

flooding the arealeaving the local peasants once again the primary victims. In the meantime, a Taiping relief expedition had taken Linqing, the battles there leaving that great city in ruins.51 Throughout the 1860s, the Nian rebels raided into Shandong (eventually killing Seng-gerinchin) first largely in the Southwest where they several times threatened to take Jining, then with considerable devastation in the Northwest, especially in 1868. These great rebellions, and the often heavy-handed efforts of the authorities to defend the old order, sparked a series of lesser uprisings and tax-resistance movements throughout the western part of the province, most notably the rebellion of White Lotus rebels allied with the martial artist Song Jing-shi, which was centered in the region west of the Canal. Few areas of the Northwest were immune from destruction in this period, though not necessarily at the hands of the rebels. The people told one traveller through this area that "they suffered quite as much from the Imperial soldiers as they did from the rebels, and at times even worse."52

This was the Boxer homeland: a poor agricultural region, densely populated, but particularly prone to both natural and human disasters. Many of these attributes it shared with the Southwest, which certainly suffered as much from the mid-century rebellions, and almost as much from natural disasters. Since the Southwest was the home of the Big Sword Societythe Boxer's predecessor in the anti-Christian incidents of the 1890swe must turn now to a closer comparison of these two regions, showing how they differed from each other, and also how the two areas differed from the rest of the province. In doing so, we must shift our focus from the natural environment and local economics of these regions to their social formations.

Social Formations of West Shandong

As we begin our consideration of western Shandong, it is worth reminding ourselves that socioeconomic regions frequently transcend administrative boundaries, and that was surely the case in both our Northwest and Southwest regions. Northwest Shandong was scarcely distinguishable from the bordering regions of Zhiliwhich were also unbroken plain, predominantly agricultural, densely populated, and fully dependent on the same rains that watered Shandong. In fact the most systematic study of rural society in this area, Philip C. C. Huang's The Peasant Economy and Social Change in North China consistently

treats Zhili and northwest Shandong as a single ecosystem with a similar natural environment and social structure.53 The ecological homogeneity of the region is extremely important for this study, for it will help to explain why the Boxers spread so easily from northwest Shandong across the Zhili portion of the north China plain.

Southwest Shandong was comparable in almost all respects to the bordering counties in Jiangsu and Henan. This whole area was of course richer than northwest Shandongand as one moved into Jiangsu it became richer still. When travellers on the Grand Canal left the marshy regions of southern Shandong for Jiangsu, they found a new "appearance of prosperity"54 with the country "daily . . . growing more beautiful, better cultivated, and in all respects more interesting."55 In general, both the crops and the yields of northern Jiangsu were similar to southwestern Shandong but the area was somewhat less susceptible to flooding, which was probably the main reason for its relative prosperity. Of course, now that we have moved the Lower Yangzi macroregion boundary north into southern Shandong, northern Jiangsu's prosperity also makes sense in terms of Skinner's regional systems theory, for it corresponds to movement in the direction of the regional core.

If one looked for one distinguishing feature of this southwest Shandong-northern Jiangsu region in the late nineteenth century it would probably have to be endemic banditry. Caozhou, at this time, was known as a "classic Eldorado of bandits."56 The phenomenon was certainly not new, and in part it was the product of the complex intersection of provincial borders in this region. Caozhou prefecture had borders with Jiangsu, Henan, and Zhili provinces; and Anhui was only thirty miles away. Bandits in general did not prey on their immediate neighbors, and by striking across provincial borders it. was particularly easy to escape the justice of the Chinese state, whose vertical linkages were far stronger than its capacity for horizontal cooperation.57

But it is clear that the banditry also had to do with the persistent victimization of this area by the Yellow River before it shifted to its northern course. The Shandong provincial gazetteer of 1736 wrote: "In records since the Song and Yuan, 50 percent of the flooding had been, if not in Cao [county] then in Shah. Since the Zheng-[de] Jia-[jing] period [1506-1567], it has been 80 percent."58 The Cao county gazetteer described life in the region as "like sitting in the bottom of a basin. One is always fearing a deluge."59

There can be little doubt that this precarious existence at the mercy of the river significantly affected the character of southwest Shan-dong societyin part because this area was not uniformly low-lying. Some areas were repeatedly flooded, while those which escaped the floods rather prospered on the rich alluvial soil. The floods thus created victims who might turn to banditry, but also left a wealthy few who were potential targets of banditry. This wealthy elite, in part to protect itself, became increasingly militarized and tended more and more to depend on violent methods to defend its position. The 1716 Cao county gazetteer described the process in the following terms:

The gentry were pure and the people good: such were the old customs of Cao. Then during the Ming, the Yellow River repeatedly broke its banks, and bandits ran amok. There were successive years of epidemics and the corpses of famine victims piled up on the roads. People lost the comforts of house and home, and customs changed accordingly.

