Table 4. Probability of death (qx) at various ages (in %) Interval
Treppio
South 8
South 9
(Females)
(Males)
1q0
19.6
18.5
19.0
5q1
16.5
18.5
16.2
50q20
26.5
26.5
27.0
e 0
37.0
37.5
38.5
Incidentally, the data for the age distribution of mortality of the population of Treppio, the Appennine community located at high altitude to which del Panta compared Grosseto, are very similar to model life-tables with similar levels of life expectancy at birth ( Table 4). This shows that at the very same time when some Italian populations had severely atypical age-structures as a result of ¹³⁷ Arlacchi (1983: 182).
¹³⁸ Arlacchi (1983: 176–83) on the Crotonese; Bonelli (1966: 662 n. 5). On malaria in Calabria see also Douglas (1955: 293–300), a perceptive account by a traveller who realized that the physical environment has changed substantially over the last two thousand years and appreciated the importance of these changes in relation to malaria. He reached the following conclusion: ‘Malaria is the key to a correct understanding of the landscape; it explains the inhabitants, their mode of life, their habits, their history’ (p. 300). Levi (1945: 156–7) described the effects of malaria in Lucania. Genovese (1924: 56–126) described the distribution of malaria in Calabria in recent times.
162
Demography of malaria
malaria, other populations in Italy not affected by malaria had quite normal patterns.
The population of Grosseto had three distinctive features: (1) much lower life expectancy at birth; (2) much higher overall mortality; (3) an unusual and distorted age-specific distribution of mortality. The third feature merits some further analysis. Table 5
shows that age-specific mortality was higher than predicted in Grosseto from age 5 to 9 and from 20 to 50. These characteristics of age-specific mortality emerge not only from the comparison with the communities of Stia and Pratovecchio in the Casentino given by del Panta, but also from the comparison with model life-tables.¹³⁹
Table 5. Number of deaths per person-years lived between age x and x+n (m(x) ) Age-group
Grosseto
South 2
Grosseto
South 2
(Males)
(Males)
(Females)
(Females)
0–4
16.5
18.0
17.7
17.3
5–9
2.3
1.5
3.0
1.7
10–19
1.2
0.9
1.0
1.1
20–9
1.8
1.6
1.5
1.6
30–9
3.0
1.7
2.2
0.9
40–9
3.5
2.3
2.8
1.9
50–9
5.9
3.6
6.8
3.0
Note: Bold type indicates items which deviate significantly from the values predicted by the model life-tables.
Del Panta was undoubtedly right to explain the excess age-specific mortality in the 5–9 age group in Grosseto as a direct consequence of P. falciparum malaria, as in tropical African countries today. Since direct mortality among adults from malaria was low in Grosseto, del Panta explained the excess adult mortality in terms of synergistic interactions with respiratory and gastro-intestinal diseases. Very high mortality rates required very high fertility rates if the population was to have a chance of reproducing itself. Consequently populations badly affected by malaria, such as Grosseto and the Sardinian populations mentioned earlier, had both higher mortality and higher fertility levels than other populations. Del Panta showed that the marriage patterns of Grosseto favoured very ¹³⁹ Del Panta (1989: 21); del Panta (1997) on infant mortality.
Demography of malaria
163
high fertility levels. Elsewhere he has described the coastal regions of central and southern Italy with intense malaria as characterized by neolocal marriage, with simple nuclear families and a predominance of agricultural wage labour.¹⁴⁰ Gregorovius made the following observation in Latium:
They marry very early in these parts—a young fellow of twenty one chooses frequently a girl who has only numbered fifteen summers.¹⁴¹
Table 6 demonstrates that an age-specific mortality pattern with some similarities to the data from Grosseto can be identified in the malarial parishes in the marshlands of south-east England.¹⁴²
Table 6. Number of deaths per person-years (m(x) ) for various age-groups Age-group
Marsh parishes
Model West 6
(Females)
0–4
9.5
10.0
5–9
0.9
0.9
10–14
1.1
0.7
15–19
1.3
1.0
20–9
2.0
1.3
30–9
2.7
1.6
40–9
4.2
1.9
50–9
4.7
2.9
60–9
5.9
5.7
Note: Bold type indicates items which deviate significantly from the values predicted by the model life-tables.
The demographic pattern found by Dobson in the English
marsh parishes is not dissimilar to the pattern of Grosseto, but with differences in detail; this is not surprising taking account of the fact that P. falciparum malaria was absent from England, not to mention numerous other environmental differences between England and Italy. In the English marsh parishes there was no deviation of the mortality level from the model’s expectations in the 5–9 age group.
This is comprehensible, since no significant degree of mortality produced directly by P. vivax in this age-group is to be expected.
