The Great Wall and Conceptualizations of the Border Under the Northern Song

Since the publication of the seminal work of Arthur Waldron, it has been widely assumed that the location of the Great Wall of China was not fixed over time, that the Wall as it exists today is a Ming dynasty construction situated far to the south of the original wall of the Qin dynasty. Moreover, according to Waldron, it was not until relatively recently that the Great Wall came to represent a timeless cultural and ecological border between China and the "barbarians" to the north. In his view, the location of the ancient Great Wall of the Warring States and early imperial periods had long been forgotten by the Northern Song. If it survived in Chinese consciousness, it was only as a symbol of the tyranny and excesses of the Qin First Emperor.¹

The appearance of new archaeological and cartographic evidence over the past decade and a half makes it possible to revise certain aspects of Waldron's otherwise incisive scholarship. It is now clear that remnants of older border fortifications survived into the Song and that the Great Wall was intimately tied to novel Song conceptions of the Chinese oikumene. Maps of China drawn up during the Song period are usually very general in nature—indicating only the most important rivers and the names of administrative prefectures. Yet the Great Wall is often prominently depicted, sometimes even anachronistically, as on one set of twelfth-century maps, where it appears for periods long before the first border walls were ever erected (see Figure 1).² In fact, by Song times, the Wall was already seen as a timeless natural boundary between the Chinese and the northern barbarians.³ In this sense, the view popularized by Owen Lattimore—that the Great Wall was constructed within an ecological transition zone between the Eurasian steppe and the cultivated lands of China Proper—may have been first conceived under the Song dynasty.⁴ An important goal of this paper is to explore Song encounters with physical wall remains, the use of the Great Wall to demarcate the border between Song China and its northern neighbors, and the reasons why the Ming Great Wall followed a course remarkably similar to that of the much older Northern Qi fortifications. A future study will examine how notions of the Wall were integrated into classical models of the cosmos, contributing to the development of a medieval Chinese sense of identity.⁵

A second goal of the present paper is to contribute to the critique of the old "tributary model" of Chinese foreign policy. According to this model, developed notably in the work of John King Fairbank, the Chinese emperor sat at the apex of a universal moral social order. Non-Chinese regimes could interact with China only on China's terms, by accepting their own subordination and presenting tribute to the Chinese ruler.⁶ Although many if not most Chinese did subscribe to this sinocentric world view, Wang Gungwu, Tao

Figure 1.

Map of Shang-dynasty China, with anachronistic depiction of the Great Wall along the northern border. From Song ben Lidai dili zhizhangtu, 18–19.

Jing-shen, Morris Rossabi, Christian Lamouroux, Irene Leung, and others have demonstrated that the landmark Treaty of Shanyuan (Chanyuan) of 1005 ushered in a period in which relations with the Khitan Liao (907–1125) were marked by particular pragmatism and innovation. These scholars have emphasized diplomatic parity as the foundation of nearly 120 years of peace between Northern Song (960–1127) China and its neighbor to the north; the regularized use of diplomacy in place of military action to resolve disputes; and the unprecedented attempts at a comprehensive, ethnographic understanding of foreign lands and their people.⁷ However, Song relations with its western neighbor, the Tangut Xia, have not received the same attention. To be sure, Song officials never acknowledged the equal diplomatic status of the Xi Xia state (c.982–1227) and often explicitly differentiated the Tangut Xia from the Liao regime.⁸ Nevertheless, I will argue that, at least in the minds of some eleventh-century Song officials, both Song-Liao and Song-Xi Xia relations were subject to a similar "legal" framework—based on a body of bilateral treaties and other diplomatic correspondence—that superseded any idealistic notion that the Chinese emperor's authority was without bounds.

A third goal of this paper is to call into question assumptions that date the emergence of precisely demarcated border lines, "border consciousness," territorial sovereignty, and strictly-defined notions of the "geo-body" to post-Westphalian Europe or post-seventeenth-century Eurasia, theories propounded by Michel Foucher, Peter Sahlins, Peter Perdue, Thongchai Winichakul, and others.⁹ The present study will demonstrate that, already in the "pre-modern" era, Chinese at times turned to these "modern" approaches to territorial delineation. Military circumstances as well as the practical needs of the Song centralization project necessitated both the "linearization" of the border—that is, the reformulation of the border as a definitive line (at times demarcated by the Great Wall)—and the elimination of frontier marginal zones of ambiguous political control.

Early History of the Great Wall

Before embarking on a survey of the Great Wall under the Northern Song, it is worth briefly reviewing the early history of the wall and the particular lines of fortification that Song witnesses might have encountered. One of the first individuals to attempt a multidynastic history of the Great Wall was the seventeenth-century scholar Gu Yanwu 顧炎武 (1613–1682). More recently, in the 1930's and 1940's, Wang Guoliang 王國良 and Shou Pengfei 壽鵬飛 (1873–1961) produced exhaustive studies that culled extant literary records to date and map the courses of early border walls.¹⁰ Ultimately, however, all of these studies were based on the inconclusive analysis of a series of very obscure place names and elusive literary references. What has truly revolutionized scholarship of the early Wall are the contributions of archaeologists, whose discovery and mapping of datable Wall remains have at times corroborated and at times refuted earlier research. These archaeological investigations have culminated both in impressive syntheses on Han and pre-Han border fortifications—most notably by Xu Pingfang 徐蘋芳 of the Chinese Institute of Archaeology—and in several province-wide comprehensive surveys of extant Great Wall remains.¹¹ Unfortunately, as of yet, border walls dating to the post-Han Period of Disunity have received far less attention from archaeologists, most likely because extant Ming dynasty fortifications were often built directly on top.

The earliest lengthy and continuous border walls were built during the Warring States period, in the fourth and third centuries b.c.e. Many of these walls were not intended for defense against northern "barbarians," frequently serving rather to protect against attacks from neighboring "Chinese" states. In the centuries prior to the Qin unification of China, the states of Qi, Chu, Wei, Yan, Zhao, Zhongshan, Qin, and Wu were all involved in important wall-building projects.¹² The Yan and Qin walls have been the subjects of particularly careful archaeological surveys.¹³ In 215 b.c.e., a few years after the Qin unification of China, Qin Shihuang sent his general Meng Tian 蒙恬 to drive out the northern barbarians and construct a long wall fortifying the border. Although it has been argued that Meng's project merely linked together a number of older walls built along the northern frontier, archaeological evidence suggests that the Qin walls frequently consisted of entirely new structures.¹⁴ Some decades later, the Western and Eastern Han dynasties (202 b.c.e.–220 c.e.) repaired the Qin wall, and also embarked on new wall-building efforts of their own, both to the north and to the south of the old Qin constructions.¹⁵ Most of these Han and pre-Han border fortifications were built far to the north of territory controlled a millennium later by the Song. As such, they are not the focus of the present study. However, the Warring States Qin and Yan walls, depicted on Map 1, both play a role in the subsequent discussion.

