CULTURES IDENTIFIED WITH the semiarid Northern zone were traditionally deemed semicivilized and disparaged as “barbarian” by the denizens of Imperial China because they were perceived as lagging far behind Hua-Hsia material and intellectual levels. However they might be interpreted, the interminable steppe-sedentary clashes that would plague both realms throughout Chinese history commenced in the Shang, if not earlier. However, vestigial evidence for martial threats having troubled the Northern zone itself exists in defensive works that date back prior to the currently known conflict horizon.

Ditches constituted the primary defensive measures in the Hsing-lung-wa (6200/6000 to 5600/5500 BCE), Hung-shan (3500-3000), and Hsia-chia-tien (2000-1500) cultures.1 However, recent decades have witnessed the discovery of numerous Lungshan villages where the inhabitants chose to erect walls instead of relying on ditches, including a group of twelve sites in central and southern Inner Mongolia characterized by protective walls and dwellings constructed from stone rather than earth.2 Originally occupied between 3000 and 2300 BCE, an interval during which localized ecological constraints prompted the initiation of new settlements by splinter groups, they were abandoned by 1500 BCE because the climate cooled below the point of sustainable agricultural yields.3

The segregated quarters, variation in building size, large sacrificial altars, well-developed pottery, and handful of bronze artifacts discovered in these twelve towns are interpreted as evidence of growing class differentiation and the emergence of localized chiefs. The sites themselves vary in size from a minimal 4,000 square meters to a very substantial 130,000 but are primarily small and must have been inhabited by limited populations of perhaps a thousand.

Although the sites may be geographically divided into three groups, they all appear to have functioned as military citadels because their walls are not only constructed of stone but also display significant defensive characteristics. For example, gate openings are generally protected by screening walls, and the sites with the fewest natural advantages of terrain often employ parallel doubled walls augmented by external moats or ditches. Such extensive, determined measures, undertaken in an era of simple tools, can only be interpreted as evidence of a pervasive fear of aggressors.

These fortified settlements differ in their exploitation of the terrain’s configuration. The five sites in the Pao-t’ou area, despite being situated on comparatively raised ground, are all located on the lower slope of Mt. Ta-ch’ing in the region south of Mt. Yang and thus are marked by north-to-south declinations.4 They are also fully walled except where precipitous cliffs obviate any need for further defensive works and generally comprise aggregates of smaller citadels positioned in close proximity. For example, Wei-chün has three distinct walled enclaves, whereas A-shan, the largest in the group, and Sha-mu-chia each have two. (Spaced some 250 meters apart, the square one at A-shan measures 260 meters by 120 meters for a total 31,200 square meters; the second tapers from 120 meters to 50 meters along a length of 240 meters.)

In contrast, the four towns studied in the Tai-hai area essentially assume the shape of a traditional Chinese bamboo basket, three of the sides being high and the remaining one low, the equivalent of the basket’s opening.5 Rather than multiple bastions and semisubterranean dwellings, they are internally marked by smaller, segregated walled areas and vary in size from 20,000 square meters to Lao-hu’s surprising 130,000 square meters.

Insofar as they average roughly 4,000 square meters each, terrain considerations must have heavily constrained the six small sites clustered along the southern flow of the Yellow River.6 Although they all exploit local heights and take full advantage of the Yellow River’s confluence with nearby valleys and ravines, they also augment their defensive posture with doubled exterior walls, external ditches and moats, and protective galleys on both the interior and exterior of entrances.

Another recently excavated western Liaoning hillside site marked by integrated defensive measures provides further evidence that fear of attack had become a crucial factor in village design and location even in the north during the first half of the second millennium BCE. Dating to the lower Hsia-chia-tien culture (2000 to 1500 BCE or roughly contemporaneous with the Hsia), the 40,000-square-meter village is defined by extensive ditches on three sides and a steep cliff on the remaining or southern border. The ditches on the east and west, both of utilized preexisting ravines, are an astonishing 10 meters deep, and a double stone wall marked by a few crenellations and 17-meter internal spacing runs down the western side. Additional walls isolate a section in the north, and other internal walls provide interior barriers in the remaining portion, completing the formidable bastion.7

Walls constructed from larger stones stabilized with smaller stones and plugged with pebbles wedged into the gaps define these sites.8 Since remnants of tamped earth walls may still be found in this area after thousands of years, and the soil in the vicinity of the Yellow River is particularly conducive to rammed earth construction, the inhabitants must have deliberately opted to employ stone. Insofar as stone walls more readily resist scouring by floodwaters than does pounded earth, the inhabitants could have positioned their settlements along the local rivers, which would have been far more convenient for drawing water. Defense against human threat rather than flood avoidance therefore must have been the paramount factor in selecting somewhat distant, moderate heights for their villages.

