Escape From Rome

by Walter Scheidel

  35. Sedentism was a precondition for large-scale farming, mining, and manufacture, and agrarian states arose to protect and acquire such resources. Quote: Khodarkovsky 2002: 5, and cf. 7–75 on “why peace was impossible” on the (Russian) steppe frontier.

  36. Lattimore 1988 [1940] on agriculturalists’ interaction with steppe peoples and the impossibility of the latter’s integration. Theory: Turchin 2003: 50–93, esp. 53–55, 79–81. For an empirical test for post-Roman Europe, see Turchin 2003: 79–93, esp. 83–89.

  37. Lieberman 2009: xxi–xxii, 85, 92–93, 97, 100–101n144. Quote: 97.

  38. For the Pannonian plain, see chapter 6 in this volume and in the present chapter.

  39. See Barfield 1989 and more briefly 2001. On the distinction between primary and shadow empires, see 2001: 28–39 (quote: 33). Correlation of complexity: Barfield 2003: 461 (“In general nomadic polities tend to match the degree of political complexity of the sedentary societies with which they interact”); Turchin 2009: 194. Challand 2004 gives a broad overview of steppe empires, and see more generally Golden 1998. Sinor 1990 covers Central Asian history prior to the Genghisid expansion.

  40. Beckwith and Nicola di Cosmo are among the most vocal critics of the “shadow empire” model. Both sides seeking resources from the other: Di Cosmo 2002a: 169–70. Encroachment: Di Cosmo 2002a: 142–43, 174–75, 186–88. Both note that the unwillingness of agrarian societies to grant steppe peoples access to resources led to violence: for nomad raids as a response to encroachment, aggression, and trade restrictions, see Di Cosmo 1999b: 12; Beckwith 2009: 330, 334–38, 345–48. Beckwith 2009: 320–22, in an impassioned diatribe against the denigration of nomads in ancient and modern sources, accepts that pastoralists were familiar with horses and use of the compound bow (322), greatly relied on trade (325, 345), resorted to shock tactics because they had to be careful to preserve manpower and could not easily seize fortified cities (339–40), and that both sides sought to expand at the expense of the other (350–51). None of this contradicts Barfield’s reasoning: I fully agree with T. Hall 2010: 109–10, who notes the “broad overlap” of Barfield’s and Beckwith’s positions, notwithstanding the latter’s “very puzzling” attacks on the former. Shared responsibility for conflict along the steppe frontier is a given: see, e.g., Jagchid and Symons 1989: 23 for lack of trust and stability at the frontier, and 24–51 for the options of raiding and trading; likewise Golden 2002: 106.

  41. Correlation: Bai and Kung 2011; Zhang et al. 2015. Cf. the earlier prescient piece by Hinsch 1988, written before much of the data became available. Turks: Bulliet 2009: 96–126; Peacock 2010: 45; cf. also Peacock 2010: 53–61 for context. Better resources: Büntgen et al. 2016: 235 (Arabs); and chapter 6 in this volume (Mongols).

  42. Denunciation by literati: Beckwith 2009: 355–62, who rightly objects to the use of the term “barbarian.” Farming at steppe rims (Inner Mongolia, Northern Mongolia, Southern Siberia, Manchuria, Xinjiang): Di Cosmo 1994; Beckwith 2009: 341–43. See also Di Cosmo 1999b: 8–13 on the formation of steppe empires.

  43. Khodarkovsky 2002: 76–183; Sunderland 2004 (steppe colonization by Russians); Perdue 2005: 133–299 (Qing suppression of nomads).

  44. For the emergence of Eurasian pastoralism, see Di Cosmo 2002a: 13–43. Abstract model: Turchin 2009: 196–97. Quote: Deng 2012: 339. The same is true of the notion that steppe invasions were so catastrophic that they acted as a “centripetal force” on China (Ko, Koyama, and Sng 2018: 290), which overestimates their impact: see chapter 10 in this volume.

  45. Lattimore 1988 [1940] 542–51 (“marginal zone,” for China), 238–51, 547 (“reservoir” and “marginal zone”), 541 (conquests from that zone). Critical importance of borderlands: see, e.g., Barfield 1989: 100, 104, 2001: 22; Skaff 2012: 15–17. With respect to China, Mair 2005: 81 singles out the arid loess- and sand-covered Ordos plateau as the locus of the most intense interaction: “As the zone of consummate interface between the settled and the steppe, the Ordos is the true homeland of ‘China.’ ” Skaff 2012: 16 rejects Fletcher’s 1986: 41 claim that geography separated pastoralists from farmers (“At the eastern end of the steppe zone, where the lines between nomad and sedentary were most sharply drawn, Mongolia and China confronted one another through much of history as worlds apart”). In reality, the contact zone was an essential meeting ground.

