Hostname: page-component-5c6d5d7d68-wtssw Total loading time: 0 Render date: 2024-08-25T13:53:06.109Z Has data issue: false hasContentIssue false
  • English
  • Français

Human Mobility in Roman Italy, I: the Free Population*

Published online by Cambridge University Press:  14 March 2012

Walter Scheidel
Walter Scheidel*
Affiliation:
Stanford University
*
scheidel@stanford.edu
Get access Rights & Permissions [Opens in a new window]

Extract

How did the relentless spread of Roman power change people's lives? From military mobilization, urbanization, slavery, and the nexus between taxation and trade to linguistic and religious change and shifting identities, the most pervasive consequences of empire all had one thing in common: population movements on an unprecedented scale. Yet despite its pivotal role in social and cultural change, the nature of Roman mobility has never been investigated in a systematic fashion. In this study, I develop a comprehensive quantitative model of population transfers within, to, and from Italy, from the late fourth century B.C. to the first century A.D. Owing to the diverse and complex character of these movements, I develop my argument in two steps. The present paper deals with the demographic context, scale, and distribution of the migration of free persons. I argue that the total population of Italy in the early imperial period was of the order of five to six million rather than fourteen to twenty million (Section II); that state-sponsored re-settlement programmes dramatically increased overall levels of mobility on three occasions (during the Italian wars in the late fourth and early third centuries B.C., in the aftermath of the Second Punic War in the early second century B.C., and in the period of constitutional transition from the 80s to the 10s B.C.) (Section III); and that in the last two centuries B.C., colonization programmes and urban growth in Italy required the permanent relocation of approximately two to two-and-a-half million adults (Section IV).

Type
Articles
Information
The Journal of Roman Studies , Volume 94 , November 2004 , pp. 1 - 26
Copyright
Copyright ©Walter Scheidel 2004. Exclusive Licence to Publish: The Society for the Promotion of Roman Studies

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Footnotes

*

I wish to thank Nathan Rosenstein and three anonymous referees for valuable comments and criticism, and Alison Sharrock for convincing me to convert the unwieldy original draft into two more manageable instalments.

References

1 ‘Human mobility in Roman Italy, II: the slave population’, JRS 95 (2005)Google Scholar.

2 A. J. Coale and P. Demeny, Regional Model Life Tables and Stable Populations (1983). I use Model West Level 3 Females (e0 = 25, r = 0) as a generic template (ibid., 57). Any reasonable alternative would not make a significant difference to my results.

3 Scheidel, W., ‘Roman age structure: evidence and models’, JRS 91 (2001), 1–26Google Scholar.

4 For example, according to Model West Level 3 Females, 63 per cent of the population is seventeen or older, compared to 62 per cent in the ‘male’ variant of the same model. Levels 1 (e0 = 20) and 5 (e0 = 30) Females predict shares of 60 and 65 per cent, respectively. Thus, the probable margin of error is unlikely to exceed 5 per cent, even if we allow for non-standard ratios of minors to adults.

5 See below, Section IV and Scheidel, op. cit. (n. I). For computational purposes, I reckon with a balanced sex ratio unless otherwise specified.

6 Beloch, J., Die Bevölkerung der griechisch-römischen Welt (1886), 370–8Google Scholar; Brunt, P. A., Italian Manpower 225 B.C.–A.D. 14 (1971, repr. 1987), 113–20Google Scholar.

7 Frier, B. W., ‘Demography’, CAH XI 2 (2000), 811–14Google Scholar.

8 Frank, T., ‘Roman census statistics from 225 to 28 B.C.’, CP 19 (1924), 329–41;Google ScholarWiseman, T. P., ‘The census in the first century B.C.’, JRS 59 (1969), 59–75,Google Scholar at 72–5; Cascio, E. Lo, ‘The size of the Roman population: Beloch and the meaning of the Augustan census figures’, JRS 84 (1994a), 23–40;Google Scholar ‘La dinamica della popolazione in Italia da Augusto al III secolo’, in L'Italie d'Auguste à Dioclétien (1994b), 91–125; ‘The population of Roman Italy in town and country’, in J. Bintliff and K. Sbonias (eds), Reconstructing Past Population Trends in Mediterranean Europe (3000 BC–AD 1800) (1999), 161–71; Morley, N., ‘The transformation of Italy, 225–28 B.C.’, JRS 91 (2001), 50–62Google Scholar.

