H. E. EWING. 



TABLE V. 



GIVING LENGTH MEASUREMENTS OF INDIVIDUALS IN ISOLATION 8, THE AVERAGE 



LENGTHS OF THEIR ANTENNAE, THE RATIO OF BODY LENGTH TO ANTENNAL 



LENGTH, AND THE RATIO REPRESENTING THE FRATERNAL MEAN. 



The results of the selections in this isolation are in themselves 

 too meager to show anything, but taken together with the results 

 obtained in previous isolated sublines they simply furnish addi- 

 tional proof that selection in a parthenogenetic pure line does 

 not affect hereditarily any of the ordinary individual variations. 

 The meager results obtained for isolation 9 are shown more fully 

 in Fig. 6 and Table VI. 



The Effects of Continuous Selection for Forty-four Generations on 



a Character that is Known to be Easily Modified by Selection 



in Animals that Reproduce Sexually. 



All the selections thus far reported were not continued over a 

 very large number of generations, and the results obtained with 

 them may not be convincing to some of our adherents to the 

 belief that selection is effective when made from fluctuating 

 variations and continued for a large number of generations. 

 But if selections are continued for a much longer period do we 

 discover any appreciab'e effects? In order to give a more thor- 

 ough test to the effectiveness of selection in a pure line, I decided 

 to make selections for a great many generations using a character 

 that is well known to be inherited in higher animals and plants 

 where cross fertilization takes place. Few characters have been 

 more thoroughly studied or more generally employed in selection 



