THE 



POPULAR SCIENCE 

 MONTHLY. 



NOVEMBER, 1872. 



THE STUDY OF SOCIOLOGY. 



By HERBERT SPENCER. 

 V. Objective Difficulties {continued). 



ANOTHER common cause of very serious perversion of evidence 

 is the unconscious confounding of observation with inference. 

 Everywhere, a fertile source of error is the putting down as something 

 perceived what is really a conclusion drawn from something perceived ; 

 and this is a more than usually fertile source of error in Sociology. 

 Here is an instance : 



A few years ago Dr. Stark published the results of comparisons 

 he had made between the rates of mortality among the married and 

 among the celibate ; showing, as it seemed, the greater healthfulness 

 of married life. Some criticisms made upon his argument did not 

 seriously shake it ; and he has been since referred to as having con- 

 clusively proved the alleged relation. More recently I have seen 

 quoted from the Medical Press and Circular the following summary 

 of results supposed to tell the same tale : 



" M. Bertillon has made a communication on this subject (the Influence of 

 Marriage) to the Brussels Academy of Medicine, which has been published in 

 the Revue Scientiftque. From 25 to 30 years of age the mortality per 1,000 in 

 France amounts to 6.2 in married men, 10.2 in bachelors, and 21.8 in widows. 

 In Brussels the mortality of married women is 9 per 1,000, girls the same, and 

 widows as high as 16.9. In Belgium, from 7 per 1,000 among married men, the 

 number rises to 8.5 in bachelors and 24.6 in widows. The proportion is the 

 same in Holland. From 8.2 in married men, it rises to 11.7 in bachelors, and 

 16.9 in widowers, or 12.8 among married women, 8.5 in spinsters, and 13.8 in 

 widows. The result of all the calculations is that from 25 to 30 years of age 

 the mortality per 1,000 is 4 in married men, 10.4 in bachelors, and 22 in widow- 

 ers. This beneficial influence of marriage is manifested at all ages, being always 

 more strongly marked in men than in women." 



VOL. II. 1 



