54. For the Manchu banners, see Mark C. Elliott, The Manchu Way; The Eight Banners and Ethnic Identity in Late Imperial China (Stanford, CA: Stanford University Press, 2001).
55. For the descent of the Han (Poliuhan) family, see Chen Lianqing 陈连庆, Zhongguo gu dai shao shu min zu xing shi yan jiu: Jin Han Wei Jin Nan bei chao shao shu min zu xing shi yan jiu 中国古代少数民族姓氏硏究: 秦汉魏晋南北朝少数民族姓氏硏究 (Changchun: Jilin wen shi chu ban she, 1993), 43–44. This is just one of a number of transcriptions of the same Xiongnu name. See Yao, Bei chao Hu xing kao, 136–38.
56. WS 51.1127. It needs noted here that Poliuhan Han “bore the pennon for the central army” 为中军持幢, from which we can infer that “pennons” did not just mark 100-man companies, but larger military units as well.
57. Yan, Bei Wei qian qi zheng zhi zhi du, 154. See Zhang (Jin wei wu guan, 2: 680–81) on apparent earliest dates for establishment of the Huben. Though we don’t have a systematic description of the organization of the guards, Zhang has labored to recover what he can from anecdotes and snippets taken from various sources, and general statements can be given about their interrelations, such as that the Yulin were more prestigious than the Huben.
58. Zhang, Jin wei wu guan, 2: 697–98; on the fact that this was a separate, distinct group, not under the Yulin or earlier units, see 2: 702.
59. See, for instance, the story of “Big Thousand” Lai: WS 30.725.
60. Zhang, Jin wei wu guan, 2: 696, 702; Yan, Bei Wei qian qi zheng zhi zhi du, 56. And see NQS 57.985, where the functions of the palace steward position are described. On the number of stewards, see Zhang, Jinwei wuguan, 2: 698.
61. Tang, Wei Jin Nan bei chao Sui Tang shi san lun, 193; Yan, Bei Wei qian qi zheng zhi zhi du, 154.
62. Such practices are found in other societies as well. See the comments by Armin Hohlweg on the appearance and disappearance of such units within the late Byzantine state (Beitr·ge zur Verwaltungsgeschichte des Ostr·mischen Reiches unter den Komnenen, 61), quoted in Marc C. Bartius, The Late Byzantine Army: Arms and Society, 1204–1453 (Philadelphia: University of Pennsylvania Press, 1992), 272: a new ruler might doubt the loyalty of groups; there might be difficulties recruiting guards for certain units (particularly those based on ethnicity); an emperor might wish to create his own personal guard division.
63. Zhang, Jin wei wu guan, 2: 692, points out that a series of titles in Chinese translation (in this case, not transcription)—nei xing ling 内行领, nei xing zhang 内行长, nei xing zhang zhe 内行长者, nei xing a gan 内行阿干—all apparently translate/transcribe the same original guo yu term.
64. See Zhang Qingjie 张庆捷 and Li Biao 李彪, “Shanxi Lingqiu Bei Wei Wenchengdi ‘Nan xun bei’” 山西灵丘北魏文成帝《南巡碑》 WW (1997.12): 72; and Zhang, Jin wei wu guan, 2: 713–45.
65. NQS 57.985; Zhang Qingjie 张庆捷 and Guo Chunmei 郭春梅, “Bei Wei Wenchengdi ‘Nan xun bei’ suo jian Tuoba zhi guan chu tan” 北魏文成帝《南巡碑》所见拓跋职官初探, Zhongguo shi yan jiu (1999.2): 59–61. The reconstruction of huluozhen is by Shimunek, Languages of Ancient Southern Mongolia and North China, 151–52. Zhang Jinlong (Jin wei wu guan, 2: 717–20) has suggested that this term may refer to the Yulin or Huben; Boodberg (“Language of the T’o-pa Wei,” 227), on the other hand, has suggested that this was a reference to the “officer who girdled (the ruler·) with weapons.”
