Bryn Mawr Classical Review 2007.04.58
Mogens Herman Hansen, The Shotgun Method: The Demography of the Ancient Greek City-State Culture. Columbia and London: University of Missouri Press, 2006. Pp. ix, 140. ISBN 978-0-8262-1667-0. $39.95.
Reviewed by Peter Hunt, University of Colorado, Boulder (Peter.Hunt@Colorado.edu)
Word count: 3387 words
Table of Contents
This is a short book, but an important one. In it Mogens Herman Hansen (H.) deploys a new method based on urban habitation to calculate the "shotgun" spread within which the population of the Greek world in the fourth century B.C. must have fallen. H.'s eventual range goes from 7.5 to 10 million people for the "ancient Greek city-state culture," which includes colonies and Hellenized cities, or about 4 to 6 million for the Greek homeland, both larger population numbers than previous scholars had estimated.1 Although I have a couple of reservations, H's argument is generally persuasive and his calculations conducted in a admirably transparent and clear fashion. H. also argues that his population results imply a more modern Greek economy than had previously been thought--in particular by Moses Finley and his followers, as H. points out on several occasions (33-34, 64-66, 75). Finally, H.'s high count of city dwellers shows decisively that classical Greece was more urban than most historians have thought--even if many of these city dwellers, especially in the smaller cities, were farmers who walked out to their fields each day. Thus, not only H.'s population numbers, but their implications, are likely to shape debate both in Greek social history and beyond for years to come.
The book is based on H.'s Fordyce W. Mitchell Lectures of 2004 at the University of Missouri--Columbia. It consists of four chapters: the long first chapter lays out the whole of H's argument and treats in detail his extrapolation from cities with known urban areas to all cities. The second and third chapters justify in more detail two crucial steps in his calculation. The fourth comes back to the notion of Greece's "carrying capacity." Given the importance of the results involved and the originality of H.'s method, I will first consider his argument in some detail--though without the back-and-forth of his chapters. I will pay special attention to his minimum numbers, since, if these hold, H. will have refuted those prior scholars whose methods led to smaller estimates. The two steps that strike me as most vulnerable to objection will be my last subject: the old problem of using reported army strengths to calculate total population figures--which H. uses to confirm his results--and the step in H.'s new calculation where he extrapolates from cities whose urban areas are known to all cities.
H. begins by surveying four previous ways of estimating the population of the Greek world. H. argues that each of these methods tends to underestimate the population of Greece.
First, Karl Julius Beloch used the totals of soldiers mentioned in our literary texts to estimate the population of Greece: he would multiply the number of soldiers by four to arrive at the total free population of a city. But H. has shown elsewhere that not all men would be physically capable of serving, especially in a high-mortality society, and a fair number of men would have had to carry on government functions even in wartime.2 Consequently, he argues that Beloch's figures should be increased by at least 25%. Although H. represents himself as poking holes in all four previous methods, he actually considers the correct use of army figures, i.e. Beloch plus 25%, as trustworthy: he uses the figures as an argument against methods 3 and 4 (11, 13-14) and uses it again to corroborate his own results, especially in Appendix 1 (93-96).
Second, H. criticizes the method of Eberhard Ruschenbusch. Ruschenbusch derived the ratio between the tribute levied by the Athenian Empire and the population of a city in the couple of cases where we may know both figures. He came up with one talent per eight hundred people, and from this tried to calculate the population of the Athenian empire from the total tribute and then to calculate the population of the whole of Greece by extrapolating from the Athenian empire. Although there must have been some correlation between size and tribute--since big cities paid more than small ones--the ratio was not at all consistent and the evidence for Ruschenbusch's original ratio is flawed in ways that make his total too small (9-10, citing previous scholarship also to this effect).
A third method, deployed by Nicholas Corvisier, is "to establish the average number of persons per square kilometer and then multiply by the total area of mainland Greece."(10).3 But wherever Corvisier's numbers can be checked against population figures derived from army totals, they come in too low (11). This method is similar to H.'s in its relative independence from literary sources and in that Corvisier multiplies an area by a population density. But the population density of the countryside is archaeologically less visible and thus harder to estimate than that of a city, the basis of H's own method--not to imply that the latter is an easy matter either.
