INFLUENCE OF MARRIAGE. 285 



perance, poverty, or whatever cause, have not (on the 

 average) so good a chance of marrying as others. And 

 few who consider the facts carefully can doubt that 

 both these inferences are sound. But the statistics 

 cannot show whether one or the other is the true 

 ^explanation ; or, if both causes operate, still the sta- 

 tistics do not show in what degree each is effective. 

 The mistake made by Stark, Bertillon, Drysdale, and 

 others, is in assigning the observed relation to one 

 cause, or rather in saying this is cause and that is 

 effect, when in reality cause and effect may, for aught 

 the statistics show, be interchanged. It is, for instance, 

 at least as probable that a tendency to commit crime 

 diminishes the likelihood of a man's marrying as that 

 his marrying diminishes the likelihood of his com- 

 mitting crime. I do not say that the former explana- 

 tion of the asserted infrequency of crime among married 

 folk is necessarily the true one. To assert this would 

 be to make a mistake of precisely the same kind as 

 that which mars Dr. Drysdale's reasoning. At least, if 

 the assertion were based solely on the statistics, for 

 these can prove nothing of the sort. On a priori 

 grounds the assertion has far more in its favour than 

 the other, for it is difficult to conceive any reason why 

 matrimony should diminish criminal tendencies to any 

 marked degree (though a man's marrying might be 

 evidence that he had been tamed) ; whereas one can 

 see many reasons for believing that a man with criminal 

 tendencies would be regarded as an undesirable suitor. 

 If the argument of Dr. Drysdale, though demonstrably 



