¹ Hopkins, K., Conquerors and Slaves (1978), 9Google Scholar.

² For a wide-ranging discussion see Frederiksen, M. W., ‘The contribution of archaeology to the agrarian problem in the Gracchan period’, in DdA (1970/1), 330–57Google Scholar.

³ The same point is made by Carandini, A., L'Anatomia delta Scimmia (1979), 226Google Scholar.

⁴ Pansiot, F. P. and Rebour, H., Improvement in Olive Cultivation (Food and Agriculture Organization of the U.N., Agricultural Studies no. 50) (1961), 214Google Scholar.

⁵ See for examples of this type of approach Billiard, R., La Vigne dans l'Antiquité (1913)Google Scholar; White, K. D., ‘The productivity of labour in Roman agriculture’, in Antiquity 39 (1965), 102–7CrossRefGoogle Scholar.

⁶ P. Garnsey, ‘Non-slave labour in the Roman world’ and J. E. Skydsgaard, ‘Non-slave labour in rural Italy during the late Republic,’ both in Garnsey, P. (ed.), Non-slave Labour in the Greco-Roman World (CPhS suppl. vol. 6) (1980)Google Scholar.

⁷ Dyson, S. L., ‘Settlement patterns in the ‘Ager Cosanus’: the Wesleyan University survey, 1974–1976’, in JFA 5 (1978), 251–68Google Scholar.

⁸ Carandini, A. and Settis, S., Schiavi e Padroni nell' Etruria Romana. La Villa di Settefinestre dallo Scavo alia Mostra (1979)Google Scholar. This also contains a useful bibliography relating to the ager Cosanus’ in general.

⁹ There is not space here to argue for my interpretation of Cato's figures but I hope to publish a separate paper on this topic. For the early twentieth-century A.D. average see Duncan-Jones, R., The Economy of the Roman Empire (1974), 45.Google Scholar

¹⁰ I ‘vilicus’, I ‘vilica’, 13 vineyard workers, 2 for the olives, 1 swineherd, 1 shepherd and 1 for luck. As yet we cannot be sure about the location or number of the slave quarters in the villa, and thus this estimate lacks archaeological confirmation.

¹¹ I here assume that in the villa's prime its most important commercial products were wine and olive oil. The results of the 1980 season of excavation should illuminate greatly the role of animal husbandry in the economy of the villa, but although they may make qualification of my assumption necessary, it will almost certainly remain true that the extent of the estate and the labour-demand of the villa related primarily to its cultivation of the vine and the olive.

¹² Such as can be found in White, K. D., Roman Farming (1970)Google Scholar and in Duncan-Jones, op. cit. (n. 9). Obviously these figures are open to various doubts, and calculations based on them can have only provisional validity pending a comprehensive study of Roman agricultural statistics.

¹³ Varro, RR 1.50.3; Columella 2.12.11.

¹⁴ Exactitude is impossible to attain with this type of figure and one can only play for safety, that is for a minimum free labour requirement. Full discussion of the relevant statistics is impossible here, but take, as one example, the estimate of Dalmasso, G., ‘Problemi economici di agricoltura astigiana’, in Ann. R. Accad. Agric. Torino 53 (1910), 194Google Scholar f., that I hectare of ‘vigneto specializzato’ required for the harvest between 25 and 32 man/woman-days’ labour, equivalent to from 625 to 800 man-days for 100 iugera. This figure, which is comparable to other Italian and Central European statistics, should be at least doubled to allow for the other operations, especially the extremely laborious pressing process, and then a maximum of 400 man-days deducted from the total to allow for work done by the 13 slaves.

¹⁵ Excluding the ‘vilicus’ and ‘vilica’, see n. 10 above. Columella's estimate that 200 iugera bearing grain and legumes require the permanent employment of 8 men (2.12.7), on the basis that 1 iugerum of grainland requires a labour input of 10½ man-days per annum (2.12.1), implies an expected working year of 262½ days.

¹⁶ This figure may seem high but compare, for example, a large Hungarian estate of the early twentieth century A.D. with mixed cultivation, where casual labour made up from 28% to 44% of the total annual labour input—figures from Warriner, D., Economics of Peasant Farming ² (1964), 148.Google Scholar In fact the figure for the Settefinestre model should be even higher since I have here taken no account (largely because of difficulties of quantification) of the casual labour employed for other operations such as pruning the vines and olive trees.