The "clear laws and fair administration" of the Qing, the gazetteer obsequiously continued, had brought a restoration of the honest customs of old; but "the tendency for the wealthy to rely on cruelty and violence is well established."60

There is little evidence that the Qing ever broke the bandit hold on this area. Even in the middle of the generally peaceful eighteenth century, one gazetteer spoke of the Caozhou people's "tendency to fight and bring lawsuits. In the southeast [of the prefecture], there is the evil of banditry."61 In the 1820s imperial edicts complained of growing lawlessness in the region.62 But the problem was not only banditry. Salt smuggling was also endemic;63 and in the late nineteenth century this became a prime area for the illicit cultivation of opium. In this it was quite unlike northwest Shandong. A traveller from Beijing in the 1870s was approaching Confucius' home in Qufu (in our Jining region) before he first sighted the opium poppy: "from this point to the old Yellow River, three hundred miles to the south, the cultivation of this plant continues at intervals the whole way."64 Northern Jiangsu, and especially the county of Dangshan, was a prime opium-producing area.65

The impression one forms of the Shandong-Jiangsu-Henan border region is of an outlaw society, which sought as much as possible to live without interference from the state. The salt smuggling and opium growing were certainly illegalbut they were not only a source of livelihood for many, they also brought a necessary commodity (in the

case of salt) or considerable wealth (in the case of opium) into an area where commerce had previously been quite undeveloped. Once the Yellow River shifted to a northern course, the danger of flooding passed and some areas became reasonably prosperous.66 The banditry did not disappear with the floods, of course. The habit was too well ingrained; and public order was declining everywhere amid the nineteenth century rebellions and attendant militarization of local society. Most importantly, the prosperity of some was not the prosperity of allmore likely it provided a tempting target for outlaws. In any case, the area had long since learned to live with banditry. Maps from the early Republic show substantial walls around all but the most insignificant settlements. This was a highly militarized society which had learned to fend for itself in a fairly brutal struggle for survival of the fittest.

With banditry such an important part of southwest Shandong society, it behooves us to look for a moment into its nature, and the robbery cases (tiao-an ) recorded in the Board of Punishments archives in Beijing yield some interesting insights. First, Shandong as a whole was not unusually victimized by banditry. In a sample of 514 memorials from 1890 to 1897, only 25 related to robberies in Shandong. Zhili, with 114 cases (quite probably reflecting stricter security near the capital) and Sichuan with 101 far outstripped the Boxer homeland. But not far behind these leaders was Jiangsu, with 98 cases40 of which were in the single northern prefecture of Xuzhou.67 If we look more closely at cases from Shandong and areas on its borders, we find that though Shandong did not itself suffer inordinately from banditry, it exported a goodly number of bandits to its neighbors. In fact it had a very favorable balance of trade in banditry. Only 3 percent of the bandits caught in Shandong came from outside the province, while in heavily bandit-stricken Xuzhou, 45 percent came from outside of the province, 36 percent from Shandong alone. Along the Zhili border, 47 percent came from Shandong.

Banditry was clearly a much larger-scale business on Shandong's southern border than it was either within the province or in Zhili. In both Zhili and Shandong, bands averaged about 8 members. The Zhili bands, which specialized in highway robbery of merchants traveling across the plains, were quite well armed in the limited number of cases I examined; but in Shandong only half had any firearms at all. Along Shandong's southern border, the average band had 13 members, and two-thirds had at least one gun. But what most strikingly distin-

guished the cases on the southern border was the inclusion of opium in the loot stolen. Only 4 percent of the Shandong cases involved opium, and none in Zhili; while south of the border, 43 percent of the cases mentioned opium among the loot.

Clearly the widespread opium growing in this region both created an addiction that was often satisfied through a life of crime, and created a target population that had considerable wealth to steal. Society throughout this Shandong-Jiangsu-Henan border region was highly stratified, and the wealthy households were inviting targets. Jewelry and foreign (machine-woven) cloth and silk were often stolenitems rarely seen in cases from poorer regions of Shandong. In the opium growing center of Dangshan, every bandit case examined involved a take of over 100 taels (a tael was a Chinese ounce of silver), as did 16 of the 29 cases (55 percent) in the Xuzhou counties bordering Shandong. Only 10 of 49 Shandong cases (20 percent) involved such sums.

But border region banditry did not just derive from a history of natural disasters and the presence of ready targets of opportunity. A fascinating document from Shan county in Caozhou, written by some village literati in 1896, graphically describes the circumstances that pushed so many into banditry.

The wealth of Shan is not the equal of Cao [county], but throughout the prefecture there is always enough. But the wealthy join acre after acre of land, some several thousands of mu , while the poor have none at all. These rich folks care only for their summer rooms or stone arches. . . . With the humanity of the past gone, how can we hope for pity for the poor? These ills are not confined to Shan: all of Caozhou is the same. Powerful families look on poor relatives or neighbors like strangers on the road. They will lend them neither cloth nor grain. They treat their tenants and hired laborers particularly cruelly, arousing a hatred so strong that these people are apt to turn to a life of danger. . .. This all tends to invite banditry. Every few weeks they will come to the door, bearing sharp knives or foreign rifles. When they repeatedly demand a loan, people are quick to assent. With things like this, how will bandits not increase day by day? If the grain of the county were equitably distributed, there would be enough to eat. But there are many without any land. None can make enough to eat through commerce, so they can only hire out their labor. But north China agriculture uses little hired labor: at most 30-40 days per year. There is no other source of foodonly collecting dung. You cannot eat and drink your fill once a year. When they hear of the fun of being a bandit, who is not tempted? Some scatter to settle the frontier in Shanxi, Shaanxi, or the Northeast; or to work as hired hands in northern Anhui and Henan. But many more stay than leave.