The fourth method H. examines involves applying the concept of the "carrying capacity" of the land to the city-states in the classical period and to modern Greece in the 19th century. (He revisits this subject at length in chapter four, but I will collect his arguments here.) According to H., calculations based on carrying capacity depend upon several assumptions: for example, they assume similar agricultural productivity in the two periods and that the importation of grain reveals that the limits of the land's productivity had been reached (12). These assumptions allow the conclusion that Greece, both in the 4th century B.C. and in the 19th century A.D., had reached the carrying capacity of the land. Consequently the populations at both times should be about the same. And since we know the population in the nineteenth century, we can estimate the classical population (13-14, 33). But H. shows that the scholars who have claimed to substantiate the claim that Greece in the 4th century B.C. had a population similar to that in the late 19th century have typically engaged in circular reasoning (82-83). For example, H. argues that any figure between 10,000 and 30,000 is possible for the ancient population of the city-state of Mantinea. Björn Forsén's choice of a number in the lower end of this range is motivated, H. claims, by the 1896 census that shows 13,200 inhabitants of this area.4 The resulting correspondence thus cannot be taken to show that Greece had a similar population in 1896 and the classical period.
The difficulty in escaping such circularity is that there are few areas of ancient Greece where we can determine the likely ancient population with enough certainty to produce an independent comparandum for the 19th century figures. Indeed, H. argues that we have reliable population numbers for only two ancient regions, Attica and Boeotia. The uniqueness of Athens in both periods--it became the capital of modern Greece in 1834 and was notoriously atypical in the classical period--rules it out as a representative test case, so H. considers Boeotia in some detail. A calculation from army numbers and based on H.'s assumptions leads to an estimate of 200,000 to 250,000 inhabitants; John Bintliff using the same original figures comes to a total of 165,000 inhabitants.5 What is startling about both these figures is how much greater they are than figures for more modern periods of Greece. A calculation based on an Ottoman tahrir (registration of households) suggests that Boeotia had fewer than 50,000 and more likely 38,000 inhabitants at a peak of population in 1570 (88). Even more astonishingly, the modern census in 1889 showed a population of 40,000-42,000 (88). Both reveal that the population in 16th and 19th century Greece was a mere fractions of the ancient population. The explanation for this contrast may lie in a higher carrying capacity--perhaps due to greater agricultural productivity--and some importation of grain in the 4th century B.C. (89-91). What H. has established with certainty is that the equation of 19th century and classical carrying capacity cannot help us estimate ancient populations--nor provide an objection to H.'s larger numbers.
H.'s own method, the book's greatest contribution, is based on the advance and accumulation of archaeological information, both about cities and from surveys of the countryside (5). The data collected in the Copenhagen Polis Centre's Inventory of Archaic and Classical Poleis are also crucial to the H.'s calculations (4, 16-24). H.'s method basically involves multiplying the inhabited area of all the cities in the Greek world by the density of population in these inhabited areas and by a factor to reflect the fact that not all Greeks lived in cities.
The first challenge is to determine the urban area--that is, the area within the walls--of all Greek cities. The urban areas of 232 of the approximately 1000 cities of the Greek city-state culture have been estimated by archaeologists. But a simple extrapolation by multiplying the sum of their areas by 1000/232 to arrive at the total area of the 1000 cities gives too high a result, since archaeological information is more likely to exist for larger than for smaller cities (18). So H. turns to the 636 cities the size of whose total territory is known. These constitute a fuller and, H. believes, a more representative sample of all the cities. These 636 have been divided into categories based on the size of the territory in the Copenhagen Inventory. Within each category, H. uses the subset whose urban area is also known to come up with the average urban area for that category of city. For example, we know the urban area of 33 of the cities in Category Three (19, table 1.5); the average urban area of these 33 is 49 hectares (19, table 1.5); Category Three cities make up 16% of the 632 cities whose territory is known and thus, as H. believes, 16% of the thousand total cities, that is 160 cities (21, table 1.6). By multiplying 49 (average hectares/Category Three city) by 160 (extrapolated number of Category Three cities) we get 7840 hectares total urban area of Category Three cities (21, table 1.7). After doing this with all six categories of cities, H. comes up with a total of over 50,000 hectares enclosed by walls in all the cities of the Greek world.
To determine how many people lived in these walled cities H. needs to address three questions: "How much of the intramural space was used for habitation? What was the average number of houses per hectare? How big was the average household?" (35) H. first notes that, in addition to public spaces and buildings, many cities set aside uninhabited areas within their walls for a variety of reasons (37-41). His list of cities whose total walled areas and the portion inhabited are approximately known shows great variation: the area within the walls could be 90% inhabited or only 20% inhabited (42, table 2.1). There is a vague correlation between size of city and percentage habitation: in general small cities tended to devote more of the area within the walls to habitation and less to public space or as an uninhabited preserve. H. eventually decides on the following estimate: in very small cities he assumes that two-thirds of the space is inhabited, in small and medium cities one-half, and in large cities one-third (46). This approximation is very rough and does not at all reflect the actual variety even among cities of the same size. That does not matter. A glance at how many large cities have an inhabited area greater than one-third (42, table 2.1) makes it clear that H. has indeed chosen a minimum proportion and that his eventual argument for a large urban population has not been not compromised.