9 My reconstruction of Italian slave numbers in Part II would be only marginally affected.

10 Polyb. 2.24; Liv., Per. 89; Phlegon FGrHist 257 F 12; RGDA 8.

11 Close variants of the final totals are also attested in other sources, all of them ultimately derived from Fabius Pictor: F. W. Walbank, A Historical Commentary on Polybius I (1957), 199.

12 For the most comprehensive list of references, see Baronowski, D. W., ‘Roman military forces in 225 B.C. (Polybius 2.23–24)’, Historia 42 (1993), 181–202,Google Scholar to which one needs to add E. Lo Cascio, ‘Recruitment and the size of the Roman population from the third to the first century BCE’, in W. Scheidel (ed.), Debating Roman Demography (2001), 111–37, esp. 129–33.

13 I use the figures in A. Afzelius, Die römische Eroberung Italiens (340–264 v. Chr.) (1942), 134–5, of 25,615 km2 for the Ager Romanus and 87,175 km2 for the allied states listed or implied by Polybius. This apparent mismatch has been tempered by adjustments such as the common but unsubstantiated assumption that the allied (but not the Roman) totals in the katagraphai comprise only iuniores (e.g., Brunt, op. cit. (n. 6), 52), but has more recently been restored by Lo Cascio's assertion that all these numbers refer to iuniores (op. cit. (n. 8, 1999), 168).

14 Brunt, op. cit. (n. 6), 44–60; Lo Cascio, op. cit. (n. 8, 1999), 168. Brunt's total, unlike the other two, includes the Greeks of southern Italy. Slaves must be added to all three estimates.

15 viz., 250,000/250,000, and 23,000/35,000. Note that the latter mirrors the ratio Polybius reports for the troops on active duty, of 2,700 Roman and 4,000 allied cavalry, or 2:3. In the context of the katagraphai, this ratio makes perfect sense if it represents a formula togatorum (cf. Brunt, op. cit. (n. 6), 545–8) — that is, if it reflects Roman demands that the allies cumulatively match Roman infantry strength and one and a half times Roman cavalry strength. This, however, cannot explain ratios in a tally that omits Sabines, Etruscans, and Umbrians.

16 52,300 vs. 158,000+, or 1 to 3.02+.

17 For a possible parallel, see P. Hunt, Slaves, Warfare, and Ideology in the Greek Historians (1998), 33–7, on Herodotus' Spartiate/hoplite ratio of 7:1 at Plataea (9.10, 28–9) which may simply be extrapolated from the conventional depth of the phalanx. Note that Fabius Pictor (FGrHist 809 F 9) also ‘knew’ of the existence of 80,000 adult male citizens under Servius Tullius, an impossible number.

18 Afzelius, op. cit. (n. 13), 98–135; Brunt, op. cit.(n. 6), 54.

19 Brunt, op. cit. (n. 6), 189; G. Bandelli, ‘La popolazione della Cisalpina dale invasioni galliche alla Guerra sociale’, in D. Vera (ed.), Demografia, sistemi agrari, regimi alimentari nel mondo antico (1999), 189–215.

20 Frier, op. cit. (n. 7), 812. Note, however, that the population of Gaul is similarly unknown.

21 See above, at the beginning of this section.

22 W. Scheidel, ‘Progress and problems in Roman demography’, in idem, op. cit. (n. 12), 55, 57.

23 Morley, op. cit. (n. 8), 62.

24 For the fading of the past, see Scheidel, W., ‘Emperors, aristocrats, and the Grim Reaper: towards a demographic profile of the Roman élite’, CQ 49 (1999), 279–80CrossRefGoogle ScholarPubMed. For the census, see, e.g., L. Neesen, Untersuchungen zu den Staatsabgaben in der römischen Kaiserzeit (27 v. Chr.–284 n. Chr.) (1980), 39–41; R. S. Bagnall and B. W. Frier, The Demography of Roman Egypt (1994), 2–5.