This document then goes on to identify four sources of banditry: "(1) border towns,68 (2) bad villages that have produced bandits for generations, (3) bankrupt households, and (4) large families with many sons." While the authors say that throughout Caozhou about half of the bandits came from outside the county, in Shah county most bandits were locals "who hide abroad while the moon is bright, and return to steal during its dark phase." The nearby provincial borders were conducive to such bandit strategies, and most of the theft was done abroadoften from hideouts in one of the neighboring provinces. Still, the bandits always kept their homes:

The bandits of Caozhou love their native villages. . . . Though their home be only a dilapidated one-room earth-walled hut, and though they own not even the smallest piece of land, still they always return. . . . The bandits of Shah hide nearby so it is easy to return, but on their return they need a place to hide so that they will not be discovered. Their lairs are opium dens, brothels and the hang-outs of yamen runnerseveryone knows of these three. But few realize that crafty sheng-yuan ["government students"holders of the lowest (county) examination degree] and evil jian-sheng ["student of the Imperial Academy"usually a purchased degree] also provide hiding places. Most surprisingly, so do rich households and even militia leaders. At first they do so unwillingly, hoping to protect themselves from trouble, but soon they cannot break the connection. . . . When things continue like this for some time, everyone regards it as inevitable. As a result, by now seven or eight out of ten poor and middle households have gotten themselves into this sort of trouble. This is worse than other counties. Households with only a little property first wish to avoid trouble. Then they gradually think of buying stolen goods cheaply. Soon they are fencing the loot to enrich themselves and listening for information [on potential victims] for the benefit of the thieves. It has been like this for a long time.69

Banditry had clearly become an integral part of the social fabric of southwest Shandong. Bandits were not just roving bands who preyed on innocent villagers. They had homes of their own, and regularly returned to them. For the vast majority, banditry was surely not a full-time job. These were peasants, who would return to their villages when there was work to do in the fields, or for festivals when they were accustomed to being with their families. Banditry was a seasonal activity, closely tied to the agricultural calendar. The Board of Punishments records show this clearly. The winter, from the eleventh to the second lunar months, was a prime time for banditsas there was no work to be done in the north China fields in this season. But the eleventh and second months were the peaks of the winter cycle, for

bandits returned home for the Chinese New Year in between. The spring was a busy agricultural season and banditry fell off markedly until the sixth month, usually beginning around mid-July. As one July 1897 account from northern Jiangsu put it: "The season when highwaymen are especially numerous and dangerous is upon us. The kaoliang [gaoliang , i.e., sorghum] is in its prime, and being 7 or 8 feet high and very thick affords a most convenient ambush. It is unsafe to travel alone even in daylight over lonely roads."70 Then for two months, banditry would fall to the same low level as the spring while the fall harvest was brought in, followed by another peak in the ninth month, and a fall to its lowest point in the tenth month when cotton and corn were harvested and the winter plowing done.71

But our Shan county informants also focus on the intimate connection to the extensive landlordism of this arealandlordism which all contemporary references describe as considerably more extensive than the 23 percent of cultivated area shown in the Republican era figures. In the aggregate, southwest Shandong was surely better off than the Northwest, but the extremes of wealth and poverty were pronounced. That is exactly the environment in which banditry is most likely to thrive. It provides both profitable targets for bandits to prey upon, and a large population of poor young males who might be tempted to fill their stomachs with a stint in the greenwood. Equally importantly, as the document above suggests, such extremes of wealth and poverty are likely to engender the sort of resentment and hatred of the rich which would lead men to prey upon the wealthyfor there is little doubt that bandits were discriminating in their targets, and the general population had little to fear from them.72

Landlordism, however, was not only a source of banditry: landlords also provided the nucleus for defensive efforts against bandits. The authors of the document above were no doubt small village landlords, and what they were advocating was the organization of local militia. As we shall discuss in more detail in chapter 3, landlord-ism in the Southwest was not the sort of cash-nexus landlordism of the Lower Yangzi, but one which still entailed mutual patron-client obligations of the parties involved. Landlords would organize local militia, but they would also expect their tenants to join in the protection of the locality. Villages became very tight-knit communities, well-fortifiled against outside threats, and typically subject to the absolute rule of some leading family.73

The three characteristics so influential in shaping the social fabric

of southwest Shandong were all absent in the Northwest. Opium was not cultivated to any extent. Except in a few counties along the Yellow River, banditry was not a major problemespecially not in the specific counties, like Chiping, from which the Boxers grew. As that county's gazetteer tellingly put it: "Before the Republic, the people didn't even know what bandits were."74 And landlordism was almost non-existent.