Despite the highly conservative interpretations offered by archaeologists, these sites in the semisteppe region of southern Inner Mongolia that date to central China’s Lungshan and Hsia periods thus show a strong martial orientation. The solidity of the stone walls coupled with a pattern of dispersed towns inhabited by single extended clans implies a high, if localized, ongoing threat level. Whether they posed a threat to each other, or the entire culture was reacting to external dangers, is unclear. However, it appears that neither larger aggregates nor great chiefs who could unify them against foreign marauders or organized forces from China’s interior had yet emerged.

Somewhat surprisingly, just as the Shang ascended to power, these settlements were abandoned when the inhabitants apparently adopted a more nomadic form of existence based primarily on pastoral rather than agricultural practices. This suggests it was not just climatic degradation but also military pressure from the southeast that ended the viability of their sedentary lifestyle.9 As a result they apparently evolved into the peoples of the so-called Northern complex, who not only came to dominate the region but may also have acted as a conduit for the chariots and distinctive weapons that suddenly appeared in the thirteenth-century Shang.

One other important northern site has been discovered in Liaoning province.10 Numbering among those that comprised a virtual defensive line along the 42nd degree of latitude, where the state of Yen would construct its border wall in the Warring States period, it dates to the late Shang or very early Chou and typifies settlements that fully exploit riverside locations and natural ravines while being oriented to open vistas. Its incomplete, badly crumbling protective walls outline an elongated, somewhat axe-shaped compound that tapers down at the top or north. Situated between two rivers that sweep protectively from the northwest to the northeast, it is bordered by a 300-meter-long, 20-meter-deep steep ravine off to the northeast.

The compound extends 430 meters on the eastern and western sides, 150 meters across the south, but only 80 meters in the north. The walls have a width of about 3.2 meters and remnant heights that vary between 1.2 and 1.8 meters. They were constructed on a 1-meter-deep foundation by erecting two facing portions from large, indigenous stones and plugging any gaps with soil and small stones before filling the interior with a composite mixture of rocks and earth and leveling the top, a highly localized technique. A 4.5-meter-wide wall about 0.8 meter high, gradual 15-degree slope, and hardened surface protected the interior. However, the steep cliff on the eastern side seems to have been deemed adequately formidable for defensive purposes, obviating any need for further fortifications.


Impressive but comparatively late walled towns have also been found in the semitropical hinterlands of Sichuan, far to the southwest of the Hua-Hsia core. Here, at a crossroads for trade and transport in the upper Yangtze River basin, lies the fertile plains area that eventually evolved into the famous states of Shu and Pa. As might be expected from the fiercely independent clans having retained localized power well into imperial times, this relatively isolated area was early on marked by chieftains capable of organizing and commanding massive work projects. A number of significant sites associated with the Pao-tun culture around Ch’eng-tu or otherwise generally dating to about 2500-1800 BCE, roughly contemporaneous with the late Lungshan and the Hsia’s projected dates, have been excavated in recent decades.11 Although not considered immediate ancestors to the now famous San-hsing-tui culture, the Pao-tun manifestations are deemed precursors.12

Even though tamped earth methods and a finely laddered profile were often employed, the techniques used to enclose these sites tended to lag behind those along the Yellow River. However, the terrain’s characteristics, particularly the nearby rivers, were fully exploited to create substantial defensive fortifications that assumed the usual basic configurations of squares, rectangles, and circles.13 Representative Sichuan sites include Pao-tun itself; Mang-ch’eng at Tu-chiang-yen, where the Ch’in would undertake its famous irrigation project; P’i-hsien, Yü-fu-t’un; Ch’ung-chou Shuang-ho; and Tzu-chu.

These sites are all marked by double concentric walls, whether constructed simultaneously or in different eras as part of a town’s expansion, with Mang-ch’eng and Ch’ung-chou Shuang-ho even being said to be the first double-walled cities in China; the employment of river pebbles on exterior wall faces to improve weathering and retard flood erosion; and deliberately planned but often rapidly executed construction, including at Pao-tun. Brief characterizations of a few sites will provide a sense of the dimensions and variations that existed within this moist, semitropical location.

The now definitive site of Pao-tun, basically a rectangle that runs 1,000 meters from north to south and 600 east to west, was constructed on the edge of a natural terrace on the Ch’eng-tu plains along the Mien River. Occupied from about 2600 to 2300 BCE, it was protected by walls ranging from 29 to 31 meters at the base to 7.3 to 8.8 meters at the top that still display remnant heights as great as 4 meters.