66. Zhang, Jin wei wu guan, 2: 716. The name lists on the stele are reproduced in many sources: see, e.g., Matsushita, Hokugi Kozoku taiseiron, 75–86. Di Nicola, “Aristocratic Elites,” 27, points out that the highest-ranking Xiongnu aristocrats had dual titles: “one was a title linked to a specific political and government post and the other was a generic military title.”
67. Zhang, Jin wei wu guan, 2: 682–83.
68. Thomas Allsen, The Royal Hunt in Eurasian History (Philadelphia: University of Pennsylvania Press, 2006), 129. I will point out that Allsen is drawing here on Mencius. More blunt about the linkage between hunting and war would be the statement made by Ernest Hemingway, in his article “On the Blue Water” (Esquire, April 1936): “Certainly there is no hunting like the hunting of man and those who have hunted armed men long enough and liked it, never really care for anything else thereafter.” One way to get a taste of this state of mind would be examination of the scenes of hunting (and war) on the carved panels of Ashurbanipal (r. 669–ca. 631 bce) that are kept in the British Museum, where showing “an image of vigor and authority,” “The King Pursues a Herd of Wild Asses with a Pack of Hounds,” or “Ashurbanipal, on Horseback, Kills a Lion with a Spear”: Gareth Brereton, ed., I Am Ashurbanipal, King of the World, King of Assyria (New York: Thames and Hudson, 2018), 72–73 figure 78, 71 figure 77. Such themes are pervasive—“The royal lion hunt,” the text tells us, “played a significant role in the artistic scheme of the carved reliefs of the North Palace.” Somewhat reminiscent of the written account of how in 391 Shegui made the north loop of the Yellow River “run red with blood,” on pp. 268–69, figure 279, of I Am Ashurbanipal, “The River Ulai Is Filled with Corpses, Horse Carcasses and Broken Chariots.” The organization and large scale of the Manchu hunt, in many ways similar to that of the Taghbach, is described in Mark Elliott and Ning Chia, “The Qing Hunt at Mulan,” in New Qing Imperial History, 66–83, esp. 77, where we are told that the hunt, as an “emblem of the warrior origins of the horse-riding, tiger-hunting Manchu nation,” had become “a sacred Qing institution,” for which, in both Chinese and Manchu, the Jiaqing emperor (r. 1796–1820) wrote a valedictory in 1807.
69. Zhang, Jin wei wu guan, 2: 401–6. Hunting would certainly fit in with “war and other types of conflict,” which according to Walter Pohl were the ways in which in early medieval Europe “a community acquired the capacity to act as a collective”: see his “Strategies of Identification: A Methodological Profile,” in Strategies of Identification: Ethnicity and Religion in Early Medieval Europe, ed. Pohl et al. (Turnhout: Brepols, 2013), 45.
70. ZZTJ 115.3624.
71. See Zhang, Jin wei wu guan, 2: 682–83, for examples; and also Liu Meiyun 刘美云 and Wei Haiqing 魏海清, “Shou lie xi su dui Bei Wei qian qi zheng quan de ying xiang” 狩猎习俗对北魏前期政权的影响, in Bei chao shi yan jiu, ed. Yin Xian (Beijing: Shang wu yin shu guan, 2005), 423–27.
72. WS 29.705. And see Zhang’s analysis of these events: Jin wei wu guan, 2: 682–83.
73. WS 113.2972.
74. BS 85.2844 (WS 87.1891). After his death, the person who seems to have killed the wise lord—Empress Dowager Wenming—bestowed upon the guardsman’s family 200 bolts of silk.
75. WS 30.725.
76. See Anthony D. Smith, “War and Ethnicity: The Role of Warfare in the Formation, Self Images and Cohesion of Ethnic Communities,” Ethnic and Racial Studies 4.4 (1981): 375–97. An excellent description of these ties is given in Junger’s Tribe, beginning with his definition of “tribe” as “the people you feel compelled to share your last bite of food with” (xvii); but it is perhaps summed up even more succinctly in the famous speech placed by Shakespeare in the mouth of Henry V at Agincourt: “We few, we happy few, we band of brothers; For he to-day that sheds his blood with me / Shall be my brother.”