25 Lo Cascio, op. cit. (n. 12), 128–9, with reference to Brunt, op. cit. (n. 6), 66, 422, 714.

26 M. H. Hansen, Three Studies in Athenian Demography (1988), 22, 27.

27 Lo Cascio, op. cit. (n. 12), 124–5, 137. City-states are different creatures: see in general H. Spruyt, The Sovereign State and its Competitors (1996), 130–50; M. H. Hansen (ed.), A Comparative Study of Thirty City-State Cultures (2000).

28 cf. now Rosenstein, N., ‘Marriage and manpower in the Hannibalic War: assidui, proletarii and Livy 24.18.7–8’, Historia 51 (2002), 163–91,Google Scholar for a new and convincing explanation of how it was possible that, in 214 B.C., only 2,000 iuniores who could not claim a legitimate exemption had not yet served in the army.

29 J. M. McPherson, Battle Cry of Freedom (1998), 306 n. 41; Haines, M. R., ‘Estimated life tables for the United States, 1850–1910’, Historical Methods 31 (1998), 149–69CrossRefGoogle Scholar.

30 Lo Cascio, op. cit. (n. 8, 1994a), 164–5. See below, Section IV, for a rebuttal of a related objection.

31 Morley, N., Metropolis and Hinterland (1996), 182CrossRefGoogle Scholar; Hopkins, K., Conquerors and Slaves (1978), 68–9Google Scholar

32 See below, nn. 83–4, for a rejection of Lo Cascio's argument for a substantially larger urban population.

33 Ginatempo, M. and Sandri, L., L'Italia delle città (1990), 148–9, 190–1, 227Google Scholar; Bellettini, A., La popolazione italiana (1987), 25Google Scholar.

34 For the share of cities of under 5,000 in the total urban population, cf. de Vries, J., European Urbanization 1500–1800 (1984), 49–77Google Scholar.

35 M. H. Hansen, ‘The concept of the consumption city applied to the Greek polis’, in T. H. Nielsen (ed.), Once Again: Studies in the Ancient Greek Polis (2004), at 11–16. I am grateful to M. H. Hansen for thism reference.

36 Tac, Ann. 11.25, with Beloch, op. cit. (n. 6), 371–2.

37 A 10 per cent undercount would translate to 6.6 million adult men or 21 million altogether, noncoverage of 20 per cent of all citizens to 7.5 and 24 million, respectively. My very generous assumption of 5 million citizens in the provinces is more than two and a half times as high as the corresponding tally for A.D. 14 posited by Brunt, op. cit. (n. 6), 265.

38 Bellettini, op. cit. (n. 33), 35.

39 ibid., 176. Cf. Panta, L. Del, ‘L'Italie’, in Bardet, J.-P. and Dupâquier, J. (eds), Histoire des populations de l'Europe II (1998), 513–16.Google Scholar

40 Morley, op. cit. (n. 8), 56–9, esp. 59 n. 64: 17 million Italians could have been fed by planting 80 per cent of the arable with barley and 20 per cent with grain, and by suppressing fallow on one-third of the land. 19 million consumers outside Rome would require a switch to 100 per cent barley.

41 Proctor, D. L., Grain Storage Techniques (1994)Google Scholar; www.cropstorage.com (storage losses); M. S. Spurr, Arable Cultivation in Roman Italy c.200 B.C.–c. A.D.100 (1986), 10–17, who points out that barley was so unpopular that it was distributed as punishment rations in the army; fed to slaves; and grown either as a back-up crop or as animal fodder (14–15). See also L. Casteletti, ‘Contribute alia ricerche paleobotaniche in Italia’, RIL 106 (1972), 331–74. Morley, op. cit. (n. 8), 56, acknowledges that Italy's wheat could not feed nearly as many people; see Scheidel, op. cit. (n. 22), 54 n. 216 for independent confirmation.

42 cf. Frier, op. cit. (n. 7), 814.

43 Vague allusions to the ‘posizione di primato dell'economia italica nell'ambito del Mediterraneo’ and ‘la posizione di primato politico dell'Italia’ in Lo Cascio, op. cit. (n. 8, 1994b), 119–20, do nothing to explain Italy's presumed demographic exceptionalism.

