Comparisons of mortality from childhood to old age

9In Table 1 we list the ratios of values of q_x for men to those for women from birth to ages 75 to 80 for most Western European countries and for Geneva during the past, taking the figures as far back as the data permit : the seventeenth century for Geneva, the eighteenth for France and Sweden, the mid-nineteenth century for England and Wales, Belgium, Norway and the Netherlands, 1875 for Germany and Italy, and the very end of the nineteenth century for Spain. In the case of France, Geneva and Sweden [1], we show similar ratios for more recent periods to make it possible to assess earlier movements during the beginning and the middle of the nineteenth century, as the later development in these countries is better known [2].

10The most striking feature of Figure 1 in which the data from Table 1 are plotted is the similarity of the curves for most of the countries. The ratios decline beginning with the age group 0-4 years, where they always exceed unity, to reach a minimum which, though situated at different ages in different countries or periods, always occurs before the age of 20 years. After that age the curves rise again and generally reach a maximum value between the ages of 20 and 24 ; this is followed by a renewed fall which continues generally speaking from the age group 25-29 to that of 40-44 years. A second maximum is normally found between the ages of 45 and 55 years, and from then onwards the curves fall again up to the highest ages considered.

Table 1

Ratios of mortality rates for men to those for women (5-year age groups) for various countries and periods (p. 100)^{(1),(2),(3),(4),(5),(6),(7),(8),(9),(10),(11)}

Age (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Cohort 1740-1789 1840-1859 1625-1699 1700-1790 1841-1850 1840-1851 1838-1854 1856-1865 1751-1800 1851-1900 1871-1881 1865-1875 1871-1880 1900 0 107 109 109 114 108 109 111 110 111 115 110 111 107 105 5 99 96 123 109 87 98 101 97 107 104 101 105 96 97 10 110 85 118 85 102 73 94 95 111 99 92 100 89 87 15 105 90 101 79 85 113 95 106 110 103 108 103 94 94 20 120 145 112 105 81 111 96 133 129 126 117 121 110 117 25 96 109 83 83 126 104 95 122 119 115 99 110 89 117 30 86 93 73 78 116 87 96 105 106 111 97 100 88 106 35 86 93 77 91 116 81 98 99 112 111 103 101 94 104 40 100 101 135 87 98 108 104 110 111 120 120 111 113 117 45 115 107 116 81 92 114 111 113 130 133 134 119 126 50 113 104 105 112 91 129 120 120 130 133 131 114 114 126 55 105 104 127 135 126 130 114 122 126 130 122 107 119 60 100 103 100 112 152 111 111 120 115 122 112 102 111 65 98 98 96 107 119 107 110 112 113 116 108 99 106 70 98 97 88 104 99 101 109 106 106 114 105 98 102 75 101 102 92 90 101 103 108 110 106 113 102 97 101 80 102 87 89 108 106 100 106 109 102 97 101

Ratios of mortality rates for men to those for women (5-year age groups) for various countries and periods (p. 100)^{(1),(2),(3),(4),(5),(6),(7),(8),(9),(10),(11)}

Sources :
(1) France 1740-1789 : Yves Blayo, « La mortalité en France de 1740 à 1829 », Population, 1975, numéro spécial "Démographie historique", p. 123-142. France 1840-1859, Belgium, Netherlands, England, Norway, Germany : Statistique internationale du mouvement de la population d’après les registres d’état civil, des origines à 1905. Paris, Imprimerie nationale, 1905.
(2) Geneva : Alfred Perrenoud, « Surmortalité féminine et condition de la femme (xvii^e-xix^e siècles). Une vérification empirique », Annales de Démographie historique 1981, pp. 89-104.
(3) Belgium
(4) Netherlands
(5) England
(6) Norway
(7) Sweden : Historical Statistics of Sweden, part 1, Population, 2^nd edition, 1720-1967, Stockholm, National Central Bureau of Statistics, 1969, 133 p.
(8) Germany
(9) Austria : Österreich Sterbetafeln, Vienne, Österreichischen Statistischen Zentralamt, 1967.
(10) Italy : Sommario di statistiche storiche italiane, Rome, Istituto Centrale de statistica, 1968.
Note : For Italy, the ratio of ordinary age-specific rates has been computed, instead of the ratio of probabilities of death.
(11) Spain : Tablas de mortalidad de la poblacion espanola 1900-1940, Madrid, Instituto nacional de estatistica, 1952.

Figure 1

Ratios of male to female death probabilities (per 100)

Ratios of male to female death probabilities (per 100)

Figure 2

France 1740-1789. Ratios (per 100) of male to female death probabilities assuming no maternal mortality

France 1740-1789. Ratios (per 100) of male to female death probabilities assuming no maternal mortality

11The sole exception to this pattern is found in the Belgian figures for 1841-50, where the two maxima and the intervening minimum occur between five and ten years later than elsewhere. The same peculiarity is found in the table used by Tabutin for the period 1847-56, but it does not appear in the table for 1881-90, where the pattern is very similar to that of most of the other tables.

