Weber, Bloch, Finley, de Ste. Croix, McCormick, Wickham: the decline of the Roman slave system in Late Antiquity has been central to every major modern account of the transition from antiquity to the Middle Ages. Kyle Harper's monumental study of late Roman slavery hence has implications far beyond its ostensible focus on the ‘long’ fourth century (a.d. 275–425). Lucid, sophisticated, and beautifully written, it deserves the widest possible readership.

The place of slavery in the late Roman economy is the subject of Section I (1–200, chs 1–4). H. plausibly argues that slaves made up some 10 per cent of the population of the fourth-century Roman Empire (5,000,000 in total), half of them owned by élite senatorial and decurial households, the other half by ‘middling’ urban and rural households (38–60). One of the great strengths of the book is H.'s awareness of the differences in kind between these two broad classes of slave-owners: the kinds of benefits that one or two slaves brought to a well-off peasant household (128–41) had little in common with the economics of slave labour on huge dispersed decurial estates (162–79).

H.'s account of the late Roman slave-supply (67–99) makes excellent use of a fourth-century tax-register from Thera (first published in 2005) listing 152 agricultural slaves with their names and ages. The age- and sex-profile of this sample (numerous children, and a ratio tilted towards females among slaves over the age of thirty) is taken by H. to reflect an agricultural workforce ‘shaped by natural reproduction, family life, and some adult male manumission’ (76). This would offer striking support for Walter Scheidel's hypothesis (JRS 87 (1997), 156–69) that natural reproduction within the household was the most important single source of Roman slaves. However, a sceptic might reasonably point to the prevalence of child exposure and infant sale in the eastern Roman Empire (78–83, 391–423), which could easily create a ‘bottom-heavy’ age-profile of this kind without the need for large-scale slave breeding.

In his enormous chapter on agricultural slavery (144–200), H. explains the persistence of large-scale slave-ownership in the Late Empire in terms of four determinants: ‘supply, demand, formal institutions, and the dynamics of estate management’ (152). Slavery flourished because it represented the most efficient way of maximizing household revenues within a diverse labour-market: ‘land-owners could choose from three categories of labor: slavery, tenancy, and wage labor’ (156, my emphasis). When a fourth-century land-owner weighed up direct costs (slave prices vs wages) against transaction costs (efficiency, the need for oversight, the seasonality of Mediterranean agriculture), slavery would have presented itself as the ‘rational’ choice. Whatever one thinks of this unashamedly neoclassical approach — the influence of the New Institutional Economics is patent — H.'s account is the most coherent and sophisticated analysis of the economics of ancient slavery that we currently possess. It raises the bar.

Section II (201–348, chs 5–8) is concerned with the subjective experience of Late Roman slavery. It may not be news that slaves and free men ate different kinds of bread (237), that a slave was forbidden to look his master in the eye (332), or that an owner expected to know the complete sexual history of his slave-women (295). But thanks to the massive wealth of fourth-century homiletic evidence, the social realities of the Roman master-slave relationship have never been better documented, and seldom more vividly evoked, than they are here. One of the things that comes out most strongly is the fundamental compatibility of Christian ethics with slave-ownership (212–14). As H. shows, the Christian critique of slavery focused on the ethical consequences of mastery, not on the human rights of the slave (300–4). If the sexual exploitation of slave-girls was a bad thing, that was because Christianity emphasized the value of sexual exclusivity tout court; it was the spiritual welfare of the slave-owner, not of his human property, that worried Augustine, Gregory of Nyssa and John Chrysostom (346).

Legal norms concerning enslavement and manumission changed radically in the course of the fourth century, but H. is unwilling to use this as evidence of social change: ‘in late antiquity, the values behind public law changed less than the mechanisms used by the state to enforce those values’ (431). As a result, Section III (349–494, chs 9–12), on the pronouncements of the Late Roman state on slavery, is ‘more about change within the state than change within society’ (423). So the increased willingness of Constantine and his successors to countenance the sale of children into slavery does not mark a change in social relations, but ‘is simply a reflection on the material limits of the state's power and its unwillingness to exert great energy in this arena of social activity’ (414). Likewise, when Constantine permitted the re-enslavement of freedmen, he was simply bringing Roman statutory law ‘into harmony with the natural balance of social power’ (488). As a result, the legal sources add less to our picture of fourth-century slavery than one might have hoped: neither Christianization, nor changes in the Roman status system, seem to have had any significant impact on the public law of slavery.

H. is better at illustrating the vitality of the fourth-century Roman slave system than he is at accounting for its dramatic fifth- and sixth-century decline. Readers seeking a new interpretation of the end of ancient slavery, or of the transition from antiquity to feudalism, will be disappointed. For the West, H. argues (following Chris Wickham's Framing the Early Middle Ages) that the decline of estate-based agriculture and long-distance commodity exchange was the decisive factor: ‘Roman slavery was situated in the sectors of the economy most influenced by elite land-ownership and market-orientated production’ (198). Hence, when inter-regional Mediterranean exchange declined sharply in the mid-fifth century, large-scale slave-ownership simply withered away. For the East, where it is harder to show economic simplification in the fifth and sixth centuries, H. tentatively suggests — true to his ‘rational-choice’ model — that population growth among the free peasantry may have rendered slave production ‘less attractive, less necessary’ (506). But this ‘demand-side’ explanation of the end of Roman slavery only works (if at all) for large-scale estate-based agricultural slavery. What induced the millions of fourth-century ‘middling’ slave-owners, in city, town and village, to give up the weavers, nurses and pig-keepers on whom their social status depended? A supply-side crisis must surely be part of the answer; but where does that leave the thesis of a self-reproducing Late Roman slave population?

This is the most important book on Roman slavery to appear in many years. No social or economic historian of the Roman world or early medieval Europe can afford to ignore it.