77. Loosely based on the translation given by Kwa and Idema, Mulan, 3; though otherwise fine, “marched” has been removed from their translation, since the poem itself makes clear they were on horseback. This incorporation of the concept of xiao in the course of the evolution of the Mulan story within the Chinese world is an interesting example of remaking in the course of cultural borrowing.
78. For the role of members of the imperial house and cadet branches in the guard units, as both leaders and soldiery, see Liu, “Lun Bei Wei qian qi zong shi zai jin jun zhong de di wei ji zuo yong.”
79. WS 113.2974, 2.32; Zhang, Jin wei wu guan, 2: 688; Gao, Wei Jin Nan bei chao bing zhi yan jiu, 304. These groups are part of the 100,000 transported by Daowu.
80. WS 113.2983, 2985, 2988. See discussion of these recruitment issues in Zhang, Jin wei wu guan, 2: 681; and on p. 702, Zhang also cites an example of a palace steward recruiting High Carts (in this case referred to as Chile; BS 17.639 [WS 19A.450]). This expansion of recruitment, a very important issue, was, of course, true of the larger Wei armies as well, in which over the generations we see increasing numbers of Chinese recruited into the ranks. See Tang, Wei Jin Nan bei chao Sui Tang shi san lun, 190ff.
81. WS 30.730. Many other examples can be found of stewards leading their guard units out to war under the emperor’s command: Zhang, Jin wei wu guan, 2: 703.
82. WS 35.808.
83. He, “Fu bing zhi qian,” 327–30. The first major set of garrisons in Chinese territory were the “eight military headquarters” established by Daowu after the defeat of Later Yan: WS 58.1287.
84. On these see: Tang Zhangru 唐长孺, “Bei Wei nan jing zhu zhou de cheng min” 北魏南境诸州的城民, in his Shan ju cun gao (Beijing: Zhonghua shu ju, 1989), 96–109; and He Dezhang 何德章, “Bei Wei cheng ren yu nong geng” 北魏城人与农耕, in his Wei Jin Nan bei chao shi cong gao (Beijing: Shang wu yin shu guan, 2010), 346–54.
85. BS 16.617 (WS 18.429–30).
86. E.g., WS 54.1202.
87. For the garrison system in general, see Yan Gengwang 严耕望, Zhongguo di fang xing zheng zhi du shi: Wei Jin Nan bei chao di fang xing zheng zhi du 中国地方行政制度史: 魏晋南北朝地方行政制度, 2 vols. (Shanghai: Shanghai gu ji chu ban she, 2007), 2: Chapter 11; and Pearce, “Land of Dai.”
88. Tang, Wei Jin Nan bei chao Sui Tang shi san lun, 194, stresses that the decline was relative, insisting that their station was still much better than the indentured military labor of the Jiankang regimes.
89. WS 38.688. He Ziquan describes preponderance of cavalry in “Fu bing zhi qian,” 318–20; and of occasional early use of Chinese infantry units on pp. 332–33.
90. WS 110.2857, describing development of corvée service; and He, “Fu bing zhi qian,” 334–35; Tang, Wei Jin Nan bei chao Sui Tang shi san lun, 198. Sagawa, “Bei Wei de bing zhi yu she hui,” 49, suggests it was first seriously implanted in 473 under Xiaowen’s father, Xianwen.
