Minimal Demographic Impact of Roman Slavery

It's commonly thought that slaves made up a huge percentage of Italy's population during the Roman Empire, and that most were of foreign origin, eventually replacing the native Italians. Historian Nathan Rosenstein turns this conventional notion on its head by showing that previous estimates of slaves' numbers are all wrong, and that increases in urban populations were caused instead by migrants from rural areas of Italy. These findings are supported by DNA evidence that detects no significant change in Italy's gene pool from the Iron Age.

Recent studies of Italian demography have further increased doubts about a rapid expansion of the peninsula's servile population in this era. No direct evidence exists for the number of slaves in Italy at any time. Brunt has little trouble showing that Beloch's estimate of 2 million during the reign of Augustus is without foundation. Brunt himself suggests that there were about 3 million slaves out of a total population in Italy of about 7.5 million at this date, but he readily concedes that this is no more than a guess. As Lo Cascio has cogently noted, that guess in effect is a product of Brunt's low estimate of the free population in Italy in A.D. 14. That is, Brunt must assume that the slave population had come to comprise nearly 40 percent of the population of Italy by the time of Augustus because he believes that the nonservile population of Italy had only managed to stay even between 225 B.C. and A.D. 14. At the same time, however, the number of residents of cities and towns throughout the peninsula and especially of Rome itself was skyrocketing. Consequently, without the supposition that slaves made up a very high percentage of the total population, not enough people would have been left in the countryside to produce the food needed to feed those in the towns. The basis for the supposition that slaves in Italy numbered as many as 3 million by the reign of Augustus in other words really consists of nothing more than a kind of elaborate circular argument in which the low free population "explains" the high number of slaves, which in turn "explains" how there could be so few free men and women in Italy.

Brunt also advances the claim that the Romans owned about 500,000 slaves circa 212, which suggests that, in his opinion, the slave population of Italy might have seen an average annual net gain between then and the end of the first century B.C. of perhaps 12,500 individuals. But the starting point for postulating such a rapid rate of increase is also based on a similar piece of guesswork. After noting that, by his reckoning, the Romans had mobilized about 11 percent of their free population in that year and mentioning the comparisons that other scholars had made to the 10 percent of their populations that some Balkan states in 1913 and Germany in 1914 had mobilized, Brunt continues, "We have only to suppose that the Romans owned not far short of half a million slaves to reduce the proportion of men in the armies and fleets far below 10 percent, even after allowing that 20,000-30,000 slaves may have been used after 214 as rowers." In other words, a slave population of 500,000 is necessary to bring the ratio of men under arms to the civilian population down into a range that Brunt finds acceptable. He makes no attempt to discover what might constitute a maximum rate of mobilization for a society such as Rome's in this period except to state that productivity per person was lower than in Germany and the war lasted longer than the modern conflicts. Of course, the cost of equipping and maintaining an army was much lower for the Romans as were the economic requirements of the civilian population. And one might suppose that more men could be spared from a simple agrarian economy like ancient Rome's than from a complex industrial one like early twentieth-century Germany's.

Consequently, Brunt's figures offer no basis for assuming that a dramatic rise in the number of Roman slaves — and hence in the number of the plantations that employed them — was getting under way during the early second century. To be sure, Livy records a depressing litany of enslavements by Roman armies in the course of their conquests in this period. But it does not necessarily follow that these would have helped bring about the sixfold increase in the Roman slave population by the reign of Augustus that Brunt postulates. Given the usual assumption in modern scholarship that male slaves significantly outnumbered females, the slave population would have been incapable of reproducing itself at full replacement level. As a result, the Romans regularly had to import substantial numbers of new slaves just to keep the slave population from shrinking. Scheidel has shown that on the assumption that slaves in 225 numbered 500,000 and were declining by only 1 or 2 percent per year, far more new slaves would have been required simply to replace current slaves who died than to generate a net increase of 2.5 million in the total slave population by 25 B.C. As large as the enslavements of this period were, therefore, they cannot in and of themselves demonstrate a rapid rise in Italy's slave population along the lines Brunt supposes. It is also worth bearing in mind that not all of those whom Rome's armies captured will have wound up in Italy, for this by no means constituted the only market for slaves in the early second century. Agriculture and manufacture in Carthage, Sicily, and elsewhere in the Hellenistic world made extensive use of slave labor, and the same factors of imbalanced sex-ratios and low birthrates that created a very high demand for replacement slaves in Italy may well have been operating in these areas also.

