¹ Gsell, S. and Joly, A., Khamissa, Mdaourouch, Announa II, ‘Mdaourouch’ (1922), 18 ffGoogle Scholar. I have been indebted for advice to Mr. N. H. Carrier, Miss K. Duncan-Jones, Mr. M. I. Finley, Professor D. V. Glass, Monsieur J. Guey, Mr. M. K. Hopkins, Mr. A. W. James, and Professor A. H. M. Jones; but responsibility for the views expressed here is entirely my own.

² Grimal, P., ‘Les fouilles de Siga,’ Mélanges de l'École française à Rome 1937, 117.Google Scholar

³ Courtois, C., Les Vandales et l'Afrique (1955), 104 ff.Google Scholar

⁴ -Picard, G. C., La Civilisation de l'Afrique romaine (1959). 44–59.Google Scholar

⁵ cf. for instance Haywood, R. M. in Frank, T., Economic Survey of Ancient Rome IV (1938), 112Google Scholar. Another survey of African population is given by Russell, J. C., ‘Late Ancient and Mediaeval Population,’ Trans. Amer. Philos. Soc. XLVIII, 3 (1958)Google Scholar, 67 and 76; but this discussion contains too many errors of fact and method to be considered seriously.

⁶ It would be mistaken to contend that there was no noticeable internal trade in essential foodstuffs; for both grain and oil were needed everywhere, yet many cities grew only one of these crops on their own territories in any abundance (cf. Picard, o.c. (n. 4), 74).

⁷ See the criticisms made by Picard, o.c. (n. 4), 47.

⁸ These strictures of course do not apply to estimates for towns like Pompeii and Ostia, cities with a substantial urban population, whose miraculous state of preservation allows calculations of this type to be made with some prospect of success, cf. Meiggs, R., Roman Ostia (1960), 532 ff.Google Scholar

⁹ It is also difficult to see any demographic logic behind the fact that at a given town the amphitheatre usually seats considerably more people than the theatre and the circus sometimes a larger number still.

¹⁰ A rural analogy is provided by the letter in which the younger Pliny describes his plan to enlarge a temple on one of his estates, adding a separate portico for the use of the country people who come to visit the shrine, both to make transactions there and to sacrifice (Ep. IX, 39).

¹¹ This generalization applies only to the general run of secondary towns, which formed the majority. The really large towns, where private residences could be numbered in thousands rather than in hundreds (as well as some smaller coastal cities), were often closer to modern towns in their density of population.

¹² CIL VIII, 967 + 12448.

¹³ The last phrase is a direct quotation from the will; cf. ILS 6818.

¹⁴ In ‘Costs, outlays and summae honorariae from Roman Africa’, Papers of the British School at Rome XXX (1962), (hereafter referred to as CSRA), no. 259 + appendix, p. 114. One feature was omitted from the account given in that article: the donor must have left money to pay for the statue to whose dedication he refers, at least judging by the analogy of CSRA no. 262. The total bequest thus probably amounted to more than HS 50,000; all the parallels suggest that the total would have been a round sum nonetheless, probably HS 60,000, the next obvious possibility above HS 50,000.

¹⁵ The authority behind the text is good: the squeeze was seen by Wilmanns and Schmidt and Mommsen was responsible for part of the restoration.

¹⁶ Some no doubt will conjecture that ‘civibus’ here denotes the whole citizen population, men, women and children. The ample surviving evidence for ephemeral distributions in Africa (see CIL VIII, supp. V, pp. 339–340) does not support this view. I can find no African text which overtly suggests that any but male adults were included in such distributions (the serious philanthropy represented by the alimentary schemes for both sexes was something entirely different). In fact the gifts generally have an exclusive character: the cash sportulae, usually of several sesterces per head, tended to be reserved for the small body of decurions, while generosity to the citizens was often confined to gifts of oil (‘gymnasium’, see CSRA n. 151, p. III) or games, whose cost per head probably worked out at a fraction of a sesterce. The donor of the present gift was liberal enough to extend his cash distribution beyond the ordo decurionum, but in the absence of any phrase such as ‘utriusque sexus’ (found in Italy, cf. ILS 6643), it is difficult to think that he included women, still less children, in his bounty.

