¹ Harris, W. V., ‘The Theoretical Possibility of Extensive Infanticide in the Graeco-Roman World’, CQ 32 (1982), 114–16CrossRefGoogle ScholarPubMed; cf. Golden, M., ‘Demography and the Exposure of Girls at Athens’, Phoenix 35 (1981), 316–31CrossRefGoogle Scholar. Probably most infants who died from exposure were deformed and many of them would have died in infancy. Statements in Aristotle (Pol. 1335b), the Twelve Tables (quoted by Cicero, , Leg. 3. 19Google Scholar), and the so-called ‘Law of Romulus’ (cited by Dion. Hal. 2. 15. 2), indicate that, in general, the practice of exposure was restricted to deformed infants. The exposure of such infants, given preindustrial conditions of mortality, would not have substantially raised the overall death rate of the population. The issue is, to what extent were healthy children fatally exposed? In a previous article (Engels, D., ‘The Problem of Female Infanticide in the Greco-Roman World’, CPh 75 [1980], 112–20Google ScholarPubMed) I showed that earlier scholars had established that the practice of infanticide was of negligible importance and that more recent attempts to show the contrary were not convincing. The recent scholarship I cited included the excellent and highly respected book, Pomeroy, S., Goddesses, Whores, Wives, and Slaves: Women in Classical Antiquity (New York, 1975)Google Scholar, which incorporates the views of much recent research.

² For the demographic transition see e.g. De Jong, G. F., ‘Patterns of Human Fertility and Mortality’, in Harrison, G. A. and Boyce, A. J., The Structure of Human Populations (Oxford, 1972), 39–51Google Scholar.

³ Engels, D., op. cit., 112Google Scholar For the use of stable population models in historical demography see Hollingsworth, T. H., Historical Demography (Ithaca, 1969), 339–53Google Scholar. Cf. Aristotle, , Politics 1265aGoogle Scholar: regulation of population would be unnecessary in the ideal state because known states maintain level population growth (⋯νομαλιθησομ⋯νην εἰς τ⋯ αὐτ⋯ πλ⋯θος).

⁴ In this respect, the Greco-Roman population probably resembles that of other pre-transitional societies: Hopkins, K., ‘On the Probable Age Structure of the Roman Population’, Population Studies 19 (1966), 256Google Scholar; Durand, J., ‘A Long-Range View of World Population Growth’, Annals of the American Academy of Political and Social Science 369 (1967), 6CrossRefGoogle Scholar; Petersen, W., ‘A Demographer's View of Prehistoric Demography’, Current Anthropology 16 (1975), 232CrossRefGoogle Scholar; Coale, A., ‘The History of the Human Population’, Scientific American 231 (09, 1974), 43–4CrossRefGoogle ScholarPubMed; and below notes 5–11.

⁵ Russell, J. C., ‘Population in Europe, 500–1500’, in Cipolla, C. M., ed., The Fontana Economic History of Europe (London, 1972), 36Google Scholar. The four exceptions are France, the Low Countries, and Germany from A.D. 1000 to 1340 and the British Isles from A.D. 650 to 1000, which seem to have grown at a rate of 3/1,000/year. Nevertheless, the growth rate of these and other European countries averaged far less than 2/1,000/year throughout the era. Cf. Wrigley, E. A., Population and History (New York, 1969), 23, 111–13Google Scholar; Durand, J. D., ‘Historical Estimates of World Population: An Evaluation’, Population and Development Review 3. 3 (1977), 253–74CrossRefGoogle Scholar.

⁶ Russell, J. C., ‘Late Ancient and Medieval Population’, Trans. Amer. Philosoph. Soc. 48. 3 (1958), 19–21Google Scholar; Wrigley, , op. cit. 54, 62Google Scholar. This is probably the maximum for women in pre-industrial societies. Certain modern groups such as the Hutterites can now attain a higher rate than their preindustrial counterparts because the rates of mortality among husbands and wives of childrearing age, specific mortality among women because of medical complications arising from childbirth, and perinatal mortality among infants, are higher in pre-industrial societies than in modern populations. For these reasons, John Durand informed me by personal communication in 1975 that for an ancient society to attain a birth rate as high as 50 per 1,000 per year may require the abandonment of the institutions of marriage and the family by that society.

