Miscellanea 
513 
which the probable errors could be determined. I find for the probable error of the mean legiti- 
mate male mortality 1882-90 a value of + -0676 day, and for the mean legitimate female mortality 
+ '0784 day. For the illegitimates the probable errors would be of course somewhat higher, 
although as nearly as I am able to judge from a rough estimate the probable errors of the means 
for this group would not exceed ± '15 day. The probable errors of the standard deviations will 
in all cases be lower. In calculating probable errors the actual number of deaths in the given 
group and period was taken as n. 
A number of points of general interest regarding infantile mortality are brought out by this 
table. Before considering the main question for which the work was undertaken some of these 
may be discussed. 
(a) The mean duration of life is uniformly greater in the case of the legitimates than in the 
case of the illegitimates. The legitimate males have an average excess of 14'59 days, and the 
legitimate females an average excess of 13"96 days. The excess is almost exactly the same for 
each of the three periods, although the mean changes. This uniformity is remarkable, and 
indicates to what an extent those differences in " nurture " (both pre- and post-natal) to which 
the difference in mean duration of life of legitimate and illegitimate infants must be attributed 
are uniform in long periods of time. Associated with the low mean duration of life in the case of 
the illegitimate infants there is of course a high death-rate as compared with the legitimate 
OTO\\\). 
(b) The illegitimate infants are markedly less variable with res[)ect to duration of life (as 
indicated by the standard deviation) than the legitimate. The average difference in the case of 
the males is 9"88 days, and in the case of the females it is 11 '53 days. To adopt the illustration 
which has been used by Pearson, it may be said that the marksman. Death, shoots faster and 
with deadlier aim at illegitimate than at legitimate infants. This lower variability in the case of 
the illegitimates may conceivably be the result of a more sharply selective mortality than in the 
case of the legitimate infants. 
(c) The mean duration of life of those dying within a year after birth is greater in the two 
later periods considered than in the first. Apparently, between 1877 and 1896 there has been a 
gain of about a day. Does this represent a real evolutionary tendency, or is it merely a chance 
fluctuation 1 In order to get light on this question I dealt with the mortality of each of the 
following years .separately; 1882, '83, '84, '85, '86, '89, 1890, '91, '94, '95 and '96. These were all 
the years for which I could get separate data. The groups of male and female, legitimate and 
illegitimate were treated separately as in the other cases. The results obtained were very 
interesting in many particulai's, but as they fall outside the scope of this note I shall not consider 
them in detail here. The general tendency from about 1884 on is for tlio mean duration of life 
to decrease, with considerable fluctuations from year to year. Thus in 1886 and in 1890 there 
was an unusually high duration of life. These two years explain why the 1882-1890 group in 
Table II. shovps such high means. In general, a study of these individual year data makes it 
very clear that there is no steady tendency towards lengthening of the mean duration of life of 
infants under 1, within the period under consideration. 
Another interesting point brought out by tlie single year records is that in the case of both 
legitimate and illegitimate infants, there is a definitely marked tendency for an increase in the 
mean duration of life in any year to be associated with an increase in the death-rate for that year. 
This appears to indicate that in general there is a tendency for any increase in the infantile 
death-rate to be the result of an increased number of deaths of older rather than younger 
infants (under 1). It will be noted, however, that this positive relation between death-rate 
and mean duration of life which appears within both legitimate and illegitimate groups is 
exactly reversed when the two groups are themselves compared. Thus in the legitimate group 
as a whole we have a condition of low death-rate and high mean duration of life, while in the 
illegitimate group as a whole the opposite condition obtains. 
Biometrika iv 65 
