Demography of the Roman Empire

For the lands around the Mediterranean Sea and their hinterlands, the period from the second millennium BCE to the early first millennium CE was one of substantial population growth. What would become the territory of the Roman Empire saw an average annual population growth of about 0.1 per cent from the 12th century BCE to the 3rd century CE, resulting in a quadrupling of the region's total population. Growth was slower around the eastern Mediterranean, which was already more developed at the beginning of the period, on the order of about 0.07 per cent per year.[1] This was stronger growth than that seen in the succeeding period; from about 200 CE to 1800 CE, the European half of the empire only saw about 0.06 to 0.07 per cent annual growth (Europe as a whole saw 0.1 per cent annual growth rates), and the north African and west Asian parts of the empire saw almost no growth at all.[2]

By comparison, what is now the territory of China experienced 0.1 per cent annual growth from 1 CE to 1800 CE. After population decline following the disintegration of the western half of the Roman state in the fifth and sixth centuries, Europe probably re-attained Roman-era population totals in the twelfth and thirteenth centuries and, following another decline associated with the Black Death, consistently exceeded them after the mid-15th century.[2]

There are no reliable surviving records for the general demography of the Roman Empire. There are no detailed local records, such as underlie the demographic study of early modern Europe, either. Large numbers of impressionistic, moralizing, and anecdotal observations on demography survive from the literary sources. They are of little use in the study of Roman demography, which tends to rely instead on conjecture and comparison, rather than records and observations.[3]

When the high infant mortality rate is factored in (life expectancy at birth) inhabitants of the Roman Empire had a life expectancy at birth of about 25 years. However, when infant mortality is factored out, life expectancy is doubled to the late-50s. If a Roman survived infancy to their mid-teens, they could, on average, expect near six decades of life, although of course many lived much longer or shorter lives for varied reasons. Although this figure relies more on conjecture than ancient evidence, which is sparse and of dubious quality, it is a point of general consensus among historians of the period. It originates in cross-country comparison: given the known social and economic conditions of the Roman Empire, we should expect a life expectancy near the lower bound of known pre-modern populations. Roman demography bears comparison to available data for India and rural China in the early 20th century, where life expectancies at birth were also in the low 20s.[4]

About 300 census returns filed in Egypt in the first three centuries CE survive. R. Bagnall and B. Frier have used them to build female and male age distributions, which show life expectancies at birth of between 22 and 25 years, results broadly consistent with model life tables.[5] Other sources used for population reconstructions include cemetery skeletons, Roman tombstones in North Africa, and an annuities table known as "Ulpian's life table". The basis and interpretation of these sources is disputed: the skeletons cannot be firmly dated, the tombstones show non-representative sample populations, and the sources of "Ulpian's life table" are unknown. Nonetheless, because they converge with low Roman elite survival rates shown in the literary sources, and because their evidence is consistent with data from populations with comparably high mortality rates, such as in 18th century France, and early 20th century China, India, and Egypt, they reinforce the basic assumption of Roman demography: that life expectancies at birth were in the low 20s.[6]

As no population for which accurate observations survive has such a low life expectancy, model life tables must be used to understand this population's age demography. These models, based on historical data, describe 'typical' populations at different levels of mortality. For his demographic synopsis of the Roman Empire, Bruce Frier used the Model West framework, as it is "the most generalized and widely applicable".[7] Because it is based on only one empirical input, the model life table can provide only a very approximate picture of Roman demography. On two important points, the table may seriously misrepresent the Roman situation: the structural relationship between juvenile and adult mortality, and the relative mortality rates across the sexes.[8] In any case, Roman mortality should be expected to have varied greatly across times, places, and perhaps classes.[9][notes 1] A variation of ten years would not have been unusual. A life expectancy range of between 20 and 30 years is therefore plausible,[11] though it may have been exceeded in either direction in marginal regions (e.g., malarious urban districts on one end; high-altitude, low-density settlements on the other).[6]

The specifics of any ancient age distribution, moreover, would have seen heavy variation under the impact of local conditions.[6] In pre-modern societies, the major cause of death was not the chronic, end-of-life conditions that characterize mortality in industrialized societies, nor primary malnutrition, but acute infectious disease, which has varied effects on age distributions in populations. Pulmonary tuberculosis, for example, characterized much of the Roman region in antiquity; its deaths tend to be concentrated in the early twenties, where model life tables show a mortality trough.[12] Similarly, in pre-modern societies for which evidence is available, such as early modern England and early eighteenth-century China, infant mortality varies independently of adult mortality, to the extent that equal life expectancies at age twenty can be obtained in societies with infant mortality rates of 15% to 35% (life table models omit this; they depend on the assumption that age-specific mortality ratios co-vary in uniform, predictable ratios).[13] No ancient evidence can gauge this effect (the sources have a strong tendency to overlook infant death), and the model life tables might overstate it, but comparative evidence suggests that it is very high: mortality was strongly concentrated in the first years of life.[14]