12In Geneva the pattern is disturbed by random fluctuations due to the small numbers involved, but the general pattern remains more or less the same as it does in other populations. In England and Wales, the maximum between the ages of 20 and 24 is hardly noticeable, and the impression is one of a constant ratio between ages 10 and 34 [3].

13The similarity of the ratio of the mortality of men to that of women for different periods and different countries suggests that this ratio exceeds or falls short of its average at roughly speaking the same stages of life. This similarity of form is, however, consistent with differences in the value of the ratio. We need to answer the question : at what ages does the mortality of one sex exceed that of the other, and at what ages are their mortalities the same ? As it is rare for mortality of the two sexes to be equal we need only distinguish two cases : those in which there is and those in which there is not a higher mortality of women. In the latter case, there will almost always be a higher mortality for men.

14We shall successively examine mortality in the age groups 5-19, 20-24, 25-44 and 45-64 years. In looking at the first of these age groups we shall denote areas in which the mortality of women is significantly higher (a ratio lower than 0.97) by X, those in which it is only slightly higher (0.98 or 0.99) by x, and leave blank cases where the mortalities of the two sexes are equal or where the mortality of men is higher.

The age group 5-19

15In the 14 observations retained, the mortality of women was higher in 11 cases, but in two of these (France 1740-89 and Sweden 1851-1900) the excess was negligible and occurred in only one of the quinary age groups which make up this group. The three cases in which there was no excess mortality of women are : Geneva during the seventeenth century, Sweden between 1751 and 1800, and Austria between 1865 and 1875. In the 11 tables in which the mortality of females before the age of 20 was higher, an excess was found between the ages of five and nine years in seven cases, between the ages of 10 and 14 in nine cases, and between the ages of 15 and 19 in six cases.

16Higher mortality of women in only one of the quinary age groups is found in three tables, in two quinary age groups in five tables, and in all three quinary age groups in three tables : France 1840-59, Italy about 1875, and Spain at the end of the nineteenth century.

17At first sight it would seem that higher mortality of females in at least one of the quinary age groups between the ages of 5 and 19 years is the rule. However, the earliest observations we have throw doubt on whether any such rule really exists. Alfred Perrenoud has drawn attention to the fact that the trend of excess mortality was reversed between the seventeenth and eighteenth centuries, and he sees this change as a consequence of "changes in the condition of women" [4].

18The figures in Table 2 show that excess mortality of women in France increased between 1740-89 and 1840-59. In Sweden, where there was no excess mortality in any of the age groups during the second half of the eighteenth century, it only became apparent during the second half of the nineteenth, and this conclusion is confirmed by Tabutin’s paper which we have already quoted. In Tables 4 and 6 given in the Annex to his paper, an excess mortality of girls between the ages of 5 and 9 is shown in the Swedish table for 1891-1900, and at ages 10-14 in the table for 1901-10. It is even more clearly shown in the tables covering the period 1871-80 to 1921-30. Between the ages of 15 and 19 a higher mortality for women is found in the tables for the period 1881-90 to 1901-10. In France, excess mortality between the ages of 5 and 14 first appears in the table for 1815-17 and increases up to the end of the century. In contrast, Table 5 in the Annex for England and Wales shows no increase in women’s excess mortality during that period. In a study of the higher mortality of little girls in Belgium [5], this excess declined in the period after 1846, the earliest year mentioned in the study.

Table 2

Cases of higher mortality for women

Age group (1) (2) (3) (4) (5) (6) (7) (8) (10) (11) Cohort 1740-1789 1840-1859 1625-1699 1700-1790 1841-1850 1840-1851 1838-1854 1856-1865 1851-1900 1871-1881 1871-1880 1900 5-9 x X X x X x X 10-14 X X X X X x X X X 15-19 X X X X x X 20-24 X X 25-29 X X X X x X 30-34 X X X X X X X X 35-39 X X X X X X x X 40-44 X X 45-49 X X x Insignificant X Significant (1) France (2) Geneva (3) Belgium (4) Netherlands (5) England (6) Norway (7) Sweden (8) Germany (10) Italy (11) Spain

Cases of higher mortality for women

19However, in another study it has been suggested that in England and Wales, too, the mortality of little girls was no higher than that of boys of the same age before the beginning of the nineteenth century [6]. An analysis based on a sample of eight parishes chosen from villages and towns in very different regions shows that little boys suffered higher mortality than little girls between the ages of five and nine years during the period 1550-99 and 1600-49. This was followed by an excess mortality of girls in 1650-99 and 1700-49. The ratios of the probabilities of dying for men to those for women come to 1.14, 1.05, 0.93 and 0.95. In London, excess mortality of boys was clearly apparent between the ages of five and nine years with a ratio of 1.09, and between 10 and 14 years with a ratio of 1.55. The ratios tended to be higher in poorer than in wealthier parishes. In the latter there was an excess mortality of girls between the ages of five and nine years, the ratio being 0.83, compared with 1.31 in the poorer parishes.