⁷ Herlihy, D., ‘Life Expectancies for Women in Medieval Society’, in Morewedge, R. T., ed., The Role of Women in the Middle Ages (Albany, 1975), 6–7Google Scholar; Russell, J. C., ‘Population in Europe, 500–1500’, in Cipolla, C. M., ed., The Fontana Economic History of Europe (London, 1972), 57–61Google Scholar (although his evidence must be used with caution since it comes from tombstones); Russell, J. C., ‘Late Ancient and Medieval Population’, Trans. Amer. Philosoph. Soc. 48. 3 (1958), 13–17Google Scholar.

⁸ Mols, R., Introduction à la démographie historique des villes d'Europe du XIV au XVIII siècle, Vol. 2 (Louvain, 1955), 183–222Google Scholar; Herlihy, , op. cit., 12–13Google Scholar; Russell, , Late Ancient, 16–17Google Scholar.

⁹ Coleman, E. R., ‘L'infanticide, dans le haut moyen âge’, Annales ESC 29 (1974), 315–35Google Scholar; Dickemann, M., ‘Female Infanticide, Reproductive Strategies, and Social Stratification: A Preliminary Model’, in Chagnon, N. A. and Irons, W., eds., Evolutionary Biology and Human Social Behavior (North Scituate, Mass., 1979), 360–1Google Scholar.

¹⁰ Wrigley, E. A., op. cit. 47, 124–5Google Scholar; Dickemann, , op. cit. 352, 356Google Scholar. Rates of infanticide seem to have increased in Europe during the demographic transition, as in China.

¹¹ Ho, Ping-ti, Studies on the Population of China, 1368–1953 (Cambridge, Mass., 1959), 3–35CrossRefGoogle Scholar. Cf. Croll, E., Feminism and Socialism in China (Boston, 1975), 23–5Google Scholar; Dickemann, , op. cit. 341–50Google Scholar.

¹² Harris, , op. cit. 115Google Scholar also believes that a population with a life expectancy at birth of less than 25 years might have a birth rate of 40/1,000/year and a death rate of 36/1,000/year. However, For these factors, it is best to consult a life table. For example, Coale, A. J. and Demeny, P., Regional Model Life Tables and Stable Populations (Princeton, 1966), 27–8Google Scholar give for a population with a life expectancy at birth of 20 years and a growth rate of 5/1,000/year a BR (birth rate) of 57 and a DR (death rate) of 52/1,000/year; and for a population with a life expectancy of 22·5 years and a growth rate of 5/1,000/year, a BR of 50·5 and a DR of 45·5. Other life tables yield similar figures. Furthermore, a young age of marriage for girls, about 14–15 years, would not necessarily be favourable to high fertility; even Aristotle knew that young adolescent girls have difficulty bearing children (Pol. 1335a 4–6, 1335b 11), and any handbook on demography (e.g. Petersen, W., Population [New York, 1975], 193Google Scholar) will discuss subfecundity in adolescent girls before the age of 17 or 18. Harris is correct in maintaining that ‘no historian’ would claim that any society would deliberately increase the BR only to increase the DR through infanticide, in order to maintain a stable population (why not make it easy and just stop the killing?). (However, two paragraphs earlier [p. 115], he states that Roman girls deliberately married early to increase fertility, that is the BR, because the DR was so high from infanticide.)

¹³ Polygyny was not practised by the Greeks nor was concubinage, that is the simultaneous taking of a concubine with one's legal wife: Pomeroy, , op. cit. 70Google Scholar; Harrison, A. R. W., The Law of Athens: Family and Property (Oxford, 1968), 15–17Google Scholar. An exception might be the troubled time just after the Peloponnesian War in Athens, although this is doubtful. Positive checks are those that control existing populations – disease, malnutrition, and killing. Preventive checks are those that a population uses to prevent births, such as abortion and contraception. Infanticide, that is the killing of infants, is therefore a positive check for population growth. See Thompson, W. S. and Lewis, D. T., Population Problems (New York, 1965), 18Google Scholar: ‘Positive checks (are) all factors that operated chiefly as determiners of the death rate, i.e., as destroyers of life “already begun” … The second kind of check to the growth of population (was) called the preventive or “prudential” check because it operated to reduce the birth rate; … the postponement of marriage was and would remain the chief preventive check.’

