182 



BIOLOGY OF DEATH 



are just as prevalent in industrial as in rural communities, and could be 

 measured by other means if the data were forthcoming. 



LO I- 



^ 



5| 



& 



■S 



.3 - 



.a — 



.1 — 



MALES 



EXPEiCTCD CORRELATION - NO SELECTION 



± 



1670 



± 



1871 



167a 



1681 



laaa 



LNGLISH 



PRUSSIAN 



Fig. 45. — Snow's results on selective death rate in man. The cross-hatched area may 

 be taken, in comparison with the small clear area at the bottom, as indicating the influence 

 of the selective death rate in increasing the correlations. 



Our investigation substantiates for a general population the results 

 found by Pearson and Ploetz for more restricted populations, and disagrees 

 with many statements of health officers. It is with great reluctance that 



