Geographical contrasts
271
not lost any of the members of his household whom he had brought with him. There is an implicit contrast with two other places where he spent much of his time, Laurentum and Rome, where he was presumably not so fortunate:
also my slaves have never lived in a healthier environment: so far I have not lost any of those whom I brought with me⁶
The contrast between Tifernum, on the one hand, and Laurentum and Rome on the other, in antiquity, is a typical example of the extreme regional variation in demographic patterns which is now known to have been the norm in early modern Europe. It exactly parallels the contrast made by del Panta between Treppio and Grosseto in the nineteenth century, for exactly the same fundamental reason, namely the absence or presence of malaria. In a review of a book on medieval demography Johannson made a devastating critique of a series of assumptions that are also frequently made by modern historians writing about the demography of the Roman Empire. She concluded that ‘early modern Europe was characterized by extreme variability with respect to its mortality patterns’, after noting that England in the seventeenth century contained places with life expectancy at birth as high as 50 and other localities with a life expectancy at birth as low as 20, as demonstrated by the parish studies.⁷ These data destroy the view regularly expressed by Roman historians that the Roman population (it would be better to say Roman population s) could not have had a life expectancy at birth higher than 25 or 30. The inhabitants of Tifernum would have been astonished to learn of that hypothesis. Johannson then observed that ‘long-run growth rates of 1% to 2% per year (on average) are not unusual for those villages which were free of malaria, sheltered from frequent epidemics, and spared major famines. In malarial parishes, in contrast, death rates persistently exceeded birth rates’.⁸ This, again, destroys the widely held view among ancient historians that ancient populations could not possibly have grown at more than about 0.5% per annum.
Since the attempt by Lo Cascio to overthrow the Roman demography of Beloch and Brunt depends on such assumptions about life ⁶ Pliny, Ep. 5.6.46: Mei quoque nusquam salubrius degunt; usque adhuc certe neminem ex iis, quos eduxeram mecum . . . ibi amisi.
⁷ Johannson (1994: 528).
⁸ Ibid., 531.
272
Geographical contrasts
expectancy and growth rates, and more generally on defining too narrowly what Braudel called ‘the limits of the possible’, it must be regarded as unconvincing in detail. Nevertheless Lo Cascio was right to draw attention to the problems of interpreting the Roman census data, problems which will probably never be completely resolved without the discovery of new evidence.⁹
One of the conclusions of this book is that the figure of 25, commonly assumed by historians as an appropriate figure for the life expectancy at birth of the Roman population, is both too low and too high. Healthy areas could very easily have had substantially higher life expectancy at birth than that. It is worth quoting some of the exact words of Herlihy and Klapisch-Zuber about late medieval Florence to hammer home this point:
About 1300 the average duration of a human life at Florence was approximately 40 years. But over the following 100 years, amid the fury of pestilence [sc. the Black Death], it collapsed to only one half that figure (an average of only some 20 years). In the fifteenth century the average duration of life grew more extended, especially after 1450, to regain the summit of 40 years . . . A perusal of the Florentine family memoirs leaves the impression that men lived to an old age in the thirteenth century.¹⁰
Florence was certainly not unique. In the course of a discussion in which he cited numerous examples of historical populations drawn from all over the world which had a life expectancy at birth significantly higher than 25, the Italian demographer Massimo Livi-Bacci drew attention to research showing that the nobility of Milan had a life expectancy at birth of about 40 in the seventeenth century .¹¹ However, the words of Herlihy and Klapisch-Zuber also illustrate the downside: in an unfavourable disease environment life expectancy at birth could easily be significantly less than 25. The city of Rome during the period of the Roman Empire did not have bubonic plague, but had P. falciparum malaria instead.
⁹ Lo Cascio (1994).
¹⁰ Herlihy and Klapisch-Zuber (1985: 83–4).
¹¹ Livi-Bacci (1983). Historical populations with a life expectancy at birth of more than 25
are now known from all over the world, e.g. Zhao (1997) suggested a life expectancy at birth of about 34 (Coale–Demeny Model East Level 8) for a Chinese clan in the period 1000–1750, relying on data for adult mortality; Farris (1985: 43) suggested life expectancies at birth ranging from 27.5 to 32.5, with growth rates of over 1% per annum, for some villages in Japan in the early eighth century . Barker and Rasmussen (1998: 102) observed that numerous very high ages are recorded on Etruscan funerary inscriptions, although these records are of course not necessarily reliable (Hopkins (1966) ).
Geographical contrasts
273
Once P. falciparum malaria has become endemic, its effects are nowhere near as spectacular as those of a sudden, major epidemic of plague ( Yersinia pestis), such as the Black Death or the plague of Justinian. Nevertheless the depression of life expectancy at birth which it caused in the long run in historical European populations appears to match the magnitude of the effects of plague, since life expectancy at birth in Grosseto in the nineteenth century was apparently not dissimilar to that in Florence during the period of major plague epidemics. Shortly after the end of antiquity, the Rome of Pope Gregory the Great was unfortunate enough to have both bubonic plague and endemic malaria at the same time, the end of the