¹⁴ Thus Golden, , op. cit. 328Google Scholar rightly rejects a ‘marriage squeeze’, a surplus of girls of marriageable age, as an ‘explanation’ for a hypothetical high rate of female infanticide in classical Athens. But his contention (p. 329) that an oversupply of widows was a problem because ‘a large number of such widows could even endanger the male domination of Athenian society’ is unconvincing.

¹⁵ Petersen, W., op. cit. 558Google Scholar.

¹⁶ Petersen, W., ‘A Demographer's view of Prehistoric Demography’, Current Anthropology 16 (1975), 228–9, 234CrossRefGoogle Scholar.

¹⁷ Chagnon, N. A. et al. , ‘Sex-Ratio Variation among the Yanomamö Indians’, in Chagnon, and Irons, , op. cit. 319Google Scholar.

¹⁸ Ibid. 296; Lizot, J., The Yanomami in the Face of Ethnocide (Copenhagen, 1976), 26–9Google Scholar.

¹⁹ Golden, , op. cit. 317Google Scholar. This is why we should reject the casual remark in Plato's Theaetetus, 149a–151c, 160e–161a. Here, Socrates compares an idea to a newborn baby and if the baby is found to be a mere image and not real, it is cast out. Few would conclude on the basis of this passage that healthy children born within marriage could be exposed. Golden also shows that there are as many scattered casual remarks in literary sources that suggest the age of marriage for men in Athens was 30, as there are that suggest the age was about 18: Dem. 40. 12–13; Lysias, frag. 24; Xen, . Sym. 2. 3, 4. 23Google Scholar; Menander, , Samia 60 (57)Google Scholar. Furthermore, Hopkins, K., ‘The Age of Roman Girls at Marriage’, Pop. Stud. 18 (1965), 309–27CrossRefGoogle Scholar shows a modal age for pagan men at marriage of 17–20, and a median age of 24, and Hopkins, K., ‘Brother-Sister Marriage in Roman Egypt’, Comp. Stud, in Soc. and Hist. 22 (1980), 316–20CrossRefGoogle Scholar, shows a late age of marriage for Egyptian girls. It seems clear that there was probably a wide variation in the ages of marriage in the classical world.

²⁰ For sex ratios cf. e.g. Plato, , Laws 780d–781aGoogle Scholar: an Athenian criticizes the Spartans and the Cretans for failing to include women in their practice in dining in common. As a result, half (ἠμισν) of the population is neglected. Aristotle, , Politics 1260b 12Google Scholar: women and children ought to be educated in regard to the constitution; ‘for women are half (ἥμισν μ⋯ρος) of the free population’. Aristotle, , Politics 1269b 5Google Scholar: ‘For just as man and wife are part of a household, it is clear that the state is also divided nearly in half into its male and female population (⋯γγὺς το⋯ δίχα διηρ⋯σθαι) so that in all constitutions in which the position of the women is badly regulated, one half of the state (τ⋯ ἤμιου) must be deemed to have been neglected in framing the law.’ Cf. Plato, , Laws 805aGoogle Scholar; 806c; Aristotle, , Rhetoric 1361a 6Google Scholar. The observations of these great social thinkers ought to be accorded at least equal parity with the anecdote from the third-century comic poet Poseidippus. I am grateful to Lawrence Bliquez for pointing out these passages to me. It must be noted, however, that there are problems in interpreting sex-ratio evidence. AsChagnon, et al. , op. cit. 306 f.Google Scholar, and Tarver, J. D. and Lee, C., ‘Sex Ratio of Registered Live Births in the United States, 1942–63’, Demography 5 (1968), 374–81Google Scholar, have shown, in some modern populations, younger mothers are more likely to bear sons and first pregnancies at all ages are more likely than later pregnancies to produce sons. Therefore, a society whose females marry young and experience high maternal mortality might have an imbalanced sex-ratio without any infanticide. Hence in such societies (which seem to include Greece and Rome), sex-ratios showing somewhat more males than females (if indeed these ratios can be demonstrated and not merely asserted) cannot be used as evidence for female infanticide.

²¹ As Golden, notes (op. cit. 330–1)Google Scholar, no Greek source mentions the practice of infanticide. Roman sources (e.g. Tert, . Apol. 9. 6–8Google Scholar) seem to mention the practice in connection with deformed infants.