¹⁷ All values in the calculations are expressed in sesterces. The assumed land price of HS 1,000 per iugerum is Columella's price for undeveloped land, thought to be much too high by Duncan-Jones, op. cit. (n. 9), 48–52. The assumed price for wine of HS 10 per amphora is probably nearer reality; cf. Duncan-Jones, op. cit., 46–8. For the average slave purchase price of HS 2,000 see Duncan-Jones, op. cit., 50 n. 2 and Appendix 10. Amortization is calculated over 20 years following Hopkins, op. cit. (n. 1), 110 n. 23. I base my assumption that the average hired free labourer was paid HS 2 per diem on Cato 22.3 (cf. Duncan-Jones, op. cit., 54). For the wine yield I use the early twentieth-century A.D. average of 1.17 cullei per iugerum (cf. p. 12 above).

¹⁸ For the purpose of my argument I need only show that I have not assumed a proportionately higher cost for slave than for free labour. Obviously from time to time in the Republic the influx of war captives will have caused the average price of slaves to fall sensibly, but for this general model such unquantifiable fluctuations are fairly irrelevant (although would one not in any case expect that a fall in the cost of slave labour would tend to reduce the demand for and hence the cost of free labour?). I see no reason why agricultural slaves should have normally cost less than the average; Columella allowed HS 6–8,000 for the purchase of a decent ‘vinitor’, a figure perhaps intended to shock, but unlikely to be over three times the usual outlay. The use of home-bred slaves was a possibility, but it is dubious whether they were really cheaper than bought slaves; Duncan-Jones, op. cit. (n. 9), 50 thinks so, but the agronomists' comments do not prove his point (do they rather indicate the difficulty of making home-breeding pay?); Wallon, H., Histoire de l'Esclavage dans l'Antiquité (Paris, 1879) I, 158,Google Scholar esp. n. 3 quotes figures for French plantations in Guadeloupe where rearing slaves was many times more expensive than buying them. Clearly much depended on local factors, but I think it is reasonable to skip a tricky attempt to quantify this possibility here. The pricing of 1 man-day of free agricultural labour at HS 2 is certainly generous; in Cato 22.3 this figure includes the hire of oxen. Cicero, pro Rose. Com. 28 gives HS 3 as the maximum conceivable daily pay for unskilled labour in Rome, where labour costs presumably tended to be higher than in the countryside. Compare too the rate of pay for legionaries in the Republic, which after Caesar's doubling of it was HS 900 per annum.

¹⁹ See n. 15 above. This may seem low, but presumably even slaves were underemployed in the winter (or kept occupied with non-essential unproductive tasks), and we must allow for the time they spent producing crops for their own consumption.

²⁰ For the purchase price of slaves see n. 17 and n. 18 above. the calculation here also takes account of the annual cost of upkeep of the slave. To determine this one can take the low allowance for boys in Trajan's alimentary scheme of HS 16 per mensem (cf. Duncan-Jones, op. cit. (n. 9), 144) and increase it by 50% to allow for the slaves being men. Or one can cost out the allowances in Cato 56 f.: 50 modii of wheat @ HS 4 (cf. Duncan-Jones, 145 f.) = HS 200, 7 amphorae of wine @ HS 8 (cf. Duncan-Jones, 46 f.) = HS 56, clothes etc. at say HS 50. Both work out at around HS 300 per annum, the figure used here.

²¹ See, for example, Hopkins, op. cit. (n. 1), 106 f.; Carandini and Settis, op. cit. (n. 8), 39 f.

²² Pansiot and Rebour, op. cit. (n. 4), 215.

²³ The literary sources contain a fair number of references to the hiring of casual free labour at the peaks of labour demand in the agricultural year, but there is not a single reference, to my knowledge, to recruitment of this labour from the urban poor. Martin, R., ‘«Familia rustica»: les esclaves chez les agronomes latins’, in Actes du Collogue 1972 sur l'Esclavage (1974), 269Google Scholar, takes Pliny, Ep. IX. 20.2 to mean that Pliny used his ‘familia urbana’ instead of ‘mercenarii’ for the vintage, but the passage in fact shows he had brought some of his ‘familia urbana’ to supervise the ‘rustici’ at work. Cato's advice (RR 4) ‘Vicinis bonus esto … Si te libenter vicinitas videbit … operarios facilius conduces’ implies very local hiring of this extra labour.