P. vivax does not produce death directly itself in the same way that ¹⁴⁰ Del Panta et al. (1996: 162–4); Livi-Bacci (2000: 98–9, 145–6).
¹⁴¹ Gregorovius (1902: 90).
¹⁴² Data for the parishes of Canewdon, South Benfleet, Burnham and Tollesbury, which Dobson (1997: 169) compared to Coale and Demeny Model West Level 6.
164
Demography of malaria
P. falciparum does among children in tropical Africa today. The deviation in the English marsh parishes from the model pattern started at age 10, not 20, as in Grosseto, and steadily increased from the age of 30 onwards. Coale and Demeny Model West Level 6
gives a very good fit to English data up to the age of 20. However, Table 7 shows that this comparison is unsatisfactory from age 20
onwards.
Table 7. Number of people aged 20+ who die between ages x and y Interval
Marsh
West 6
West 2
West 1
parishes
20–9
18.3
14.1
19.1
20.7
20–39
36.8
28.8
37.7
40.4
20–49
58.6
44.1
53.5
56.7
20–59
75.9
61.2
69.9
73.0
20–69
87.4
80.3
87.0
89.1
Table 7 shows that from age 20 onwards the mortality rates predicted by Model West Level 6 are
far too low. The attested rates of attrition are roughly consistent with Levels 1 and 2, with a life expectancy at birth of between 20 and 22.5, rather than 32.5 as in Level 6. In so far as the Coale–Demeny models are of any relevance at all, the English data indicate a drop from Level 6 mortality in the first ten years of life to a lower level from ages 10 to 20, followed by a sharp drop down to Level 1 or 2 from age 20 onwards.
Consequently life expectancy at birth in the English marsh parishes was probably rather lower than 33, the figure suggested by Level 6.
This would not be surprising in view of the exceedingly high crude death rates for the marsh parishes, up to 80 per 1,000. Nevertheless a more important conclusion, for the purposes of this chapter, is that when all the obvious environmental differences between the English marshlands and western central Italy are considered, the mortality patterns produced by P. vivax in England and the combination of P. falciparum and P. vivax in western central Italy were remarkably similar. Both were characterized by very excessive age-specific adult mortality relative to the prevailing levels of infant mortality.
Similarly Tognotti described deviations in the age-structure of mortality on Sardinia, which had some of the most intense malaria in the western Mediterranean. In Sardinia infant mortality in the Demography of malaria
165
first year of life was actually below the average of all the various regions of Italy (including regions where malaria did not occur at all).¹⁴³ In fact, infants in the first few months of life seem to be less severely affected by malaria than older infants. A variety of possible explanations have been offered for this phenomenon.¹⁴⁴ One possibility is that infants sleeping alongside their mothers have a much smaller surface area than their mothers, and move around more even when asleep, and so are less likely to attract mosquito bites.
Infants may also be carrying antimalarial antibodies derived from their mothers in the first few weeks after birth, although this may simply indicate a high transmission rate of malaria and have little effect on infections. Malarial parasites grow much more slowly in erythrocytes with foetal haemoglobin than in cells with the adult form of haemoglobin. Another possible explanation, noted in Chapter 5. 3 above, is that human breast milk contains an extremely low concentration of para-aminobenzoic acid, a chemical required by malaria parasites. These factors probably all interacted to reduce mortality and morbidity from malaria in very young infants on Sardinia. Nevertheless after the first year of life on Sardinia, the risk of mortality increased progressively until in the 10–15 age-group (normally the healthiest segment of any human population) mortality was higher in Sardinia than in any other part of Italy. The mortality regime as a whole of the human population of Sardinia was worse than that of any other region of Italy. The situation on Sardinia was fundamentally the same as in Grosseto and the English marshlands, namely that infant mortality was not a reliable guide to mortality levels in older age-groups. Human populations which are severely affected by either P. falciparum or P. vivax or both under the transmission rates and seasonality typical of temperate to subtropical climates exhibit distinctive and much more severe adult-mortality patterns which distinguish them from populations unaffected by any species of malaria. Demographic regimes in history characterized by excess adult mortality relative to infant mortality can also be produced by causes of death other than malaria. The recently published family reconstitution studies of English parishes have shown that until the eighteenth century the English population as a whole had adult mortality levels higher ¹⁴³ Tognotti (1996: 81–2 n. 11).
¹⁴⁴ Brabin et al. (1990); Riley et al. (2001) surveyed the possibilities.
