64. Taylor 2017: 169–70.
65. Polybius 32.13.6–8. Cf. also Livy 10.1.4, on a much earlier incursion into Umbria “lest they should pass the whole year entirely without fighting”: Oakley 1993: 16. Neither one of these passages betrays a sense of irony.
66. Harris 1985: 10 (at most thirteen years without war between 327 and 101 BCE); Oakley 1993: 14–16 (eleven years without war from 410 to 265 BCE; but cf. twelve years from 440 to 411 BCE, when the record is even poorer but war-making might also have been less intense before Rome ramped up its capabilities). China: data for Qin (186 years) and Chu (185 years) are drawn from a database compiled and kindly shared by Dingxin Zhao. Years of peace 722–221 BCE: Hsu 1965: 56, 64. Cf. Deng 2012: 336, table 4.1 for the disproportionate belligerence of the Qin state. (Adding the entire first century CE to the Roman count to match the 502 years of the China count would not substantively alter the picture.) The claim that Hellenistic powers were (also) always at war (Chaniotis 2005: 5–12) is facile in its imprecision: it can be empirically substantiated only for a century or so of Macedonian history. On Roman continuous war in general, see Hopkins 1978: 25–37.
67. Thus, evidence of warlike behavior of other ancient states does not show that they were more or less the same (contra Eckstein 2006: 118–80; and cf. also Eich and Eich 2005: 1, 4, 14, 33), even if the vitriol Harris 2016: 42 expends on this notion is unhelpful. For Rome’s greater bellicosity, see also Harris 2016: 37. From this perspective, Eckstein’s 2006 critique of “unit-attribute explanations” of Roman success that focus on domestic conditions must be judged excessive, even though his emphasis on systemic relations is well taken (see esp. 35). Domestic conditions did in fact account for Rome’s ability to project power on an unprecedented scale. More specifically, the Schumpeterian “war machine” dynamics of Roman conduct deserve more attention than they have generally received.
68. See Scheidel (in press-a) for this pattern.
69. See chapter 3 in this volume, and esp. Doyle 1986: 88–92.
70. Campaigning beyond Italy had always bestowed greater autonomy to generals (Eckstein 1987; Woolf 2012: 74): coupled with political clout, their powers grew with increasing distance from the center. Vervaet 2014: 214–92 offers a formalistic perspective of this shift.
71. See, e.g., Löffl 2011: 133–34 on the Alpine campaigns, which were risky and unprofitable and had therefore been unappealing to more vulnerable Republican commanders (135–36).
72. From 61 to 59 BCE, the Roman state got by with 15 legions, some 80,000 men—4 legions in Spain, 4 in Provence and Northern Italy, 3 in the Balkans, and 4 in the eastern Mediterranean (Brunt 1987: 449). If we move 3 legions from Spain (1 legion remained after conquest there had been completed) to North Africa and Egypt and put an extra one in the East, we end up with 16 legions, compared to the 28 that actually existed under Augustus, the 25–28 from 9 CE to the 160s CE, and the 30–33 in the late second century CE. This suggests that without the massive expansion into the European periphery that commenced with Caesar and Augustus, military forces only a little over half as large as they came to be could have secured the empire if it had remained confined to the Mediterranean basin.
73. Service: Scheidel 1996: 93–94. Roman emperors’ exposure to challenges from within the military domain is just one of these consequences, a theme I explore in forthcoming work.
74. Economic perspective: Harris 2016: 130–32.
75. Not rational: Mattern 1999: 123–61. Question: contra Cornell 1993: 145–49. Quotes: Strabo, Geography 4.5; Appian, Preface 5. In the second century CE Britain received around 15 percent of total Roman military spending (Mattingly 2006: 493). If second-century-CE Britain accounted for 10 percent of Rome’s military budget, or 6–7 percent of overall public expenditure, if two-thirds of the troops stationed there would not otherwise have been deployed elsewhere, and if the province comprised 3 percent of the population of the empire (see Mattingly 2006: 368 for a population of 2 million) and was slightly poorer than the imperial average, regional military expenditure could easily have been double the amount of regional tax revenue. Its share in Rome’s military budget may have been even higher (Mattingly 2006: 493 reckons with 15 percent); and although its population may also have been larger (Millett 1990: 185 posits 3.7 million), if that had been the case it would have been even more rural and poorer. Therefore, quite simply, in no way could Roman Britain ever have paid for itself. Limits: Mattingly 2006: 491–96.
