One of the assumptions that I've made on this weblog repeatedly based on ancient literary references is the idea that before 1900 urban areas were population sinks. In Blessed Among Nations: How the World Made America the historian Eric Rauchway asserts that ~1900 in the USA urban health and life expectancy surpassed that of rural areas, arguably a first in the history of the world (the reason for this was a public health revolution prompted by fear of the diseases which Southern and Eastern European immigrants were presumably bringing to American cities). It's not that difficult to find data on early modern urban areas, which does confirm the lower fertility vis-a-vis rural areas, and often an excess of deaths over births (some scholars have argued that this second is an artifact of the transient nature of urban residency for a large segment of the population). Most of the older research seems based on extrapolation from archaeology and physical anthropology. But I did find one paper from 1913 which looked at inscriptions on tombstones in the Roman Empire to calculate curves of life expectancy at a given time from three locations: Rome, the Iberian provinces, and Roman North Africa. The first was an urban location, and the latter two less urban ones. The paper is ON THE EXPECTATION OF LIFE IN ANCIENT ROME, AND IN THE PROVINCES OF HISPANIA AND LUSITANIA, AND AFRICA. Some caveats: - The author notes than underrepresentation of lower status individuals in marked graves. This is expected. But, one presumes that life expectancy curves calculated from slaves and urban proletarian would probably be very different from what you see below. Days of birth and death listed on a tombstones indicates a level of functional literacy in the class from which the individual is derived (even if the individual was an illiterate, their family and friends are likely to have included literates). - Some of the ages at death recorded are almost certainly fictitious exaggerations. I don't think the author expresses proper skepticism of recorded ages in ancient Rome. - All that being said, the key is to focus on differences between locations. Unless there are systematic biases which vary by location (e.g., more marked graves among lower status individuals in urban Rome) we're getting a sense of differences of mortality. As a reference the author naturally uses the English of circa 1900. The charts are in order those of Rome, the two Iberian provinces, and Africa.
What you see on the x-axis are ages of individuals, and on the y-axis the expected years of life at that age. The lower life expectancy for women at all ages in the ancient world except for among the very old is in line with our intuitions. In a pre-modern environment there was no necessary expectation of "women and children first" in times of want, and women had to bear offspring which would be a net drain on their resources. Older women do not bear as many of these costs. At least the one of having to bear offspring. The author explains the higher life expectancy of older individuals in the Rome Empire as being due to a process of selection; the healthiest made it to more advanced ages. This sounds plausible, but I suspect part of it also has to do with inflation of ages for those who make it do an advanced age. Finally, you see the noticeable difference between Rome and the provinces. The author sees this as confirmation of the contemporary literary sources which point to the lack of healthfulness of ancient urban life. He also adds an addendum that after the paper was drafted a correspondent pointed him to another analysis which included a Gallic sample, which confirmed again that provincials seemed to have a higher life expectancy. I don't think that our argument can hinge on this one analysis, and I'll be digging further. To get a better sense though of ancient Roman urban areas soon I'll look to see what there is on Alexandria, since Egypt is one area because of climate that ancient documents have been preserved. It's interesting that this idea that cities are demographic sinks seems widely accepted in the literature, but there's been little cliometric exploration of the ancient data...though who can blame them given the nature of the ancient textual records.