44 20 million Italians equals 2.6 times Frier's estimate for A.D. 164. Applied to the entire Empire, this multiplier yields a grand total of 160 million. Extrapolation from comparative data suggests an even higher total. The ratio of the French to the Italian population from 1300 to 1800 is usually put at a stable value of 1.5–1.6 to 1; if this multiplier is applied to the Roman period, 20 million Italians imply 30–32 million Gauls (compared to 23 million according to the Frier multiplier), equivalent to the population of that region around 1800. Roman Spain ought to have been inhabited by 12–15 million, just as in the first half of the nineteenth century.

45 McEvedy, C. and Jones, R., Atlas of World Population History (1978)Google Scholar; Elvin, M., The Pattern of the Chinese Past (1973), 84Google Scholar.

46 Neesen, op. cit. (n. 24), esp. 68–70, 128, 137–9. Cf. Hopkins, K., ‘Rome, taxes, rents and trade’, Kodai 6/7 (1995/1956), 46–7,Google Scholar with Scheidel, op. cit. (n. 22), 76.

47 For the latter, see Rathbone, D. W., ‘Villages, land and population in Graeco-Roman Egypt’, PCPS 36 (1990), 134Google Scholar. Lo Cascio, E., ‘La popolazione dell'Egitto romano’, Studi Storici 40 (1999), 425–47,Google Scholar assigns some 9 million people to Roman Egypt, compared to Frier's 5 million. If we multiply Frier's other estimates for the Roman East by 1.8, Roman population number in Anatolia and Syria reaches mid-twentieth-century levels.

48 Extrapolated from Frier, op. cit. (n. 7), 814: 2.6 times 38.2 million in the ‘Latin’ West versus 23.1 million in the ‘Greek’ East.

49 I know of no historical parallels for contraction on this scale in large populations outside the Americas after 1492 whose previous isolation from Eurasia had rendered them exceptionally vulnerable to new infectious diseases. For population fluctuations in China, cf. Scheidel, op. cit. (n. 22), 69 n. 284.

50 G. Forni, Il reclutamento delle legioni da Augusto a Diocleziano (1953), with W. Scheidel, Measuring Sex, Age and Death in the Roman Empire (1996), 95–6 n. 18. Roman soldiers ate better: A. C. King, ‘Animal bones and the dietary identity of military and civilian groups in Roman Britain, Germany and Gaul’, in T. Blagg and A. King (eds), Military and Civilian in Roman Britain (1984), 187–217.

51 Morley, op. cit. (n. 8), 52 n. 14, employing Hayden White's once-trendy taxonomy.

52 ibid., 62.

53 Contrary to Morley's contention that the ‘high’ count offers the best explanation for conflict over land in the late Republic (op. cit. (n. 8), 61), L. de Ligt shows in a forthcoming paper that even if we reject Brunt's inflated slave numbers (see Scheidel, op. cit. (n. I)), the ‘low’ scenario is perfectly compatible with increased demand for land.

54 Brunt, op. cit. (n. 6), 116, 262–5. Brunt's some-what lower totals are marred by his arbitrary assumptions about the proportion of adult males in the total population.

55 ibid., 115.

56 Contra Lo Cascio, op. cit. (n. 12), 120–1. See Bagnall and Frier, op. cit. (n. 24), 334 for partial neglect of young girls in the Egyptian census declarations.

57 The total of 900,000–910,000 reported for 69 B.C. is compatible with this figure. Brunt's (highly) conjectural analysis yields a presumptive total of 1,155,000 adult men (21 per cent of whom went uncounted), for a total of 3.7 million (op. cit. (n. 6), 97). A somewhat lower undercount of 15 per cent would yield 1,070,000, or 3.4 million. If the enfranchisement of the Transpadana generated an additional 950,000 citizens in Italy, between 250,000 and 750,000 Italian citizens of old stock would have been lost between 69 and 28 B.C. This tallies well with the transfer of some 250,000–300,000 adult settlers and soldiers to the provinces from the 40s to the 208 B.C. (see below, Section III), and makes it unnecessary to posit any significant natural decrease in this period (contra Lo Cascio, op. cit. (n. 8, 1994a), 37).

58 Cornell, T. J., The Beginnings of Rome (1995), 381 tab. 9, adapted from Afzelius, op. cit. (n. 13), 133Google Scholar.

59 Cornell, op. cit. (n. 58), 380. Only a small proportion of them were settled in the six to eight citizen colonies founded in that period: cf. Salmon, E. T., Roman Colonization under the Republic (1970), 70–81Google Scholar.