²² For classical Greece: Germain, L. R. F., ‘Aspects du droit d'exposition en Grèce’, Rev. Hist, de droit fr. et étr. 47 (1969), 177–97Google Scholar; for Greco-Roman Egypt: Biezunska-Malowist, I., ‘Die Expositio von Kindern als Quelle der Sklavenbeschaffung im griechisch-römischen Ägypten’, Jahrbuch für Wirtschaftsgeschichte (1971), part 2, 129–33Google Scholar; for the Roman Empire: Pliny, , Epist. 10. 65, 66, 72Google Scholar.

²³ Germain, L. R. F., op. cit. 177–97Google Scholar. He concludes (p. 183) that there is no evidence for the practice of infant exposure in Athenian law from the classical era: ‘II n' y a done à Athènes, à notre connaissance, aucune trace sûre de législation en la matière, à l'époque classique.’

²⁴ Biezunska-Malowist, I., op. cit. 129–33Google Scholar.

²⁵ Harris, W. V., ‘Towards a Study of the Roman Slave Trade’, MAAR 36 (1980), 121–4Google Scholar. In his second article, ‘The Theoretical Possibility’, op. cit., he takes a different view.

²⁶ First Apology, 27.

²⁷ Harris, , ‘Towards a Study’, op. cit. 125Google Scholar, has convincingly shown that there were probably shortages of slaves (less supply than demand) in the first and second centuries A.D. Hopkins, K., Conquerors and Slaves (Cambridge, 1978), 158–63Google Scholar notes that prices paid for slaves increased during the second and first centuries B.C., which also indicates a shortage in supply. Cf. Croix, G. E. M. de Ste., The Class Struggle in the Ancient Greek World (Ithaca, 1981), 226–59Google Scholar.

²⁸ The translation is from Paton's Loeb text.

²⁹ Coale, A. J. and Demeny, P., op. cit. 31, 72, 78Google Scholar; cf. Coale, A. J., ‘The History of the Human Population’, Scientific American 231 (09, 1974), 44–5CrossRefGoogle ScholarPubMed.

³⁰ Gomme, A. W., The Population of Athens in the Fifth and Fourth Centuries B.C. (Oxford, 1933), 81Google Scholar. Gomme's critical attitude towards ancient anecdotal sources is in marked contrast to that of some modern scholars. It is also surprising that the immense Greek migration to the Hellenistic kingdoms is ignored as a cause for population decline in Hellenistic Greece.

³¹ It is difficult to know precisely what the term εὐγενο⋯ς means in this context, but it is more likely to mean well-born or free-born than nobility, and presumably would include those with equestrian status. For the best discussion of the problem see Brunt, P. A., Italian Manpower, 225 B.C. – A.D. 14 (Oxford, 1971), 558 fGoogle Scholar. For the difficulties in interpreting sex-ratio information see above nn. 7, 20.

³² οἱ μ⋯ν γ⋯ρ π⋯νητες οὐ τρ⋯ϕουσι τ⋯κνα, ϕοβο⋯μενοι μ⋯ χεῖρον ἢ προσ⋯κει τραϕ⋯ντα δουλοπρεπ⋯ κα⋯ ⋯παίδευτα κα⋯ τ⋯ν καλ⋯ν π⋯ντων ⋯νδε⋯ γ⋯νηται.

³³ Harris, , ‘The Theoretical Possibility’, op. cit. 116Google Scholar.

³⁴ Alternatively, one could translate ‘The poor (or poor men) do not rear their children when they fear’, the meaning is the same.

³⁵ Hombert, M. and Préaux, C., ‘Recherches sur le recensement dans l'Égypte romaine’, Papyrologica Lugduno-Batava 5 (1952), 155–6Google Scholar, record the sex-ratio in the Arsinoite, Oxyrhynchite, and Prosopite nomes as 107 males per 100 females in the general population, a quite normal ratio. Cf. Hopkins, K., ‘Brother-sister Marriage’, op. cit. 316–20Google Scholar.

³⁶ Golden, , op. cit. 330–1Google Scholar; Germain, , op. cit. 183Google Scholar. I am grateful to Lawrence Bliquez, Vincent H. Whitney, Keith Hopkins, and the editors of Classical Quarterly for their useful help and advice. They are not responsible for the views expressed or any errors.