Mortality on this scale: (1) discourages investment in human capital, hindering productivity growth (adolescent mortality rates in Rome were two-thirds higher than in early modern Britain); (2) creates large numbers of dependent widows and orphans; and (3) hinders long-term economic planning. With the prevalence of debilitating diseases, the number of effective working years was even worse: health-adjusted life expectancy (HALE), the number of years lived in good health, varies from life expectancy by no more than 8% in modern societies; in high-mortality societies such as Rome, it could be as much as one-sixth (17%) beneath total life expectancy. A HALE of less than 20 years would have left the empire with very depressed levels of economic productivity.[15]

To maintain replacement levels under such a mortality regime—much less to achieve sustained growth—fertility figures needed to be very high. With life expectancies of twenty to thirty, women would have to give birth to between 4.5 and 6.5 children to maintain replacement levels. Given elevated levels of divorce, widowhood, and sterility, however, the birth rate would have needed to be higher than that baseline, at around 6 to 9 children per woman.[16] Fertility could not long have either fallen below or outstripped replacement levels. A population which maintained an annual growth or decline of 0.7% would double or halve itself every century. Such rates are feasible locally or over a short period of time, and deaths could consistently outstrip births during epidemics, but, in the long term, convergence to maintenance levels was the rule.[17]

The surviving census returns from Roman Egypt speak to a population that had not yet undergone the "fertility transition"; artificial fertility controls like contraception and abortion were not widely used to alter natural fecundity in the Roman period. Only family limitation, in which couples ceased procreating after they had attained an acceptable level of children, could have been widespread.[18] There is no indication that even this limitation was widespread, however; the recorded distribution shows no evidence of being governed by parity or maternal age.[19]

Imperial Rome largely conforms to what is known as the "Mediterranean" pattern of marital fertility: men married late and women married early.[21] The evidence on marriage age is fairly robust for Roman elites: men in the senatorial class were expected to marry in their early twenties; women were expected to marry in their early teens. According to the most plausible interpretation of the evidence from funerary commemoration, in the lower classes, women married in their late teens or early twenties, and men married in their late twenties or early thirties.[22]

The Roman pattern thus stands in contrast to the "Eastern" (i.e., East Asian) pattern, in which both men and women married young.[21] China, the major example of the "Eastern" pattern, also had lower levels of fertility than Rome. This was apparently achieved by a combination of prolonged breastfeeding, female infanticide, and male celibacy, though the details are controversial.[23] Roman families share some features of the "Eastern" pattern. Roman Egypt, for example, had a custom of extended breastfeeding, which may have lengthened birth spacing. Egyptian fertility levels are comparable to those recorded in the early modern Japanese village Nakahara, where about half the population practiced family limitation. On the historian Walter Scheidel's judgment, this speaks to the incidence of family limitation even in what are supposedly "natural fertility" regimes.[24]

Roman and Greek literary and legal tradition also makes frequent reference to the "Eastern" demographic features infanticide and child exposure. Although the extent of these practices is unlikely to have been small, it is nonetheless impossible to quantify (nor can reported gender ratios permit judgment on the prevalence of femicide). These "Eastern" features did not prevail in medieval or modern Europe, where there were cultural and structural factors directly discouraging them or diminishing their effects on childhood mortality (religious doctrine, legal enforcement, institutions of foundling care, child labor, wet-nursing, etc.). These constraints were weak or absent in Greek and Roman society.[25]

According to the Cavalli–Sforza reconstruction of genetic history, there was little migration in Europe after the Iron Age. Most population growth can therefore be ascribed to the gradual expansion of local populations under conditions of improving fertility, rather than inter-regional transfer. That said, local migration from village to village may have been substantial; for the successful dedication and expansion of new settlements, it would have been necessary. The geography of the Mediterranean made this fairly convenient;[26] at the beginning of the empire, about 750,000 Italians lived in the provinces.[27] Julius Caesar, Mark Anthony and Octavian–Augustus settled many of their veterans in colonies, in Italy, and the provinces. Those established in Italy up to 14 BCE have been studied by Keppie (1983). In his account of the achievements of his long reign (Res Gestae), Augustus stated that he had settled 120,000 soldiers in twenty colonies in Italy in 31 BCE, then 100,000 men in colonies in Spain and southern Gaul in 14 BCE, followed by another 96,000 in 2 BCE.[28] Brian Campbell also states "From 49 to 32 BCE about 420,000 Italians were recruited" – which would thus be the Veteran (citizen) stock that was largely sent to the provinces (colonies) during Augustus; The Lex Calpurnia however also allowed citizenship to be granted for distinguished bravery – as example the 1,000 Socii from Camerinum after Vercellae 101 BCE (Plutarch Mar. XXXVIII) or the auxiliary (later Legio XXII Deiotariana) after Zela. By the time of Augustus the legions consisted mostly of ethnic Latins/Italics and Cisalpine Gauls.[29] Historian Theodore Mommsen estimated that under Hadrian nearly 1/3 of the eastern Numidia population (roughly modern Tunisia) was descended from Roman veterans.[30]