²⁴ Cardarelli, R., ‘Confini fra Orbetello e Marsiliana; fra Port’ Ercole e Monte Argentario (28 dicembre 1508–2 marzo 1510)’, in Maremma (Bollettino delta Societa storica maremmana) 1 (1924), 131–42, 155–86 and 205–24;Google Scholar 2 (1925), 3–36, 75–128 and 147–213.

²⁵ See Livy 39.55.9; Pliny, NH 3.52.

²⁶ This figure is derived from the map. It is possible that the lower hills in the territory which are today covered with ‘macchia’ and are of minimal agricultural worth were then suitable for some arable farming, especially if terraced, but it is more likely that the greater part of them was left uncenturiated as communal ‘ager compascuus’.

²⁷ The other Latin colonies (with their dates) for which we know the number of colonists are Cales (334) 2,500. Luceria (314) 2,500, Interamna (312) 4,000, Sora (303) 4,000, Carseoli (298) 4,000, and Cremona and Placentia (218) 6,000 each.

²⁸ F. Castagnoli, ‘La centuriazione di Cosa’, in MAAR 24 (1956), 147–65.

²⁹ On the size of the blocks within centuriated systems see Castagnoli, F., Ricerche sui Resti della Centuriazione (1958), 24 f.Google Scholar No good evidence for the size of allotments at Latin colonies is available until the second century B.C., when they were often very generous.

³⁰ It is clear that the city of Cosa can have accommodated only a small proportion of its original colonists. For a general discussion see P. D. A. Garnsey, ‘Where did Italian peasants live?’ in PCPhS n.s. 25 (1979), esp. 13–15.

³¹ Dyson, op. cit. (n. 7), 259.

³² Livy (29.15.5) says that in 204 B.C. the Roman Senate summoned to Rome the magistrates and the ten leading citizens (‘principes’) of each of twelve defaulting Latin colonies, which implies some previous socio-economic differentiation. After the Hannibalic War we can see the Roman state implanting this class differentiation ready-made into new colonies at their foundation, the evidence being Livy's accounts of the foundation of Copia in 193 B.C., Vibo Valentia in 192 B.C., Bononia in 189 B.C. and Aquileia in 181 B.C. (35.9.7–9; 35.40.5f.; 37.57.7f.; 40.34.2). Copia had 300 equites to 3,000 pedites, Vibo 300 equites to 3,700 pedites. At Copia and Vibo the equites received allotments twice the size of those of the pedites; at Bononia the equites received 70 iugera, the pedites 50 iugera; at Aquileia the equites received 140 iugera, the centurions 100, and the pedites 50. According to Salmon, E.T., Roman Colonization under the Republic (1970), 25Google Scholar, larger allotments for equites only became normal after the Hannibalic War, but this is an argument from silence. The magisterial class had probably always had some economic privileges, although we can suppose that official recognition of what had become the customary scale of wealth-differentiation had lagged somewhat behind the actual development. The estimate of 6% ( = 1:16) for the wealthy class is derived from Polybius' figure (2.24.10) for the total Latin military manpower in 225 B.C.: 5,000 equites and 80,000 pedites.

³³ cf. Brunt, P. A., Italian Manpower 225 B.C.–A.D. 14 (1971), 57Google Scholar: ‘it looks as if there was little or no natural increase in the population of Latin cities after their foundations’.

³⁴ Harris, W. V., War and Imperialism in Republican Rome 327–70 B.C. (1979), 44Google Scholar, estimates a norm of six or seven.

³⁵ cf. Brunt, op. cit. (n. 33), 84: this war ‘had surely brought about a decline in Latin population at least proportionate to that of 17 per cent among Roman citizens, perhaps as much as 20 per cent, and probably greater’.

³⁶ Latin colonies were closed societies in that the recruitment of new citizens was outside their normal powers: Cosa, for example, required the Senate's special permission in 190—197 B.C. The obvious comparison here is the Spartan state.

³⁷ Harris, op. cit. (n. 34), 44–8. See too the conclusion of Skydsgaard, op. cit. (n. 6), 69—71, that recruitment under the Republic suggests that many of the rural poor found fighting a more attractive proposition than farming.