20The same inequality is found in Geneva between the ages of 10 and 19 during the earliest period (Table 6 on p. 96 of the article cited). The ratio of probabilities comes to 0.74 among the bourgeoisie, 0.86 among artisans and skilled workers, and 1.23 among employees and unskilled workers. But this order is reversed during later periods when the figures are 1.17, 1.08 and 0.72 respectively. During the earlier period the variation between different groups in the 5-9 age group was the reverse of that found in the 10-19 age group ; however, during recent years the differences have become less important.

21These results suggest that the mortality of little girls and adolescent women was not always higher than that of boys of the same age.

22It is necessary to investigate the reasons for this excess mortality, but it is difficult to obtain information about cause-specific mortality for earlier periods. According to a study quoted by Tabutin [7] mortality of women in England and Wales between 1848 and 1872 between the ages of 5 and 14 and 15 and 24 years (the age group 15-19 was not studied separately) was higher than that of men and was caused mainly by deaths from infectious diseases, of which the most important was pulmonary tuberculosis (Table 3). At that time deaths from infectious diseases accounted for 62 % of men’s deaths and for a little less than 70 % of those of women in these age groups. Pulmonary tuberculosis caused 8.8 % of all deaths of boys between the ages of 5 and 14 years and 40.4 % of those between the ages of 15 and 19. The corresponding percentages for women’s deaths were 12.8 and 47.9.

Table 3

Ratios (p. 100) of mortality rates for men to those for women, by cause of death (England and Wales, 1848-1872)

All causes Infectious diseases Other causes Age All Pulmonary tuberculosis 5-14 101.5 90.8 69.3 125.1 15-24 96.9 88.5 81.9 114.7

Ratios (p. 100) of mortality rates for men to those for women, by cause of death (England and Wales, 1848-1872)

23About 20 years later, between 1870 and 1890, the excess mortality of women from tuberculosis in Belgium was definitely more pronounced than in England between 1848 and 1872. The ratio of probabilities of dying for men to those for women was 0.54 between the ages of 7 and 15 and 0.73 between 15 and 21 [8]. At a later date, between 1895 and 1910, the values of these ratios were more like those found earlier, 0.58 and 0.68 respectively. For the latter years, we also know the proportion of deaths from pulmonary tuberculosis, 14.8 % between the ages of 7 and 15 and 22.6 % between 15 and 21 among men, the corresponding figures for women being 34.8 and 44.9 % respectively.

24Having presented these figures, we must consider the problem of data quality, as well as the probable reasons for the increasing excess mortality of young women during the period considered. As the same problems may be met in the higher age groups, we shall comment on the results first and return to the problem of quality at the end of the paper.

The age group 20-24

25It is shown in Table 2 that the mortality of men between the ages of 20 and 24 almost always exceeds that of women, and that exceptions to this generalization are rare. In Belgium women’s mortality may have exceeded that of men of these ages in the tables for 1841-50 and 1847-56. Evidence for the first statement can be found in Table 1, and evidence for the second in Tabutin’s paper ; however, in the same paper and on the same page there is clear evidence of an excess mortality of men for the period 1881-90.

26Higher mortality of women in this age group is also found in England and Wales in tables for the period 1838-54, but the tables used by Tabutin show a slightly higher mortality for men in 1841-45 and 1846-50, and a slightly higher mortality for women in 1851-55 and 1856-60. It would seem that at the time in England and Wales, women’s mortality was higher between the ages of 10 and 39 years, and that the excess mortality of men between the ages of 20 and 24, which seems to have been the rule elsewhere, cannot always be found in that country. But it is possible that deaths of members of the armed forces who died in British overseas territories did not figure in the English statistics.

27Excess mortality of men between the ages of 20 and 24 was particularly pronounced in France between 1840 and 1859 ; this could have been caused by military casualties, particularly in the Crimean war which took a heavy toll of lives. In other countries, military service which, as J. Houdaille has shown, leads to excess mortality even in times of peace could have been a reason. It is surprising to find such pronounced higher mortality of men even in countries with small armies.

28Accidents at work may have been another reason for this excess, but it is difficult to explain why they should have been concentrated in this age group at a time when young men started work before their fifteenth birthday and when it might have been expected that the victims would have been boys who had just started in the labour force and lacked experience rather than men who had already been working for some time.

29Statistics on causes of death might have helped to explain this phenomenon, but unfortunately Logan does not show these separately for the age groups 15-19 and 20-24.