Of greater pertinence are the series of walls built after the fall of the Han. Sources record that the Western Jin (265–316), Northern Wei (386–534), Eastern Wei (534–550), Northern Qi (550–577), Northern Zhou (557–581), and Sui (581–617) dynasties all engaged in wall-building or wall-repairing projects.¹⁶ New lines of fortification followed courses significantly further south than the Qin and Han limes, in many cases running along the route of the present-day Ming wall. The Northern Wei, followed by the Northern Qi—in an even more comprehensive building project—initiated the bipartite wall system adopted by the Ming court nearly a millennium later. An "Outer" Wall followed a course north of Datong, while an "Inner" Wall ran further south along the Hengshan range in central Shanxi Province (see Map 1).¹⁷

Not long after Li Shimin 李世民 (599–649) and his father overthrew the second and last Sui emperor to found the Tang dynasty, Li turned his attention to the Turkish tribesmen on the northern frontier. Some officials at court encouraged the new emperor to draft corvée laborers to repair the aging Great Wall fortifications of earlier dynasties. But Li adamantly opposed such a defensive posture. In an oft-quoted critique of the last Sui emperor, he observed that "Sui Yangdi made the people labor to construct the Great Wall in order to defend against the Turks, but in the end this was of no use. I need merely to establish Li Shiji 李世勣 (594–669) in Jinyang for the dust on the border to settle."¹⁸ The Tang emperor's activist border policy of sending talented generals like Li Shiji to the frontier helped the Tang build one of the largest of all Chinese empires and put into question the very desirability of

Map 1.

Remains of the Great Wall under the Northern Song (opposite)

notes: Wall locations have been reconstructed on the basis of references cited in the footnotes. The course of the Song-Liao border is adapted from Tan Qixiang 譚其驤, Zhongguo lishi ditu ji 中國歷史地圖集 (Shanghai: Ditu chubanshe, 1982), vol. 6; the course of the Song-Xi Xia border is adapted from Li Huarui, Song Xia guanxi shi, 224. Note that both Tan and Li base their maps largely on literary sources.

border walls. Under the Tang, fortifications on the frontier were limited to a series of walled cities, including the famous "cities for accepting surrenders" (受降城), in which troops were garrisoned, forming the backbone of what Jonathan Skaff has called a "defense-in-depth" military strategy.¹⁹ Whereas most major regimes occupying north China since the third century b.c.e. had engaged in ambitious wall-building projects, after the reign of Li Shimin, no such massive enterprises were undertaken for a period of several hundred years. Following the Tang example, the Song likewise refrained from extensive wall building.²⁰ Yet, as is clear from abundant references in Song sources, the Great Wall survived in the collective memory of eleventh-century Chinese. Moreover, as described below, surviving remnants of Great Wall border fortifications played an important role in diplomatic negotiations with Song China's northern neighbors.

The "Old Great Wall" and Northern Song Border Demarcation

In order to explore Song encounters with the Great Wall, I have perused a variety of primary sources, including not only dynastic and unofficial histories, but also poetry, maps, embassy diaries, geographical and military treatises, and archaeological reports. Here, I will survey the places where the Wall was observed by Song witnesses. On the one hand, these encounters confirm that the locations of old fortifications were well known in the eleventh century; on the other hand, the use in Song times of the Qin and Northern Qi Great Walls to mark international borders provides insight into medieval Chinese conceptualizations of the frontier and of their neighbors to the north.

Song cartographers, geographers, travelers, and generals in the field were well aware of extant remains of the "old Great Wall" (古長城). Song sources describe four important stretches of Great Wall remains (often incorrectly dated to the Qin period by the Song scholars):²¹ 1) the Northern Qi wall connecting Qianzhou on the Bohai Sea to several important passes in the vicinity of Yanjing (modern-day Beijing); 2) the Warring States Yan wall running parallel to the Song-Liao border in central Hebei; 3) the Northern Qi Inner Wall that followed the crest of the Hengshan mountains in Hedong; and 4) the Warring States Qin wall that traced a southwesterly path below the great bend of the Yellow River (see Map 1).

As is well-known by visitors to China today, the surviving Ming Great Wall meets the Pacific Ocean at Shanhaiguan, where a strategic narrow stretch of coastal flatland links the North China Plain to the fertile lowlands of southern Manchuria. The first wall built at this site dates to the Northern Qi dynasty; it was later refurbished under the Northern Zhou.²² Although Shanhaiguan was already deep in Liao territory by the time of the founding of the Song state, Chinese diplomats traveling here in the twelfth century noted the remains of the Northern Qi Wall just east of Qianzhou prefecture. In Xu Gengzong's 許亢宗 account of his 1125 embassy to the Jin, he writes: "Passing through the eastern gate of the prefectural city [of Qianzhou] and continuing several dozen paces, there lies the old Great Wall. The remains of its structure are unmistakable."²³ Although the precise course of this wall heading west away from the sea is not traced in surviving sources, it appears to have followed a path slightly to the north of the Ming-period Great Wall. According to Wujing zongyao 武經總要, an eleventh-century military treatise, it passed 220 li north of Yutian 玉田 County (in Jizhou) and 45 li north of Miyun 密雲 County.²⁴ Eventually it rejoined the course of the present-day Ming wall at Gubei Kou 古北口, a key strategic pass leading into Manchuria. It was at Gubei Kou that Su Song 蘇頌 (1020–1101) encountered remains of the "ten-thousand li wall" while on a diplomatic mission to the Liao capital.²⁵ Archaeological evidence indicates that, further to the west, the Northern Qi wall more or less followed the path of the Ming wall, passing by the present-day tourist attractions of Mutianyu and Badaling before heading south beyond Beixiling.²⁶

In addition, Song sources describe another wall running southwest from Gubei Kou, descending from the mountains northeast of Shunzhou, intersecting the main road from Yanjing to Manchuria somewhere between the Wangjing 望京 and Sunhou 孫侯 post stations, then heading south across the North China Plain. Two early eleventh-century Song embassies, led by Lu Zhen 路振 (957–1014) and Wang Zeng 王曾 (978–1038), respectively, both crossed this wall soon after setting off from the Liao Southern Capital of Yanjing.²⁷ Nearly a century later, in 1094, Zhang Shunmin 張舜民 (c.1034–c.1100) came upon the "old Great Wall" just north of Shunzhou, an encounter that inspired him to compose a "Rhapsody on the Great Wall."²⁸ Shen Gua 沈括 (1029–1093), famous for his development of rigorous methods of cartography, describes in detail its course in his embassy diary of 1075:²⁹

As for the Wangjing Post, Youzhou lies 30 li to the southwest. From this post, walking for a little over 20 li to the east and a bit to the north, the old Great Wall appears, and after another 20 li, one reaches the halfway post. When one passes this post and then crosses the Sunhou River, after another 20 li, one reaches Shunzhou. As for the old Great Wall, looking off towards it [from Shunzhou], it appears amid the northeastern mountains, and upon reaching Shunzhou, then it turns southward…. As for Shunzhou, the Wangjing Post lies 60 li to the west, and a little to the south. The post here [in Shunzhou] is named Huairou. The city borders on the old Great Wall…. From the prefectural city [of Shunzhou], walking northeastward for several li, the old Great Wall appears.³⁰

