It is true that steppe conquerors sometimes received land allotments, the Mongol forces in China and Iran among them. This, however, did not end taxation: tax-farming was reintroduced under the Ilkhans, and while the Timurid system was based on land grants coupled with fiscal immunity to local elite allies and senior Turkic military, this did not relieve subject populations from obligations (Potts 2014: 209–13). What we do not see in these cases is a creeping assimilation of the tax status of these populations to that of their privileged conquerors. One explanation would be the overall degree of difference, in culture and mode of subsistence, between these groups: the gap between mounted steppe warriors and sedentary farmers was larger than that between Germans and locals elsewhere. I discuss the impact of these differences in more detail in chapter 8 of this volume.
31. Halsall 2003: 45–110.
32. Halsall 2003: 119–33 (army size), 134–62 (nature of campaigning).
33. Contra Rosenthal and Wong 2011: 32, the succession system Frankish lack of primogeniture was not the key determinant of this process: even though this custom contributed by encouraging internal divisions among princely brothers, the roots of local empowerment were deeper than that.
34. Nelson 1995b: 385–86 (quote: 386, from Nithard, Histories, 4.3); Costambeys, Innes, and MacLean 2011: 319; Wickham 2016: 11.
35. Wickham 2009: 529–51 on the “caging of the peasantry”; Halsall 2003: 31 on armies as assemblies.
36. Airlie 1995: 448–49; Roger Collins 1999: 361–62, 395–96; Wickham 2001: 91–94.
37. Wickham 2001: 88–89; Wickham 2005: 154–258; Wickham 2009: 508–28; Wickham 2016: 102–9.
38. Strayer 1970: 15 (quote); Wickham 2009: 555–56, 564 (quote). See also Wickham 2016: 78–79, 254–55 for the time around 1000 or the eleventh century as a crucial tipping point.
39. Graff 2002: 39–51; M. Lewis 2009a: 51, 59–62. Inflow of “barbarians” under the Jin: Holcombe 2001: 121–22.
40. Graff 2002: 54–69; M. Lewis 2009a: 73–85.
41. Graff 2002: 59–60; M. Lewis 2009a: 77.
42. Graff 2002: 61; M. Lewis 2009a: 78–79, 114, 118.
43. R. Huang 1997: 76–77 (quote: 77).
44. See Scheidel 2011b: 197, based on Bielenstein 1987: 12 (Han: 9.2–10.8 million households with 47.6–56.5 million residents), 15–17 (third century CE). For strong evidence in support of the notion that state (registration) capacity was the critical variable, see Scheidel 2011b: 203n18, and cf. also Graff 2002: 35–36, 127.
45. Bielenstein 1987: 17 (census), with Scheidel 2011b: 197–99; Graff 2002: 67 (offensive). Europe: McEvedy and Jones 1978: 41–118.
46. Dien 2001: 108–13 (Western Liang). Some exam questions (from fifth- to sixth-century northern China) survive in literary texts (105–7); the examination system continued in southern China (103–4). M. Lewis 2009a: 145–48 (Xiongnu and Murong Xianbei).
47. Graff 2002: 69–73, 97; M. Lewis 2009a: 79–81, 148. The Tuoba were probably a Turkic-speaking group: S. Chen 2012: 183–91.
48. Eberhard 1949: 209–11; Yang 1961: 146–47; Huang 1997: 89. Tuoba military strength attracted Chinese collaborators to their administration: Barfield 1989: 118–19. Von Glahn 2016: 175 suspects that the reclamation of deserted land reduced the state’s pressure on these clan communities.
49. Graff 2002: 72; M. Lewis 2009a: 149.
50. Huang 1997: 90–91; M. Lewis 2009a: 139–40; von Glahn 2016: 173–74. The census total of 3.4 million households reported for 528–530 CE (Bielenstein 1987: 17) cannot represent more than a fraction of the actual number; Graff 2002: 127 and 13n19 provides a reference to a more respectable tally of 5 million households and 32 million people at about that time.
51. Kang 1983: 116–75; Huang 1997: 93; Graff 2002: 98–100; M. Lewis 2009a: 81.
52. Kang 1983: 176–216; Graff 2002: 100–106; M. Lewis 2009a: 82–83.
53. Pearce 1987 is fundamental: see esp. 463–74, 480–83, 507–18, 561–66, 734. See also Kang 1983: 222; Graff 2002: 107–10. 114; M. Lewis 2009a: 84–85; von Glahn 2016: 180–81. Army size: Graff 2002: 108–9.