60 Brunt, op. cit. (n. 6), 29; compare Cornell, op. cit. (n. 58), 367.

61 Brunt, op. cit. (n. 6), 70 n. 1, 279.

62 Salmon, op. cit. (n. 59), 95–111.

63 Brunt, op. cit. (n. 6), 342. Salmon, op. cit. (n. 59), 161–3 lists forty-six certain and thirty probable colonies in Italy founded from Sulla to Augustus.

64 Brunt, op. cit. (n. 6), 589–601. Cf. R. MacMullen, Romanization in the Time of Augustus (2000) for updated totals.

65 Brunt, op. cit. (n. 6), 244–61. Colonies based on existing conventus of Roman citizens were rare (6 out of 106), and those wholly comprised of enfranchised peregrini even rarer (at best 2 or 3, if any): ibid., 244, 246.

66 ibid., 262–5.

67 For legionary deployments in A.D. 14, see B. Campbell, War and Society in Imperial Rome 31 BC–AD 284 (2002), 19 (plus the three Varian legions prior to A.D. 9). For the provenance of legionaries in this period, see above, n. 50. Base population: Frier, op. cit. (n. 7), 812, for 25.1 million in the West and 45.5 million in the entire Empire in A.D. 14. I adjust his guesstimates for intervening growth between 48–14 B.C. and A.D. 14.

68 Given a mean life expectancy of about thirty-three years at age seventeen, this implies a mean lifetime probability of relocation of 20 per cent. Cf. below, Sections IV and V.

69 Salmon, op. cit. (n. 59), 79–81, 112–25.

70 cf. Brunt, op. cit. (n. 6), 78–81.

71 ibid., 233 for the total (italics in the original), 209–24 for settlement patterns. The Mithridatian casualty figures are doubtless hugely inflated (224–7).

72 For the ratio of recruits to soldiers, see Scheidel, op. cit. (n. 50), 117–24; for their provenance, 95–6 n. 18.

73 Frank's rival ‘high count’ guess of 400,000 adult males outside peninsular Italy in 90 B.C. (op. cit. (n. 8), 333) produces a NROM of 0.4 per cent for that region for the years from 200 to 90 B.C. if we accept the ‘low’ count scenario, but drops to 0.2 per cent in the context of Frank's own ‘high’ count of 1.6 million adult male citizens in Italy at that time.

74 The literature is large and repetitive. See now Lo Cascio, E., ‘Le procedure di recensus dalla tarda repubblica al tardo antico e il calcolo della popolazione di Roma’, in La Rome impériale: démographie et logistique (1997), 3–76Google Scholar, for the most detailed recent discussion.

75 e.g., Rickman, G., The Corn Supply of Ancient Rome (1980), 157–97.Google Scholar

76 Scheidel, op. cit. (n. 1), Section I.

77 Noy, D., Foreigners at Rome (2000), 15–29.Google Scholar

78 This is broadly in line with existing guesses: e.g., Brunt, op. cit. (n. 6), 384; Morley, op. cit. (n. 31), 39.

79 Rome may not have grown much further during the Principate: comparative evidence for the expansion of other pre-modern capital cities often points to a limited growth spurt followed by stagnation; see W. Scheidel, ‘Creating a metropolis: a comparative demographic perspective’, in W. V. Harris (ed.), Ancient Alexandria (forthcoming).

80 See above, n. 31.

81 Morley, op. cit. (n. 31), 182.

82 Bagnall and Frier, op. cit. (n. 24), 55; Rathbone, op. cit. (n. 47), 124–37.

83 Duncan-Jones, R., The Economy of the Roman Empire (2nd edn, 1982), 262–77Google Scholar. Lo Cascio, op. cit. (n. 8, 1999), 165 is wrong to claim that these records indicate a significantly larger urban population than that proposed by Hopkins and Morley (above, n. 31):in most cases, it is simply impossible to know if these benefactions were restricted to urban residents. Most texts vaguely refer to municipes or populus, whereas only one text specifies universus populus, two others plebs urbana (see below). There is nothing to suggest that free farmers were commonly excluded from the populus, and that universus populus was therefore not just an embellishment of this term.

