2.88, 90.
¹⁷⁶ Judson (1968); Potter (1979: 120–37) recorded large numbers of villas by c. 100 (with slave labour forces), but few significant towns occupied by a free population.
Ecology of malaria
113
Substantial erosion and redeposition of sediments has also occurred since the classical period. The consequence is that the modern landscape is quite different from the ancient one. Recent research in the Fiora, Marta, Treia, and Valchetta river valleys has drawn a very sharp contrast between the ancient and the modern river systems:
The sedimentology of the coarse deposits suggests that Roman rivers and floodplains were very different to those existing today. They were characterized by shallow channels, actively migrating, depositing bars of gravel.
Some reaches of the contemporary streams are trying to re-establish this condition but only to a limited extent due to confinement by high cohesive sandy-silt banks.¹⁷⁷
The ancient fluvial regime had very important consequences for malaria. The laterally mobile Roman rivers in shallow channels would have been more liable than their modern counterparts in deeper channels to flood the surrounding countryside. As the flood waters retreated, leaving pools here and there, mosquito breeding sites would have been created.¹⁷⁸ Thus it is likely that the rivers of Latium and Tuscany in the Roman period had a greater propensity to create mosquito breeding sites than they do today. Tibullus was right to fear the rivers of Latium and Tuscany in summer.
The overall effect of all the landscape changes which have been described was to provide more breeding sites for mosquitoes in general in Roman times. In coastal areas, if the water was brackish, those species of Anopheles mosquito that are the most dangerous vectors of human malaria had an advantage over other species of Anopheles. Even in Lazio, suitable breeding sites for mosquitoes can be scarce. In 1928–9 the lake at Nemi was partially drained to uncover Caligula’s ships, inadvertently providing new breeding sites for mosquitoes. In 1929 a sudden epidemic of P. falciparum malaria infected over half the population of the town, which had no previous history of endemic malaria in its hilly location (320 metres above sea level). Prior to the drainage operations there were not enough mosquitoes around to sustain the transmission of malaria in Nemi, given that the chances against transmission by any individual mosquito are extremely high, since most mosquitoes do not ¹⁷⁷ Brown and Ellis (1995: 69).
¹⁷⁸ An example of periodic flooding at a specific site is the Roman mausoleum at Fosso della Crescenza along the Via Veientana in southern Etruria (S. Judson in Fentress et al.
(1983: 70–2) ). On the rivers of southern Etruria see also Rendeli (1993: 122–5).
114
Ecology of malaria
live long enough to transmit the disease.¹⁷⁹ P. falciparum, the most dangerous species of human malaria, is a very ancient parasite which was probably present in Mediterranean countries on a localized basis from the Neolithic period onwards, as has already been argued. It can certainly continue to exist in very small foci without spreading. It was not a new disease in classical times. Nevertheless the conclusion reached here is that the frequency of all types of malaria increased very substantially in Roman times primarily because of human modifications of the landscape which unintentionally favoured the vector mosquitoes. The demographic effects of this expansion on human populations demand detailed examination now.
¹⁷⁹ Hackett and Missiroli (1931: 64–5) on Nemi.
5
The demography of malaria
5. 1 D
Bonelli rightly observed that in order to understand the impact of malaria on Italy in the past it is not enough simply to consider its demographic effects in purely quantitative terms.¹ It also influenced the entire lifestyle of people, even if it did not kill them, by forcing them to live away from the fields in which they had to work. In other words, malaria altered settlement patterns.² Old Salpi in Apulia was the most dramatic example of this in antiquity (see Ch. 10 below), but malaria also influenced settlement patterns within the city of Rome itself, as will be seen later (Ch. 8 below).