Thus, the "shotgun method" has allowed H. to proceed to establish a minimum inhabited area despite our incomplete and varied data. The average number of houses per inhabited hectare does not vary quite as much: 26 to 50 but with most cities grouped between 31-33 houses per hectare (Table 2.3). Accordingly, H. takes 30-33 as his range for houses per hectare (51).6 H.'s discussion of the number of people per household is sophisticated and persuasive (52-60). He takes into account the number of births per women required to maintain a population with a high mortality rate, the prevalence of orphans living with relatives, and variations during the lifecycle of a household. Studies of the families mentioned in Athenian lawcourt speeches confirm his total of something over 3 children per married couple (57). For his minimum number he naturally takes a minimalist view of the number of slaves (.5 per house). H. settles for a range of 5-7 people per house. Again one must admit his minimum is indeed a minimum--indeed, he notes that some archaeologists have gone as high as 10 (61). By multiplying the houses per inhabited hectare by the people per house, he arrives at 150 to ca. 200 people per inhabited hectare (61). Studies of medieval and early modern cities confirm that this estimate is not unreasonable (62-63).
H. has now established his range for the number of people living within the walls of Greek poleis, about 3.3 million (23), but how many people lived in the countryside? To get at this issue, H. summarizes the results of seven archaeological surveys that have attempted to estimate the proportion of urban and rural residence (66-71). For the purpose of his calculation, H. estimates that half of the population of large city-states lived within the walls and that two-thirds of the population of smaller city-states lived there (64, 71). Only in the largest city-states did most of the people, two-thirds in his calculation, live in the countryside. (23-24, 64) This conclusion is at odds with scholarship that emphasizes the rural nature of Greek society: for example, Horden and Purcell in The Corrupting Sea claim that less than ten percent of the ancient population lived in towns.7 H. is quite aware of the difficulties in extracting rural population data from survey archaeology, but, at this point in the argument, after he has established an urban population in the millions, H.'s position does not depend only on controversial evidence from survey archaeology. Rather his results support the inferences from the surveys, since he has the ruralists in an almost inescapable dilemma: either the proportion of rural to urban inhabitants cannot have been that great or the total population of Greece will end up being much higher than H.'s estimate, an alternative unpalatable even to the most ardent of modernists; Horden and Purcell's proportion of rural to urban inhabitants, for example, would lead to an ancient Greek population of the order of thirty million!
The result of these complex calculations is that there were more than 7 million inhabitants of the Greek world in the 4th century. H. argues that this figure needs to be raised even further since Epirus and Macedonia were far less urbanized than the rest of the Greek world and thus their population is underestimated by his method. Indeed, H.'s method of calculation gives a population of 50,000 for Epirus while Corviser estimates 425,000 (25). With a correction to account for this anomaly, the "shotgun" range comes out at from 7.5 to 10 million inhabitants of the "ancient Greek city-state culture." H.'s total figures are not as radical as they seem, because he includes all the Greek cities throughout the Mediterranean and Black Sea, something that other scholars have not done. Nevertheless, H.'s minimum for the "Greek Homeland" is just under 4 million people, whereas Beloch, Ruschenbusch, and Corvisier all came up with estimates around 3 million (4, 9, 11). H. is also able to say something about the distribution of city sizes: although most cities were small, 40% of the Greek population was concentrated in the largest 10% of the cities, those with territories of 500 square kilometers and populations near to 30,000: the average Greek did not live in an average-sized polis.
In the course of this long series of estimates and calculations, there are many points at which objections are possible. Most of these either would not affect the final range or would require only a slight adjustment. For example, Greeks with more than one residence will be over-counted by H.; nor can archaeologists vouch for the continuous occupation of every house in a city. Nevertheless, only two steps really worried me.