The age group 25-44

30In almost all the countries and periods shown in Table 1 and Figure 1, the curve of the ratios reaches a minimum in this age group. The only exception is Belgium, both in 1841-50 and in the tables given by Tabutin. We shall not discuss this case, as we have no explanation for this aberrant phenomenon. Nine of the other 13 tables show higher mortality for women in at least part of that age group ; the four exceptions are the two tables for Sweden and those for Austria and Spain. In addition, excess mortality of women is only very small in Norway, and seems to have been limited to the age group 35-39 ; in the German table, too, excess mortality is slight between the ages of 25 and 29, but a little more marked between 30 and 34.

31In the remaining seven tables, a higher mortality is found for women everywhere between the ages of 30-34 and 35-39 years ; it extends to ages 25 to 29 in five of the seven cases, and in the other direction (40-44 years) in a single instance, the table for the 1700-1790 cohort in Geneva, where women between the ages of 45 and 49 seem also to have been affected, though this may be a chance effect.

32In the six cases where the mortality of women is no higher than that of men, or the excess is very low, the curves of the ratios of men’s to women’s mortality reach a minimum between the ages of 30 and 40 years, and this minimum extends to the 40-44 year age group in the Swedish table for 1751-1800.

33The age group 25-44 covers women’s reproductive period, and as the proportion of married women only tends to become significant after the age of 25, and because maternal mortality was not negligible in the past, it seems a priori almost certain that some of the excess mortality is the result of deaths associated with pregnancy and childbirth. It remains to be determined whether these causes are solely responsible for the excess mortality of women or whether other factors were operating as well.

34This question can only be answered when sufficient information is available on the size of the population and the numbers of births and deaths. Data are available for France for the period 1740-89 from an enquiry undertaken by INED. They can be supplemented by Swedish data on maternal mortality, as some maternal deaths, particularly those caused by pregnancy which did not result in the delivery of a child, were hardly recorded in the French parish registers [9].

35We have computed two indices of women’s mortality from causes other than maternal mortality. They are based on two different definitions of maternal mortality : the first, very restrictive, based on the French observations, in which a death is considered a maternal death only when it occurs during the confinement or is a direct consequence of confinement, together with a few other cases [10] ; the second, more comprehensive though it is, is not a maximum estimate [11], and includes Swedish data on deaths caused by pregnancy which did not necessarily end in confinement. The calculations are shown in the Annex [12].

36We have taken men’s probabilities of dying in Table A3 and corresponding probabilities for women in Table A6, and calculated their ratios on the assumption of zero maternal mortality. The results are shown in Table 4 and plotted in Figure 2.

Table 4

Ratios (p. 100) of mortality rates for men to those for women, for causes other than maternal mortality (France, 1740-1789)

Age Low estimate High estimate 15-19 106 106 20-24 130 131 25-29 114 118 30-34 103 110 35-39 100 107 40-44 108 115 45-49 115 116

Ratios (p. 100) of mortality rates for men to those for women, for causes other than maternal mortality (France, 1740-1789)

37If maternal mortality is assumed to be zero, the excess mortality of women disappears and is replaced by an excess mortality of men, except when the restrictive definition of maternal mortality is used in the age group 35-39, when the mortalities of the two sexes become equal. Given that the true rate of maternal mortality is likely to be higher and nearer to that given by the less restrictive definition, we would conclude that throughout the age group 25-44 men’s mortality would have been higher, in the absence of maternal mortality or if it had been as low as it has been during the modern period.

38It would be tempting to draw a similar conclusion for other cases in which the mortality of women was higher, especially when the minimum value of the ratio is lower than that shown for France. There is evidence to support such an interpretation from the English data. In a group of 13 English parishes, maternal mortality between 1800 and 1849 was 5.5 deaths per 1,000 births. From these data and from what is known about the development of maternal mortality in that country from the middle of the nineteenth century ownwards, we could estimate maternal mortality to be about 5 per 1,000 births between 1838 and 1854. This is only 43 % of the average values obtained when the less restrictive definition is used in Table A6, where it was 11.6 per 1,000. We could also compare the proportions of deaths attributed to maternal mortality between the ages of 25 and 44 years, which were 19.4 % in France and 8.4 % in England between 1848 and 1872. This ratio is the same as that between the mean rates of maternal mortality, and suggests – given that the periods covered are nearly the same – that these two indices are a good measure of maternal mortality.

39On these assumptions the proportions of deaths caused by pregnancy or confinement in each age group may be computed by multiplying the values of tg/t in the right hand column of Table A7 by the ratio 8.4/19.4. The figures given in Table 1 for England and Wales can then be divided by the complement of this ratio from 1 to obtain ratios of men’s probabilities of dying to those of women on the assumption of zero maternal mortality. The result is that excess mortality of women remains unchanged between the ages of 15 and 19 years, but practically disappears between the ages of 20 and 24 years. Thereafter, the mortality of men becomes higher.