¹⁷ CSRA nos. 290–305.

¹⁸ Digest L. 15. 4.

¹⁹ CIL V, 4922.

²⁰ Chronological concentrations of gifts are discussed in CSRA Part I, pp. 52–56.

²¹ The annual surpluses known in privately given perpetual foundations in Africa are all very small: 1·6, I, 0·31 and 0·16 per cent (CSRA nos. 258, 257, 248, 250, cf. appendix, pp. 113–14).

²² 1,000 sesterces, the amount that would have been wasted had there been only 3,000 recipients of the gift, was not a negligible sum (cf. CSRA Part I, pp. 64–5).

²³ Eighteen was the age at which decurial obligations began during the period when they were being spread most widely (Cod. Theod. XII, i, 19); the age at which eligibility had begun in the earlier period had been twenty-five (Digest L. 2. II). Legionary recruitment began at seventeen (Daremberg et Saglio, Dictionnaire, s.v. ‘Legio’).

²⁴ Close resemblances between the mortality curve for pre-Christian Roman Africa and that for India in 1901 have been established by A. R. Burn (‘Hic breve vivitur’, Past and Present, November, 1953, 2–31). Mortality alone is not a reliable index, but a factor of 3½ also corresponds closely with what we should expect from the total reproduction rate of 5–6 implied by the African figures for life-expectation compiled by Burn, l.c. It is clear that population in second-century Africa, as in India sixty years ago, was rising (Tertullian, De anima 30). A further resemblance between the two societies lies in the lowness of the age of marriage permitted to girls. (cf. Harkness, A. G., ‘Age at Marriage and at Death in the Roman Empire,’ TAPhA XXVII (1896), 35–72Google Scholar; Harkness is too sceptical of the evidence for early marriages that he collects.) Age-distribution for India in 1901 is given in ‘The Aging of Populations and its economic and social implications’, United Nations Population Studies no. 26 (1956), 112.

²⁵ NH III, 3, 28. There is remarkable consistency between the three average figures, all of which refer to places in the north-west of the Spanish peninsula, part of Hispania Tarraconensis under the Romans. The averages are: 10,909 per community for the Astures, who inhabited part of the modem inland province of Leon (twenty-two populi with 240,000 libera capita in all); 11,066 for the conventus Lucensis, the modern province of Lugo (fifteen populi with 166,000 libera capita); and 11,875 for the Bracares, who belonged to what is now northern Portugal (twenty-four civitates with 285,000 capita).

²⁶ ‘Servus’ and ‘libertus’ in index to CIL VIII.

²⁷ CIL VIII, 22721, cf. ILS 2927.

²⁸ cf. Apuleius, Apol. 17: ‘Ne illud quidem credibile fuisset, cum tribus (servis) venisse, omnis liberasse.’

²⁹ ibid. 24.

³⁰ ibid. 17.

³¹ ibid. 43–5.

³² ibid. 93.

³³ ibid. 44.

³⁴ Gsell gives an excellent discussion, citing other authors besides Apuleius, (‘Esclaves ruraux dans l' Afrique romaine’, Mélanges Glotz I (1932), 395–415).Google Scholar

³⁵ Apuleius, , Apol. 23, 75, 77.Google Scholar

³⁶ Gsell (o.c, n. 34) comes to the same conclusion.

³⁷ Compare slave-prices listed by Jones, A. H. M. in Slavery in Classical Antiquity (ed. Finley, M. I., 1960), 9–10Google Scholar, with stele-costs in CSRA nos. 226–244. The laws of the funeral college at Lanuvium allowed for the possibility that its slave-members might receive ‘iniqu(a) sepultura’ from their masters, ILS 7212, pag. II, 1.4.