84 CIL XI.2650 (A.D. 234), with Duncan-Jones, op. cit. (n. 83), 272. A colonial foundation with a sizeable territory, Saturnia was by no means one of the smallest Italian communities. The Younger Pliny's donation of HS 1,866,666 in support of 100 of his freedmen that after their death was to fund a feast for the plebs urbana of Comum (CIL V.5262 = ILS 2927) cannot be used to calculate the size of the latter for the simple reason that the per capita allocation is unknown. Duncan-Jones' suggestion of HS20 yields 4,200–5,000 adult men, for a total of 13–16,000. However, documented per capita rates for ‘the people’ (as opposed to magistrates) could reach as high as HS200 (ibid., 142). More importantly, Duncan-Jones' guess is compatible with Comum's status as one of the ‘major’ Italian cities with a mean (gross) population of 15,000 (Morley, op. cit. (n. 31), 182 but has no bearing on estimates concerning the large majority of towns in Italy.

85 Wrigley, E. A., People, Cities and Wealth (1987), 162Google Scholar: 60% c. 1600, 70% c. 1670, and 68% c. 1700. The inclusion of smaller towns might reduce this proportion to a value closer to one-half. In the late seventeenth century, London was about twenty times as populous as the next-largest city, Norwich (ibid., 160). Rome's unique status appears to have produced a similarly lopsided pattern in Italy.

86 For the servile element, see Scheidel, op. cit. (n. 1), Section I; for aliens, above, n. 78. Morley, op. cit. (n. 31), 159–83, and K. Lomas, ‘Roman imperialism and the city in Italy’, in R. Laurence and J. Berry (eds), Cultural Identity in the Roman Empire (1998), 64–78, deal with different aspects of Italian urbanization.

87 For a judicious discussion, see esp. de Vries, op. cit. (n. 34), 175–98. Annales de Démographie Historique 1990, 5–151 includes several more recent contributions. For older references, see Scheidel, op. cit. (n. 22), 28 n. 106. Galley, C., ‘A model of early modern urban demography’, Economic History Review 48 (1995), 448–69,CrossRefGoogle Scholar stresses the crucial role of the sex ratio and fertility in smaller English cities but allows for substantial excess mortality in London. His finding that low urban sex ratios can cause natural decrease may be relevant here in so far as cities in late Republican Italy attracted women who had lost male relatives: see J. K. Evans, War, Women and Children in Ancient Rome (1991), 114–44.

88 Wrigley, op. cit. (n. 85), 134–7; Jongman, W., ‘Slavery and the growth of Rome. The transformation of Italy in the second and first centuries BCE’, in Edwards, C. and Woolf, G. (eds), Rome the Cosmopolis (2003), 100–22, at 106–9Google Scholar; Morley, op. cit. (n. 31), 43–4, 49–50, and op. cit. (n. 8), 53.

89 W. Scheidel, ‘Germs for Rome’, in Edwards and Woolf, op. cit. (n. 88), 158–76; Sallares, R., Malaria and Rome (2002), 201–34CrossRefGoogle Scholar (Rome), cf. 264–7 (on the relocation of the town of Salapia in Apulia because of malaria).

90 Lo Cascio's polemic against the notion of urban excess mortality in Roman cities in ‘Condizioni igienico-sanitarie e dinamica della popolazione della citta di Roma dall'etá tardorepubblicana at tardoantico’, in Corvisier, J.-N. et al. (eds), Thérapies, médecine et démographie antiques (2001), 37–70Google Scholar, and ‘La population’, Pallas 55 (2001), 179–98,Google Scholar fails to appreciate the probable role of endemic infectious disease. I note in passing that lower urban excess mortality would make it even easier to defend the ‘low’ count attacked by Lo Cascio.

91 Morley, op. cit. (n. 31), 49–50; Jongman, op. cit. (n. 88), 107–8.

92 See already Lo Cascio, op. cit. (n. 12), 117–18.

93 For the annual number of migrants to equal the annual number of transferred live births, the age and sex structure of all migrants would need to be a representative cross-sample of the total population, On the other hand, if all migrants were twenty year olds, their number would be about half that of the underlying number of births. Actual ratios must have fallen in between these extremes.