Although no Song or pre-Song history or chronicle alludes to the construction of a border wall at this location, the wall described by these travelers was probably contiguous with the fortifications erected under the Northern Qi. The History of Liao (Liao shi) mentions a Great Wall at this precise location, and its fourteenth-century author indeed dates the remains to the Northern Qi.³¹ Recently, archaeologists have identified a lengthy section of this very same wall running north-to-south in the eastern suburbs of Beijing. Eighth-century epitaphs excavated in adjacent tombs identify this structure as the "Great Wall" of the "Qi" period.³²

Whereas these lines of fortification were entirely in Liao territory, a number of other stretches of early Great Wall remains lay farther south, including along the Song-Liao border. In the early decades of the Song dynasty, the "Great Wall Gap" (長城口) was the site of numerous border skirmishes between Song and Liao armies; a Song fortress was constructed here in 980.³³ This place acquired its name from the remains of the Warring States Yan Great Wall, situated nearby according to Song sources.³⁴ Archaeologists have ascertained that, for several dozen miles, the old Yan wall (constructed to defend against the state of Yan's southern neighbor) ran parallel to the Juma 拒馬 River, which marked the Song-Liao border subsequent to the Treaty of Shanyuan.³⁵ The Great Wall Gap was situated approximately where the Yan Wall crossed the river into Khitan territory to the north. However, the Yan wall did not actually serve to delineate the Song-Liao border, except perhaps along a short span in the very close vicinity of the Gap.³⁶

On the other hand, for much of the eleventh century, most of the Song-Liao border in Hedong farther west was demarcated by treaty along the course of extant Great Wall remains—notably, the Northern Qi Inner Wall, initially constructed in 557 as a secondary layer of frontier defense and later refurbished under the Northern Zhou.³⁷ Portions of this line of fortifications are mentioned in Song texts at several locations in Hedong, running more or less along the crest of the Hengshan Mountains. Wujing zongyao as well as the late tenth-century national gazetteer Taiping huanyu ji 太平寰宇記 both mention an "old Great Wall" passing north of Fort Xiyao 細腰 in western Ninghua Military Commandery, then crossing Langu 嵐谷 County in Kelan 岢嵐 Military Prefecture.³⁸ To the east of Yanmen, a stretch of what must have been the same wall is said to have passed seventy li north of Lingqiu 靈邱 County.³⁹ Archaeologists have found surviving remains of this line of fortifications, remains that would certainly have been discernable in Song times (see Maps 2 and 3).⁴⁰

The earliest indication that the Northern Qi Inner Wall served to mark the Song-Liao border in Hedong involves a 1049 policy proposal, in which the Song minister Bao Zheng 抱拯 (999–1062) encouraged the government to reinforce the northern border of Hedong Circuit. Bao paid particular attention to the border region stretching "south of the Great Wall behind Mount Juzhu in Yanmen."⁴¹ More details can be gleaned from the extant accounts of the Song-Liao border dispute of 1074–1075, which has been the subject of admirable studies by Klaus Tietze and Christian Lamouroux.⁴² In 1075—the same year he was dispatched on an embassy to the Liao capital—Shen Gua provided the emperor with an analysis of the disagreement at four points on the Hedong border on the basis of earlier treaties and on Shen's own careful study of relief maps:

1) In the Yuzhou region, our dynasty originally used the Qin Wang Terrace and the Old Great Wall for the border. The Northerners [Liao] claim the watershed is the border. The territory in dispute is a little over 7 li east-to-west. 2) In the Shuozhou region, it was already previously agreed upon to use the northern foot of the Huangwei Massif for the border. Today, the Northerners claim the watershed ridge of the Huangwei Massif is the border. The territory in dispute is about 30 li north-to-south. 3) In the Wuzhou region, our dynasty has used the Fenghuo Border Post for the border. The Northerners claim the watershed ridge at Wayao Hamlet is the border. The territory in dispute is over 10 li north-to-south. 4) In the Yingzhou region, our dynasty has used the Great Continuous Wall (長連城) for the border. The Northerners claim … the watershed ridge marked by Mount Xue, the Huangwei Massif, and Mount Niutou is the border. The territory in dispute is 17 or 18 li north-to-south.⁴³

An addendum to Shen's memorial added that the "Old Great Wall" ran along the northern foot of the Huangwei Massif past Chini Spring and Fort Duanjia and that several earlier documents issued by Liao representatives had set the border precisely along the course of this border wall.⁴⁴ Indeed, the archaeological remains of the Northern Qi "Inner" Wall located southeast of Wuzhou and south of Shuozhou run very near the foot of the Huangwei Massif (see Map 2). Thus, the border both east of Yanmen (that is, the portion of the border facing the Liao prefecture of Yuzhou) and west of Yanmen (that is, the portion of the border facing the Liao prefecture of Shuozhou) were marked according to earlier agreements by remnants of the "Old Great Wall," which was none other than the Northern Qi Inner Wall described above.

Perhaps more problematic is the identification of the "Great Continuous Wall" in the vicinity of Yanmen (opposite the Liao prefecture of Yingzhou).⁴⁵ Tietze suggests that it was "a kind of Great Wall of Sung," though he admits

Map 2.

Huangwei Massif and vicinity, showing watershed ridges

notes: This map was drawn on the basis of maps NJ 49–3, NJ 49–4, NJ 49–7, and NJ 49–8 (series L500, edition 1-AMS), compiled by the Army Map Service, Corps of Engineers in 1955. The course of the Northern Qi Great Wall is based primarily on Zhongguo wenwu ditu ji: Shanxi fence 上 :242–243, 256–257. The location of the Tianchi Temple and of Mt. Huangwei were confirmed using a combination of Qing-period gazetteers and modern-day tourist maps. Fort Xijing is known to have been on Mt. Niutou, though its placement southwest of the mountain peak is hypothetical.

he was "unable to find additional references … to throw more light on what it really was."⁴⁶ Shou Pengfei believes that this was a construction built in 979, though his supposition seems based on a misinterpretation of the sources.⁴⁷ In fact, there is no evidence that a continuous border wall was constructed under the Song; more likely this "Great Continuous Wall" was an extension of the Northern Qi "Inner" Wall, which is known to have run along the Hengshan range north of Yanmen. Indeed, Bao Zheng in 1049 and one Songshi account of negotiations held near Yanmen in 1074 both refer to this wall simply as the "Great Wall."⁴⁸ The alternate nomenclature used during the 1074–1075 debate was likely designed to avoid confusion with remnants of the "Old Great Wall" further to the west and further to the east, allowing these three stretches of border to be treated separately by the Song negotiators.

In summary, it has long been known that throughout the Northern Song period, a one hundred kilometer-long stretch of the Hengshan mountain range separated Song territory from the Liao prefectures of Shuozhou, Yingzhou, and Yuzhou. Pertinent to the present discussion, however, is the fact that if we are to believe the Song position as articulated in 1074, a series of bilateral agreements beginning with the Treaty of Shanyuan set the border in this region precisely along the five hundred year-old remains of the Northern Qi "Inner" Wall. Although the Song court had agreed by 1075 to shift the border south by a few kilometers to run along the "watershed," the Song negotiators continued to insist upon using the "Old Great Wall" at the foot of Mt. Huangwei to mark the boundary in the region south of Yingzhou.