54. Kang 1983: 220; Graff 2002: 107, 115.
55. Graff 2002: 111–13; M. Lewis 2009a: 62–73, 117.
56. Graff 2002: 76–96, esp. 82–83, and 127; M. Lewis 2009a: 138.
57. In the late Eastern Han period, about one-third of the registered population had resided south of the Central Plain region (demarcated by the Huai River) (Graff 2002: 76, 93n1), and about 10 million people south of the Yangzi (de Crespigny 2004: table 2). These shares are likely to have risen between the second and fifth–sixth centuries CE as southward migration occurred, the North suffered devastation, and the South became more developed. In 609 CE, the unified Sui empire produced a census tally of 8.9 million households with 46 million people (Bielenstein 1987: 20, who prefers 9.1 million households), but even that count appears to have missed households in the south, as suggested by comparison with the geographical distribution of the 140 CE census; the actual total would thus have been higher still. Consequently, the southern tallies for 464 (equivalent to one-tenth of the 609 total of households, rather than the expected value of more than one-third) and 589 (equivalent to one-eighteenth of the 609 households instead of more than one-sixth and one-twenty-third of people instead of more than one-fifth) are inevitably far too low. Xiong 2006: 251 bizarrely accepts all reported tallies at face value.
58. Graff 2002: 122–30.
59. A. Wright 1978: 139–56; Graff 2002: 132–35.
60. Von Glahn 2016: 182 (statuses); A. Wright 1978: 85 (Chang’an); Xiong 2006: 134 (Luoyang), 95–105 (palaces).
61. A. Wright 1978: 173 (Sima Guang). Canal: A. Wright 1978: 178–81; Xiong 2006: 86–93; M. Lewis 2009a: 254–55.
62. Graff 2002: 138–59, esp. 145–49 (149 for an estimate of c. 600,000 men).
63. Bielenstein 1987: 20; Graff 2002: 183, 190; M. Lewis 2009b: 44; von Glahn 2016: 187–90.
64. M. Lewis 2009a: 54. Unlike Lewis, who places military power and elite cooperation side by side, I give priority to the former, for the reasons presented in chapters 8 and 9 of this volume.
65. Census: Bielenstein 1987: 20. Decentralization: M. Lewis 2009b: 58–84. Tang collapse: Tackett 2014.
66. Standen 2009 (North); H. Clark 2009 (South).
67. Bielenstein 1987: 47–48 (the multiplier of about 2.2 people per household implied by Northern Song census totals is far too low and probably limited to adults; a mean of about 5 is more plausible and in line with earlier census tallies that aimed to register all household members); W. Liu 2015a: 266, table F-4 (tax rate); W. Liu 2015b: 64–65 (tax system); von Glahn 2016: 240–41, 245 (tax yield and capital city). Soldiers: Deng 1999: 305. For the Roman empire, cf. Scheidel 2015a: 243. On the Song economy in general (starting with late Tang), see von Glahn 2016: 208–78, alongside the classic survey by Elvin 1973: 111–99.
68. Eugippius, Life of St Severinus 6. It does not matter much what exactly Odoacer actually did: as we saw before, this story is part of a larger pattern. The shift from taxation to landowning was a defining characteristic of this period: Wickham 2016: 29.
69. Maier 2005: 62 (Ostrogoths); Wickham 2005: 105–6 (Franks).
70. For Middle Eastern fiscal systems, see Wickham 1994: 56–66. For relations between Rome, Iran, and the Arabs, see most recently Fisher 2011. Between 1984 and 2010, a monumental seven-volume history of Roman-Arab relations was published by Irfan Shahid (https://www.doaks.org/newsletter/byzantium-and-the-arabs).
71. H. Kennedy 2001: 59–95 on military compensation, esp. 59; H. Kennedy 2015: 390–97. Practices in Egypt were similar to Iraq’s, whereas those in Syria are less well known (396–97).
72. H. Kennedy 2001: 62–64, 74–78; Wickham 2009: 331–34; H. Kennedy 2015: 398–401; Haldon 2016: 75–77.
73. Wickham 1994: 57–61 remains a classic summary. For the evolution of iqta‘ and similar arrangements, see, e.g., Lambton 1968: 230–39 and Cahen 1975: 311–14 (up to the Seljuqs); Basan 2010: 169�
�71 (Seljuqs); Petrushevsky 1968: 516–20 (Il-Khanids); Jamaluddin 1995: 154 (Timur); Savory 1986: 364–66 (Safavids). The Ottoman regime in Anatolia and the Balkans likewise insisted on tax payments by peasants, which were initially assigned to military beneficiaries who were then replaced by a salaried army sustained by tax farmers. When the latter turned into lords with private armies, the center moved to suppress them (Wickham 1994: 63–64).
74. Asher and Talbot 2006: 35–41, 128, 152; Lieberman 2009: 640, 643, 740. For the Mughal tax system, see Richards 2012: 411–16.
