INFLUENCE OF ORDER OF BIRTH, ETC. 301 



excellent work on The English Convict it is claimed that crimi- 

 nality develops in the first bom to a much greater extent than it 

 does in the later born members of the stock from which the crim- 

 inals are derived. Pearson confirms the deductions of Heron and 

 Goring for insanity and criminality, and he has adduced data to 

 show that the first born are unusually liable to albinism, imbecil- 

 ity, epilepsy and cataract. 



A number of writers have attacked the findings of Pearson 

 and his colleagues on the ground that they are based upon a 

 statistical fallacy. Greenwood and Yule have arrived at a quite 

 different ordinal distribution of the relative number of individuals 

 in the members of the famiUes of the marked individuals. When 

 we are dealing with cases of insanity or tuberculosis in which we 

 start with individuals, say in institutions, it is obvious that all 

 members of the marked person's family are not equally apt to be 

 found in the segregated class. There is an age at which insanity 

 and tuberculosis is more than likely to appear and the chances are 

 decidedly against two persons from the same family being con- 

 fined at the same time, there being an especially strong bias 

 against the members who have not reached adult life. Recently 

 Pearson's methods have been attacked by Dublin and Langham 

 of the Statistical Bureau of the Metropolitan Life Insurance 

 Company of New York. These authors contend that Pearson's 

 method "is based unequivocally on the assumption that the 

 distribution according to order of birth of the pathologic com- 

 munity from which his 'marked' or affected subjects are ob- 

 tained is identical with the distribution of the sibships of these 

 subjects. For if that be the case he can use the distribution of 

 the sibships of the affected as a norm in comparing with it the 

 distribution of the affected, in the effort to show that actually 

 the early born among his subjects preponderate beyond all ex- 

 pected proportions. We shall endeavor to show that, when there 

 is no weighing according to order of birth among the individuals 

 affected, the distribution of the affected or that of the pathologi- 

 cal community represented by them is not in any case compar- 

 able with that of their sibships. We propose to take the distri- 