76. Motivation: Morley 2010: 37. Service: Harris 2016: 133–36. Political restraints on commanders: Cornell 1993: 162–64; demilitarization of elite: 164–68. See Scheidel 1996: 95–96n18 on the evolution of recruitment patterns.
77. Harris 2016: 112–36 gives a good account. Transition: Morley 2010: 36–37.
78. Gellner 1983: 9.
CHAPTER THREE
1. On the concept of semiperipherality, see Chase-Dunn and Hall 1997: 78–98. Italy conforms to their definition as a region that mixed both core and peripheral forms of organization and was spatially located between core and periphery (78); see also 19, 201, and 274 for definitional issues. Different types of networks (bulk- and prestige goods, political/military, and information networks) coexisted (52). For the emergence and expansion of the Near Eastern PMN, see 201, and the visualization 203, fig. 10.1, derived from Wilkinson 1987, who labeled this network the “Central Civilization” that eventually expanded to encompass the entire globe.
2. See Broodbank 2013, esp. 508–9, fig.10.2, for Mediterranean integration before the Roman period.
3. I. Morris 2009: 159–63, esp. 163, on capital-intensive state formation featuring mercenaries and navies but concurrent institutional underdevelopment due to autocrats’ (i.e., tyrants’) lording it over fragmented citizen communities; and I. Morris 2013a: 283–86 for Syracuse as a hybrid case. The Sicilian boom-bust cycle of war and peace that was sustained by large levies until funds ran out “reflects shallow state capacity” (292).
4. Ameling 1993 is the most penetrating account of the nature of the Carthaginian state, stressing its militarism and domestic mobilization capacity against earlier (ancient and modern) myths of mercantile proclivities.
5. Hansen 2006: 98 reckons with 750,000 people living in poleis (or Greek-style city-states) from Spain to the Adriatic, most of them located in Sicily and southern Italy. This fits De Angelis’s 2006: 139 suggestion of a possible Sicilian population of 600,000, compared to a potential maximum of 1.2 million or more. This number would have been equivalent to maybe 15 percent of the (highly mobilizable) population of the area under Roman control in the mid-third century BCE (see above). For Sicilian troop sizes, see Beloch 1886: 290–94, who rightly discards the unique outlier of 83,000 troops mentioned by Diodorus 14.47 that is still kept alive by I. Morris 2013a: 291.
6. The population of the Roman Maghreb in 14 CE may have been around 3.5 million (Frier 2000: 812, table 5), which need not have been very different from conditions in the Carthaginian period: compare the 3 million–4 million conjectured by Beloch 1886: 470 and Hoyos 2010: 199 for the time of the First Punic War. By 218 BCE, both sides had increased their populations by maybe another half, in Northern Italy, Sicily, Sardinia, and Corsica in the case of Rome and in the Iberian peninsula in the case of Carthage. In this regard they were quite evenly matched. Concerning Carthage’s martial disposition, Ameling 1993 is fundamental: see pp. 155–81 for elite militarism and pp. 190–210 for citizen participation. For citizen levies of 20,000, 30,000, and 40,000, see pp. 190–94. Normal naval strength was 120–130 ships, which could have been manned by fewer than 40,000 citizens (198).
7. Ameling 1993: 210–12 for allies (Libyans). For the various layers of Carthaginian rule, see Huss 1985: 467–74. Either navy or army: Ameling 1993: 209–10. Carthage supposedly lost at least 450 ships and 100,000 crew in the First Punic War, fewer than Rome if reported figures (or at least the ratios they imply) are to be believed, but it also lost the war. On mercenaries, see Ameling 212–22, and in great det
ail, see Fariselli 2002.
8. Hoyos 2010: 145, 185–86, 190–92 on Libyan risings in 396, 309–307, 256–255, and 241–238 BCE, the last case compounded by a mercenary revolt that pushed Carthage to the brink. Unlike in Rome, there is no sign of an integrative reward system beyond the metropolitan citizenry: Ameling 1993: 225, 262. See Barcelo 1988 for involvement in Spain before 237 BCE.
9. Ameling 1993: 199–203 argues for a maximum of 200 vessels, while Lazenby 1996: 82–87 sides with the maximalists. Either way, Carthage’s naval levies were very large by the standards of the time. My own estimates in unpublished work, based on a mix of ancient sources and conjectures, suggest, very crudely, shifts from around 25,000 mean military strength in the fourth century BCE, to around 50,000 in the First Punic War, to around 75,000 in the Second Punic War, and peaks of 50,000+, 70,000–80,000, and 100,000, respectively. Figures given by Roman sources (such as 120,000–150,000 in the Second Punic War, Hoyos 2010: 199, 205) tend to be inflated, just as they were later for the Hellenistic kingdoms.