Further west, a much older wall, constructed under King Zhaoxiang 昭襄 (r. 306–251 b.c.e.) of Qin, played perhaps an equally significant role in Song–Xi Xia relations. This ancient line of fortifications is still discernable today and its entire course from Inner Mongolia, across Shaanxi and Ningxia, clear to western Gansu Province has been mapped by archaeologists.⁴⁹ Needless to say, remains of the Warring States Qin Great Wall would have been identifiable in the Song period as well and are indeed mentioned in historical texts and geographical treatises.⁵⁰ In at least three locations, this wall appears to have marked the Song-Xi Xia border.

First, there was the "Great Wall Slope" (長城坂), in modern-day Shaanxi Province. In 1072, in the midst of skirmishes with Song, Xi Xia horsemen crossed the Quye 屈野 River in the vicinity of Linzhou 麟州 to raid Chinese border towns. The Tangut attack was rebuffed by Chinese troops, who went on to pursue the enemy "as far as the Great Wall Slope."⁵¹ The Quye River (called the Kugouye 窟溝野 River today) runs very near extant remains of King Zhaoxiang's fortifications, leaving little doubt that the Great Wall Slope referred to a stretch of the old Warring States Qin fortifications.⁵² The fact that the Chinese cavalry chased the enemy raiders only as far as the remains of the old Great Wall suggests that they probably considered that the Wall marked the extent of Chinese-held territory.

A second stretch of this same wall still stands today at a place referred to in Song times as the "Great Wall Ridge" (長城嶺).⁵³ This ridge is mentioned as the site of Xi Xia-Song border skirmishes in 1066 and 1081.⁵⁴ More to the point, for much of the eleventh century, it served to mark the Tangut-Song border. In 1069, after a Xi Xia envoy sought territorial concessions, his Chinese counterpart referred to a diplomatic letter from the former Tangut ruler Li Deming 李德明 (983–1032), Prince of Xiping, effectively setting the border at the Great Wall Ridge. This contention was sufficient to quiet the Xi Xia envoy.⁵⁵

Finally, Song historical sources make reference to a stretch of King Zhaoxiang's Wall in Zhenrong Military Prefecture (鎮戎軍). Remnants of these Warring States fortifications stand to this day in the vicinity of modern-day Guyuan 固原, Ningxia Province, site of the prefectural capital of Song-period Zhenrong.⁵⁶ In this case, Song sources are explicit on the role of the wall in marking the border. In 1005, amid Song-Xi Xia border skirmishes, the Zhenrong military commander Cao Wei 曹瑋 (973–1030) requested that "eastward from Mount Long along the Old Great Wall we dig a trench to serve as border."⁵⁷ The mid-eleventh-century military treatise Wujing zongyao confirms that this border trench was still maintained a few decades later and names eight fortresses along its course.⁵⁸ Not atypically, Song contemporaries erroneously believed that these stone remains represented "ancient relics of the Qin-Han."⁵⁹

In summary, Song observers were well aware of the location of important stretches of ancient Great Wall remains, especially those lines of fortification constructed by the pre-imperial Qin state and the Northern Qi dynasty. In some places, most notably along the Song-Liao border in Hedong and parts of the Song-Xi Xia border further west, the Great Wall was even used to delineate the border in bilateral diplomatic agreements. The implications of these bilateral agreements go well beyond issues strictly related to the Great Wall. Christian Lamouroux has observed that the Song-Liao border negotiations of 1074–1075 reflected "mutual acknowledgement of the two states' political equality and the will to avoid military confrontation."⁶⁰ What is notable is that Song-Xi Xia border negotiations were conducted in similar ways, emphasizing what one might call "legal" arguments (rather than Heaven's mandate) to justify territorial claims. In 1069, a Song-Xi Xia dispute was resolved when Chinese negotiators brought up an earlier diplomatic communiqué sent by Li Deming that agreed to a border set at the Great Wall Ridge. As in the case of Shen Gua's analysis of the Song-Liao border in Hedong, documents preserved in government archives constituted the "evidence" to bolster China's position when determining the course of the border. But such legal arguments were not limited to interactions with foreign representatives—when language based on the "tributary" model may have been deemed particularly inopportune. A few years earlier, while developing a case for Chinese claims to land "beyond the trench" (壕外) in northwest China, Deputy Commissioner of the Bureau of Military Affairs Hu Su 胡宿 (996–1067) contended at court that this was territory that Tibet had offered to China two decades earlier.⁶¹ Chinese claims to this land thus stemmed from what was essentially a legal transfer of property rather than from rights granted by Heaven to the Chinese emperor.

Needless to say, the notion of a legal transfer of property as well as the very demarcation of linear borders constituted implicit acknowledgment that the Song state was finite and coexisted with Xi Xia, Liao, and perhaps even other states. In fact, in another memorial regarding the Hebei border, Hu Su went so far as to admit that "the land has constant terrain but the state does not have constant territory" (地有常險而國無常地).⁶² Bilateral negotiations and court discussions regarding the Song borders with Xi Xia and Liao together reveal that the Tanguts, Khitans, and Chinese had begun to formulate a body of extra-military principles that could serve as tools for legitimating territorial claims. Older agreements and other archived documents constituted "proof" of China's right to control a particular frontier zone; negotiating on the basis of these documents, then, supplemented military conquest as a means of settling disputes. To be sure, Song was not at peace with Xi Xia. In fact, Wang Anshi's administration in the 1070s aggressively promoted military expansion into the northwest.⁶³ Nevertheless, Hu Su's admission that "the state does not have constant territory" represented an important alternative model of interstate relations—one ultimately harking back to the Warring States period (as Tao Jing-shen has argued with regard to Song-Liao interactions) when rival states recognized each other's right to exist.⁶⁴ The fact that this "legal" approach might apply to Xi Xia as well as to Liao suggests that it was more than just a unique instance of pragmatic thinking; it represented a conceptualization of foreign relations that could in theory apply to multiple coexisting regimes.

Military Defense and the Linearization of the Border

Why were Great Wall remains used to mark the border in the Song, even though they served no apparent military function? In Zhenrong, for example, it was deemed necessary to dig a trench parallel to the old Qin fortifications, themselves presumably inadequate for defensive purposes. Moreover, why did the Ming Great Wall more or less follow the course of much older Northern Qi fortifications? Here I will argue, on the basis of a careful analysis of topographic conditions and debates at court, that particular military and political circumstances during the Song brought about the linearization of the border. The border was fixed along strategic lines of defense that, over multiple dynasties, would exploit the exact same geographic and topographic positions.

