M. HOSNY 
28.5 
Many European towns with half the above birthrates have considerably greater 
excesses of births over deaths. 
Unfortunately the infantile mortality is only recorded in Egyptian toiuns and 
not in the governments at large or in the rural districts. We are obliged, there- 
fore, to deal with these only when considering the relation of education to infantile 
mortality. From p. 49 of the Annuaire Statisttque de VEgyj^te, 1912, we obtain 
the infantile mortality for 1911, and note at once how extraordinarily high it 
stands. From p. 286 of the Census of Egypt, 1907, we take the percentage of 
male literates in the total male pojjulation, and from the Statisttque Scolaire, 
1912-1913, p. 74, the percentage of scholars in the total popidation*. As it 
was possible that the density of town population might influence the results, we 
took the number of persons per house, which was about the only social factor 
available. This will probably represent fairly closely the average size of family. 
This was taken from the Census, 1907, p. 286. The mean, .5'64 persons per 
house, suggests that the average number of living children can hardly exceed 
three. The marked relationship that occurs in European towns between gross size 
of family and infantile mortality cannot be satisfactorily tested on the Egyptian 
data, because we cannot ascertain the infantile mortality in each size of family. 
The number of persons to the house is indeed rather a measure of net than 
gross family, and we only know this as an average value for each town. It does 
not follow that a town with a low number of persons per house is one with 
small gross families ; the low number may be due to the heavy infantile mortality 
itself Accordingly the correlation between persons per house and infentile 
mortality is not necessarily even a measure of the influence of overcrowding on 
infantile mortality (although this is often supposed to be the case) ; it is con- 
ceivable that a high infantile mortality might be the source of a low number of 
persons per house, and the unravelling of cause and effect is only possible 
where we know not only the number of persons per house, but its relation to 
both the gross and net family of that house. 
Let / = Infantile mortality, L = Literacy, S = Scholars, P = Persons per house. 
Then we have the following results : 
Means Standanl Deviations Correlations 
Mj = 29-30, a I = 7-608, rjj^ = - -1040 + -1500, 
Mj^ = 21-96, (Tjr = 5-137, = + -5309 + -1278, 
Ms= 5-569, cr,,- 2-951, = + -0093 + -1508, 
Mp= 5-643, o-p= 1-428, 
Correlations : 
= + -1675 ± -1487, r^^ = - -3421 + -1437, = -I- -0296 ± -1508. 
* The scholars were taken for 1910-1911, the year of infantile mortality, but this involved 
the assumption that the foreign scholars were the same in numbers in 1910-11 and 1912-13, probably 
not a very inaccurate assumption, which in any case affects little more than Cairo, Alexandria, Suez 
and Ismailia practically. It is the number of Egyptian scholars that is rapidly changing and the 
scholars dealt with in our ratio are Egyptian only. 
Biometrika x 37 