Malaria influenced agricultural systems by preventing intensive farming practices (Ch. 9 below). Nevertheless consideration of the demographic effects of malaria in purely mathematical and statistical terms is essential to understand the scale of its impact. The most fundamental weakness of nearly all recent literature on the question of the effects of malaria in Italy in antiquity is the absence of any appreciation in quantitative terms of the demographic effects of malaria on human populations in Europe in the past. This lack of knowledge about the demographic effects of malaria on human populations has permitted many researchers from different branches of scholarship to minimize the importance of malaria.
A good starting-point is the recent debate between Scheidel and Shaw concerning the explanation of excess seasonal mortality in the city of Rome in antiquity, as revealed by funerary inscriptions.³
Scheidel opted for malaria, in synergistic interactions with other diseases, as the principal cause. Shaw attempted to minimize the role of malaria. It is instructive to consider the flaws in Shaw’s argument. Shaw relied on national statistics which show that malaria ¹ Bonelli (1966: 662).
² Pinto (1982: 30) saw malaria as a determinant of settlement patterns in Tuscany in the medieval period.
³ Scheidel (1996); Shaw (1996).
116
Demography of malaria
accounted for 2.2% of all deaths in Italy in 1887–9. He concluded that malaria was less important in Italy than in Greece, where 5.6%
of all deaths were attributed to malaria as recently as 1921–32.⁴ The most detailed statistics can be extremely misleading if they are not handled very carefully indeed. In this particular case, there are two major reasons why these national statistics for cause of death are unlikely to give an accurate estimate of the contribution of malaria to mortality in the city of Rome or in the Roman Campagna in antiquity.
The first rather elementary reason is quite simply that these national statistics lumped together mortality data from localities with intense malaria with data from places where there was no malaria at all to yield meaningless averages which may be completely inappropriate if applied to any particular locality. It is always possible in demography to aggregate data from two or more different regions and calculate the average, but the crucial question is whether the average so obtained is meaningful. As del Panta said in an article on infant and child mortality in early modern Italy: No satisfactory explanations can be given for the evolution of child and infant mortality in Italy without considering territorial differences. In fact, national values result from quite different regional and provincial levels.⁵
In localities in Italy where malaria was endemic the proportion of deaths which were directly attributed to malaria (the significance of this phrase will be seen shortly) was far higher than the national statistics for mortality indicate. Bonelli gave as examples for 1882
Rossano in Calabria, Grosseto, and Paola in Calabria where 24.8%, 23%, and 19.5% respectively of all deaths were directly attributed to malaria. An earlier, less complete database of causes of death yielded a figure of 19% of all deaths directly attributed to intermittent fevers in Grosseto in 1840–1. This result is of the same order of magnitude as the result for 1882. However, it is probably an underestimate since the surviving records for 1840–1 excluded almost entirely causes of death for infants and children under the age of five, who would be expected to suffer severely from P. falciparum malaria when it is endemic.⁶ Bonelli stated that ‘in the 1880s ⁴ Shaw (1996: 133 and note 108); Balfour (1935: 302) on Greece.
⁵ Del Panta (1997: 10).
⁶ Bonelli (1966: 662); del Panta (1989: 48–9 n. 23), using the statistics of Salvagnoli Marchetti.
Demograp
hy of malaria
117
it was quite common in the Mezzogiorno for malaria to account for 20–30% of all deaths . . . in numerous other localities malaria accounted for 10–15% of all deaths’.⁷
There are two lessons to be drawn from this. First, in any kind of comparative history it is essential to look for appropriate parallels.