10. Manpower: Chaniotis 2005: 20–24, 46–51. On professional warfare, see 78–101. Doyle 1986: 89, 91, and 134–36 defines these polities as “fractionated republics” that were a source of collaborators. Aitolians: Grainger 1999: 202–14, esp. 213. In 279 BCE, Boiotia, Phokis, Aetolia, Lokroi, and Megara sent 22,100 men to Thermopylae to stop a Galatian invasion (Chaniotis 2005: 23). In 225 BCE, Rome mobilized no fewer than 160,000 troops to deal with a similar event.
11. Doyle 1986: 89, 91, and 133 defines them as patrimonial monarchies.
12. This is consistent with the maximum number of Macedonian soldiers that were available after the end of Alexander’s reign, of some 40,000 in 321–320 BCE (Bosworth 2002: 86).
13. Armies: Aperghis 2004: 191; Fischer-Bovet 2014: 73. Navies: Aperghis 2004: 197–99; Fischer-Bovet 2014: 71–72. As noted above, Rome lost some 500 to 700 ships in the First Punic War alone.
14. Grainger 2010 for the Syrian Wars, esp. on competitive development (89–115), Seleucid internal strife and fragmentation from the 240s to the 220s BCE (171–94), and domestic turmoil in Egypt from 207 BCE onward (219–43).
15. Parallel wars: Grainger 2010: 245–71.
16. Sekunda 2007: 336 for risk of losing core troops.
17. Bar-Kochva 1976: 7–19 (army size), and 20–48 on military settlers as “hard core.” See also Fischer-Bovet 2014: 199–237 for Egypt. Sekunda 2007: 335–36 for this vital contrast.
18. Armies: Brunt 1987: 423–24. These Roman numbers might be a bit too high (p. 423), but then so are the Hellenistic ones (cf. Grainger 2002: 322, 359–61 on the battle of Magnesia). Potential: Brunt 1987: 422. Manpower was lower right after 200 BCE than it had been in 225 BCE, but recovery was swift: Hin 2013: 142–54. In any case, the original total pool may well have been closer to 900,000.
19. Numbers from Afzelius 1944: 47, 78–79, 89. Ships: Afzelius 1944: 83–84 and W. Murray 2012: 209–25. Even if Afzelius’s numbers are a little inflated (cf. Brunt 1987: 423), this does not (much) affect ratios between theaters.
20. Roman citizen infantrymen were paid the equivalent of one-third (or perhaps rather only 0.3: Taylor 2017: 146) of a drachma per day (which covered food as well), and those of the Italian allies received 34.5 liters of grain per month, equivalent perhaps to one-tenth of a drachma per day (Taylor 2017: 150 estimates one-twelfth). Hellenistic soldiers could expect five-sixth or a full drachma per day plus another one-third of a drachma for food (Launey 1950: 725–80, at 763; Aperghis 2004: 201–5). This yields a ratio of approximately 1:6 (if Roman forces were evenly split between citizens and allies), or closer to 1:3 if Romans were granted double (combat) pay and grain was dearer in wartime. Taylor 2015 discusses military finances more generally.
21. Fischer-Bovet 2014: 76, table 3.5. See T. Frank 1933: 145 and Taylor 2017: 165 for average annual military outlays of between 1,550 and 1,800 talents in Rome (200–157 BCE), compared to 5,000 talents for the Ptolemies and 7,000–8,000 talents for the Seleucids in peacetime and 10,000+ talents for each in time of war (Fischer-Bovet 2014: 76, table 3.5). Orders of magnitude are all that matter here.
22. Reported fatalities on the Roman side for these four engagements add up to merely 1,350, about 1.3 percent of the forces deployed, compared to up to 90,000 enemy deaths. Even if actual Roman totals were higher and the others lower (cf. Grainger 2002: 249–50, 328 for rather arbitrary adjustments), the overall imbalance is not in doubt.
23. Sekunda 2001 on reforms in the 160s BCE, though their actual extent remains unclear; and see 115 on manpower. Quote: Livy 9.19.
