EVIDENCE OF GERM CELL SELECTION 



403 



First, as to brachydactyly : it was found that upon adding all 

 the control data together there were 464 individuals of which 

 39 per cent were brachydactyl. Since this is about the fre- 

 quency of booting in comparable crosses recorded in the litera- 

 ture, it is assumed that 39 per cent represents the normal in- 

 cidence of brachydactyly under conditions such as obtain in 

 the control experiments. Next the total number of cases in 

 each group is found and the probable error calculated on the 

 assumption that 39 per cent represents the true incidence. This 

 gives the fourth column in the table — headed 'brachy dactyl 



TABLE 4 

 Percentage distribution of characteristics . Compare table 8 



expected.' The values in the third column represent the per- 

 centages actually observed in each case with the probable errors 

 calculated for the respective percentages and magnitudes. Thus 

 in 1-A, for example, the chances, as it is generally expressed, 

 are even that in a random sample of this magnitude the per- 

 centage of brachydactyly would fall between 36 and 42 (39 ±3). 

 The observed percentage is 46.7 with a probable error of 3.1, 

 giving a range of from 43.6 to 49.8. In other words, in the 

 hypothetical case the chances are 7.5 : 2.5 that the value would 

 fall below 42, while in the actual case observed the chances are 