91. WS 73.1638.
92. Müller, “Horses of the Xianbei,” 185.
93. WS 9.237.
12
狼主
尽管前一章所述的军事力量在明元帝统治时期持续增强，但这一时期整体上处于主动征战的间歇期。第二位魏国君主将于423年去世，时年仅31岁。1若其死因存在非自然因素，很可能与明元帝自身行为相关；与其父王类似，他长期服用寒食散。而现代学者张金龙提出更直接的病因可能源自其在洛阳地区军事考察期间感染的疾病——该区域某些地点爆发的疫病曾对拓跋军队造成重大伤亡。2若此说成立，他便成为又一位被地方疾病击倒的殖民者。
太武帝（426-452年在位）是北魏第三位皇帝，自始祖拓跋力微以降第十七任拓跋可汗。3其父（当时已显衰态）早在前年便妥善安排继承事宜：先封其子为太平王，继而立为副帝（vice-emperor），赋予摄政权力，实质上建立二元统治以加速权力交接。4该计划取得相当成功，明元帝至太武帝的权力过渡成为北魏早期唯一未引发宫廷内斗的案例。太武帝生母出自独孤氏旁支，已于420年（即确立继承人前两年）去世，现有证据无法证明其死亡存在非自然因素。5这位继承人自出生起便未由生母抚养，而是由乳母照料——该乳母后来成为北魏朝廷首位重要女性政治人物，最终被擢升为皇太后，此举至少在华夏历史上尚无先例。6
作为"武皇帝"（或称狼主，B·ri Beg），太武帝重启大规模军事行动，最终重新统一汉帝国北方疆域——这一成就距前秦苻坚短暂统一北方仅隔半个世纪。其首次重大战役针对拓跋部的宿敌匈奴。刘卫辰之子赫连勃勃重夺河套地区建立夏政权后，于418年从东晋军队手中夺取长安并自立为帝。7 425年，太武帝听闻赫连勃勃死讯及夏国后续权力斗争，亲率两万轻骑突袭并焚毁夏国要塞都城统万（位于今陕西榆林西南约65英里处，沿无定河而建）。8与此同时，另一支拓跋军队攻占渭河流域的长安。
427年初，赫连勃勃的继承人赫连昌在重新占据统万城后，派遣军队南下试图夺回长安，导致北方防务空虚。太武帝此时率领规模大得多的军队——约十万人——缓慢南下进入河套地区。9初夏时节，他们在统万城以北的某条河流处停驻，皇帝在此向腾格里（Tengri）天神献祭，向祖先禀告作战计划，并与将士共同立誓；这段记载或许比编年史家通常提供的记录更完整地展现了早期魏国君主如何为作战整备军队。一周后他们攻陷该城——这座城池对于仍保持游牧性的赫连氏统治者而言，似乎主要被用作储存积累财富（此处更准确说是重新积累，考虑到该城两年前刚遭洗劫）。10赫连昌的丰厚财物被太武帝缴获，并依照惯例"按军阶分赐将士"。11缴获物中包括三十万匹马；其中至少大部分被留在河套地区，该地自此成为拓跋部的巨型皇家牧场（如同两个世纪后之于唐朝的情形）。12然而在关中地区，赫连氏仍持续抵抗塞北势力的入侵，导致驻守当地的魏军将领最终放弃长安。13出于根除匈奴宿敌的决心，以及对部将失利的震怒，太武帝下令处决了该将领。
但在彻底消灭夏国前，太武帝选择先打击平城后方迅速形成的重大威胁——北方草原地带已完全崛起为支配势力的柔然。429年，再次以祭祀腾格里为开端，太武帝率军北越大漠击溃柔然，将其可汗驱逐至西部；该可汗在势力大衰后患病身亡。14柔然部众的畜群此时无人照管散落草原，旧日附庸部落开始归降。最终约有五十万营帐群落（落）臣服，而拓跋部则获得超过一百万匹战马。15事实上，由于如此大量牲畜突然涌入代郡市场，导致牲畜及毡革价格急剧下跌。16至于俘虏方面，这些被剥夺财富的人群被安置在阴山以北从承德地区延伸至五原的广阔草原带。当局在这些群体中设置军镇，以维持控制。这些据点同时也承担边防功能——因为柔然不久后就将重组势力。这正是前一章提及的北方六镇之起源，它们将在北魏末期扮演极其重要的角色。此举标志着的与其说是实力彰显，不如说是防御态势的开端，因为草原控制权正逐渐从拓跋可汗手中流失。或许为部分表达对此事的挫败感，太武帝——显然通晓汉语双关（可能还掌握其他数种语言）——此时开始改动其草原对手名称的汉字转写。当秃发氏后裔已采用汉字"柔然"记录其部族名时，太武帝将其改为"蠕蠕"，即"蠕动的虫豸"，以此作为蓄意贬低。17