  46. Turchin et al. 2013. For criticism, see R. Thomas 2014, who suggests that seeding of traits in lowland civilizational cores that had already attained high complexity by 1500 BCE—Egypt, Mesopotamia, the Indus Valley, and northern China—might produce identical simulation results. Turchin et al. 2014 fail to address this point, which receives some support from preliminary modeling (James Bennett, personal communications May 28 to June 1, 2017). This underlines the need for concrete case studies of how equine warfare and state formation were related in order to assess the validity of Turchin et al.’s interpretation, which is discussed later in this chapter.

  47. This tendency, as well as a profusion of anthropological jargon de jour, bedevils the attempt by Honeychurch 2015 to offer an alternative to existing reciprocal feedback models.

  48. Adapted from Turchin 2009: 193, table 1. Ming is an outlier in more ways than one: its rise was the only indigenous (Han) reaction to outright conquest from the steppe (by the Mongols) rather than from the border zone. This was the only time steppe forces had intruded more deeply, effectively pushing the steppe frontier into southern China, as it were. This might make us wonder to what extent the Ming regime, which responded to the Mongol presence within China, should count as a genuine exception. Even so, this might be taking it too far: after all, revolts against established dynasties were common regardless of the latter’s ethnic origin.

  49. Mair 2005: 56–64 (quotes: 46, 64). Only modern China broke with that paradigm: its leadership had southern roots (83). For the heavy Turco-Xianbei element of the Tang dynasty (“a Särbo-Chinese regime”), see S. Chen 2012: 4–38 (quote: 36).

  50. Di Cosmo 1999a: 902–8; Mair 2005: 69–75; Turchin 2009: 198.

  51. Pastoralism: Di Cosmo 1999a: 909–14, 924–26; Di Cosmo 2002a: 13–43; Graff 2016: 153. Emergence of mounted warfare: Drews 2004: 31–98. Interstate warfare: M. Lewis 1999; Zhao 2015a: 222–61. See M. Lewis 1990 and Zhao 2015a: 169–221 on the internal restructuring prompted by these conflicts.

  52. Offensives: Deng 2012: 336, table 4.1: from 475 to 221 BCE, Qin launched 102 of 244 major offensives, compared to 35 by Zhao, 26 by Chu, and 23 by Wei. Very briefly but to the point: Zhao 2015a: 99, esp. n101 on Qin’s origins. See also Teng 2014 for the early weakness of local kinship ties. Conflicts with Rong from the ninth through seventh centuries BCE: Di Cosmo 1999a: 921–24. In the end, the Western Zhou fought and were brought down by Rong and had to abandon the Wei valley.

  53. Di Cosmo 1999a: 947–51 (loss of buffer). Warring States–era contacts: Di Cosmo 1999a: 951–52, 960–62; Di Cosmo 2002a: 127–58, esp. 140–43 on attacks and border walls; also Deng 2012: 339; Graff 2016: 154–55. Xiongnu expansion in response to Qin aggression: Di Cosmo 2002a: 174–75, 186–88. The question of whether the Xiongnu empire predated this particular conflict (thus Di Cosmo 2002a: 167–74) remains controversial: e.g., Barfield 2003: 462–63.

  54. See Chang 2007 for an account of the enormous cost incurred by these campaigns, even if he takes the sources too much at face value. For the attendant domestic reforms, see Loewe 1986: 152–79, and generally Lelièvre 2001. Xiongnu federation: Barfield 1989: 32–84. Di Cosmo 2002a: 205 acknowledges that steppe demand for Han resources became critical once the Xiongnu empire had been established.

  55. Han Chinese ran Former and Western Liang, Northern Yan, and Former Qin; and the Ba ruled Cheng Han. Military contingents from the steppe had already become prominent under the Jin, which they then overthrew: Holcombe 2001: 121–22.

  56. Barfield 1989: 123–24; Graff 2002: 72; M. Lewis 2009a: 149; Graff 2016: 145. The Di state of Former Qin must already have enjoyed similar access: large cav
alry forces were reported for its failed invasion of the South up to 383 CE (Graff 2002: 67).