75. For this taxonomy of developmental correlations, see chapter 9 in this volume. States that collected tax were generally more stable than those based on gift exchange or control over farmland: Wickham 2016: 11, 29.
76. For China, see Lorge 2005: 178. Relaxation: Deng 2015: 314–21 (Qing); M. Lewis 2015: 300–304 (Tang).
CHAPTER EIGHT
1. Quote: Montesquieu 1750 [1748]: 384 (De L’Esprit des Lois, book XVII, ch. VI).
2. Adapted from Diamond 1997: 414 (quote). The “peninsular” portion of Korea is roughly three times the size that of Shandong. Cf. also Keay 2000: xxiii, contrasting Europe’s mountains and coastlines with India, which lacks real barriers. Cosandey 2008: 499–581 considers the coastline critical for Europe’s development, as will be discussed later.
3. Criticism: Hoffman 2015: 112–14. European coastline: Cosandey 2008: 509–33.
4. Cosandey 2008: 533–69, esp. 558: Europe (46 percent peninsulas, 10.2 percent islands), Islamic zone (0.9 percent peninsulas and no islands), India (1.7 percent peninsulas and 1.9 percent islands), and China (1.1 percent peninsulas and 2 percent islands).
5. Cosandey 2008: 561, 567–68. For him, European development is rooted in the resultant system of smaller states: 75–312.
6. Mountains: Montesquieu 1750 [1748]: 384 (De L’Esprit des Lois, book XVII, ch. VI) (quote); Diamond 1997: 414. Europe not more rugged: Hoffman 2015: 109–12. Elevation in China: Marks 2012: 17, map 2.4; Auyang 2014: 339 (terrain above 400 and 1,500 meters).
7. Diamond 1997: 414 (quote). See also very briefly Lang 1997: 24.
8. See Tvedt 2010: 37 for the weakness of China’s river-coast connections.
9. J. McNeill 1998: 32 (quote), 36.
10. J. McNeill 1998: 36–37 (quotes). See Tvedt 2010: 35–37 for very perceptive observations on the nature of the Chinese river and irrigation system, stressing the risk of flooding, and the drainage and tributary focus of the canal system. See chapter 12 in this volume. Note the partial decay of this system between the Song and Qing periods: W. Liu 2015a: 94–95.
11. Wittfogel 1957. Review of Google Scholar yields studies, too numerous to list here, both supporting and conflicting with Wittfogel’s main claims. Europe: Chirot 1985: 183; Diamond 1997: 414 (size and connectivity). For the importance of riverine connectivity in general, see Scheidel and Meeks 2014.
12. Marks 2012: 29, with map 2.7.
13. See the maps of census results in 2, 140, 609, 742, and c. 765 CE in Bielenstein 1987: 193–94, 199, 202–3. Only in the Northern Song period do we see a more even distribution of the registered population between North and South: Bielenstein 1987: 207–12. See, e.g., Lieven 2000: 37. De Crespigny 2012 asserts that “the wealth and population of the North China plain means that the ruler who holds it may dominate East Asia, and this fact of geography … explains much of China’s continuing capacity for reunification.”
14. E. Jones 2003: 105–7, 226; earlier J. Hall 1985: 111; P. Kennedy 1987: 17; and later Lieberman 2009: 738. Quote: Montesquieu 1750 [1748]: 384 (De L’Esprit des Lois, book XVII, ch. VI). See also Kiesewetter 2006: 47–52. Lieven 2000: 33 misses the point by noting that China was divided into semi-enclosed regions: comparative metrics are key.
15. E. Jones 2003: 105–7 (quote: 107). Remarkably, the Roman empire is not even mentioned in the book’s index.
16. Keay 2000: xxv (farm belt); Lieberman 2009: 738–39.
17. Keay 2000: xxv–xxvi; Asher and Talbot 2006: 12; Lieberman 2009: 739.
18. My estimate for the Roman empire schematically draws on the length of Hadrian’s Wall, of the Rhine and Danube, and Turkey’s, Syria’s, and Israel’s eastern borders.
19. Needham 1969: 151, fig. 20 (quote, with map visualization). For another map visualization considering altitude and precipitation levels, see Ko, Koyama, and Sng 2018: 288, fig. 1. Elvin 1973: 21 and Pines 2012: 8 note isolation as a contributing factor to China’s imperial cohesion.
20. Hui 2005: 160, despite her overall resistance to geographical arguments, notes that the European state system was “more expansive and dispersed” than China’s and that Napoleon “faced the tyranny of distance in the Russian campaign.” Capitals in the Warring States period—which ended in imperial unification—were much closer together than Paris and Moscow (160).
21. For Rome, see chapter 2 in this volume. In 9 BCE, the Roman commander Drusus reached the Elbe, and the short-lived trans-Rhenanian Roman province of Germania appears to have reached as far as the lower Elbe. For Charlemagne’s expansion, see chapter 5 in this volume.