40Things are not as simple in Geneva. Perrenoud, in his paper, used cause-specific statistics of deaths for 1730-39 and obtained a figure of 9.3 % for deaths due to complications following confinement, compared with 12.9 % for France. Moreover, maternal mortality rates obtained by the method of family reconstitution in France in which only deaths following confinement were used, are of the order of 15 per 1,000, compared with the figure of 8.5 which is obtained by using the same definition (average values in Table 4).

41It is not impossible that in two populations the proportion of deaths due to childbirth may be lower in the population with the higher maternal mortality rate ; it would be sufficient for this to happen if the proportion of mothers were lower in that population. However, it is extremely unlikely that that proportion in Geneva could only have been half of that found in France. Perrenoud believes that the cause-specific mortality statistics greatly underestimate maternal mortality, because they only yield a figure of 7.3 deaths per 1,000 births, a little less than half the value obtained from family reconstitution. Deaths due to complications following confinement should have been of the order of 19 %, 1.47 times the proportion found in France according to Table A4. By using this proportion, as we have done in the case of England and Wales, we obtain an average value for two large cohorts which shows that probabilities of dying for men were lower than those for women between the ages of 15 and 19, higher between the ages of 20 and 44 years, and roughly the same between the ages of 45 and 49 years.

42Thus, in spite of the approximations which are made necessary by the absence of appropriate data, the results confirm the impression obtained by a first look at the curves ; it seems reasonably probable that the excess mortality of women between the ages of 25 and 44 years was caused by maternal mortality alone.

43Comparisons between men’s and women’s mortality would be incomplete without considering causes of death which are specific to men. We are attempting to determine whether one sex is more resistant to mortality than the other throughout life, or at certain stages of life, and for a fair comparison we must eliminate causes of death which in pre-industrial society were linked to the role of each sex : occupational hazards for men and maternal mortality for women. Of course, women were also subject to and may have died from occupational hazards, but this is likely to have happened less frequently than among men, and it is, therefore, necessary to eliminate the difference between the mortality of men and women due to accidents at work.

44Unfortunately, very little information is available about mortality caused by accidents at work in the past. Perrenoud quotes some figures for Geneva for 1730-39 [13], which show that between the ages of 20 and 44, 5.7 % of the deaths of men and 1.4 % of those of women were caused by accidents at work, an excess of 4.3 % for men. This means that if mortality from these causes were eliminated, the ratios of men’s to women’s mortality would, on average, be multiplied by a factor of 0.957. In his article on England and Wales, Logan used a category called "violent deaths", but gave no further particulars, except deaths from burns, which were probably the main cause of death from industrial accidents. Unfortunately, his figures relate to the period 1848-72, when the process of industrialization was already advanced in England and when the chances of death from industrial accident were likely to be considerably higher than in pre-industrial societies.

45Consider the situation in England and Wales between 1848 and 1872. Between the ages of 25 and 44 the ratio of the annual death rate for men (11.415 per 1,000) to that for women (10.891 per 1,000) was 1.048 for all deaths. Subtracting violent deaths (1.050) from the figure for men, and violent deaths and deaths due to pregnancy and childbirth (1.041) from that for women, we obtain death rates from other causes amounting to 10.365 for men and 9.850 for women, with a ratio of 1.052, nearly the same as that for deaths from all causes. In other words, having eliminated the mortality due to causes specific for each sex, mortality of men remains higher than that of women between the ages of 25 and 44 during the period 1848-72, and is nearly the same as that found for deaths from all causes. As maternal mortality in England and Wales hardly changed between the mid-nineteenth century and 1930, it seems likely that the distribution of deaths by cause in 1838-54 was similar to that in 1848-72. This would suggest that elimination of deaths from causes specific to each sex would result between the ages of 25 and 44 in a slight excess mortality of women in 1838-54, because this excess mortality is found for deaths from all causes. This would, however, be slight, the ratio of men’s mortality to that of women being 0.988. It remains possible, therefore, that during the first half of the nineteenth century death rates of women from disease may have been higher than those for men between the ages of 25 and 39, as women’s mortality from all causes was higher in these age groups.

46We next consider France between 1740 and 1789. As France was not industrialized at the time, it is likely that the proportion of deaths from accidents at work was more like that in Geneva than in England and Wales. Let us consider the two cases. In Table 5 we show the proportion of violent deaths in the age groups given by Logan, and the difference between the proportions for the two sexes. By graphic interpolation we obtain the values for these differences in the age groups 20-24 to 40-44 years. Assuming that changes in those differences with age were similar in Geneva and England and Wales, we estimate their values for Geneva. These figures are shown in Table 6. By multiplying the ratios in Table 4 by the complement of these figures to unity, we obtain the ratios of men’s to women’s mortality, after causes of death specific to each sex have been eliminated.