However, one piece of negative evidence, to which Scheidel has also drawn attention, provides an intriguing hint that conventional estimates of slaves making up as much as 40 percent of Italy's population by the late first century B.C. may be far too high. An analysis of the genetic makeup of Italy's modern population argues that the various distinctive genetic combinations currently found in different regions within the peninsula by and large track the linguistic distribution that resulted from the migrations of the Iron Age. No data indicate the subsequent large-scale infusion of new genetic material into the populations of these regions except in the case of southern Italy and eastern Sicily, which is explained by the well-documented Greek migrations there. If this finding is correct, then the slave population of Italy even at its greatest extent must have been far smaller than Brunt imagined, perhaps no more than a million. Otherwise, one must suppose that a very large number of slaves existed but made no contribution to the peninsula's genetic composition because they simply failed to reproduce themselves. Yet a very large number of slaves, on the order of 3 million, presupposes that this population was fairly successful at reproducing itself because it could never have reached that size in the first place and then maintained those numbers for centuries through imports alone. As already noted, the majority of new slaves brought into a servile population that was not reproducing itself completely would only have replaced old slaves who had died. But if a population of 3 million slaves, representing as much as 40 percent of Italy's inhabitants in the first century B.C., was successfully reproducing itself, it would surely have left its mark on the genetic makeup of contemporary Italians. That it did not argues strongly for a very low rate of natural reproduction among Italy's slaves, which in turn is difficult to reconcile with the hypothesis that the number of slaves ever grew large enough to comprise 40 percent of the Italian population.

If a dramatic rise in Italy's servile population during the second and first centuries is beginning to appear increasingly questionable, the decline in the numbers of free men and women that is supposed to have been its corollary is also being viewed with a growing skepticism. The census returns of 70 and 28 B.C. represent the linchpin for this pessimistic assessment of the condition of Italy's smallholders. For many years Brunt's powerful defense in Italian Manpower of Beloch's view that these totals demonstrate a drop in the free population of Italy remained unchallenged, even though the numbers themselves, around 900,000 in 70 and over 4 million in 28, would seem to reflect precisely the reverse. But Beloch and Brunt argue that the latter figure represents free men, women, and children, whereas the censors in 70 had counted only adult male citizens. When the totals are adjusted and allowances made for enfranchisements between 70 and 28 and citizens overseas, the result is a net decline in the free population. When these figures are in turn compared with the census returns of 225, the general regression in Italy's free population becomes patent, a regression that Brunt traces to the damage that Rome's wars and the importation of slaves inflicted on Italy's farmers.

In a provocative article, Lo Cascio has asked how it is possible to make demographic sense out of the Beloch-Brunt thesis. The argument they advance must assume that the population between 70 and 28 was declining annually by .5 percent, and the implications of such a decline, Lo Cascio believes, are unacceptable. Beyond question, the urban population of Italy increased dramatically during the middle of the first century, and any rise in urban numbers, with the possible exception of Rome itself, had to come from the rural population. In the preindustrial world, however, an urban population does not grow without a sustained growth in the rural free population whose economic products support it. Thus Lo Cascio argues that unless we are prepared to suppose that the ratio of urban to rural dwellers in Italy between 70 and 28 was far in excess of preindustrial norms — and there is no good reason to do so — the Beloch-Brunt interpretation of the census total for 28 cannot be made plausible. For it must assume that a dramatic and unparalleled drop in Italy's nonurban population was occurring at a time of unprecedented urban growth. Consequently, the figure of 4 million must represent only adult, male citizens just as had been the case in earlier republican censuses. If that is so, then as Tenny Frank long ago argued, the free population of Italy must have been growing vigorously during the second and first centuries.

Lo Cascio's article certainly will not be the final word on the controversy surrounding Beloch's and Brunt's thesis, but the mere fact that this critical prop is now being challenged renders claims about a crisis among Italy's small farmers due to war and the introduction of plantation agriculture all the more open to question. From a different perspective, Morley, too, has raised additional doubts about the conventional view. He notes that the populations of early modern cities generally could not reproduce themselves; they depended instead upon a large, steady influx of immigrants from the countryside to reach and then to maintain their size. Rome, he believes, would have been no different. Therefore the swelling of the city's inhabitants to nearly a million over the course of the second and first centuries B.C. and the stability of their numbers at roughly that level over the ensuing centuries cannot be attributed to a single, discrete event like the displacement of smallholders after the Hannibalic War. Such an episode would create a temporary increase, but then the process would slow, perhaps even reverse course, and the city would shrink as its population gradually died off.

Nathan Rosenstein. Rome at War: Farms, Families, and Death in the Middle Republic. Chapel Hill: University of North Carolina Press, 2004.