  57. Graff 2016: 161.

  58. See esp. W. Liu 2015b, and chapter 10 in this volume.

  59. See chapter 6 in this volume.

  60. Barfield 1989: 266–96; Perdue 2005: 133–299.

  61. Deng 1999: 363–76, in a survey of mass rebellions since the Qin period, identifies 103 cases over 2,106 years. With very few exceptions, they broke out close to the end of dynastic eras: in 208 BCE (fall of the Qin regime), 17 CE, lasting for seven years (end of the intra-Han interlude of the Wang Mang usurpation), 184–191 CE, lasting for up to three decades (decline and fall of the Eastern Han), 294–311 CE, lasting for up to half a century (fall of Western Jin), 611–623 CE (fall of Sui), 863 and 875 (decline of the Tang), 1119–1130 (multiple cases at the end of Northern Song), 1351–1353, lasting for up to 16 years (expulsion of the Mongols), 1641–1646, lasting for up to 2 decades (fall of the Ming), and finally an uptick in the 1780s followed by bigger disruptions in 1795–1796, lasting for up to 12 years, and the Taiping and other rebellions 1851–1861, lasting for up to 15 years, which seriously challenged the late Qing.

  62. For the structural link between unity/disunity in China and the presence of steppe empires, see esp. Barfield 1989: 10–11 and 13, table 1.1, and Barfield 2001: 23, table 1.1; and see 1989: 12–16 on parallel cycles. This elaborates on Lattimore’s observation of the existence, over 2,000 years, of “two cycles, distinct from each other as patterns but always interacting on each other as historical processes—the cycle of tribal dispersion and unification in the steppe and the cycle of dynastic integration and collapse in China” (1988 [1940] 512); see Barfield 1989: 11 for acknowledgment of his indebtedness to “Lattimore’s tradition.” See also Jagchid and Symonds 1989 for an overview of the key features of 2,000 years of interaction between China and the steppe.

  63. Fletcher 1986: 21; Di Cosmo 1999b: 29–37. As discussed, dual administration was already pioneered in the Period of Disunion.

  64. Turchin 2009: 199 notes the underlying feedback loop.

  65. Mote 1999: 26; cf. also Barfield 2001: 15.

  66. See, e.g., Golden 2002: 130–32; Graff 2016: 154. For female involvement, see Golden 2002: 130–31; Mayor 2014: 395–410.

  67. Bai and Kung 2011: 974, table 1, with 975, fig. 3 (broad survey; adopted by Ko, Koyama, and Sng 2018: 300–304); Skaff 2012: 40–41, table 1.2, with a complete list 301–12 (599–755 CE, based on Sima Guang’s Zizhi tongjian).

  68. Border walls have already been discussed. City walls: Ioannides and Zhang 2017: 79–81 (taller, n = 1,436; thicker, n = 934). Unity: Ko, Koyama, and Sng 2018: 300–304.

  69. Gaubatz 1996: 13–44, esp. 19–21 (20 for the term).

  70. Shifting control: Skaff 2012: 17, 26, 50–51, 291–94. It was rare for the border zone to be controlled by one separate power that did not engage in further expansion, as was the case with the Qidan pastoralists that formed the Liao state in the tenth century CE and sought to exercise but failed to maintain hegemony over northern Chinese states that rose and fell in rapid succession. See Standen 2007 on Liao frontier state formation. Cession: Huang 1997: 124; Standen 2009: 86–87.

  71. Perspective: see, most recently, e.g., M. Lewis 2009a: 151; Skaff 2012: 3–4; Rawski 2015: 225.

  72. For the shadowy Rus’ khaganate, see Franklin and Shepard 1996: 31–70, and the recent summary in Neumann and Wigen 2018: 165–72 (and 163–98 on the general context). Large empires: Lithuania rivaled Carolingian Francia and Germany in size, and Rus’ claimed territory twice as large. Russia gradually grew to be much larger still: by the late fifteenth century, it exceeded the size of the two earlier polities, and by the late sixteenth century it was as large as all of Latin Europe (see Taagepera 1997: 494, 497–98).

  73. Franklin and Shepard 1996: 81–83, 97, 113, 143–45; Golden 2002: 115, 119.

  74. Franklin and Shepard 1996: 170 (quote), 170–72 (500 kilometers of ramparts and 100 forts), 326 (power), 369–70 (lack of unity); Golden 2002: 121–23 (Cumans); cf. also Neumann and Wigen 2018: 174–78 for Rus’ enmeshment with steppe peoples and practices.

  75. See chapter 6 in this volume. Hartog 1996: 55–56 (tax), 162–67 (influence); Ostrowski 1998: 36–63 (borrowings), 133–248 (“Tatar yoke” ideology in the fifteenth and sixteenth centuries); Knobloch 2007: 173, 177 (borrowings); Neumann and Wigen 2018: 184–86 (early Tatar support); Lieberman 2009: 113 (size), 212–38 (evolution).