94 See above, n. 57.

95 Del Panta, op. cit. (n. 39), 515.

96 e.g., Galloway, P. R., ‘A reconstruction of the population of North Italy from 1650 to 1881 using annual inverse projection comparisons to England, France, and Sweden’, European Journal of Population 10 (1994), 223–74CrossRefGoogle ScholarPubMed. The motley sample of historical growth rates in Morley, op. cit. (n. 8), 53 is of questionable relevance; cf. Scheidel, W., ‘The Greek demographic expansion: models and comparisons’, JHS 123 (2003), 120–40,CrossRefGoogle ScholarPubMed esp. 127–8.

97 This is because in the second century B.C. transfer rates were much lower (c. 0.15–0.2 per cent p.a.) for rural non-Roman/Latin Italians.

98 For the latter, see now Rosenstein, N., Rome at War (2004.)Google Scholar.

99 In addition, the latter may arguably have triggered a modest and temporary contraction of the free population of Italy: see above, n. 57.

100 W. Scheidel, ‘Sex ratios and femicide in the ancient Mediterranean world’ (in preparation).

101 Jongman, op. cit. (n. 88), 109.

102 See below, Section V, fig. 1.

103 cf. also below, Section V.

104 Contrast Morley, op. cit. (n. 8), 53, for a growth rate of 0.8 per cent required by ‘high’ count conditions, and Jongman, op. cit. (n. 88), 118, on manumission.

105 Jongman, op. cit. (n. 88), 106 and passim.

106 Scheidel, op. cit. (n. i), Section III.

107 Nicolet, C., The World of the Citizen in Republican Rome (trans. 1980), 2Google Scholar.

108 Hopkins, op. cit. (n. 31), 33–5.

109 Scheidel, op. cit. (n. 50), 93–4.

110 Brunt, op. cit. (n. 6), 200 n. I.

111 Scheidel, op. cit. (n. 50), 95–6 n. 18, based on Forni's study of the provenance of soldiers in epitaphs.

112 For discussion, see above, Sections III and IV. 200–51 B.C.: ¼ of 330,000–420,000 townward transfers plus 50,000–80,000 adult male colonists and 40,000 emigrants from 200–101 B.C. (assuming a Roman/Latin ratio of 3:1), and ¼ of 670,000 townward transfers, 80,000 colonists, and 50,000 emigrants from 100–51 B.C. 50–1 B.C.: ¼ of 465,000–590,000 townward transfers plus 510,000 colonists and soldiers (340,000 of them emigrants), A.D. 1–100: ¼ of 900,000 townward transfers and 315,000 soldiers. In the category ‘from overseas’, I add a guesstimate of 2,000 annual manumissions of foreign-born male slaves (for ⅓to ¼ of all male manumissions: cf. Scheidel, op. cit. (n. 1) for totals) since 50 B.C., and half as many before. The actual rate may well have been significantly higher.

113 See Fig. 1, assuming that most male migrants were iuniores. This estimate does not account for private rural migration and may therefore still be too low.

114 Tilly, C., Coercion, Capital, and European States, AD 990–1992 (rev. edn 1992).Google Scholar

115 I hope to address this deficit in my current project on Coercion, Capital, and Ancient Mediterranean States.

116 cf. H. Mouritsen, Italian Unification (1998) for the previous lack of integration.

117 Marius' attempt to obtain land for his veterans is consistent with the principle. The Gracchan programme need not have been particularly sweeping:see above, n. 70.

118 Mattingly, J., ‘Vulgar and weak “Romanization”, or time for a paradigm shift?’, JRA 15 (2002), 537–8Google Scholar lists nine different terms that have been used in recent work.

119 Keay, S. and Terrenato, N. (eds), Italy and the West (2001) is representative.Google Scholar

120 G. Woolf, Becoming Roman (1999) offers the most detailed survey of the numerous facets of emulative culture change in a Roman province. In this context, the notion of a ‘complementary’ Roman identity is also useful (G. Bradley, Ancient Umbria (2000), 268), in that it allows us to speak of ‘Romanization’ without envisioning it as an exclusive process.

121 M. W. Doyle, Empires (1986), 128–38.

122 For examples of the scarcity of emulative culture change in several provinces after conquest but before mass colonization, see P. van Dommelen, ‘Cultural imaginings: Punic tradition and local identity in Roman Republican Sardinia’, in Keay and Terrenato, op. cit. (n. 119), 68–84; S. Keay, ‘Romanization and the Hispaniae’, ibid., 117–44.

