56 H. E. EWING. 



These sublines I will refer to as isolated sublines, or simply 

 isolations. In each of these, selections were carried on for a few 

 or for many generations to see if there was produced any in- 

 heritable effect, due to the picking out of individuals showing 

 extreme variation in the character considered. 



Selections Made in an A ttempt to Change the Ratio of the Length of 



the Third Segment of the Antenna to the Fourth Segment. 



Previous Results. Isolations Nos. i and 2. 



The first character used for observing the effects of selection 

 was that of the ratio existing between the lengths of the third 

 and fourth segments of the antenna. For fifteen generations 

 selections were made in an attempt to change the mean of this 

 ratio for the line, which was found to be 1.80 to i,i. e., it was found 

 that on the average that the third segment was 1.80 times as 

 long as the fourth segment. Selections were made for the first 

 10 generations in the attempt to increase this ratio. At the end 

 of the tenth generation the fraternal mean was 1.66 to I, or 

 decidedly less than the mean for the pure line. Selection had 

 here failed utterly to produce any inheritable effects. Various 

 other selections were made, and the results of the selections with 

 this character have been reported in the two short papers pre- 

 viously referred to. In the writer's conclusion, published in the 

 first of these papers, the following statement was made which 

 summarizes the results so far as the effects of selection are con- 

 cerned in changing the ratio of the length of the third to the 

 fourth segment of the antenna: "Selections from among extreme 

 variants do not alter the mean as obtained for the strain without 

 selection. The fraternal mean of any generacion may show a 

 great fluctuation from the mean of the strain, but this fluctuation 

 is not transmitted to following generations." 



Selections Made in an Attempt to Increase the Length of the Cornicles. 



Isolation No. j. 



After making selections for fifteen generations using the ratio 

 existing between the length of the third and fourth segments of 

 the antennae as a character, much time was spent in studying the 

 specimens obtained for characters that were sufficiently variable, 