There are several different ways of conceptualizing a frontier. Under the Tang, especially during the expansionist period of the seventh and eighth centuries when China successfully extended its control well into the Eurasian steppe, the northern and northwestern frontier consisted of a series of non-contiguous walled cities and forts, as well as the inhabited land in the close vicinity (Figure 2a). To describe Tang border defense, Jonathan Skaff has used the notion of "defense-in-depth," a term borrowed from Edward Luttwak's analysis of Roman military strategy following the reforms of Diocletian.⁶⁵ According to Skaff, the Tang made no attempts to construct frontier lines of fortification. Instead, armies of mobile horsemen were garrisoned in walled cities, where they could be mobilized rapidly (sometimes in response to signals from a network of beacon towers) in the event of an attack. Some of these garrison cities appear on Song-period maps as the famous "Cities for Receiving Surrenders" (受降城).⁶⁶ Such a frontier was not ultimately mappable using fixed boundary lines. It was not clear, for example, how far Chinese territorial control extended outside of the walled city, nor was such a question meaningful from either a military perspective (if there were no extramural fortifications or troops) or a political perspective (if land far from the city walls was neither farmed nor taxed). Indeed, modern attempts to delineate Tang-controlled territory in the northern and northwestern border regions using a fixed boundary line are misleading and presume that control of any two neighboring prefectural cities implied control of the territory in between.⁶⁷ A more accurate representation of the Tang frontier might make use of a series of dots or small circles (representing cities and forts held by the military) connected together by lines (representing roads). Peter Bol has suggested on the basis of medieval geographic texts that pre-Song "spatial ontologies" were based on networks of points rather than on "polygons."⁶⁸ In this sense, Tang conceptualizations of space may have coincided with the de facto military and political circumstances on the frontier.

But such a discontinuous boundary is just one possible model of the frontier. Two alternate conceptualizations became important under the Song: the frontier zone and the border line.⁶⁹ Both reflect what might be termed territorial notions of sovereignty, where the state controls not just fortified pockets of land, but rather large expanses of contiguous space. In the case of the frontier zone (Figure 2b), the border between two political entities is linear in the sense that the frontier zone itself is a contiguous buffer region between two neighboring states. The precise limit of state control, however, remains ambiguous. There may be a no man's land in the middle or a region of joint administrative control, occupied by populations of questionable or shifting loyalties. Alternatively, as was the case in ninth- and tenth-century Hebei, frontier generals or warlords might maintain their quasi-independence by playing both sides and switching allegiances when necessary to preserve their semi-autonomy.⁷⁰ Although fortifications may be present, these constitute defensive constructions built well within the territory being defended.⁷¹ Crossing these fortifications would not put one immediately in the territory of the foreign state, but rather in a liminal zone of ambiguous political control. The border zone should be distinguished from the border line (Figure 2c), where a precisely defined boundary between two states could be demarcated by means of border stones, a fence, or a wall. Either side of such a border line is unambiguously the territory of one state or the other.

Figure 2.

Three frontier models

notes: These figures depict three types of frontiers. In (a), two neighboring states control pockets of territory in the vicinity of fortified border cities, as well as the roads linking these territories to the interior. In (b), the two states control contiguous expanses of territory on the frontier, separated by a liminal zone of ambiguous political control. In (c), a fixed border line (perhaps defined by treaty) clearly and unambiguously defines the territory of the two states. For an alternative approach to the categorization of frontier types, see Naomi Standen, "Nine Case Studies of Premodern Frontiers," in Frontiers in Question, eds. Daniel Power and Naomi Standen (New York: St. Martin's Press, 1999), 23.

On some levels, it is tempting to see these three frontier types as evolving sequentially over time, with the most "primitive" state defining its frontier by means of non-contiguous cities and forts; the more advanced state developing a well-defined border zone; and the modern nation state conceiving of the frontier as a fixed line. Such a teleological interpretation, however, is over-simplistic and neglects the possibility that generals, policymakers, and other historical actors might make practical decisions in response to the particular circumstances they faced. Indeed, the linearization under the Song of the unbounded and non-mappable Tang border was the result of specific military and political requirements.

Chinese policymakers of the eleventh century were well aware of the fundamental difference between Chinese armies and the armies of the steppe tribes to the north: whereas Chinese forces were dominated by large-scale infantries, the northern nomads fielded light and mobile cavalry troops. This discrepancy reflected the great cultural and ecological differences between the agricultural zone of China Proper and the pastoral-nomadic zone of the Eurasian steppe.⁷² Agriculture, as a general rule, is much more efficient than animal husbandry (with regard to calories produced per acre).⁷³ Consequently, the Chinese state encouraged agricultural production as a means of maximizing tax revenue. However, the great Eurasian Steppe, extending from Manchuria in the east, across north China and Central Asia, to the lower Danube in the west, is an arid zone where agriculture is difficult, but where rich soils and minimal rainfall readily support the extensive grasslands upon which mobile pastoral herders depended.⁷⁴ Because of fiscal imperatives favoring agriculture, Chinese horse-breeding programs never produced the vast herds of livestock raised on the steppe, resulting in perennial shortages of horses. Over the course of multiple dynasties, attempts to remedy this problem led to short-lived and usually ineffective breeding programs as well as to large-scale horse trading.⁷⁵ Moreover, because raising animals was the way of life for women and men of the steppe, the northern nomads were always more proficient at horse riding (a skill they honed from a very young age) and at training war horses. In the end, native Chinese cavalry armies were usually ineffective, due both to inadequate training on horseback and to insufficient and inferior mounts.

Although the Tang made good use of mercenary armies of steppe horsemen—an essential requirement of Tang "defense-in-depth"—this practice was more limited in scope under the Song. Consequently, Northern Song strategists obsessed over the implications of using foot soldiers against mounted cavalry armies. On broad and flat plains, infantry armies were easily outflanked by their horse-riding enemy. In the words of the eleventh-century military manual, Wujing zongyao, "The cavalry's advantage is on flat plains; China, with its many foot soldiers, benefits from strategic barriers" (騎利在平地, 中國多步兵, 利於險阻).⁷⁶ Because slow-moving Song troops were easily eluded, "defense in-depth" was no longer viable; instead, it was essential under the Song to establish continuous linear fortifications that could not be circumvented by fast-moving horses.

In Hedong and regions further west, the Song exploited advantageous terrain when establishing their linear defenses. One dramatic natural border consisted of the Hengshan Range, which separates the Sanggan River watershed from the Fen and Hutuo River watersheds (Map 2). Establishing a line of defensive fortifications along this range was important for blocking access to the Fen and Hutuo river valleys, which both flowed south deep into Chinese territory. Military strategists understood that mountainous terrain gave horsemen virtually no advantage over foot soldiers, whereas a cavalry attack could quite easily sweep around any forts or walled cities in the river plains below. It was for this reason that, as early as the Northern Qi, a lengthy continuous wall was erected in these mountains. A millennium later, the Ming "Inner" Wall followed a course in the same range. Although the Song never controlled northern Hedong (the region south of the "Outer" Wall), Song China tenaciously clung to its strategic position in the critical Hengshan range. In the 1074–1075 border dispute, the Liao sought to renegotiate the course of the border in order to gain control of the watershed ridge atop the Huangwei Massif.⁷⁷ The Song court was agreeable to a southward shift in the border further east, thus reestablishing the official boundary at the Sanggan-Hutuo watershed ridge. However, it refused to budge from its position that the border further west was set by treaty along the northern foot of the Huangwei Massif. Mount Huangwei was on a promontory extending well into the Hutuo River watershed. Control of this secondary ridge would have given the Liao army complete control of an important tributary of the Hutuo River, precisely the route used by an invading Liao army in 936 (see Figure 3).⁷⁸