The weakness of landlordism is particularly striking: only 10.6 percent of the land in this region was rentedthe lowest for any part of Shandong. The most sophisticated attempt to reconstruct landholding patterns at the end of the Qing confirms these Republican figures. In a 1957 survey of 60 villages in the Northwest region, two Chinese scholars found that in the 1890s, 44 (73 percent) of the villages had no landlords renting over 50 mu . By contrast, on the peninsula, only 31 percent of the villages were without landlords.75

Philip C. C. Huang has analyzed at some length and with great insight the "persistence of small-peasant family farming" on the north China plain. There had been some large estates on the plain during the Ming, but most were devasted by the rebellions at the end of the dynasty, and the Qing made every effort to encourage a smallholder economy as the best fiscal base for a centralized imperial regime. The normal bases for significant land accumulation were exceedingly weak in this region. In contrast to the Lower Yangzi, and especially to the Canton delta, lineages never developed as corporate landowners. At most they controlled a few mu of burial land. Nor did commerce provide the basis for the accumulation of wealth and land. The most important commercial development was the spread of cotton cultivation beginning in the Ming. Several counties in Dongchang prefecture (especially Gaotang and En) and the area around Linqing produced significant quantities of raw cotton, but through most of the Ming this production only made the north China plain a dependent periphery to the rapidly developing lower Yangzi core: Jiangsu and Anhui merchants carried off the raw cotton, and brought finished cloth back to the north. Clearly the commercial profits did not significantly accrue to the northern plain.

In the Qing, by the eighteenth century, this relationship was substantially altered, and north China peasants began spinning and weaving their own cotton, exporting much of it to the northern frontier. But by this time, population pressure had driven north China agriculture to such an involuted state that handicraft labor in the

agricultural slack season (and hiring out to wealthier farmers) only served to protect the small peasants' slender margin of subsistence, and thus perpetuate the small holder economy. Some occasionally managed to become "managerial farmers" profiting from commercial crops, but there were severe limits to the amount of land they could manage with existing technology; and within a few generations, family division under the Chinese custom of partible inheritance would reduce these larger farmers once again to smallholder status.

Huang has persuasively argued that the village-level elite in this region was not a landlord elite, but was instead predominantly composed of managerial farmers, who worked their plots, usually of 100-200 mu , with hired labor. Peasants spoke of "the rich" [literally: "wealth-lords" (cai-zhu ), not "landlords" (di-zhu )] when they referred to the village elite.76 Huang stresses the gap between this village elite and the "upper tier" of gentry-official elite, whose much greater wealth was necessarily based on commerce or public office.77 I would suggest that the limited commercial potential of the northwest Shandong plain made it extremely difficult for anyone in this region to pass into that "upper tier" of the elite. Few could rise by farming alone; but the above-cited 1957 study of late Qing landlordism suggests that there were relatively few alternatives in this region. A survey of 131 "managerial landlords" from throughout the province found that in the northwest, 65 percent acquired their wealth by farming, 31 percent from commerce, and only 4 percent from public office. By contrast on the more commercialized North Slope, only 33 percent rose through farming, 58 percent through commerce, and 10 percent from public office.78

Huang has aptly characterized the "conjuncture of low-yield, disaster-prone dry farming with high population density that laid the basis for severe scarcity in this area."79 Since some have associated the small number of tenant-farmers with the absence of an agrarian crisis,80 it is essential to remind ourselves that the weakness of landlordism bespoke not the well-being of the peasantry but its pervasive poverty. The fact is, this region was simply not wealthy enough to support a substantial landlord class; or, to look at it from the standpoint of the wealthy, crops were too vulnerable to the vagaries of weather and other imponderables to warrant any investment in land in this region. The result was a generalized poverty of the peasantry of northwest Shandong. Though wealth and resources were certainly not distributed equally, the extremes were not nearly so glaring as in

the Southwest. People were not so much victimized by rapacious landlords as by the very harshness of the physical environment.

It is essential to realize the extent to which poverty left peasants literally at the margin of subsistence, even in normal times. This account of daily life from the twentieth century provides a graphic appreciation of just what poverty meant on the Shandong plains:

They live in a small humid thatched hut which usually does not let in enough light. The window is exceptionally small. Inside, all ornamentation is extremely simple. If they are a little poor, this single room serves a number of purposes. The stove, pots and bowls occupy one corner: and when they cook black smoke spreads like a thick fog until you can hardly make out a person's face. For fuel in the villages, there are only the leaves of trees or sorghum, wheat or soybean stalks. . . . Their food is also extremely simple. In a whole year, there are very few times when they eat any meat. Usually it is a few vegetables and coarse grain. When the wheat has just been harvested, there is a short time when they can eat vegetables and a rough sort of noodle. They do not eat all of the vegetables in their gardens. Some they take into the towns to exchange for a little grain to sustain themselves. The oil, salt and soy sauce that are used so casually in the city are to them terribly precious things. If they add some sesame oil for seasoning, they use a copper cash coin attached to a stick passed through the hole in the middle of the coin, and with this draw a drop of oil from the jar to flavor their food. Usually, they will just boil a little garlic, peppers or onions in water, and drink that for a meal. Except for weddings, funerals and the New Year festival, they rarely see any meat at all. This is true even of old people. Their normal drink is just boiled water. Sometimes they will put some bamboo or oak leaves in it just to change its color. When they are very busy, they may even drink cold [unboiled] waternot even worrying about whether it will make them sick. . . .Their clothes are all made from native cloth that they have woven themselves. Most are blue or black.81