Table 5

Proportion (p. 100) of violent deaths (England and Wales, 1848-1872)

Age group 15-24 25-44 45-64 65 et + Men 11.8 9.2 5.7 2.2 Women 1.5 1.2 1.3 1.3 Difference 10.3 8.0 4.4 0.7

Proportion (p. 100) of violent deaths (England and Wales, 1848-1872)

Table 6

Difference between the proportions (p. 100) of violent deaths for men and women (aged 20-24 to 40-44)

Age group 20-24 25-29 30-34 35-39 40-44 England-Wales 9.9 9.1 8.4 7.5 6.7 Geneva 5.1 4.7 4.3 3.8 3.4

Difference between the proportions (p. 100) of violent deaths for men and women (aged 20-24 to 40-44)

47Using the more restrictive definition in Table 4 we obtain an excess mortality for women between the ages of 30 and 39, but it is distinctly higher if the rates of violent death for England are used than those for Geneva. Using the less restrictive definition neither the English nor the Geneva rates would yield an excess mortality of women. As the less restrictive definition is probably closer to the truth and as the frequency of deaths at work is probably lower than that in England, it is likely that mortality of men in France between the ages of 20 and 44 from causes other than accidental or maternal deaths, i.e. deaths from disease, was higher than that of women but that it fell between the ages of 30 and 39. It would not, however, be right to exclude the possibility of a small excess mortality of women from disease between the ages of 30 and 39, although this seems unlikely.

The age group 45-64

48Except in Belgium, the mortality of men is always higher than that of women in this age group, and the excess often reaches a maximum value in the first half of the group. The effect of maternal mortality in this group is very small and only affects the age group 45-49. Moreover, the proportion of deaths due to accidents is too low to cause the excess mortality of men. This excess mortality must, therefore, be due to mortality from disease.

Table 7

Ratios (p. 100) of mortality rates for men to those for women for causes other than maternal mortality and violent deaths

Age group Using proportion of violent deaths for : England Geneva With low maternal mortality 20-24 117 123 25-29 104 109 30-34 94 99 35-39 92.5 96 40-44 101 104 With high maternal mortality 20-24 118 124 25-29 107 112 30-34 101 105 35-39 101 105 40-44 108 112

Ratios (p. 100) of mortality rates for men to those for women for causes other than maternal mortality and violent deaths

49In Table 8 we show the ratio of men’s to women’s death rates at different ages for England and Wales in 1848-72 for major disease groups. This table shows all but one of the groups [14] used by Logan in his article. Some of these groups are of no great importance, either because they account for only a small proportion of deaths, or because of their imprecision. This is the case for deaths from diseases of the genito-urinary system, where the very high excess mortality of men in more recent periods has only continued in the 45-64 age group, and for cancer which is more easily diagnosed in women and where women’s mortality remained higher than that of men between the ages of 25 and 44 years, but where the proportion of deaths from this cause was low, and where in the age group 45-64 there was a higher mortality rate among men.

Table 8

Ratios (p. 100) of mortality rates for men to those for women, by cause of death (England and Wales, 1848-1872)

Disease Age group 5-14 15-24 25-44 45-64 Infectious disease 91 89 100 127 (incl. pulmonary tuberculosis) 69 82 96 134 Nervous system 108 98 136 113 Circulatory system 98 89 99 93 Respiratory system 96 115 147 135 Digestive system 106 84 89 102 Genito-urinary system 174 140 165 235 Cancer 33 42 All 94 91 104 111 All but pulmon. tuberculosis 98 101 111 107

Ratios (p. 100) of mortality rates for men to those for women, by cause of death (England and Wales, 1848-1872)

50Infectious diseases were particularly important for females between the ages of 5 and 24 years and for men aged between 45 and 64 years ; among them pulmonary tuberculosis, which was a disease of girls and younger women, tended to wreak more havoc among men aged over 45. For all the diseases shown here, an excess mortality among women between the ages of 5 and 24 gave way to a higher mortality of men between the ages of 25 and 64. If deaths from pulmonary tuberculosis are excluded, women’s mortality is higher than that of men only between the ages of 5 and 14 years, and the excess is not large. In England and Wales during the period considered, therefore, pulmonary tuberculosis was the main cause of the higher mortality from disease of females in that age group.

Overview and conclusion

51The principal results we have obtained are as follows : During part of the nineteenth century, the mortality of females was higher than that of males in almost all Western European countries in some or all of the age groups between 5 and 19. It was only in Austria that in one instance the ratio of men’s to women’s mortality was at a minimum between the ages of 10 and 14, and where the mortality of the sexes was almost equal, but even in this case there was hardly any divergence from the pattern in other tables.

52Tables for periods before the nineteenth century suggest that this higher mortality of girls and young women could not be found in the more distant past. This hypothesis is strengthened by the fact that during the nineteenth century the disadvantage suffered by women increased and that it only became apparent in Sweden at that time.