Since the ancient medical authors Asclepiades and Galen provide positive evidence that P. falciparum, the most dangerous species of human malaria, was common in ancient Rome (see Ch. 8 below), the appropriate parallels are places in more recent periods of Italian history where malaria was common, such as Grosseto, not places where it did not occur at all. A good example of a comparison that should be used as a source of contrasts with the city of Rome in antiquity, not as a source of parallels, is Florence at the time of the famous Catasto in 1427.⁸ The Catasto is of course a very important historical document, to which further reference will be made in Chapter 11 below. However, the demographic situation in Florence then was fundamentally different from the situation in ancient Rome for (at least) two major reasons: (1) Florence in the late medieval period was affected by epidemics of bubonic plague, for which there is no evidence in Rome in antiquity; (2) Florence had no history of endemic malaria.⁹ Since there were significant differences in the causes of death between imperial Rome and late medieval Florence, there is no reason whatsoever for expecting the mortality patterns of ancient Rome and late medieval Florence to be similar. It follows that it is essential to investigate the causes of death in detail at the local level, before choosing comparisons for demographic purposes. Shaw did not consider the important evidence of Asclepiades and Galen. He stated that ‘it seems clear that it is the factor of temperature variability that marks the main separation between “northern” and “southern” regimes of seasonal mortality’, but failed to notice that out of the various diseases which he considered malaria is the only one in fact which is strongly temperature-dependent.¹⁰ Both what he termed ‘normal’ diseases, such as tuberculosis, and epidemic diseases, such as typhus, flourish equally well in Africa and in northern Europe. In 1998 there were ⁷ Bonelli (1966: 662): negli anni ’80 era abbastanza frequente nel Mezzogiorno che il numero dei morti per malaria rappresentasse il 20–30% del totale dei morti . . . numerose altre località presentano una mortalità per malaria compresa tra il 10 e il 15% della mortalità generale.
⁸ Herlihy and Klapisch-Zaber (1985).
⁹ Carmichael (1989) on mortality patterns in late medieval Florence.
¹⁰ Shaw (1996: 132).
118
Demography of malaria
outbreaks of epidemic typhus in Burundi in Africa and in Russia.¹¹
Similarly Shaw noted the gradual disappearance of the excess summer mortality in Italy from the 1860s to the 1960s, but did not notice the almost exact correlation of this trend with the gradual eradication of malaria. In contrast Brown found a very strong statistical correlation between mortality trends in Sardinia and malaria eradication over the very same hundred-year period.¹² Del Panta observed that ‘the official statistics can only reveal, however, the phase in which mortality from malaria was declining, starting in 1887’. The data for causes of death upon which Shaw relied postdate the beginning of the demographic transition towards the modern low-mortality and low-fertility demographic regime in Italy. Consequently these data are of limited value for the pre-modern period. Del Panta stated that the transition towards the modern low mortality demographic regime commenced in Tuscany in the period 1875–80.¹³
The second lesson to be drawn from examining Shaw’s use of national statistics is that in demography all valid studies must commence at the local level, with, for example, the sort of parish studies which formed the foundation for the monumental population history of England by Wrigley and Schofield. The failure to appreciate the importance of local variation in demography, such as the differences between Grosseto and Treppio noted at the beginning of this book (see also Ch. 5. 4 below), is the single biggest deficiency of the bulk of research in ancient Roman demography carried out over the last thirty years or so. One of the main conclusions reached in this book is that it is in fact impossible to generalize about the demography of a region as small as central Italy in antiquity, never mind the whole Roman Empire.
There is a second very important reason why the national statistics for mortality in 1887–9 in Italy are an unreliable guide to the likely impact of malaria on mortality patterns in antiquity. As a specialist in malariology put it, ‘reporting of deaths by cause has been of limited use for the investigation of the impact of malaria and its control on mortality’.¹⁴ The reason for the unreliability of ¹¹ Raoult et al. (1998) and Tarasevich et al. (1998) on typhus.
¹² Shaw (1996: 112); Brown (1986: 857).
¹³ Del Panta et al. (1996: 292 n. 10): le statistiche ufficiali possono cogliere, purtroppo, solo la fase dis-cendente della mortalità per malaria, a partire dal 1887; del Panta et al. (1996: 152); Sori (1984: 542–3).
¹⁴ L. Molineaux in Wernsdorfer and McGregor (1988, vol. 2, p. 974).