暂且将"蠕动的虫豸"搁置一旁，太武帝此时将注意力转回南方持续恶化的局势——赫连氏继承者为实现匈奴对西北地区的控制，与建康（现为刘裕宋政权都城）达成战术同盟，向建康君主提议共同摧毁平城并瓜分其领土：宋政权取得恒山以东所有土地，匈奴则获得以西全部地域。18此次图谋未获成功，430年秋，太武帝再次率军南渡河套展开胜利战役，至431年已稳固控制大致相当于今陕西省全境的区域。夏国就此灭亡，拓跋部此时占据长安——这座数百年来作为华夏世界重要政治文化中心的城市，尽管其地位已不复往昔，但此前在后秦统治下曾是中亚僧人鸠摩罗什（344–413）主持佛经汉译重大工程的所在地。19随后首次大规模迁徙西北人口至平城的行动展开，部分移民将在北魏政府担任要职，并成为佛教信仰的传播者。
次年（432年），太武帝挥师东进满洲终结另一个割据政权燕国。此时已非慕容氏统治的这个政权——通称"北"燕——以极度衰微且形态剧变的状态存续，现由名为冯弘（430–436年在位）者掌控。冯氏家族虽自称源自晋帝国，但至此已世代效忠慕容氏，并显示出深厚的鲜卑文化烙印。或许最合理的推测是——如同这个时代的许多人——他们在血统与文化层面皆属混融状态。20在432至436年持续不断的系列进攻中，魏军持续打击这个东北政权，夺取领土并将人口迁返平城地区。早期叛降者包括冯弘之子冯朗，他于433年归顺。21（此事虽属微小插曲，但正如后文将述，当冯朗之女在魏廷掌权时，此事将产生重大影响。）最终攻势始于436年。当魏军逼近冯弘都城和龙（今朝阳）时，燕国东邻高句丽派军救援。22冯弘此时随援军东撤，临行焚毁都城。又一位拓跋对手就此消失。
太武帝此时再次将目光转向西方，聚焦于北凉（397–439年）——这是过去一个世纪间在河西走廊（连接塔里木盆地与渭河流域的关键沙漠-山脉通道）崛起的多个政权之一。尽管这些国家疆域狭小，但战略地位处于丝绸之路要冲。北凉由一族群统治，其名称在汉文转写中作"沮渠"，属"卢水胡"支系——可能源自移居河西走廊后与当地各族混融的匈奴后裔；该族群亦被称为"杂胡"。23至5世纪初叶，北凉势力已扩展至敦煌，定都姑臧（今甘肃武威）。
433年，太武帝授予北凉新主沮渠牧犍封号，双方实行和亲。24但太武帝此时已怀吞并之志，在和亲时私下评论道："不需多时朕必克凉"。25自435年起，他通过越过北凉与塔里木盆地绿洲城邦建交的方式施压沮渠君主。26部分北魏使团甚至远行更甚——数年后约451年，"大魏"使者在巴基斯坦吉尔吉特-巴尔蒂斯坦（Gilgit-Baltistan）地区某块岩石刻下铭文。27太武帝显然怀有超越北凉的雄心，439年当沮渠朝廷试图毒害北魏公主并与柔然结盟侧击魏军的传闻出现时，这种雄心转化为了行动。28
众多太武帝谋臣反对征讨沮渠氏，认为河西走廊是片干旱区域，缺乏维持大规模骑兵所需的牧草与水源。29然而皇帝的立场获得崔浩支持，他引用《汉书》记载该地区水草丰美。30对太武帝而言，某位国人侍卫的谏言或许具有同等或更大分量——该侍卫在《魏书》转写中的称谓被简化为"伊馛"。无论其真名为何，这位受皇帝敬重的智谋之士在群臣退朝后独留进言："若凉州（甘肃）无水草，彼辈何以立国？"31遂于439年夏，太武帝在西郊整军（可能进行过未载于史册的腾格里祭祀），而后率部自平城出征。除用以说服将士师出正义的投毒指控与通柔然罪状外，皇帝还提出个有趣议题——要求自由贸易：声称北凉通过过境税破坏了国际商贸。32