53Between the ages of 20 and 24, the mortality of men is almost always higher than that of women. If this difference were caused mainly by mortality due to accidents at work, it is suprising that it did not appear in the age group 15-19 at a time when most young men started working before their 15^th birthday.

54In the age group 25 to 44, the earlier tables, as well as those relating to the nineteenth century show a higher mortality of females in countries and during periods when limitation of births was not widespread. This higher mortality is particularly noticeable between the ages of 30 and 39 years, and the curves in Figure 1 show a marked trough in this age group in Austria, Spain and Sweden.

55French data confirm that the higher mortality of females observed between 1740 and 1789 was entirely due to mortality from causes associated with pregnancy and childbirth. Similar calculations for England and Wales and for Geneva with rather less suitable data suggest similar conclusions, and it is likely that the higher mortality of females in all cases shown in Table 1 can be explained by maternal mortality. Where the mortality of women is not higher at these ages, maternal mortality accentuates the trough in the curve of mortality ratios which is evident even in the absence of maternal mortality, as is the case for the curve for France in Figure 2.

56Between the ages of 45 and 64 it is common for the mortality of men to exceed that of women. It is likely that the statistics of cause-specific mortality for England and Wales in 1848-72 give a reasonably accurate picture of the situation : a higher mortality of men from disease, both infectious and non-infectious. This excess mortality of men tends to diminish after the age of 65, when it levels off or gives way to a slightly higher mortality among women.

57The fact that mortality among girls and young women was not higher than that of males of the same age during the seventeenth and eighteenth centuries and that it did not become apparent in Sweden until the nineteenth century presents a problem. It will be necessary to try to explain why such excess mortality appeared at a time when death rates in general were falling, or, in other words, why the mortality rates of girls declined less than those of boys.

58It is generally agreed that the female organism is sturdier than the male and more resistant to mortality. If this statement is accepted, the reason for the differences must be related to social and economic conditions.

59It has been suggested that when little girls fell ill they were traditionally given less care than little boys and that, for this reason, their illnesses more often turned out to be fatal. But this would not explain why excess mortality of girls appeared at a time when medicine was as yet insufficiently developed to provide a cure for many diseases.

60It may also be thought that girls profited less than boys from the improvement in conditions which led to a general reduction in mortality, particularly as regards nutrition. Richard Wall has investigated this situation in England [15] and has come to the conclusion that the hypothesis that girls were less well nourished than boys in nineteenth-century England cannot be proved. On reflection, this hypothesis could not explain why the excess mortality of girls should have increased at a time when nutrition was improving, because the weakest tend to go to the wall mainly during periods of famine.

61Changes in working conditions could have exposed adolescent girls and young women to infections rather more often than previously. It has been shown in a number of studies that mortality was lower in isolated villages and that concentration, crowding and promiscuity were associated with mortality. Those affected may not have been conscious of these changes, nor is it possible to regard them as a deterioration in their circumstances. A peasant’s daughter may have been well content to obtain work as a seamstress in a workshop, even though she was more likely to contract a fatal disease there than if she had remained at home to look after the cattle. Moreover, changes in working conditions would not explain why girls between the ages of 5 and 9 should suddenly have become subject to an excess mortality which did not exist earlier. This phenomenon can only be explained by conditions specific to that age group, e.g. a higher school attendance rate or more frequent boarding education. However, such changes would have placed girls on an equal footing with boys of the same age and therefore the risks should have been equal as well. If they were subject to a higher mortality, this would suggest that they were less sturdy than boys and contradict the hypothesis that the female sex was more resistant to mortality than the male.

62Similar objections might apply to other explanations. Many cultural norms are resistant to social and economic change, and it seems unlikely that customs which regulated the roles of each sex in childhood and adult life would have changed sufficiently quickly to allow little girls, adolescent females or women in general to be less well treated during the nineteenth century than previously. It is probable that some of the changes in the conditions of women to which several authors have alluded will have resulted from urbanization and industrialization. As these changes affected both sexes, it is difficult to accept the suggestion that girls would have been more affected by these changes than boys, given that a priori they were considered to be more resistant.

63One is, therefore, led to a different hypothesis which suggests that at certain ages females are more susceptible to particular diseases than males. Wall mentions this hypothesis and states that it has not been widely accepted, but points out later that B. Ehrenreich and D. English might have found a reason for the difference between the mortality of the two sexes by relating the chance of contracting tuberculosis to hormonal changes among women during puberty and childbirth.

64As tuberculosis was a frequent cause of death between the ages of 5 and 19 and as the mortality of females from this disease exceeded that of males by a significant amount, the question may be asked whether the higher mortality of women from all causes may not have been due to an increase in the frequency or severity of tuberculosis during the last 200 or 300 years.

