INFLUENCE OF INFANTILE MORTALITY ON BIRTHRATE, 249 
but will nevertheless disclose something of the nature of 
the relative frequence of different rates of infantile mor- 
tality for the entire population group represented. The 
thin lines in Fig. 3 shew the results as given, while the 
heavy line indicates the curve (of same total area) which 
may be taken to represent the general result. 
10. It will be seen from Figure 3, that in so far as the 
graduated result can be taken to exhibit the frequency of 
any particular rate of infantile mortality, it would appear 
that the dominant rate (the mode) is 146°2, and the median 
is 165°6. The centroid vertical of the curve corresponds 
to a rate of infantile mortality of 178°8. 
Asalready pointed out, however, the various populations 
are not subdivided so as to constitute homogeneous groups, 
and hence the curve should be regarded as furnishing only 
a rough indication of the true characteristics of the dis- 
tribution of infantile mortality in the group of countries 
under review. 
11. In deducing the linear equation representing the 
influences of infantile mortality on birthrate, the crude 
rate, should strictly be corrected for differences in the sex 
and age constitution of the population, in order that the 
results may be strictly comparable. The irregularities 
indicated however are so large in comparison with any 
correction for reduction to a common basis, that such cor- 
rection would have very little significance, and in fact, 
may be ignored. The linear variations of birthrate as 
related to variations of infantile mortality deduced herein, 
are abundantly accurate for the purpose of eliminating this 
element forcomparisons requiring such elimination. Secular 
changes in the correction factor b in (8) and (9) are of 
course to be expected, but can be deduced only by obser- 
vation of very long periods. These may perhaps disclose 
that the theoretical linear law is only an approximation to 