  76. See chapters 5 and 6 in this volume, and very briefly Challand 2004: 53–58.

  77. See chapter 6 in this volume.

  78. See in the next section.

  79. See chapters 5 and 10 in this volume.

  80. Remoteness: see also Lévy 1997: 36–40. Golden 2002: 108 and Graff 2016: 157–59 note Rome’s limited contact with steppe warriors. In addition, imperial allies kept steppe groups at arm’s length. The Bosporan kingdom on the Crimean and Taman peninsulas, formed by Greek colonies in the fifth century BCE in response to steppe tribes but increasingly accommodating them, was sustained as a buffer client state by the Romans, propped up by garrisons and occasional military intervention. In the Sarmatian-populated Hungarian Plain and Wallachia, diplomacy alternated with episodic clashes, but the limited grasslands did not support any larger confederations. Later, the Byzantine empire proved adept at containing steppe powers through diplomacy and subsidies.

  81. Taaffe 1990: 37 (conduits); Asher and Talbot 2006: 12 (pastoralists); Roy 2015: 50 (Maurya).

  82. Roy 2015: 56–64, 70–73, 87–90.

  83. Asher and Talbot 2006: 26–28 (critical role of cavalry); K. Roy 2015: 93.

  84. K. Roy 2015: 95–99, 104–6. Chagatay nomads formed the backbone of Timur’s forces: Jamaluddin 1995: 165–67.

  85. B. Stein 1989: 22; Asher and Talbot 2006: 54–59; K. Roy 2015: 117–19.

  86. Asher and Talbot 2006: 123–28, 152; K. Roy 2015: 120–26. Mughal identity focused on descent from Timur in particular and from Mongols in general: Foltz 1998: 21–27.

  87. Lieberman 2009: 102 (quote), 637, 643. See chapter 1, figure 1.7 in this volume.

  88. K. Roy 2015: 212–16. Mughal imports: Foltz 1998: 63. As noted later, this practice first appeared in the Neo-Assyrian empire.

  89. Cf. Lattimore 1988 [1940] 242–47 for the relevance of the “reservoir”/“marginal” zone model for India’s northwestern tribal frontier. South Asian empires tended to be less durable than those in China for reasons that need not concern us here (see chapter 10 in this volume); but that they existed at all owed much to these direct and indirect steppe inputs.

  90. Bulliet 2009: 101 (rivers); Potts 2014: 47–87, esp. 78–81, 84–85. Drews 2004: 123–38 is essential.

  91. Potts 2014: 88–119, esp. 93, 98, 113–17. Cf. also Challand 2004: 35–36 (italicized in original): “Responding to pressure from the nomads in the north was one of the constant features of Iran’s external policy.”

  92. Potts 2014: 124–56, esp. 144–48, 156; Payne 2016: 7–10; Rezakhani 2017: 87–146.

  93. Iran/Turan: Payne 2016: 5. Historiographical bias: Rezakhani 2017: 4.

  94. Arabs: Potts 2014: 169–72. Turks and Mongols: Barfield 2002: 61. See more specifically Peacock 2010: 72–98 and Basan 2010: 165–69 for the characteristics of the Seljuq military. Rezakhani 2017: 187–91 argues that Sasanian-Central Asian interactions at the eastern frontier laid the foundations for Khorasan’s pivotal role in state formation after 750 CE. Central Asian steppe people proved more accepting of hierarchy than the Arab Bedouin, in no small part thanks to more sustained contact with state-level polities: Barfield 2002: 65–66.

  95. See Kuhrt 1995: 71, 75, and esp. Porter 2012, on the identity of the Amorites. Mitanni: Kuhrt 1995: 283–89, 298.

  96. See Spalinger 2005: 8–9, 12–15, and 32–69 on New Kingdom militarism; and also briefly Kuhrt 199: 189–90.

  97. Kuhrt 1995: 393–401; cf. also 432 for the Israelites.

  98. Barfield 2002: 65; Turchin 2009: 207.<
br />
  99. Territorial evolution: summaries in Lieberman 2003: 23–31; Lieberman 2009: 12–22. Secular shift: Lieberman 2009: 22–30.

  100. Lieberman 2009: 763–894. On horses in Southeast Asia, see Bankoff and Swart 2007, esp. 33, 68 for their size.

  101. I am not aware of academic comparisons between Roman and Andean state formation. For culture change, see DeMarrais 2005. For a later New World steppe empire (that of the Comanche) sustained by Old World horses, see Hämäläinen 2008.

  102. Hoffman 2015: 115 (block quote), 142 (“nomads were China’s enemy, for like most big states, China had expanded into areas where nomads lived”), 172 (“different political geographies”), 206–7 (politics vs. geography). He refers in passing to Turchin’s “steppe effect” model (115n15) but appears to misunderstand its gist, and notes only casually that “China would have been different without the steppe” (115) without elaborating on this crucial point. The statistics in Ko, Koyama, and Sng 2018: 302–3 show that nomadic attacks on China were not “merely … a product of the tendency for the autocratic Chinese state to expand till it bordered the steppe and encountered conflicts with the nomads.”