123 See below, n. 138, and Lee, J. Z. and Feng, W., One Quarter of Humanity (1999), 117–19.Google Scholar

124 Thus Woolf, G., ‘Becoming Roman, staying Greek: culture, identity and the civilizing process in the Roman East’, PCPS 40 (1994), 116–43,Google Scholar esp. 131.

125 Wallace-Hadrill, A., ‘Rome's cultural revolution’, JRS 79 (1989), 157–64;Google Scholar G. Woolf, ‘The Roman cultural revolution in Gaul’, in Keay and Terrenato, op. cit. (n. 119), 173–86, esp. 175–6.

126 For the latter, e.g., J. P. Vallat, ‘The Romanization of Italy’, in Keay and Terrenato, op. cit. (n. 119), 109. Mass enfranchisement was limited to Italy; Spain, Sardinia, and Africa were ‘old’ provinces, while Gaul was not; colonization may be the only shared variable.

127 Curti, E. et al. , ‘The archaeology of central and southern Roman Italy: recent trends and approaches’, JRS 86 (1996), 173–5, 185–8Google Scholar (quote: 186). See also p. 188 on the absence of visible ‘Roman’ material culture prior to the second century B.C.

128 E. Benelli, ‘The Romanization of Italy through the epigraphic record’, in Keay and Terrenato, op. cit. (n. 119), 8.

129 Most recently, Bradley, op. cit. (n. 120), 193;J. H. C. Williams, Beyond the Rubicon (2001), 215–16 (Po valley). MacMullen, op. cit. (n. 64) puts the most emphasis on colonization but does not explicitly argue for a causal connection between settlers and culture change.

130 N. Terrenato, ‘A tale of three cities: the Romanization of northern coastal Etruria’, in Keay and Terrenato, op. cit. (n. 119), 54–67.

131 Proper consideration of mobility may require an adjustment of the kinship simulations in Saller, R. P., Patriarchy, Property and Death in the Roman Family (1994), 43–69CrossRefGoogle Scholar, to approximate the actual (i.e., lower) accessibility of relatives. Ex-slaves and veterans ought to have been particularly strongly affected.

132 These similarities will be explored by the Stanford Ancient Chinese and Mediterranean Empires Comparative History Project (www.stanford.edu/~scheidel/acme.htm).

133 Finley, M. I., Economy and Society in Ancient Greece (1981), 51–2Google Scholar; Hansen, op. cit. (n. 26), 14–28 (population). This entails the assumption that all kleruchs physically moved to their new possessions, which may not have been the case.

134 Rathbone, op. cit. (n. 47), 113.

135 For sex ratios, see W. Clarysse and D. Thompson, Counting the People (in press).

136 G. M. Cohen, The Seleucid Colonies (1978), 14–19 lists c. 70 of ‘the more important Seleucid settlements’; cf. 30–2 on the ethnic composition of their inhabitants. J. D. Grainger, The Cities of Seleukid Syria (1990), 95–100 reckons with a minimum of 60,000 adult male settlers in the ten principal foundations of Seleukos I but allows for an overall total of up to half a million. Cf. R. A. Billows, Kings and Colonists (1995) 155, 306–9 for constraints on Macedonian emigration after Alexander.

137 In op. cit. (n. 96), 131–5, I estimate a NROM of about 0.05–0.1 per cent for mainland Greek overseas settlement from 750 to 600 B.C. The overall mean for the entire Aegean would be much lower.

138 Lee, J., ‘Migration and expansion in Chinese history’, in McNeill, W. H. and Adams, R. S. (eds), Human Migration (1978), 20–43, esp. 21–5.Google Scholar

139 Horden, P. and Purcell, N., The Corrupting Sea (2000), 377–91Google Scholar.

140 See above, at n. 123.

141 Diamond, J., Guns, Germs and Steel (1997)Google Scholar shows that overall levels of civilizational development are contingent on ecological conditions. However, the gradual extension of imperial state formation from the Fertile Crescent to the west and north-west suggests that there was nothing essentially ‘Mediterranean’ about the Roman Empire: large territorial states could emerge anywhere in the temperate zones of Eurasia.