The military significance of linear defenses was not limited to the Song-Liao border. Further west, Chinese foot soldiers were no match for Tangut Xi Xia horsemen. Song policymakers well understood that a failure in the main defenses in Qinfeng 琴鳳 and Jingyuan 涇原 Circuits would give an invading army access to the Wei River valley and the Guanzhong region, once the great metropolitan core of the Zhou, Han, and Tang dynasties.⁷⁹ Fortunately, the terrain south of the Ordos was rugged and generally unconducive to rapid cavalry movements. But because the topography did not present the same sharp and distinct natural border found further east in Hedong, it is less clear why segments of the old Great Wall were utilized here to fix the Song-Xia border. In principle, military considerations would have permitted establishing defensive lines along several possible strategic ridges. Presumably, when establishing their defenses well over a millennium earlier, Qin military men surveyed a course that made maximal use of elevated ridges.⁸⁰ Song tacticians may have deemed it advantageous to employ the same lines of defense in some locations.

One important route of attack that the Xi Xia sought to exploit was the relatively broad Qingshui 清水 River valley, where Song's Zhenrong Military Command was headquartered. Because this important defensive position

Figure 3.

Looking east across Mt. Huangwei, showing 936 Liao invasion route

notes: This bird's eye view of the vicinity of Mt. Huangwei was produced using Google Earth, ©TerraMetrics 2006, ©Google 2005. The hypothetical observer is situated at an altitude of 20.0 kilometers. On this image, elevations are accentuated three-fold to enable visualization of mountainous terrain even at such a high altitude. The Liao invasion of 936 made use of the Yangwu 揚武 valley (see ZZTJ 280:9148), coinciding with the valley just above Yangwu 陽武 Village, which appears southwest of Guoxian 崞縣 on map NJ 49–8 (series L500, edition 1-AMS), compiled by the Army Map Service, Corps of Engineers in 1955.

occupied the plains, one general in the field assessed the situation here to be "convenient for cavalry warfare and not to China's advantage."⁸¹ It was here that Cao Wei had a trench built along the course of the old Qin Great Wall, thus establishing a linear line of fortifications where the natural topography was insufficient in itself to slow a cavalry attack. This deep trench, between fifteen and twenty meters in depth and width, seems to have been formidably effective as a linear obstacle to cross-border raids.⁸² In 1002, one Tangut incursion was initiated only after the northerners "spied out a slacking in the preparedness of the [Song] patrols and, one evening, filled in the long trench and crossed beyond the Old Great Wall."⁸³ A more telling illustration of the effectiveness of the trench occurred at the expense of a Chinese army. In 1042, the Song general Ge Huaimin 葛懷敏 and 10,000 men were surrounded by the Tanguts near the fortress of Dingchuan 定川. Ge attempted to flee on horseback into Song-controlled territory, but the enemy had destroyed the bridges across the Great Wall trench, effectively blocking his path of escape.⁸⁴

Unlike the situation in Hedong and regions further west, the Song northern border in Hebei was neither a natural geographic border nor was it blessed with rugged terrain. Nevertheless, Northern Song tacticians and policymakers expended great energy establishing an artificial linear border that could match in effectiveness the Hengshan range farther west. After Emperor Gaozu of the Later Jin gave up sixteen prefectures in northern Hebei and Hedong to the Liao in 936, the new Chinese-Khitan border was situated over a hundred kilometers south of the strategic mountain passes where remains of the old Northern Qi Great Wall still survived. As recounted in 1044 by Fu Bi 富弼 (1004–1083):

Formerly, before the lands of Yan and Ji were lost, [the passes of] Songting Guan, Gubei Kou, and Juyong Guan served as critical strategic positions for the Central Plain, keeping out the Xiongnu [i.e. northern enemies], who did not dare cross southward. Emperors and kings through the ages were fixated on defending [these passes] and never once neglected them. Ever since Gaozu of Jin abandoned the entire land of Yan, the northern strategic passes have fallen completely into the hands of the Khitans.⁸⁵

Now that the border traversed the North China Plain, which was as flat as a "straw mat" in the words of Song Qi 宋 (998–1061), there were no natural barriers that could hamper a cavalry invasion.⁸⁶ According to Zhang Ji 張洎 (933–996), "southward from Yan and Ji, flat ground extends for 1000 li, devoid of barriers provided by famed mountains or great rivers…. This is what is known as losing the advantage of terrain and causing difficulties for China."⁸⁷ Hebei's strategic vulnerability was evident in the winter of 1004, the first year of the Jingde era, when a Liao army charged south across the Juma River near Shanyuan. Before the end of the year, Khitan troops had reached the banks of the Yellow River, the last natural barrier protecting Henan and the Song capital of Kaifeng. According to Hu Su, who served as Deputy Commissioner of the Bureau of Military Affairs from 1061 to 1066:

Formerly, in the north, the strategic territories of Lulong and Baitan separated north from south. When China controlled these, we were able to restrain the northern enemies. By contrast, today, these two strategic spots are in enemy hands and the topography of Hebei provides no strategic spots that can provide protection. Northward from Henan to Shanyuan, flat terrain extends for 2000 li. In the Jingde era, northern horsemen reached the Yellow River in no time.⁸⁸

To secure Hebei, a complex hydraulic defense system was constructed along the border, under the initial supervision of He Chengju 何承矩 (946–1006). Rivers were dammed and redirected to form a network of pools and marshes that were "too deep to ford and too shallow to cross by boat."⁸⁹ Integrated into these defenses was a series of paddy and other fields that supplied the border troops with food. Deep, wide irrigation channels between fields constituted additional obstacles to shield against northern cavalry incursions. This entire defensive system occupied a band dozens of miles wide, protecting the border from the Great Wall Gap in the west to the ocean in the east.⁹⁰ The hydraulic defenses received great praise from a number of officials at court. According to Lü Tao 呂陶 (1029–1105):

I have seen how the prefectures of Heshuo [i.e., Hebei] come right up to the northern border. Thirty li past Xiongzhou and you have crossed beyond the border. But the terrain is completely flat, with absolutely no strategic positions where we can obstruct or restrain [enemy troops]. Although prefectural city walls have been completed and troops have been trained in order to prepare against an emergency, if ever the warning beacons were to flare up [alerting troops to an impending invasion], in the end, it would be easy to overrun our defenses. It is only the [network of] shallow pools that can really provide defense against attacks.⁹¹

Ouyang Xiu 歐陽修 (1007–1072) also remarked on the importance of the defensive waterworks in Hebei:

Thus, the Khitans have complete control of the hills of You and have cut off the impediment at Gubei [i.e., Gubei Kou]. Roving armies invade us, crossing the Changshan range, treading over all of Wei [in southern Hebei]. After the Shanyuan campaign [in 1004], their horses drank from the waters of the Yellow River and our populace had no means with which to survive. The situation is such not because the northern caitiffs [i.e., Khitans] are brave and numerous, but because we have lost our strategic strongholds. Today, as there are no mountains and hills in which to establish strategic defenses, that on which we can depend consists only of the embankments that channel the flow of the rivers and consolidate the accumulated waters in order to produce a terrain of strategic waterways. As for the weak points [in these hydraulic defenses], we can garrison these with crack troops.⁹²

However, hydraulic defenses were not favored by all policymakers. At court, there was recurring criticism regarding their effectiveness. On the one hand, the pools and swamps could be outflanked, notably by using routes through the western hills, at elevations unattainable by the dammed waters. As observed by Song Qi:

China has lost the impediment at Hubei [i.e., Gubei Kou]; southward from You, the land is [flat] like a mat. He Chengju first piled up the embankments and established military farms, delineating their fields with irrigation channels, in order to form a line against enemy cavalry charges. But [the hydraulic defenses] terminate after a few hundred li, without reaching the ocean to the east, nor the Changshan mountains to the west. Beyond this, as there is nothing with which to impede them, enemy soldiers frequently arrive with their yurts, whipping their brawny horses forward across [our] flat lands.⁹³

To remedy this defensive weakness, a dense palisade of three million trees was planted in the foothills west of the Great Wall Gap, with tree trunks close enough together to block incoming horses.⁹⁴ The defensive lines were further reinforced with horse traps, which included pits and a type of cheval-de-frise known as "deer antlers."⁹⁵ Finally, as has been revealed recently by archaeologists, a series of underground chambers and passageways just south of the border apparently served to conceal Song troops.⁹⁶

Nevertheless, criticism of the hydraulic defenses did not abate. Many at court wondered why it was necessary to sacrifice so much valuable farm land to establish a line of defense of questionable effectiveness. It was argued that the pools and swamps frequently dried up in summer and autumn and occasionally froze over in winter.⁹⁷ The danger of weaknesses in the Hebei border became more critical after the Yellow River breached its levees at Shanghu in 1048, subsequently altering its course dramatically to the north. The great river now merged with the hydraulic defenses and the Juma River on the border, eventually draining into the Bohai Sea in a broad and shallow delta that many feared a cavalry army could quite easily ford.⁹⁸ According to Song Qi:

After the Jingde era (1004–1007), officials [in Hebei] broadened the [extent of the] embankments to form pools that extended to the ocean. In addition, the Yellow River constituted a [second] line of separation to its south…. Ever since the Yellow River breached its banks at Henglong and [then] Shanghu, its wandering waters have meandered over Beiqiu, flowed across Yongjing, returning northward towards the sea. Penetrating Qianning, [the waters] run freely without restraint into the sea. Upon flowing to the frontier, the [resulting] shallow marshes constitute several hundred li of flat land. Thus, the prefectures of Bin, Di, Zi, and Qing [in Shandong] have lost the strategic defenses of the Yellow River, and have nothing upon which to depend.⁹⁹

Song Qi's contemporary Hu Su went so far as to argue that after the breach at Shanghu, China had lost the last natural barrier shielding the Song capital of Kaifeng:

The topography of Heshuo [Hebei] has no strategic positions that can serve for defense; southward from Xiong and Mo [prefectures], there is 1000 li of flat ground…. Since the Yellow River breached [its banks] at Shanghu, no longer following the old Henglong channel, river flow in Hebei meanders without restraint. China has lost the strategic barrier provided by the great river and cannot separate itself from the [northern] horsemen. In the depth of winter, when the waters freeze over, armored horses can cross over. If crack cavalry troops of the northerners were to charge into the unprotected plains of Cang and Jing prefectures [in the eastern portion of Song-controlled Hebei] and were to gallop southward, then the Eastern Capital [i.e., Kaifeng] would be threatened.¹⁰⁰

Although unsuccessful in convincing the court to attack Liao until just before the Jurchen invasions of the 1120s, Northern Song revanchists always maintained that the only way to guarantee the safety of the northern frontier was to reconquer the lost prefectures in Hebei. The northern and western hills that rise dramatically above the flat plains of modern-day Beijing convinced some Northern Song writers that they had been established by Heaven as a natural boundary between China and its northern pastoral-nomadic neighbors. Nowhere was the contrast in terrain more impressive than at Gubei Kou, the mountain pass through which numerous Song envoys made their way to the Supreme Capital of the Liao. In 1039, while on an embassy mission and standing near where Su Song had once observed remains of the Northern Qi Great Wall, Han Qi 韓琦 (1008–1075) wrote a few lines of verse:

East and west, the layers of mountain peaks stand profuse and tall,Numerous able horsemen cross through the pass.It was Heaven's will that north and south would be divided here,But when it comes to today, what is Heaven's will?¹⁰¹

For Han Qi, it appeared counter to "Heaven's will" (天意) that Khitan horsemen could freely cross into the plains below.

In summary, whereas the Tang at its height was able to implement a "defense-in-depth" military strategy, garrisoning mounted troops in fortified cities without additional linear fortifications, the Song was far less confident in the effectiveness of its cavalry forces. Eleventh-century policymakers were convinced that preventing an invasion by northern horsemen required establishing defenses that could not be outflanked by a far more mobile enemy. In Hedong and south of the great bend of the Yellow River, the Song court and its generals and diplomats fought tenaciously to maintain control of key mountain ridges. Because previous dynasties had faced similar threats from the north, it was not coincidental that the border largely corresponded to the course of the older Qin and Northern Qi Great Walls. On the other hand, in Hebei, where China had lost control of the critical mountain passes north of modern-day Beijing, ministers at court were obsessed with the vulnerability of their infantry armies. Military circumstances required establishing a lengthy linear zone of hydraulic defenses across the flat expanses of the North China Plain, defenses that could not be outflanked by a Khitan cavalry charge.

Song Centralization and the Linearization of the Border

Above, I have proposed that military requirements were responsible for significant differences between the Tang and the Song in terms of the nature of the frontier. Whereas the Tang could base its frontier defense on the control of critical garrison towns, Song policymakers needed to establish a linear border to prevent Khitan and Tangut horsemen from outflanking Chinese infantry positions. In terms of the tripartite categorization proposed earlier (see Figure 2a–c), military needs did not require a precise border line. Lines of defense in Hebei, in particular, consisted of a zone that, although linear (in the sense of contiguous), occupied land many kilometers in width that combined marshes and shallow pools with an assortment of other types of fortifications.