If these were the conditions in which people lived in ordinary times, one can imagine their plight in bad years. These were truly the classic Chinese peasants of R. H. Tawney's descriptionstanding up to their necks in water, ready to be drowned by the slightest ripple on the surface.82 The situation was particularly severe in northwest Shandong because of the frequency of drought, flood and other natural disasters and the lack of any substantial landlord class which might provide relief to poorer relatives or neighborsrelief offered out of some Confucian sense of moral obligation or, more likely, out of fear that grain not given freely would likely be seized by those made desperate by hunger. As a result, when disaster struck this region, people were very likely simply to pick up and leave. One gazetteer

speaks of people fleeing during the Ming because of high taxes.83 An early Qing edict says that because families often had many sons and little land, "the people of Shandong easily leave their homes without a second thought."84 In the worst cases whole villages would take to the road to begcarrying entire families, men, women and children with them.85

Such mobility was not simply a product of major disasters. In normal times there was extensive movement of young males hiring themselves out as agricultural laborers. Most of this was within a village or between nearby villages, but there was also a certain amount of long-distance travel of what amounted to migrant labor. Some from northwest Shandong moved into neighboring (apparently less populated) areas of Hebei.86 There are even reports of great bands of thousands of peasants from the Shandong-Zhili border near Dezhou travelling south to the Jining area to glean wheat from the fields,87 an activity which was not quite hired laborbut was not quite begging either.

This long-established pattern of migration and mobility was an important aspect in the social structure of northwest Shandong villages. With people constantly moving in and out, villages were much more heterogeneous. Both in southwest Shandong and on the peninsula, single-surname villages (or villages with one clearly dominant lineage) were quite common.88 But on the northwest plain, villages were overwhelmingly a mix of many different surnames, and it was not at all uncommon for new families to come and settle in a villageperhaps first working for a time as someone's hired hand.89 Obviously this makes for less cohesive villages, and more acceptance of mobility among the general population. The northwest Shandong villager does not fit the stereotype of the peasant tied to his ancestral soil.

If we compare, then, northwest and southwest Shandong we find some rather fundamental contrasts. The Southwest was clearly a richer region than the Northwestboth because the soil was better and because the shift in the Yellow River left it somewhat less subject to natural disasters. Being richer, it was also able to support a fairly high rate of landlordism, the agrarian economy of the Northwest, by contrast, was dominated by owner-cultivators and agricultural laborers. The south was also a troubled border region with extensive extra-legal cultivation of opium, and endemic banditry; neither of these was true in the more peaceful north. The insecure environment of the Southwest caused villages to turn in on themselves, and become close-

knit cohesive communities. The villages of the Northwest were more open, the population more mobile, and the villages more heterogeneous. People moved around in the Southwest as well, to be sure. We have seen accounts of them emigrating to the Northeast (Manchuria) or leaving more temporarily to rob or do agricultural labour south of the borderbut the mobility was always out of the villages in this region. Few ever seemed to move in: outsiders were rarely welcome. But in the Northwest, it was much easier for outsiders to gain acceptanceand as we shall see, that was an important factor in the acceptance of those who would teach the new Boxer magic.

The differences between northwest and southwest Shandong were both numerous and fundamentaland these differences do much to account for the different natures of the Big Sword Society in the Southwest and the Boxers of the Northwest. But the two regions had one important similarity: in both the gentry was very weak. This would be critically important because the sparse orthodox Confucian gentry presence in this region was clearly one factor facilitating the spread of such heterodox rituals as were espoused by the Big Swords and the Boxers. It is essential, therefore, to account for the distribution of gentry strength in Shandong.

The Geographic Distribution of The Shandong Gentry

The Chinese gentry were the holders of official degrees conferred by the imperial government, ideally by virtue of their having passed the examinations in the Confucian classics. Toward the end of the Qing dynasty, the sale of examination degrees became increasingly common, affording an avenue into the official elite for those with money but perhaps less capacity for book learning. But the regular path was always the examination route, and only holders of these degrees were recorded in the official gazetteers. Basically, there were three levels of examination degree: the first combined exams at the county and prefectural level, next came the provincial examination, and finally there was the metropolitan examination in Beijingculminating in a palace examination before the emperor himself. At each level there was a quota for the number of successful candidates. As a consequence each county has a number of sheng-yuan (holders of the lowest degree) roughly proportionate to its population, so this degree tells us nothing about the geographic distribution of gentry. Furthermore,

the sheng-yuan was technically only a "government student," a member of what is usually called the "lower gentry" and not yet qualified for official appointment. It was the provincial ju-ren degree which qualified one for official position but as the quota was province-wide, ju-ren were not evenly distributed across the province. Furthermore, because quotas for the ju-ren degree were relatively high (usually between seventy and eighty for each triennial session in Shandong),90 they yield a larger and more representative sample than do metropolitan jin-shi degrees.