22. Heather 2009: xv (quotes); 386–451 (spread of Slavs), 520–76 (later state formation and development); Wickham 2009: 472–507, esp. 480–91, 505–7, 556 (tenth century); Wickham 2016: 81. Brown 2003 is a classic account of the “rise of Western Christendom” up to 1000 CE.
23. Bartlett 1993, esp. 295, 298–99, 303–4, 306, 308–9; and also 24–59 on the “aristocratic diaspora.” Quote: Heather 2009: xv.
24. For a crude guide, see https://www.euratlas.net/history/europe/index.html. Latin Christendom: Bartlett 1993: 292.
25. See, e.g., Lieven 2010.
26. Gat 2006: 394 makes the same point in his (highly superficial) discussion of environmental factors (391–95). In his massive survey of modern explanations of the (second) “Great Divergence,” Vries 2013: 153–61, though unsympathetic to geographical reasoning, offers no compelling arguments against it. I fully agree with his view that geography, being static, can never be more than a necessary precondition, and not a sufficient one (413). At the same time, I object to his classification of geography and natural resources as “proximate causes” (409), except insofar as this term is used in a technical economic sense as denoting endogeneity to the economy (as Peer Vries kindly informs me it is). I also find it hard to understand why Hoffman 2015: 109–14, in his rejection of geographical arguments for explaining European exceptionalism, limits himself to mountains and coastlines but ignores features such as compactness and natural cores.
27. Steppe ecology: e.g., Barfield 1989: 20–24; Taaffe 1990: 33–37.
28. See Turchin 2009: 202–3, table 2, for a list of empires, and Scheidel in press-b: table 2.1 for a lightly augmented and revised tabulation that I use here. The fifty-four empires are Abbasid, Almohad, Almoravid, Ayyubid, Bactria, Buyid, Chagatai, Delhi, Fatimid, Former Qin, Ghaznavid, Ghurid, Gokturk, Golden Horde, Hephthalite, Huns (Attila), Il-Khans, Jin, Jurchen, Khazar, Khorezm, Kushan, Kiev, Liao, Lithuania-Poland, Mali, Mamluk, Media, Ming, Mongol, Mughal, Northern Song, Northern Zhou, Ottoman, Parthia, Qara Qitai, Qin, Qing, Rouran, Russia, Safavid, Saka, Samanid, Seljuq, Sui, Tang, Timurid, Tufan, Uigur, Umayyad, Western Han, Xianbei, Xiongnu, and Zhungar. The nine intermediate cases are Achaemenid, Assyria, Egypt, Gupta, Harsha, Maurya, Pratihara, Sasanian, and Shang; and the remaining ten are Axum, Byzantine (high medieval), Franks, Holy Roman empire, Inca, Khmer, Macedonian, Maratha, Roman, and Seleucid.
29. Turchin 2009: 201; Scheidel in press-b.
30. Turchin et al. 2013: 16386 provide analogous heat maps for states of greater than 100,000 square kilometers.
31. Scheidel in press-b: table 2.2. The twenty empires are Delhi, Former Qin, Former Yan, Ghurid, Hephthtalite, Jin, Jurchen, Kushan, Later Zhao, Ming, Mongol, Mughal, Northern Song, Northern Wei, Northern Zhou, Qin, Qing, Tang, Western Han, and Umayyad. The seven intermediate cases are Achaemenid, Gupta, Harsha, Maurya, Nanda, Pratihara, and Sasanian; and the five remaining ones are Ma
cedonian, Maratha, Pala, Roman, and Seleucid. By contrast, the largest South Indian polities, Chola and Vijayanagara, each captured only around 4 percent.
32. Quote: Turchin 2009: 201.
33. Antagonistic relationship: Turchin 2009, esp. 191, 194–97. On conflict giving impetus to state formation, key contributions include Carneiro 1970, 1988; Lane 1979; H. Lewis 1981; W. McNeill 1982; Tilly 1985, 1992; B. Porter 1994; Spruyt 1994; Kasza 1996; Ertman 1997; Turchin et al. 2013; I. Morris 2014. I summarize this concisely in Scheidel 2013b: 11–12 (conflict theories of state origin), 20–22 (role of violence in state formation), 33–38 (warfare as a driver of state formation).
34. See, e.g., Sunderland 2004: 5–8 for a standard summary of central features, with reference to earlier literature on the emergence of steppe pastoralism. See also Khazanov 1994. Lack of steppe self-sufficiency: Jagchid and Symons 1989; Golden 1998: 20–21. Christian 1998 covers the broad sweep of premodern Central Asian history.
Escape From Rome Page 63