But whereas this militarized zone was maintained throughout the eleventh century, a concurrent strand of opinion at court, seeking to enhance the power of the central government, advocated a more carefully delineated border—a border line as opposed to a broad border zone. On the one hand, the Song administrative apparatus, in seeking to maximize its tax base, could ill afford to maintain an unproductive no-man's land even on the frontier. Numerous men at court complained that the hydraulic defenses in Hebei effectively eliminated vast acreages from the tax rolls. Bao Zheng, who had served in office in Hebei, estimated that as much as one half of the land in this province was dedicated to pasturage (to support a fairly ineffective horse-breeding program) and hydraulic defense.¹⁰² One proposal circulated in 1072 recommended replacing the network of pools and marshes with a single tree palisade that would cross Hebei from the western mountains to the ocean in the east. Not only was a palisade deemed more dependable in summer (when the swamps and pools might dry out) and winter (when they might freeze over), it would also have allowed all of the flooded land to be farmed again.¹⁰³ In Hedong farther west, one important stimulus of the Song-Liao border dispute of the mid-1070s involved Liao households who had crossed the border to farm depopulated land in Song territory.¹⁰⁴ Christian Lamouroux has noted that, as early as 1044, Ouyang Xiu had warned of the risks of establishing such a no man's land along the Song-Liao border. Territorial encroachments by Liao farmers and herders as well as Liao-sponsored temple construction projects in Song territory were both used by the Liao government as rationales to proclaim sovereignty over this border zone.¹⁰⁵

Second, neither the Song nor its northern neighbors benefited from disloyal border populations. By readily shifting allegiances from one side to the other, frontier warlords became playmakers who could substantially hinder the centralization projects of the Song, as well as of the Liao and Xi Xia states. Naomi Standen has proposed that the Song-Liao Treaty of Shanyuan represented the culmination of a century-long struggle to eliminate the autonomous warlords that had dominated Hebei since the mid-eighth century.¹⁰⁶ One term of the treaty required a fixed border line, leaving no ambiguous space in the middle where bandits could thrive out of the reach of central government authorities. A second term required that both sides return cross-border fugitives. In the tenth century, roving armies in the tow of fugitive military commanders could maintain enormous political influence at the expense of the central government.¹⁰⁷ Eliminating this destabilizing force was to the advantage of both Song and Liao. Although the text of the earliest Song-Xi Xia treaty does not appear to have survived, we have seen that this border was fixed as well, at times along the Warring States Qin Wall. Moreover, a letter from the Tangut leader to the Song court explicitly cites a clause in an earlier agreement requiring the forcible return of cross-border fugitives.¹⁰⁸

In addition, Song policymakers recognized that central government actions could negatively impact frontier populations, encouraging rebellious activity. The large armies stationed in Hebei wreaked havoc on border villages, occupying land once cultivated by local farmers.¹⁰⁹ The hydraulic defenses imposed an even larger burden. Reports circulated of peasants destroying newly-constructed dikes after the resulting pools and marshes flooded their fields and ancestral tombs.¹¹⁰ Such a population of questionable loyalty might join forces with the enemy or at the least provide the enemy with critical intelligence information. Partly for this reason, many men at the Song court insisted on preserving a hermetically-sealed border, eliminating to the extent possible all cross-border movement. Hu Su openly criticized frontier administrators who had allowed locals to fish in the border river in violation of Song-Liao treaties.¹¹¹ Thus, eliminating the ambiguity of the frontier zone by establishing a precise border line bolstered Song centralization efforts in two ways: by expanding the base of taxable farmland and by significantly reducing the potentially destabilizing threat of frontier populations.

Conclusion

Three important conclusions arise from the above discussion. First, I have brought into question claims that the Great Wall was of no significance in the Song and that the present course of the Ming Wall was fixed no earlier than the fifteenth century. In fact, long stretches of the Song border with the Liao and Xi Xia states coincided with remains of the Qin and Northern Qi Great Walls. Given that the Ming and the Northern Qi Great Walls followed nearly identical courses in Hebei and Hedong, it now seems that China's border was recurrently situated along a more or less fixed line for well over a millennium. This proposition is not unexpected given the particular geography and topography of the northern frontier. As I discuss elsewhere, the notion that the Great Wall marked a timeless ecological and cultural border between China and the northern barbarians was already enunciated in Song times. The Great Wall was integrated into a cosmotopographic model that combined empirical knowledge of the locations of mountain ranges and of zones of ecological change with commonly-held beliefs regarding the essential structure of the universe.¹¹²

Second, I have provided further grounds for questioning enduring claims that the "tributary model" defined Chinese foreign policy for "over one and a half millennia."¹¹³ In fact, there is a near consensus among specialists that Chinese diplomacy in the eleventh century was not limited to such a model. But scholarship so far has emphasized the pragmatism of Northern Song relations with Liao. Although many Song Chinese may have found the Khitans to be more "civilized" than the Tanguts of the Xi Xia state, foreign policy towards the latter group was at times marked by a practical bilateralism very much reminiscent of Song-Liao diplomatic interactions—as when Chinese politicians resorted to "legal" arguments on the basis of archived documents to bolster territorial claims in the northwest. It would be an exaggeration to claim that northeast Asia of the eleventh century consisted of a mature interstate system. Yet it is notable that the Song implicitly recognized the right to exist of its two northern neighbors and that all three regimes had begun to formulate certain legal principles that provided tools and approaches for the non-military resolution of territorial disputes. Given recent scholarship demonstrating the importance of pragmatism in Tang-Tibetan and Qing-Russian relations, it seems quite possible that, more often than not, the sinocentric tributary ideal did not preclude a flexible approach to foreign policy.¹¹⁴

Finally, I have sought to put into question commonly-held assumptions that border linearization and territorial conceptions of sovereignty emerged only in the "modern" era. Already in the eleventh century, facing the threat of Khitan and Tangut cavalry attacks, China was forced to construct linear military defenses, effectively abandoning an earlier Tang "nonbounded" notion of territorial sovereignty (to use Thongchai Winichakul's terminology).¹¹⁵ Moreover, with the desire to expand central government control over distant frontier regions (a desire probably shared by its northern neighbors), Song China negotiated fixed linear border lines to demarcate unambiguously Song, Liao, and Xi Xia territory. Simultaneously, by reducing the extent of the border—from a wide border zone to a precisely-defined line—the burgeoning Northern Song administrative apparatus could expand its tax base into liminal border regions, sometimes at the expense of border populations.

What is striking is that modern and pre-modern Chinese regimes responded to similar geo-political circumstances in nearly identical ways. Peter Perdue has emphasized that the Sino-Russian treaties of 1689 and 1727 served to establish a precisely demarcated border line (marked by border stones) and to enforce a bilateral policy of repatriating cross-border fugitives. These terms of the treaties effectively eliminated the power of Mongol tribal chiefs occupying the zone in the middle, who until then had succeeded in preserving their autonomy through a policy of shifting alliances (between China and Russia).¹¹⁶ In this sense, precise border demarcation was a technology used by the Qing and Russian states (as well as by Northern Song, Liao, and Xi Xia centuries earlier) to expand their power over "borderland elites" (to use the terminology of Michiel Baud and Willem van Schendel).¹¹⁷ Similarly, Peter Sahlins' description of the "territorialization of the fiscal frontier" in France bears a striking resemblance to Northern Song efforts to expand its tax base into the border zones.¹¹⁸ The parallels between Northern Song approaches to the frontier and much later approaches demonstrate that linear borders and conceptions of bounded, territorial sovereignty do not neatly fit into teleological models of modernity. Rather, alternate conceptualizations of borders and frontiers were determined by the practical needs arising from particular political and military circumstances.