³⁸ In the second century B.C. the average ‘assiduus’ probably served for from twelve to fourteen seasons, Harris, loc. cit.

³⁹ One might compare the effect on English agriculture of the Black Death, which is estimated to have killed up to 50% of the population but even so was by no means the sole cause of subsequent developments; see, for example, E. Miller and Hatcher, J., Medieval England. Rural Society and Economic Change 1086–1348 (1978), ch. 9, e.g. 240Google Scholar: ‘even the fullest appreciation of the immediate and longer-term consequences of the Black Death does not necessarily mean that we must discount the effects of circumstances a generation and more earlier’.

⁴⁰ Brown, F. E., ‘Cosa II. the temples of the Arx’, in MAAR 26 (1960), 43 fGoogle Scholar. et passim; Plutarch, Tiberius Gracchus 8; Marius 41.2.

⁴¹ Carandini and Settis, op. cit. (n. 8), Pannelli 5 to 7.

⁴² The average ‘small site’ in the ‘ager Cosanus’ is not prolific of material. To quote Dyson, op. cit. (n. 7), 259: ‘The sample of datable material was generally small, often fewer than 10 sherds. Generally the condition of the sherds was poor making precise placement within the black-glaze chronology impossible ’. Since hardly any Roman ‘small sites’ in the whole of Italy have been excavated for comparison with field survey results, it must be a matter of faith to claim that a site with say ten black-glaze sherds and no Arretine was abandoned by the imperial era. For example, we should allow for fluctuating availability of any one indicator in both economic and chronological terms: a ‘decline’ in the number of sites with ‘type A’ pottery to the number of those with the later ‘type B’ could signify that ‘type B’ was less available (i.e. available for a shorter period of time or more expensive), not that there were really fewer occupied sites in the later period. On the other hand, the presence of both black-glaze and Arretine could mask a whole series of abandonments and reoccupations. At best this evidence provides dating parameters, not actual dates or even termini post/ante quem.

⁴³ To take one example, out of Dyson's 68 type ‘C’ and ‘D’ sites which date to the Republic and/or early Empire, 31 date to the Republic only, 28 bridge both periods and 9 date no earlier than the early Empire. Thus the statistics of an overall decline from 59 to 37 small sites would conceal the demise of 9 Republican sites—but also the birth of 9 new sites in the early Empire, almost 25% of all the small sites in that period (calculations based on the summary list in Dyson, op. cit. (n. 7), 266–8; I am grateful to Professor Dyson for making available to me the more detailed inventory from which this list is taken).

⁴⁴ See, for example, Macfarlane, A., The Origins of English Individualism (1978)Google Scholar, on mediaeval English peasants.

⁴⁵ This figure relates to my estimate for Settefinestre (p. 12 above). Settefinestre itself may have had a larger estate, but the visible remains of around fifteen other villas suggest it was one of the largest villas in the territory.

⁴⁶ The equites will always have had greater capability and inducements for emigration (especially to Rome) than the peasantry, and thus their numbers are more likely to have suffered drastic contraction. An internal agglomeration of landed wealth in most Italian cities is implied by the late Republican evidence for the emergence of a fairly high property qualification for decurions; see Duncan-Jones, op. cit. (n. 9), 147 and 243.

⁴⁷ Information kindly provided by M. G. Celuzza and E. Regoli. Two models of development are possible: (1) expansion of the original centre, as appears to be the case with the yia Gabina site II villa (see W. M. Widrig, ‘Two sites on the ancient Via Gabina’, in Brit. Mus. Occas. Papers no. 24)—in this category we should perhaps put ‘Le Colonne’, of which the visible remains date to the 60s B.C., but it occupies an intersection of the centuriation which suggests an occupation dating back to 273 B.C.; (2) change of centre—perhaps Settefinestre, built c. 70–60 B.C., replaced the earlier Republican villa in the valley below it as the centre of the same estate.

⁴⁸ Hopkins, op. cit. (n. 1), 4.

⁴⁹ Assuming that agricultural slaves comprised half of the two million slave population estimated for Italy in the first century B.C. by Crawford, M. H., ‘Republican denarii in Romania: the suppression of piracy and the slave-trade’, in JRS 67 (1977), 123Google Scholar.

⁵⁰ Harris, op. cit. (n. 34), 59.