Since the Chinese bureaucracy was in fact quite small in size, there were always fewer posts available than there were degree-holders qualified to fill them. It is these degree-holders without official appointment that we think of as the local gentry. They would be the respected members of their communities, people who would be asked to manage public works or organize militia in times of crisis, and people with ready access to political power in the person of the county magistrate or even his superiors. Thus a substantial local gentry presence meant two things: it meant that the local elite would be chosen by its mastery of the Confucian classics and thus presumably quite proper and orthodox in its beliefs and behavior; and it meant that the local population would have spokesmen with the credentials necessary to give weight to local grievances, and to receive a favorable hearing not only at the local level, but even at provincial and national levelsfor if an area produced many degree-holders, it would also produce more higher-level officials through whom influence could be exerted in the national bureaucracy.

With this introduction, we are prepared to look at the distribution of gentry across our six regions. Let us begin with the long view, and examine the distribution as it changes over the course of the Ming and Qing dynasties. Table 2 summarizes the data, giving the percentage of ju-ren from each region for each fifty-year period.

Several trends are evident in these figures. The most striking is the secular rise in the Peninsular and North Slope totals, and the decline in the two western regions beginning in all cases around the middle of the sixteenth century. (The period before 1550 is sufficiently distant from the Boxer Uprising as not to concern us very much. Furthermore, the 1368-1400 period is both shorter than the others, and immediately follows the expulsion of Mongol rule, and is thus sufficiently anomalous to be disregarded.) The rise of the Peninsula and the North Slope seems to reflect the progressive commercialization of this region

TABLE 2 DISTRIBUTION OF SHANDONG PROVINCIAL DEGREE-HOLDERS (JU-RBN ), PERCENTAGES BY REGION, 1368)1900
	Region
Year	Penin- sula	North Slope	South Hills	Jining	South- west	North- west	SW+ NW
1368-1400	12.5	26.3	10.7	10.0	8.7	32.1	40.8
1401-1450	12.5	18.5	11.7	14.5	7.4	35.7	43.1
1451-1500	12.6	15.0	6.8	13.3	8.9	43.5	52.4
1501-1550	10.6	20.0	3.8	12.6	9.9	43.5	53.4
1551-1600	15.9	23.9	3.8	10.9	6.3	39.7	46.0
1601-1650	23.9	22.3	5.7	12.1	6.0	30.3	36.3
1651-1700	26.9	23.5	3.2	12.8	7.1	26.9	34.0
1701-1750	35.4	17.6	3.9	10.6	7.8	25.0	32.8
1751-1800	30.2	19.8	6.1	10.6	6.9	26.6	33.5
1801-1850	26.6	28.0	7.1	13.7	4.1	20.8	24.9
1851-1900	29.5	28.4	8.6	11.2	3.1	19.7	22.8
(1870-1900)	31.8	27.3	8.8	10.5	2.9	18.8	21.7
(Population, ca 1900)	24.4	15.7	15.6	9.4	9.1	25.9	35.0
SOURCE : Shandong tong-zhi .

from the mid-Ming on. The nineteenth century rise of the North Slope is particularly striking, and must be associated with the rise of Wei county as the leading trade center in this area. In the last half of the eighteenth century, Wei produced only 15 ju-ren , but in the first half of the nineteenth the figure jumped to 107.91

The decline of the West is equally striking. In the first half of the sixteenth century, the two western regions supplied more than half of all ju-ren in the province. By the end of the nineteenth century, they produced barely more than one-fifth. The fall, however, was not entirely even and the pattern suggests some of the causes. The sharpest drops tended to coincide with periods of dynastic decline and changefirst in the early seventeenth century, and then in the nineteenth century. Both of these were times of natural and human disasters which, as we have seen, particularly afflicted the West. In a sense, we can see the converse of this process if we look at the

TABLE 3 JU-REN TOTALS ALONG THE GRAND CANAL, 14001900
	Years
County	1401- 1500	1501- 1600	1601- 1700	1701- 800	1801- 1900
Jining	115	84	94	115	153
Liaocheng (Dongchang)	25	34	41	85	56
Linqing	74	102	63	25	13
Wucheng	33	21	18	13	10
Dezhou	82	82	62	74	47
SOURCE : Shandong tong-zhi .

Southern Hills. Here was a region relatively protected from invasion by either Manchu invaders or Chinese rebels. Its percentage of ju-ren accordingly rises when the West suffers its sharpest decline, from 1600 to 1650 during the Ming-Qing transition, and again during the last years of the Qing.