Yü-fu-t’un, located on a plain 1,700 feet above sea level, lies about twenty kilometers southwest of Ch’eng-tu, two kilometers from the Chiang-an River, and seven kilometers from the Min-chiang River. Constructed about 2000 BCE, the now badly damaged fortifications originally protected some 320,000 square meters within an irregularly shaped enclosure closely configured to the terrain, whose walls varied between 400 and 500 meters in extent.14 Most of the perimeter’s 2,110 meters of walls stand between 2 and 3.7 meters high and are characterized by the usual trapezoidal shape as they expand from 19 meters at the top to nearly 29 meters at the base.

The walls were constructed on unimproved ground from discrete layers that generally range from 10 to 35 centimeters, some of the thicker ones actually consisting of numerous unpounded, thin layers. River pebbles were also intermixed to form an external protective layer. The core wall was further inured against deterioration and weathering by two half walls on the interior and exterior. A lengthy depression found to the north may be the remains of a defensive ditch that gradually filled in over time.

Essentially a rectangle of approximately 620 by 490 meters that enclosed 300,000 square meters, the walls at P’i-hsien vary between 8 and 40 meters in width and display remnant heights of 1 to a surprising 5 meters.15 Although generally dating to the Pao-tun culture and employing the same technology, they were apparently built in two stages, the original walls having been thickened by overlaying. (For example, a section that was originally 1.9 meters at the top, 10 meters at the base, and 2.4 meters high was expanded to 7.1 meters at the top, 20 meters at the base, and raised to a height of 3 meters.)16

Yen-mang-ch’eng, Ch’ung-chou Shuang-ho, and Tzu-chu all exploited the defensive advantages of two concentric walls that coincidentally created an isolated killing zone that would entrap anyone who managed to penetrate the outer perimeter. At Yen-mang-ch’eng an intervening ditch further augmented this formidable defense. The inner rectangle, approximately 290 to 300 meters north to south and 240 to 270 meters east to west, consisted of walls between 5 and 20 meters in width that retain a remnant height of 1 to 3 meters. Those of the outer compound, which vary between 7 and 15 meters in width and still protrude 1 to 2.5 meters above the terrain, extended 350 meters from north to south and at least 300 from east to west. Two moats were incorporated for additional protection, and the entire edifice was apparently erected in a single effort, though the inner walls show evidence of subsequent reconstruction.

Although now partially destroyed by the nearby P’ang-hsieh River, Ch’ung-chou Shuang-ho’s rectangular inner citadel was about 450 by 200 meters and originally had walls ranging between 5 and 30 meters wide that still display heights of 2 to 5 meters.17 The badly damaged exterior wall varied in width between 3 and 10 meters but retains a height of 0.5 to 2 meters. Additional protection was provided by an unusual moat interspersed between the walls that varied in width between 12 and 20 meters.

The interior and exterior walls at Tzu-chu Ku-ch’eng were similarly separated by an intervening ditch. The 400-meter interior wall varies between 5 and 25 meters in width and stands 1 to 2 meters high. Those on the exterior, although displaying the same height, are a relatively narrow 3 to 10 meters in width. However, they were constructed by simply mounding the soil.

In recent decades two major cities less than thirty miles apart, dating back to the late Hsia or early Shang dynasty, have been partially excavated in Sichuan: San-hsing-tui at Kuang-han, site of the later state of Shu, and Ch’eng-tu, tentatively identified with Pa.18 Somewhat different in character, they clearly interacted with the Shang but were neither submissive nor externally controlled, even though the Shang may have viewed them as at least nominally integrated into their sphere of influence. However, excavations at San-hsing-tui have produced a number of astonishing finds, primarily large bronze figures, trees, and enormous masks unlike anything previously recovered in the Shang and Chou cultural areas. Shang casting techniques were employed, but the dramatic stylistic and thematic differences marking these creations attest to the inhabitants’ indigenous might, cultural power, and an ability to withstand Shang military and political power.19

San-hsing-tui was probably a theocratic center that developed coincident with the emergence of a new local ruling clan or tribe. Although thought to be a precursor of Shu culture, tremendous controversy revolves around almost every question concerning its origins and nature. The very identity of San-hsing-tui’s progenitors also remains a matter of debate, some analysts claiming that they were descended from the Yellow Emperor’s clan and originally either were coterminal with the Hsia or contributed to Erh-li-t’ou culture, others attributing crucial formative influences to the Hsia and Shang.20 However, for the purposes of military history it is the existence of a flourishing, fortified city providing additional evidence of a powerful alternative culture that is paramount.