Modern estimates of the population of the Roman Empire derive from the fundamental work of 19th-century historian Karl Julius Beloch.[31] His estimates of the area of different components of the empire, based on planimetric estimates by contemporary military cartographers, have not been challenged by any more modern analyst. By providing a check to population densities, these area figures compel a baseline level of plausibility.[32] Beloch's 1886 estimate of the population of the empire in 14 CE has withstood contemporary and more recent criticism, and underlies modern analysis (his 1899 revision of those figures is less esteemed). Only his estimates for Anatolia and Greater Syria required extensive revision; Beloch estimated population figure, 19 million, produced population densities not otherwise achieved in those areas until the 20th century. Bruce Frier, in a recent estimate of the population of the empire, suggested a figure of 12 million as "considerably more plausible".[33] Beloch's figures for Spain and Africa have also been revised downwards.[34]

This estimate produces a population density of 13.6 inhabitants per square kilometer, a very low figure by modern standards (the United Kingdom, for example, has a population density of 254.7/km²). The population density in the Greek East was 20.9/km², twice as dense as the Latin West at 10.6/km²; only the Western provinces of Italy and Sicily had a density comparable to the East.[35]

Slaves constituted about 15 percent of the Empire's total population; the proportionate figure would be much higher in Italy and much lower in Africa and Egypt.[37]

There are few recorded population numbers for the whole of antiquity, and those that exist are often rhetorical or symbolic. Unlike the contemporaneous Han Dynasty, no general census survives for the Roman Empire. The late period of the Roman Republic provides a small exception to this general rule: serial statistics for Roman citizen numbers, taken from census returns, survive for the early Republic through the 1st century CE.[39] Only the figures for periods after the mid-3rd century BCE are reliable, however. Fourteen figures are available for the 2nd century BCE (from 258,318 to 394,736). Only four figures are available for the 1st century BCE, and are feature a large break between 70/69 BCE (910,000) and 28 BCE (4,063,000). The interpretation of the later figures—the Augustan censuses of 28 BCE, 8 BCE, and 14 CE—is therefore controversial.[40] Alternate interpretations of the Augustan censuses (such as those of E. Lo Cascio[41]) produce divergent population histories across the whole imperial period.[42]

The enfranchisement of the Cisalpine provinces and the Italian Allies after the Social War would account for some of the population growth of the 1st century BCE.[43] Alternate readings of the Augustan census both accept the basic accuracy of the figures, but assume different methods on the part of the census-takers. The standard interpretation assumes that the census-takers included all citizens—men, women, and children—in the Augustan censuses; the revised interpretation assumes that the census-takers only counted adult men, as they had during the Republican period. The standard interpretation is not supported by any evidence internal to the text, but reduces the implied population totals for 28 BCE Italy from 10 million to a more plausible 4 million.[44] The high total earns support from recorded conflict over land in the late Republic and other indications of population pressure, but does not accord well with comparative evidence from other periods and other parts of the empire.[45]

Estimates

Beloch's 1886 estimate for the population of the empire during the reign of Augustus:[46][47]

Region	Population (in millions)
Total Empire	54
European part	23
Asian part	19.5
North African part	11.5

Russell's 1958 estimate for the population of the empire in 350 CE:[47]

Region	Population (in millions)
Total Empire	39.3
European part	18.3
Asian part	16
North African part	5

Recent demographic studies have argued for a population peak ranging from 70 million (comparable to the contemporaneous and similarly sized Han empire in China), with one-tenth of them located in Italy itself, to more than 100 million.[48]

The cities of the Roman world in the Imperial Period. Data source: Hanson, J. W. (2016), Cities database, (OXREP databases). Version 1.0. (link).

Cartogram of the estimated populations of cities in the Roman world in the Imperial period (after Hanson 2016 and Hanson and Ortman 2017).

By the standards of pre-modern economies, the Roman Empire was highly urbanized.

According to recent work, there were some 1,400 sites with urban characteristics in the Roman world in the Imperial period.[49] At its peak, the city of Rome had at least one million inhabitants, a total not equaled again in Europe until the 19th century.[49][50] As the imperial capital, Rome was sustained by transfers in kind from throughout the empire; no other city could be sustained at this level. Other major cities in the empire (Alexandria, Antioch, Carthage, Ephesus, Salona etc.) had populations of about a few hundred thousand.[49][50] Of the remaining cities, most were quite small, usually possessing only 10–15,000 inhabitants. The cumulative urban population of the empire is estimated at around 14 million (using a population threshold of 5,000 individuals), indicating an urbanization rate of at least 25–30% to be consistent with conventional estimates for the total population, comparable to those in the 19th century.[49][50]

High mortality rates and pre-modern sanitary conditions made urban regions net population sinks, with more local deaths than births. They could only be sustained by constant immigration.[51] The large cities provided a major stimulus to demand, and not only for agricultural products, but for manufactured goods and luxury items as well.[52]

Russell estimated the urban population in Late Antiquity, as follows.[47]