Demography of malaria
119
national statistics is that ‘medical science shows that for each death attributed to malaria there are several other deaths, which are attributed to other causes, but nevertheless are directly linked to malaria or indirectly caused by the debilitating effects of malaria infections’.¹⁵ The empirical demographic evidence for this conclusion is that it has frequently been observed, both in Italy in the past and in tropical countries today, that after the eradication of malaria the reduction in total mortality rates is much larger than the reduction that would have been predicted from the proportion of deaths that were directly attributed to malaria before its eradication. For example, Hackett described the town of Sermoneta on the edge of the Pontine Marshes in 1925, just before eradication started. The population of Sermoneta, which could not reproduce itself, had a crude death rate of 41 per 1,000. All the children had splenomegaly, but only a dozen individuals, mainly children, had acute symptoms of malaria, because almost all adults had developed acquired immunity earlier in life. Only 8% of all deaths were directly attributed to malaria in 1925 (in absolute numbers, three out of thirty-seven). Nevertheless the eradication of malaria in Sermoneta led to a fall in the crude death rate from 41 per 1,000 to 20 per 1,000.¹⁶
Although the problems of Sermoneta in the early twentieth century were a product of the early modern period, there is no doubt that other communities in western central Italy were affected in the same way much earlier. For example, Toubert quoted the decree of Pope Innocent IV in 1253 granting permission to the inhabitants of Mozzano to move their village to a new location, because of ‘bad air’ in summer. This text clearly describes virtually the entire population of Mozzano in the thirteenth century as severely affected by malaria, and shows malaria directly altering human settlement patterns in that period. (Since the town had been built in a place that was so pestilential and unhealthy, owing to bad air, especially in summer, almost all of its inhabitants suffer from severe disease and chronic lethargy . . . we shall allow them to move to the place called Colle Vecchio to live in the territory of the town set aside for them there).¹⁷
¹⁵ Bonelli (1966: 662): la scienza medica mostra che per ogni morto classificato come deceduto a causa della malaria ce ne sono parecchi altri classificati come morti per ‘altri’ eventi morbosi, i quali, peraltro, sono alla malaria direttamente connessi o indirettamente dovuti al processo di debilitazione fisica delle persone col-pite.
¹⁶ Hackett (1937: 237); Molineaux (1997); Brown (1986); Giglioli (1972).
¹⁷ Toubert (1973: i. 363–4 n. 3): cum castrum in tam pestifero et corrupto sit loco constructum quod
120
Demography of malaria
19. A view of Sermoneta, a hill town dominated by the Castello Caetani, with a history of contact with malaria. The following saying, quoted from the guidebook, Sermoneta: sto
ria del paese (1989) by L. Sciotti and A. Nastri, refers to malaria: Sermoneta che stai ncima a na fossa, | ntorno ntorno ci sta l’acqua puzza:| a le femmene fa cresce la trippozza, | a gli ommeni ce cala la cocozza.
When malaria eradication caused a reduction in overall mortality of more than 50%, as it did in Sermoneta, it seems reasonable to conclude that malaria did in fact dominate the mortality regime before propter aeris intemperiem, aestivo praecipue tempore, habitatores ipsius quasi omnes infirmitates graves et diutinos languores incurrunt . . . transferendi se ad locum qui Collis Vetulus [modern Collevecchio] nun-cupatur . . . in territorio castri predicti ad inhabitandum inibi . . . concedere curaremus.
Demography of malaria
121
eradication, even if the bulk of its effects were indirect rather than direct. In tropical countries similar results followed the eradication of malaria. One study estimated that in Ceylon the total number of deaths in which malaria played a role was 4.7 times that of the deaths caused directly by malaria, while in the coastal regions of British Guyana the total number of deaths involving malaria was 3.8 times that of those directly attributed to it.¹⁸ The standard view of malariologists is that ‘the mortality attributable to malaria . . .
Malaria and Rome: A History of Malaria in Ancient Italy Page 17
