The Digging into JRA blog authors’ opinions are not necessarily identical to those of the featured JRA volume contributors.
Some things in history are clearly more challenging to figure out than others. We can, more often than not, identify the origin of a scientific invention, referring to available textual evidence and artifacts. We can also, without much changing the methods, deduce the political situation prevalent in a given country at a given time. And yet, when it comes to making an educated guess, let alone drawing firm conclusions, about the size of the entire Greco-Roman population between the 4th century B.C. and 6th century A.D., who do you turn to? “50:50”; “Ask a Friend”; “Switch the Question”? If this were “Who Wants to Be a Millionaire?”, scholars John Hanson and Scott Ortman of the University of Colorado would both be millionaires. Their answer is: Math.
With the growing interest in the urbanism of the Greek and Roman world over the last few years, scholars have reached a general consensus about some of its vital statistics, among them the sizes of various settlement populations and of the urban population overall, the urbanization rate, and even the total population (urban and rural combined). Walter Scheidel in the Cambridge Economic History of the Greco-Roman World, for instance, positions the overall urban population of the Greco-Roman world in the range between 7 and 9 million, with the urbanization rate at 10-15% and a total population of 59-72 million accordingly. Yet Hanson and Ortman won’t buy it, explaining: “Although the bare outlines of this summary are almost certainly correct, there is surprisingly little evidence for many of these statements.” While the estimates for Rome, Alexandria and perhaps Antioch are backed by relatively good evidence, the data on the rest of the settlements is largely unfounded. Before suggesting an alternative solution, however, the two scholars take their time to make it clear why we all definitely need one.
For starters, the most common forms of evidence used in estimating ancient settlement populations have been textual sources, inscriptions and papyri. Not only are those hard to use because they usually refer only to specific groups (council members, taxable citizens, the military), which by default discriminates against most women, children, slaves, foreigners and other non-qualified groups, but also because many of the numbers refer to both the settlement and its surrounding countryside. Other ways of procuring evidence include counting the number of individuals buried in cemeteries, the seats in entertainment structures, the amount of water supplied by aqueducts, and other so-called “archaeological proxies.” These, too, are not to be taken at face value: after all, we can never know how many people were cremated rather than interred, and how consistently grave goods are present as an indicator of a burial that has no longer survived. Neither can we tell how many of the visitors frequenting entertainment buildings like theaters were residents of the immediate settlement and how many were those of its nearest neighbors, or how much of the water supplied by aqueducts was used for non-domestic versus domestic uses and how much was obtained through other means, such as wells and cisterns. All in all, too much bias: Hanson and Ortman won’t buy it.
The only technique that the scholars do give some credit, and which Hanson even tried using in 2016, is the ‘area x density method’: “the only method that allows one to correct the existing bias towards the most well-known sites and for which there is sufficient evidence for a large enough sample to allow one to explore overall patterns and trends, such as the size of the overall urban population, the urbanization rate, and the total population.” While the idea was first suggested by the German historian J. Beloch in 1886, it has recently gained in popularity due to the increasing amount of information about inhabited areas generated by surface surveys, aerial photography and satellite imagery, in addition to the traditional practice of using only excavated areas. The process is to measure inhabited areas and multiply them by a range of population densities. “Range” here is key, as it enables one to play with quite an array of density figures for each settlement, exploring the upper and lower limits of what sounds like the most credible answer. Alas, this feature is as attractive as ‘area x density’ gets. All would be great, were there enough evidence for how concentrated the occupation of settlements actually was. But since there isn’t, Hanson and Ortman (surprise!) won’t buy it.
So why are population densities so sought after by Hanson and Ortman and what information do they provide us with? Apparently, a better understanding of population densities can tell quite a lot about settlement history, including what conditions people lived under, how close their homes were to one another, and how large their households (did they own any slaves?). Hmmm, would not those factors also directly influence the number of human interactions that occurred, and thus how much wealth, innovation and invention, as well as crime, pollution and disease, there was to be expected in each settlement? The correlation is real, and is explained by the role of cities as “social reactors”, implying that more human interactions will produce more output (such as wealth, innovation, invention, crime, pollution and disease) with less infrastructure (such as miles of roads per capita). In fact, the principle works wherever you apply it, whether it is the Basin of Mexico in the pre-Hispanic period, medieval Europe, or the modern United States. Magic! Now let’s turn it into simple math.
When Hanson and Ortman did, they found that the relationship between the area and the population of a settlement could be expressed as: A=aN, which claims that area is population multiplied by a constant coefficient (specific to the settlement), raised to the relationship’s exponent. Applied to 52 sites from across the entire Mediterranean between the 4th century B.C. and 6th century A.D., the equation shows that the population densities of ancient settlements increased as they grew, and at rates not so different from those obtained for other civilizations. Turns out that earlier scholars were just a tad bit far off, with Scheidel’s Greco-Roman population estimate of 59-72 million against the newly estimated 150 million, and an urbanization rate of 10-15% against the new 25-30%. As can be seen, where digging alone won’t cut it, math might solve the problem.
The real math is actually outlined in the original article, J. W. Hanson and S. G. Ortman, “A systematic method for estimating the populations of Greek and Roman settlements”, JRA 30 (2017) pp. 301-324, which you can order at: http://journalofromanarch.com/order.html.