³⁸ These were the resources of Pudentilla, wife of Apuleius, , Apol. 77, 93.Google Scholar

³⁹ Galen could assume that there were as many slaves as male citizens at Pergamum in the midsecond century, De cognoscendis curandisque animi morbis 9 = V 49 Kühn.

⁴⁰ Notices of Siagu: Guérin, V., Voyage en Tunisie (1862) II, 259–261Google Scholar; Tissot, C., Géographie comparée de la province romaine d'Afrique II (1888), 129–131Google Scholar; Babelon, , Cagnat, , Reinach, , Atlas archéologique de la Tunisie (1893–1926), fe. 37, 3Google Scholar (this is the correct map-reference; the commentary is given under fe. 37, 4); Gauckler, P., Enquête sur les installations hydrauliques romaines en Tunisie (1897–1902), 233–8Google Scholar, plan 234; Poinssot, L. in Atlas historique, géographique, économique et touristique de la Tunisie (ed. Leconte, C., Despois, J., Garbe, G., Gérard, F., 1936), 34Google Scholar; Parisot, M., Tunisie (Guides Bleu, 1955), 76.Google Scholar Constructions listed in these works include baths, three aqueducts, a nymphaeum, a large reservoir, a number of wells, twenty-five private cisterns, a Christian basilica and a Byzantine fortress.

⁴¹ Pupput, which lies 4½ km. to the south-west, had a territory of about 40 sq. km. (Seston, W. in Bull. arch. du Com. des trav. hist. 1946–1949, 309–311).Google Scholar Neapolis lies 13 km. north-west of Siagu, Vina 10 km. north by west. All these towns were politically more important than Siagu, the first two being colonies, the third a municipium. Smaller neighbours were Thinissut, 4 km. north by east of Siagu, and Tubernuc, 15 km. to the north-west. See inset ‘Les grandes carrefours de la région de Carthage’ in map in P. Salama, Les Voies romaines de l'Afrique du nord (1951).

⁴² If the buildings to the west (marked G on the plan and referred to in the text as important remains) were part of the main complex.

⁴³ 50 hectares, Courtois, C., Timgad, antique Thamugadi (1951), 19.Google Scholar

⁴⁴ 60 hectares (see n. I above).

⁴⁵ About 12 hectares from plan in L. Leschi, Djemila, antique Cuicul (1953).

⁴⁶ 20 hectares (see n. I above).

⁴⁷ Roughly 20 hectares of main area from plan in C. Poinssot, Les Ruines de Dougga (1958).

⁴⁸ Muzuc (Henchir Khachoun) had an area of at least 15 hectares, Tissot (o.c, n. 40), II, 603.

⁴⁹ 500 hectares, Gauckler (o.c, n. 40), but the plan on p. 154 suggests 300 hectares as a more likely figure.

⁵⁰ 400 hectares, Tissot (o.c, n. 40), II, 484.

⁵¹ ‘200’ hectares, Gauckler (o.c, n. 40), 144–6; but again the plan (p. 146) suggests a lower figure, 150 hectares.

⁵² At least 120 hectares from plan in Babelon (o.c, n. 40), fe. 66, 7.

⁵³ Over 100 hectares, plan in Gauckler (o.c, n. 40), 259.

⁵⁴ By -Picard, G. C., La Civilisation de l'Afrique romaine (1959), 178.Google Scholar

⁵⁵ The town was still a civitas in A.D. 215, when the main series of promotions of African cities was almost over (CIL VIII, 966); there are no inscriptions dateable to later than the third century (though the total number that have been found is very small); and no bishop of Siagu appears in the abundant Christian lists for Africa that run from the third to the seventh centuries A.D. Yet the site cannot have been completely abandoned in antiquity if the Byzantines thought it worth fortifying in the sixth century.

⁵⁶ CIL V, 4922.

⁵⁷ Σιαγούλ Ptolemy IV, 3; conjunction with Neapolis in Ptolemy's list makes the identification certain.