The decline of the Grand Canal was also a factor in the weakening of the gentry in the West, but it was particularly noticeable in the cities north of Dongchang. The hundred-year totals of ju-ren in the counties with major cities along the canal are given in Table 3. Clearly Jining was in fact increasing in prominence from the sixteenth century. Jining, we have argued, really belongs to the Lower Yangzi macroregion and it continued to prosper along with that region. Dongchang continued to prosper until the nineteenth century shift of the Yellow River cut off its links to the south. But Linqing declined precipitously, Wucheng steadily, and Dezhou significantly (but only in the nineteenth century).92

But this still does not tell the full story of the Grand Canal. It was mainly a few large cities (especially Jining, Dongchang and Linqing) whose growth was significantly stimulated by the canaland which, accordingly, declined when the canal silted up. These were the ports where the tribute barges would stop to load and unload merchandise, and their commercial importance grew accordingly. What is striking, however, is the limited commercialized hinterland of the canal, for as we have seen above, the northwest Shandong region was in

general one of the least commercialized areas of the province. Nonagricultural production was almost non-existent, transport away from the canal was overland and expensive, towns were small, and (with the exception of a few areas of significant cash-cropping in cotton) the area produced very little beyond its own immediate subsistence requirements. Even the cotton production, while an important and early form of commercialized agriculture, contributed little to the economic development of the region. It first made northwest Shandong a dependent supplier of raw materials for the Lower Yangzi; and later textile production supported only an involuted peasant handicraft production. Neither of these represented a strong economic base for a wealthy gentry elite.

One explanation for this limited commercializing function of the canal lies in the impact the canal had on the ecology of the north China plain. As early as the sixteenth century, it was observed that because the canal cut across the plain from Jining to Linqing in a northwesterly direction, its dikes blocked the natural drainage of this area, where rivers flowed to the northeast. As a result, several counties south of the canal were repeatedly flooded.93 In the words of the great seventeenth century scholar Gu Yan-wu, "The [sage king] Yu followed the rivers to bring tribute rice. The Ming blocked the rivers to bring tribute rice."94 In addition, the peasantry in the counties through which the canal passed were frequently summoned for corve labor to maintain the canal, to pull the tribute barges across shallow sections or to lighten their loads by carrying the tribute rice overland.95 We should remember that the Ming and Qing maintained the Grand Canal in order to feed the imperial capital. The Canal was not built to facilitate trade in generalmuch less to aid in the development of the areas through which it passed. For much of Shandong it only increased the burden of uncompensated corve labor while bringing actual harm to the local environment.96

These points are important because it has often been assumed that the Canal spurred significant commercialization and economic development in western Shandong. From the changing distribution of the Shandong gentry, I would argue that, outside a few big cities, the Canal's impact may have been negative. Table 4 provides data on gentry (here expressed as percentages of the provincial total) from the area along the Grand Canal. Several things are clear. First, in all categories there is a steady decline after 1550, a date which appears to

TABLE 4 PERCENTAGES OF JU-REN FROM COUNTIES ALONG THE GRAND CANAL (In parentheses is an index taking the 1401-1550 percentage as 100)
		Period
Area		1401-1550	1551-1700	1701-1800	1801-1900
Counties along		25.5 (100)	19.0 (75)	15.1 (59)	12.9 (51)
	Canal: 19
Counties north		19.0 (100)	13.4 (71)	10.0 (53)	6.9 (36)
	of  Jining: 14
Counties south		10.1 (100)	6.6 (65)	4.1 (41)	3.3 (33)
	of  Canal: 9
All west Shandong		49.2 (100)	38.9 (79)	33.2 (67)	23.8 (48)
	counties: 53
	SOURCE : Shandong tong-zhi .

represent the end of the positive contribution of the Canal. Second, in the counties north of Jining, the decline is even sharper than it is for our northwest Shandong region as a whole. And third, the decline is sharpest for those counties lying on the south bank of the canalprecisely the counties which, we were told, suffered because the canal interfered with their drainage.97

The decline in the Grand Canal region was in fact but a particular example of what was probably the basic factor in the relative decline of western Shandong as a whole. The most fundamental problem for agriculture in this region (other than the unchanging dependence on timely precipitation) was waterlogging and the accompanying salinization of the soil. This problem seems to have become progressively worse over time. The Grand Canal dikes, impeding drainage, were only part of the problem. Of more general importance was the over-all increase in population, which led to progressive encroachment on marshes which used to collect excess rainwater, and on riverbeds which had provided the natural drainage for the plain. As peasants (often with powerful protectors) encroached on riverbeds which may have carried water for only a few weeks each year, they might gain fertile (and probably tax-free) land to cultivate; but the villages upstream suffered waterlogging and periodic floods after the heavy

summer rains. Fights and lawsuits arose repeatedly from such encroachments, but the problem continued and the whole area slowly declined.98

Clearly the decline in the gentry presence in western Shandong was related to the unstable environment and the low level of commercialization, especially as the Grand Canal and a burgeoning population both acted to block the natural drainage of the plain and exacerbate the waterlogging which was the principal threat to agriculture in this region. But one more interesting factor emerges if we take all our variables and enter them in a multiple regression equationattempting to predict the number of gentry in a county during the last half of the nineteenth century from various political and socioeconomic variables describing that county. The results of that multiple regression are given in table 5.