24. Rome scored decisive victories at Chaeronea (86 BCE) and Orchomenos (85 BCE) in Greece, at Tigranocerta (69 BCE) and Artaxata (68 BCE) in Armenia, and in Pontus (66 BCE and, at Zela, 47 BCE). There do not seem to have been substantial numbers of Roman soldiers in western Asia Minor when Pontus defeated Rome’s local allies and overran its province in 88 BCE (Brunt 1987: 435). Two Roman legions (or roughly 10,000 men) were defeated in 81 BCE but survived, and were later crushed at Zela in 67 BCE. For these wars in general, see Mayor 2010.
25. Later it took centuries of interaction with the Roman empire and gradual scaling-up of capabilities for Germanic populations to become capable of threatening Roman territory: see chapter 5 in this volume.
26. See Doyle 1986: 89–90, 132–33, 135 for the characteristics of tribal peripheries: fierce but small and undifferentiated, and to be ruled directly. J. Williams 2001: 187–207 on the Po Valley (where appearance of La Tène—i.e., “Celtic”—burials coincided with a general decline in nucleated settlements followed by a dispersed settlement pattern: 198–99, 205; and see 207–18 on the impact of Roman conquest); Richardson 1986 on Spain, esp. 16–17 on the indigenous population (ranging from pastoral seminomads to large villages and fortified towns on hilltops).
27. See Johne 2006: 83–198.
28. Summaries in Doyle 1986: 91, 130, table 3.
29. Roman scores: until the late third century BCE, the Warring States of China and then the Qin empire had belonged in the same category. The early Western Han empire maintained a smaller active military than its predecessors.
30. The underlying calculations are necessarily tentative. I focus on overall population size and maximum troop numbers (excluding rowers), and all estimates are rounded off and deliberately conservative. I reckon with 4 million people in Italy and 200,000 peak troops for Rome in the Second Punic War, and 5.5 million people including provincials and 180,000 troops for the early second century BCE; with 5 million people and 100,000 troops for Carthage; 600,000 people and 30,000 troops for Syracuse; 9 million people and 100,000 troops for the Seleucids; 4.5 million people and 100,000 troops for the Ptolemies; and 1.5 million people and 40,000 troops for the Macedonians and their Greek allies. I also assume 600,000 people for Aetolia (in analogy to Epirus as calculated by Corvisier 1991: 289) and 15,000 troops. For these numbers, see McEvedy and Jones 1978; Fischer-Bovet 2014: 76, 170 (Hellenistic military forces); Corvisier 1991: 252, 271 (Thessaly and Macedonia); above (Sicily, Rome, Carthage, Macedon). Aperghis 2004: 57 proposes a larger Seleucid population of 11 million to 15 million, but that is highly uncertain.
31. Balancing failure: Macedon joined Carthage’s war against Rome in 214 BCE but to little effect. The Ptolemies stayed neutral, perhaps to balance Rome against Carthage (thus Lampela 1998: 56–63). See my discussion of counterfactuals in chapter 4 of this volume.
32. E.g., Fischer-Bovet 2014: 65−66. See W. Murray 2012: 192, map 6.1 for Ptolemaic dependencies across the entire eastern Mediterranean and in the Aegean.
33. Terms: Lampela 1998: 33–56, esp. 33–37 on an embassy exchange in 273 BCE that led to diplomatic “friendship” or perhaps even a treaty. Geographical distance as a defining geopolitical feature: 20–21. On the Ptolemaic navy, see W. Murray 2012: 188; Fischer-Bovet 2014: 71–72. King Ptolemy IV (r. 221–204 BCE) was said to have built a 420-foot-long ship that held over 7,000 crew (W. Murray 2012: 178–85).
34. De So
uza 1999: 97–178 on the Cilician pirates, born of Seleucid state failure from the 140s BCE onward, down to their demise at the hands of Rome in 67 BCE.
35. See Erdkamp 1998: 62–62 and Roth 1999: 158–65, 189–95. In the campaign against the Seleucids, grain from Sicily, Sardinia, Numidia, and even semidependent Carthage was shipped to the Aegean to provision Roman and allied forces: Roth 1999: 161–62; 228, table V. Simulation: Artzrouni and Komlos 1996: 127–33, esp. 132.
36. I owe these observations to James Bennett, based on his independent re-creation of the Artzrouni and Komlos model (personal communication March 8, 2017). Within the specifications of this simulation, low coastal defense costs on their own do not guarantee success because the algorithm randomly selects polities across Europe to attack their neighbors. As states become larger, they have a growing chance of being selected, which means that even polities with very small coastal defense savings come to dominate only if they start early and amass enough power to succeed against Italy when they collide. A head start by Italy, on the other hand, reinforces its initial advantage.
Escape From Rome Page 60