此次战役大获全胜。凉主闭守姑臧城，期待柔然盟军救援。33援军未至，约两月后城破，沮渠牧犍投降。牧犍诸弟西逃至高昌（今新疆吐鲁番东南部），此后二十年掌控着丝绸之路这一关键枢纽。34反方向强制迁徙的数千人中包括众多僧侣，被押送北赴平城。目睹姑臧丰美草原支撑起两月围城后，太武帝一方面嘉许崔浩的远见卓识，另一方面勒令朝议中主要反对者自尽。至于《魏书》所载侍卫"伊馛"，太武帝"爱其才略，常特加优礼。厚加赏赐"。35
平定北凉标志着十六国时代的终结，北魏至此已控制汉帝国北方疆域之大部分。掌握河西走廊后，太武帝派兵深入塔里木盆地东部，445年攻占鄯善，448年夺取焉耆（今新疆巴音郭楞）。36然而实际直接控制仅维持数年。沮渠残部仍在东北方数百英里外的高昌负隅顽抗，其后方更有柔然势力。太武帝统治末期魏军退守敦煌；至460年柔然已重掌塔里木盆地霸权。37尽管如此，交流活动依然频繁。6世纪初，北魏佛教徒在于阗地区发挥显著作用。38毋庸赘言，北魏与绿洲城邦的商贸及外交馈赠持续繁荣。约50年后的一份记载称："自景明（500年始）以来，承升平之业，四疆清晏，远迩来同。于是蕃贡继路，商贾交入。倍于常课"。39
大量财富涌入北魏国库。但新征服地区的控制并不均衡。高车、汉人及其他各族频繁爆发叛乱。最顽固的抵抗区域是自山西汾河向西跨越黄河延伸至陕北泾河的广袤山地。40这些地区聚居着羌、氐、汉人及卢水胡（如我们所知沮渠氏即其分支）等众多族群。445年，该地爆发由汉文史料称为盖吴的人领导的起义，此人来自杏城（位于长安以北约100英里处，现G65高速公路沿线）。41作为当地战略要地，杏城往昔屡为叛乱策源地或军事征伐目标。
盖吴起义在姑臧陷落六年后爆发，其动机或与沮渠氏存在关联——盖吴本人即属卢水胡。其他可能因素（很可能存在重叠）包括对新确立的北魏商贸管制的不满，或对该地区近期其他起义遭残酷镇压的愤懑。42无论如何，盖吴起义规模远超此前案例并取得初期胜利。在盖氏击败魏军后，西北其他地区爆发多起响应性起义。乘胜之势，盖吴此时攻占长安，自立为王，并与建康建立联系。43
次年（446年）二月，皇帝亲率大军沿汾河南下进入渭河流域，迫使盖吴逃回山地。太武帝此时巡幸旧汉都城，并造访著名昆明池——该水利工程由五百余年前的汉武帝修建；一个世纪后，西魏继承政权的摄政者宇文泰亦将临幸此地。重掌渭河流域控制权后，他缓慢率军北返，携长安熟练工匠两千户同行。虽未被拓跋军队擒获，但赦免承诺最终促使族人诛杀叛首盖吴。44
* * *
此时占据建康皇位的是刘义隆，宋开国皇帝刘裕之子。值得注意的是，其统治时期（424年至453年）与北魏太武帝在位时期（423年至452年）几乎完全重合。两人皆是强势君主，刘义隆通过精心管理与培植行政机构来维持统治。而两人最终都在与儿子的权力斗争中身亡。
450年初，太武帝亲率十万步骑兵南下洛阳，随后带领军队在南方无人区进行狩猎——此举既为展示军威，亦作为补给手段。46魏军继续南进，接连攻陷多座城镇，最终在汝南（今河南同名县城所在地）陷入围城战。尽管可见拓跋军队攻城技术有所提升——使用移动箭塔向城内倾泻箭雨，攻城器械上的巨型抓钩摧毁了部分城墙——该城仍未攻破。在承受重大伤亡后，约一个半月后太武帝率军北返，并向刘义隆递送汉文书信，指责宋帝先前试图收买叛将盖吴建立同盟："若你真是大丈夫，何不亲自来取（这些领土），却要用财物引诱我边境子民？"47
