Questions of size have long occupied center stage in the study of ancient population, reaching back at least as far as David Hume in the eighteenth century (Hume [1752] 1998). Following a shift in focus from size to structure in the 1980s and 1990s, controversies over population number are now experiencing a comeback and force us to reconsider our most fundamental assumptions about the character of ancient societies. Studies of particular locales or groups can take us only so far, increasingly for want of anything new to say: this is certainly true of the perennial favorites, the debates about the size of the number of residents of classical Attica and the imperial metropolis of Rome. The key question for Athens is whether, when and to what extent its population outstripped local food production and relied on maritime grain imports (Garnsey 1998: 183-200), an issue that is of more than antiquarian interest since it shapes our understanding of the driving forces behind political and military developments. By contrast, Rome’s utter dependence on imported food has never been in doubt: here, the main problem is how to reconcile the huge population size implied by the recorded number of recipients of the grain dole with the limited extent of residential areas within the city boundaries: a grand total of up to a million would imply extremely high - though perhaps not impossible - settlement density (Lo Cascio 1997; Storey 1997). Much depends on the question of whether the suburbium was demographically integrated into the urban core (Witcher 2005). Attempts to gauge the numerical size of social groups face even more formidable obstacles: thus, questions about the number of slaves in Attica or in Roman Italy, to name just two of the most prominent examples, are ultimately unanswerable, and can only be addressed on the basis of probabilistic modeling of demand for labor that puts some constraints on otherwise completely free-floating guesses (Scheidel 2005). Similar problems bedevil the demographic study of religious groups such as Jews or early Christians in the Roman empire (Wasserstein 1996; Hopkins 1998).
In any case, all these debates are dwarfed in importance by more general questions about the gross population of the core regions of the ancient Mediterranean world. A new comprehensive survey of all the evidence for the size of Greek poleis has prompted a higher estimate of the total number of all Greeks (Hansen 2006a): we may now have to reckon with some 7 to 9 million people, albeit including assimilated indigenes, resident aliens, and slaves. Indeed, on some readings, parts of classical Greece such as Boiotia or Aigina were more densely settled than in any subsequent period prior to the twentieth century. These observations raise profound questions about the scale of Greek population growth between the nadir of the Early Iron Age around 1000 bc and the classical period 500 years later (Scheidel 2003b), about the relative demographic strength of the Greeks and their competitors (although the population size of the Persian empire, for instance, is also unclear: cf. Aperghis 2004: 35-58), and most importantly about economic performance. If the Greeks of the archaic and classical periods grew to be very numerous and even added to natural growth by importing lots of slaves but nevertheless experienced substantial improvements in living standards (Morris 2004), their economies must have been unusually strong by premodern standards, which makes it necessary to revisit longstanding debates about the nature of “the ancient economy” (Finley 1999; Manning and Morris 2005).
Roman population size is an even greater conundrum. While the number of Roman citizens - and hence the population of Italy as a whole - may seem unusually well documented, thanks to a series of surviving census tallies stretching from the early republic into the first century ad, on closer inspection these records create more problems than they solve. These counts, traditionally thought to have been confined to adult men, rise gently in the second century bc but jump tenfold between 114 and 28 bc (roughly from 400,000 to 4 million) before returning to much slower growth to 5 million in ad 14 and 6 million in ad 47. As this cannot possibly be read as a straightforward demographic progression, we have to choose between two ways of making sense of these figures. What we might call the “low count” assumes that the republican figures are broadly correct and the later ones are so much higher because of an undocumented switch to the registration of all men, women and children of citizen status, and allows us to read the apparent tenfold increase as a tripling or at best quadrupling of the citizenry caused by the enfranchisement of the Italian allies and residents of northern Italy and the beginning spread of citizenship to the provinces (Beloch 1886; Brunt 1971; Scheidel 2004). Conversely, the “high count” is based on the belief that coverage remained unchanged but holds that republican counts were increasingly defective and thus exaggerated the apparent scale of the first-century bc increase. In this scenario, 4 to 6 million adult male citizens (or even more allowing for some undercount), most of them located in Italy proper translate to a final Italian population of anywhere from 15 to 20 million, including women, children, aliens and slaves (Frank 1930; Lo Cascio 1994; Morley 2001; Lo Cascio and Malanima 2005; Kron 2005b), and, therefore, a regional population density not encountered again until the nineteenth century.
Unfortunately, both interpretations raise logical problems (Scheidel 2008a). Hence, the “low count” requires very high levels of popular military mobilization in the republican period (Lo Cascio 2001), very high levels of urbanization and metropolitan primacy, and high levels of mobility (Scheidel 2004), and suggests that Roman population numbers fell short of those of the high middle ages (Kron 2005b). The “high count,” on the other hand, renders Roman Italy exceptionally densely populated by premodern standards, calls for a massive demographic collapse at the end of this period, and leaves us wondering why Romans imported millions of slaves at a time of rapid indigenous population growth when they already faced conflict over access to land and underemployment - and why these conflicts ceased in the Principate even as population would at least initially have continued to grow. Perhaps most crucially, it likewise raises questions about the size of the empire’s population as a whole: while the “low count” envisions some 60 to 70 million imperial subjects (comparable to the contemporaneous and similarly sized Han empire in China), with one-tenth of them located in Italy itself, the “high count” must assume either that the imperial heartland was massively overpopulated relative to its provinces or that the entire empire was much more populous than commonly assumed, presumably in excess of 100 million (Scheidel 2004). It does not help that independent consideration of provincial population size is largely unfeasible outside Egypt, and fails to yield unequivocal results even for that province (Scheidel 2001a: 181-250). If the ancient Greek experience is anything to go by, a “super-sized” empire is by no means impossible: in fact, it is widely accepted that the Asian and African parts of the empire did not re-attain Roman population densities until the nineteenth century (Frier 2000: 814). The key question is whether the same was true of Italy as well, and how other parts of the Mediterranean measured up. However, this problem would be mitigated by the latest suggestion (Hin 2008) that the Roman census figures originally referred to household heads rather than all adult men and from 28 bc onwards, in addition to these household heads they also included the orphans and widows who had previously been counted but excluded from the official tallies: this approach yields an “intermediate” scenario (with an eventual Italian population of perhaps closer to 10-12 million) that is less burdened by unpalatable implications than the “low” and “high” counts, and merits careful consideration.
In the end, just as for the Greek poleis, if we could be certain about population size, we would be better able to compare economic performance in antiquity to conditions in the medieval and early modern periods. Archaeological data may hold the key to this issue: while field surveys can cast some light on patterns of land use in different periods, physical indices of well-being, such as body height and dietary regimes, may reflect the extent of population pressure. In this sphere, despite an abundance of published local field work, major synthetic analyses are only beginning to appear (Koepke and Baten 2005; Kron 2005a; Jongman 2007; Pelgrom forthcoming). These questions will continue to occupy ancient historians for some time to come, all the more so as they are of fundamental importance for our understanding of classical civilization: how good were very different kinds of ancient states - Greek city-states on the one hand, the Roman mega-empire on the other - at fostering economic development, how many people could these economies support, and in what style?
World History