166
Demography of malaria
than those predicted by the Coale–Demeny model life-tables for the prevailing levels of infant mortality:
Viewed in terms of the Princeton North or West model life-tables adult mortality was far too high relative to rates in infancy and childhood in the seventeenth century. If the only information available were the adult rates, and one were to extrapolate from them to estimates of expectation of life as a whole, using the Princeton tables, the result would be a radical underestimate of expectation of life at birth.¹⁴⁵
Nevertheless it must be remembered that mortality as a whole was lower in non-marsh parishes in England than in the marsh parishes ravaged by P. vivax. Malaria was the most powerful cause of these atypical patterns—atypical by modern, but not necessarily by pre-modern, standards. In the pre-modern world infectious diseases were vastly more important than they are today. The Coale–Demeny life-tables, which assume that decreasing levels of life expectancy at birth can be explained above all in terms of increasing levels of infant mortality, fail to pay enough attention to additional adult mortality caused by infectious diseases in historical populations. Similar patterns of excess adult mortality relative to infant mortality have also been observed in the historical demography of India. It has been suggested that tuberculosis was the most important cause of the atypicality in India, but malaria (especially P. vivax) has been very important historically in many parts of India.
Learmonth noted the striking correlations in Bengal until recently between, first, areas with intense malaria and areas with static or decreasing populations, and, secondly, districts with little or no malaria and districts with growing human populations. Malaria undoubtedly interacted with tuberculosis in India.¹⁴⁶ These atypical mortality patterns have had a wide geographical spread in recent times: England, Italy, India, and East Asia. It is a reasonable ¹⁴⁵ Wrigley et al. (1997: 349, cf. 261–3, 284–5).
¹⁴⁶ Mari Bhat (1989) on Indian demography, and Learmonth (1988: esp. 5–7, 206–7) on malaria in India. Even if the Sanskrit texts mentioned in Chapter 3 above do not definitely associate malaria with mosquitoes, other Sanskrit texts do describe malaria itself (a demon called takmán frequently found on lowlying land) very clearly, differentiating quotidian, tertian and quartan fevers: Zysk (1985: 34–44); Raina (1991: 1–4). Hirsch (1883: 204–7) described the distribution of malaria in India in the nineteenth century, and Klein (1972) its devastating effects in Bengal. Mari Bhat (1989: 110–11) also suggested, using the Barclay data, that the age-structure of mortality of traditional Chinese populations diverged from the model life-tables even more than that of Indian populations did, extending the patterns under discussion here even further.
Demography of malaria
167
hypothesis that they also have a long history. The balance of probability is that human populations that were affected by malaria in Latium and Tuscany in antiquity also shared these atypical patterns with high infant mortality but even higher adult mortality.
Modern quantitative knowledge of the extreme effects of malaria on human demography indicates that Toscanelli was right to suggest in 1927 that the spread of malaria from c.300 onwards did play a major role in the decline of the southern and coastal Etruscan cities, in exactly the same way that malaria led to the depopulation of the Pontine Marshes. From a methodological viewpoint, the conclusions reached here suggest that it is a mistake for ancient historians to assume that the model life-tables necessarily encompass the entire range of possibilities as far as the demography of human populations in antiquity is concerned.¹⁴⁷ It is now time to investigate the operation of some of the general principles which have been discussed so far in detail at the local level in the various environments of western central Italy. Let us start with the most notorious focus in Latium, the Pontine Marshes.
¹⁴⁷ The fact that extrapolations from data for adult mortality in historical populations often yield underestimates of life expectancy at birth is relevant not only to Roman demography, but also to the demography of classical Athens. Although this cannot be explored in detail here, it is worth noting in passing that the ratio between the ephebes and the arbitra-tors in Athens in the fourth century (a measure of adult mortality), which has often been used to yield very low estimates of life expectan
cy at birth following comparisons with model life tables, probably underestimates e0, in the light of the present discussion. This provides further support for the view advocated by Sallares (1991: 113–14) that life expectancy at birth in classical Athens has been underestimated. If infant and juvenile age-groups were in fact healthier in classical Athens than is generally supposed by historians, this helps to explain many important problems of Greek history; for example, how the Athenian citizen population was apparently able to recover very rapidly from repeated military catastrophes (as well as the ‘plague of Athens’ in 430 ) during the period of the fifth-century empire (cf. Sallares (1991: 95–9) ).
6
The Pontine Marshes
The Pontine Marshes have attracted little attention in modern historiography. The most comprehensive twentieth-century accounts of their ancient history were written by Bianchini, a rare book which attracted little attention owing to its publication at the beginning of the Second World War, and Hofmann, a very long Pauly-Wissowa article described by Brunt as ‘remarkable for its ready acceptance of annalistic details and lack of scientific data’.¹ Early modern descriptions of the region are very important.² In fact, virtually all the historical questions considered by twentieth-century historians had already been debated by writers in the eighteenth century. In the year 1800 Nicola Maria Nicolai published a very substantial work on the Pontine region consisting of four books.
Malaria and Rome: A History of Malaria in Ancient Italy Page 24