First, inferences from army figures play an important role in confirming his calculations. H. admits that he has to assume that army figures were actual combatants and not just an inflated "paper strength" (84). Misgivings on this score are not assuaged by the words Thucydides attributes to Alcibiades on the eve of the Sicilian expedition: "As for their hoplites, they have not got so many as they boast of: it is the same with them as with the rest of the Hellenes; the number never came up to the estimate made by each state of its own power; in fact the falsification was a very big one . . ." (6.17.5; trans. Rex Warner). If some literary numbers for army totals come from exaggerated estimates made by states rather than counts of soldiers actually at a battlefield, we may arrive at population figures that are too high. Fortunately, H.'s first appendix is somewhat reassuring on that count. His calculation from urban areas usually comes in well below the army size extrapolation, and this margin of safety may even cover the possibility that some army figures are based on "a very great falsification"--whatever that may mean in percentage terms.
Second, the extrapolation from the 632 cities with known territories assumes that these are representative of the whole thousand cities of the Greek city-state culture (28, assumption 1). But larger cities are probably more likely to have had the extent of their territory estimated. To illustrate the possible effect of such a tendency, let us make an alternate calculation that is simplified, but not far-fetched--and especially not for determining a minimum number. Suppose that archaeologists have in fact estimated the size of the territory of all the poleis in the two categories with the largest territories. In other words, suppose all 136 of the larger cities that existed are among the 636 cities of known territorial size. The 364 cities whose size is not known would then contain no large cities. We would not want to extrapolate by multiplying the number of large cities by anything--as H. does. We might rather assume that the 364 cities of unknown territory belong to the four categories of smaller cities. One simple way to proceed would then be to add 91 cities to the number in each of the four categories of smaller cities (since 4 x 91 cities = 364 cities) to come up with the likely distribution of different sized cities among the 1000 total cities--see table. The total urban area deriving from such a distribution would be significantly smaller than H.'s estimate, 42,135 instead of 50,510 hectares. Since larger cities have a larger proportion of rural inhabitants in H.'s calculation, the effect on total population of such a change would be even greater: we get about 5.5 million instead of H.'s 6.8 million--to which must be added the adjustment for Epirus and Macedonia. This would push the minimum total for the "Greek homeland" down to about 3.5 million people. Naturally, if H. can show that large cities do exist among those 364 cities whose territories have not been estimated then this objection would have to be withdrawn--if, for example, there are references to any of these cities fielding significant armies.
H.'s "shotgun" range is large--and arguably a touch high--but the introduction of a new method of calculating population that is largely independent of literary or historical sources puts Greek demography on a more solid basis and would be contribution enough. H's confirmation of a more urban Greece than many had imagined is a second significant result. That H. finds a larger population in classical than in 19th century Greece is a modernizing result in the strict sense of the word. The implication of large-scale trade--perhaps even one million Greeks living on imported grain (33)--and more productive agriculture will also play a role in future debates on the nature of the ancient Greek economy: "The ancient economy may to a large extent have been an agrarian economy. I do not doubt that, but it was certainly not a subsistence economy" (34).
1. His decision also to include imported foreign slaves is justified; they lived, ate, and produced in the areas where they served as slaves. His claim that this is also justified by the fact that "the typical slave either was Greek or, rather, had been Hellenized" is unnecessary for his argument and, in my opinion, improbable. I will be addressing this topic in detail in a future publication.
2. Mogens Herman Hansen, Demography and Democracy: The Number of Athenian Citizens in the Fourth Century B.C. (Systime, 1985), 16-21.
3. Jean-Nicholas Corvisier and Wieslaw Suder, La Population de l'antiquité classique (Paris, 2000), e.g. 32-33.
4. Björn Forsén, "Population and Political Strength of Some Southeaster Arkadian Poleis" in Pernille Flensted-Jensen (ed.), Further Studies in the Ancient Greek Polis (Stuttgart, 2000), 44-54 in H. 82.
5. John Bintliff and Anthony Snodgrass, "The Boiotia Survey, a Preliminary Report: The first four years." JFA 12 (1985) 123-161 and subsequent publications by Bintliff cited in H. 85-86.
6. H. also calculates this range by assuming a two to one ratio between the area devoted to houses and to streets and by considering various average house sizes. This enables him to use the somewhat greater evidence for house size. Although he could have used Table 2.3 directly for houses per hectare, the correspondence between the two methods is reassuring.
7. Peregrine Horden and Nicholas Purcell, The Corrupting Sea: A Study of Mediterranean History (Oxford, 2000), 92 cited in H. 29. It should be noted that Horden and Purcell are generalizing about the whole course of history and the whole Mediterranean and might conceivably make an exception for the city-states of Greece in the Classical period; conversely, H. does admit that some areas, such as Epirus and Macedonia, were far less urbanized.