65An increase in frequency is likely following industrialization : work in a workshop or factory and crowding in the insalubrious quarters of the new industrial towns could have increased the risk of contracting this disease significantly, and the effect may have been greater for girls and young women than for boys of the same age.

66But this hypothesis is also subject to some difficulties. If the higher mortality of girls between the ages of 5 and 9 were solely the result of urbanization, it is difficult to understand why such higher mortality is found in the French tables for 1840-59, but not in those for England and Wales in 1838-54, nor why the excess mortality of women in Norway between 1899 and 1902 was greater in the countryside than in the towns, according to statistics quoted by Shorter. However, in the last case it is possible that advances in public health and hygiene in the towns and the attention that was given to them at the beginning of the century may have more than compensated for the increased risk.

67English and Belgian statistics confirm the importance of tuberculosis as a cause of higher mortality of girls and young women during the nineteenth century. In England and Wales mortality of girls from causes other than tuberculosis exceeded that of boys of the same age only between the ages of 5 and 14 and then by only a small amount, whereas excess mortality of women from all causes (including tuberculosis) was obvious at these ages and was even larger in the age group 15-24.

68We next consider the following age group : 25-44. As we have shown in France and in all likelihood elsewhere, maternal mortality was probably the only cause for the higher death rates of women in a large part of that age group. It would seem, but this is not absolutely certain, that the mortality of men may have been higher at these ages than that of women, if deaths from conditions associated with pregnancy and childbirth, as well as deaths from accidents, are excluded.

69This was the case in England and Wales for 1848-72. The only cause of death from which the mortality of women was higher than that of men was cancer, but this difference may have been exaggerated because of the greater difficulty of diagnosing this disease in men, and the differential was reduced, though not eliminated, in later periods. In any case, cancer deaths accounted for only three per cent of all deaths of women in this age group compared with more than 51 % of deaths which were caused by infectious diseases, and 40 % by tuberculosis alone. Anaemia is not even mentioned as a separate cause of death in the tables for 1848-72 and accounts for only 1 % of deaths of women between the ages of 25 and 44 in the tables for 1901-10. It is difficult to understand why Shorter gives such a prominent place to this cause of death in his discussion of the reasons for the higher mortality of women.

70The same author refers to the excessive workload falling on peasant women which resulted in a higher excess mortality of females in the country than in the towns. But given that in the essentially rural society of eighteenth-century France the higher mortality of women between the ages of 25 and 39 was almost entirely caused by maternal mortality, and that it is impossible to show that women’s mortality from other diseases was higher than that of men, it is difficult to accept that excessive work could have been a reason for the higher death rates of women. Moreover, among women who survived to their 45^th birthday, ages at death did not differ much between those who had undergone numerous pregnancies and those who had had smaller numbers of children, and this hardly supports the thesis that exhaustion from domestic and agricultural work was a major cause of women’s deaths [16].

71Instead of turning to the age group 45-64 in which the mortality of men exceeded that of women, we shall look at the expectation of life at the 25^th birthday. This figure is a summary index of the risks run by men and women throughout the remainder of their lives. We shall limit ourselves to discussing the situation in France and Geneva, because of the high excess mortality of women of reproductive age in those areas. In the three cases shown in Table 9, men’s expectation exceeds that of women by a little, but the difference is of the order of two months in France and between five and six months in Geneva, for an expectancy of between 32 and 35 years. The difference of six years observed in some villages and cited by Shorter must have been exceptional, and will have been more than compensated by others in which men survived for a shorter period than women.

Table 9

Expectation of life at the 25^th birthday

France 1740-1789 Geneva : cohort 1625-1699 1700-1790 Men 33.48 32.12 35.39 Women 33.32 31.70 34.88 Difference 0.16 0.42 0.51

Expectation of life at the 25^th birthday

72In spite of many studies, the problem of explaining the higher mortality of younger women in the past, and the changes in this differential between the seventeenth and eighteenth centuries and the end of the nineteenth is far from solved, and the absence of information on cause-specific mortality during earlier periods makes a solution particularly difficult. Fortunately, information about maternal mortality may be obtained from other sources. Although this mortality was perhaps lower than has been thought, it occurs at ages when deaths are infrequent and therefore represents a non-negligible fraction of deaths in the age group 25-44 which accounts almost entirely for the higher mortality of women at these ages.

73As the higher mortality of adult women can be explained in this way, attention must be given to the reasons for the higher mortality of girls and younger women. It must be recalled, however, that this difference appears at a time of life when mortality is at a minimum and is unlikely to have many repercussions ; indeed, it is probable that the population at the time was unaware of the existence of this difference.

74We return to Shorter’s book which was cited at the beginning of this article. It has the merit of reminding us of the sufferings of our ancestors, though it takes no account of the sufferings of the male sex, but it is defective through a careless use of sources, particularly medical sources, which, even when they are of high quality, are not necessarily representative of the population as a whole.