Unlike P’an-lung-ch’eng (discussed in the next chapter), San-hsing-tui was not a Shang bastion constructed by deploying military power or dispatching Shang clan members to a border area. Rather, it was a locally constructed, walled city that employed common hang-t’u fortification techniques that may have been acquired from the Shang, whether directly or indirectly. The walls, which retain a remnant height of up to 6 meters in some places, are particularly impressive. Those on the east and west are 1,600 meters and 2,100 meters in length; the north and south walls are both about 1,400 meters; the total core area is estimated at 2,600,000 square meters; and the overall cultural site occupies an expansive 12 square kilometers. Evidence has been found of royal quarters, numerous building foundations ranging from ordinary structures covering some 10 square meters to larger ones roughly 60 square meters in extent, and segregated production and living areas, as well as very extensive ritual or religious activities.

The walls were apparently erected at the same time as the palace complex, confirming their defensive function even though aggressors might have easily scaled the gradually sloping exterior. However, squads of soldiers could also have been stationed on top to repel intrepid attackers, and the massiveness of the walls alone would have had a deterrent value in addition to reinforcing the site against recurrent flooding. The deep trench created by excavating the earth for the walls was converted into a moat that augmented their defenses and continues to provide highly visible evidence of their sophisticated grasp of contemporary fortifications technology.

Few weapons have been found at San-hsing-tui, with those so far recovered being fabricated from jade and thus more symbolic than functional, highly suggestive of a lack of external challenges.21 Despite the massiveness of the walls, the buildings and associated artifacts are said to betray a pervasive spirituality, suggesting that the city was a commercial and ritual center rather than an administrative and military enclave. Based on one commonly employed measure that allots roughly 155 square meters per household, nearly 82,000 people could have dwelled within the walled enclosure and perhaps some 250,000 in the immediate vicinity.22 The population seems to have been highly intermixed not only in terms of economic classes and occupations, including bronze workers, but also in ethnicity, because several tribes apparently migrated in from the surrounding region.

Being located on a major trade intersection, San-hsing-tui was perfectly situated to benefit from diverse external stimuli even as a powerful ruling hierarchy capable of ordering a disparate population evolved. Because most of the artifacts display unique characteristics, the clan that ultimately wrested control may have originated elsewhere and ruled by force of conquest, just as the generalized distribution of the few recovered weapons would tend to suggest. Its period of fluorescence seems to have extended from the dynastic Hsia through the late Shang.23

In contrast to San-hsing-tui, the commercial center of Ch’eng-tu apparently lacked walls despite having once deployed a wooden palisade. However, extensive effort must have been required to construct the twelve bridges required by their location in a basically wet area crisscrossed by rivers. Parts of the site have been radiocarbon dated to roughly 4010 BP or well within the putative Hsia period, but the first significant cultural layer apparently corresponds to the early Shang, suggesting a date closer to 1600 BCE. Excavated pits attest to Ch’eng-tu’s occupation down through the Shang period, just as at San-hsing-tui.

The population in the fifteen-square-kilometer area of Ch’eng-tu has been estimated at a robust 280,000, making it another powerhouse capable of fielding a massive army and therefore likely to have been totally independent of, if not actively opposed to, the Shang.24 However, Ch’eng-tu clearly enjoyed some sort of commercial relationship with the Shang, because Shang bronze vessels have been recovered for which the city lacked production facilities. Moreover, unlike many fortified settlements in the central plains that evolved into political centers before becoming commercial centers, Ch’eng-tu and San-hsing-tui seem to have been economically robust from the outset and only subsequently developed the requisite political and military apparatus.25

Finally, an early Bronze Age site about four kilometers square has been discovered at Chin-sha, some thirty-eight kilometers west of San-hsing-tui. 26 Although the numerous artifacts and divinatory practices indicate strong Shang influence, Chin-sha has been interpreted as the center of another independent, peripheral people sufficiently powerful to challenge the Shang. Just as at Ch’eng-tu, the large site lacks defensive walls. However, the Mo-ti River flowed through its midst, and the city was situated between two rivers that flowed to the north and south that would have significantly impeded aggressors. Evidence for a considerable level of martial strife is seen in a small statue of a kneeling figure who displays a contorted facial expression expressive of fear or horror and has his hands tied behind his back. Archaeologists have tended to interpret this figurine as evidence of class distinctions and internal conflict, but he seems more likely to have been a doomed prisoner of war, especially because sacrifices of this type were already prevalent in the Shang at this time.

If you find an error or have any questions, please email us at admin@erenow.org. Thank you!