For our purposes, it is the fourth column of this table that is the most crucial. The "R-squared" tells us what portion of a change in the dependent variable (the number of ju-ren ) can be accounted for by any given independent variable. In this step-wise multiple regression, the independent variable which best explains the number of ju-ren is first entered into the equation. The second step controls for the first variable, asks what is the next most important variable, and gives us the "R-squared" for the portion of variation explained by the first two variables. The third step controls for the first two variables, and so on. In this case, we see that the number of degree holders in a given department or county is most clearly related to the administrative level of the locality: whether it is a provincial or prefectural seat, a zhou (department) directly under the provincial authorities, an ordinary zhou , or a xian (county). Almost half of the variation in the number of ju-ren can be accounted for by these differences in the administrative level of the most powerful office in the county. Quite naturally, the provincial and prefectural capitals produced a disproportionate number of gentry, for politically ambitious families were attracted to these administrative centers.99 If we then control for administrative level, and enter population into the equation, we can account for almost 60 percent of the variation, since more populous counties tended to produce larger numbers of successful candidates. Neither of these results is the least bit surprising.

The interesting results begin to appear at the third step in the regression. Here we ask: if we control for administrative level and population, what variable best predicts the number of ju-ren ? The

TABLE 5 MULTIPLE REGRESSION OF JU-REN (18511900) ON KEY VARIABLES (County as unit, n = 107)
Variablea	Metric coefficient	Standardized coefficient	t-statisticb	Multiple R-squaredb	Simple correla- tion	Partial correlation at step threec
Administrative level	4.140	0.558	9.03*	0.485*	0.696
Population	0.00002	0.211	2.70*	0.596*	0.535
Number of bandits	1.230	0.138	2.24*	0.624*	0.136	0.266
Commercial tax	0.024	0.213	3.23*	0.645*	0.338	0.207
Cultivated land per capita	3.183	0.164	2.41*	0.663*	0.343	0.172
Percent of land rented	0.117	0.067	1.07	0.667	0.223	0.089
Disasters	0.005	0.060	0.97	0.670	0.111	0.127
Constant	8.944
R-squared	0.670
Degrees of freedom	99
a See note 11 of this chapter for descriptions and sources for key variables.
b Results from stepwise multiple regression. Asterisks indicate t-statistics or increments in R-squared significant at the 0.01 level.
c The partial correlation at step three indicates the correlation of each of the remaining variables after controlling for administrative level and population.

final column gives the partial correlation for each of the remaining variables at step 3 in the step-wise multiple regression. As we can see, the best correlation is to the number of bandits from a given county counted in the Board of Punishment archives. This proves to be a better predicter of gentry strength than any of the alternative variables, after we control for administrative level and population. Basically, the gentry were weak where banditry was widespreadthe negative correlation being indicated in the signs of the metric and standardized coefficients and in the simple correlation. A number of factors might explain this correlation. The gentry were most likely to concentrate in areas that were both socially and ecologically stableand be weak in the difficult-to-control border regions and disaster-prone plains which produced the most banditry. In short, stable areas produced gentry, unstable areas produced banditry. But the major decline in southwest Shandong gentry strength in the late Ming, and again after the mid-eighteenth century, at precisely the same time that banditry was described as increasing in the region, suggests that the two were not just related, in an inverse manner, to a certain ecological constantsthey were also related to each other. The causal relationship probably worked in both directions. Strong gentry discouraged bandit activity; and bandit activity led to a gentry exodus and to the orientation of local elites more toward local self-defense than toward the scholarly path to higher examination degrees.

The next variable is the commercial tax: the best contemporary measurebut still clearly inadequatefor the level of commercialization in each county.100 The t-statistic (as well as the relatively high simple and partial correlations) indicate the importance of commercial development as a determinant of gentry strength. It is clearly more important than cultivated area per capita,101 the extent of landlord-ism, or the number of natural disastersthe last two of which are not statistically significant at the 0.01 (or even the 0.05) level.

I believe that if we had better data, the results would be considerably better than we have obtained here. In particular the disaster measure is really only a measure of flooding, and a measure of all types of natural (and human) disasters (drought, locusts, rebellions, etc.) would no doubt help to explain the sparse gentry presence in the west. Furthermore, landlordism fails to correlate closely because the Southwest and Southern Hills were areas of a quite extensive but rather backward village landlordismrude folk whose lives were not oriented towards the examinations. The more commercial landlordism

of the Peninsula was correlated to gentry strengthjust as it was in the Lower Yangzi valley, the locus classicus of the gentry landlord.

But this regression analysis does perform one function: it confirms that in examining such questions as commercial development on the Peninsula and North Slope, and banditry in the Southwest, we have indeed identified the factors that are most important in explaining the distribution of gentry in Shandongand in particular the weakness of the gentry in the West. That gentry weakness was a critical aspect of west Shandong life, perhaps especially in the Northwest where there were not only few gentry, but also few landlords to lend their weight to orthodoxy and stability in the area. Lacking such a gentry presence, west Shandong had a particular proclivity for heterodox activity. It is to the heterodox sectarian history of the region, and to its popular culture, that we must now turn.

Continues...

Excerpted from The Origins of the Boxer Uprising by Joseph W. Esherick Copyright 1988 by Joseph W. Esherick. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

Edition details

• Publisher: University of California Press
• Publication date: 1988
• Language: English
• ISBN 10: 0520064593
• ISBN 13: 9780520064591
• Binding: Paperback
• Number of pages: 472