30 H. E. EWING. 



nine offspring I selected one at random, with which to carry 

 on the strain. The individual selected was Fi3 (that is, indi- 

 vidual number three of the first filial generation). This indi- 

 vidual gave birth to ten young. Her antennal index proved to 

 be 1.77:1, the same as that of her mother. From the F 2 indi- 

 viduals I selected at random again an individual which became 

 the mother of six offspring. Her index was, as later obtained, 

 1.67:1, while that for the mean of her offspring was 1.77:1. 



For the next three generations selections were made at random, 

 a prolific or healthy looking individual being taken. The meas- 

 urements of the antennal segments of all the individuals were 

 taken, and from these the fraternal means for the second, third, 

 fourth, and fifth generations obtained. For the first filial 

 generation I obtained the index of the single individual used 

 to carry on the strain. Using this index and that of the stem 

 mother I obtained a curve for the first five generations. This 

 I have plotted (see Fig. i). It will be noted here that in the 

 case of the selections of individuals representing extreme vari- 

 ations, as was done in the second and third filial generations, 

 that the mean of their offspring came back to or almost to that 

 of the strain. 



In the fourth generation an individual was selected which had 

 an antennal formula near that for the mean of its fraternity, 

 and exactly the same as that obtained for the strain; yet its 

 offspring gave a mean much higher than that of any individual 

 belonging to the fraternity from which the mother was taken! 

 (See Fig. I.) More than this, the mean of this fraternity was 

 higher than the index for any individual of any fraternity previ- 

 ously obtained. From the individuals of the F 5 generation one 

 was obtained with a low index. It gave a brood of normal 

 individuals. 



It w^as from among the individuals of the F 6 generation that I 

 began making my selections in an attempt to obtain two strains, 

 one with a high antennal formula, the other with a low antennal 

 formula, from this isolated pure line of parthenogenetically 

 reproducing plant lice. F 6 5, with a formula of 1.88:1, was 

 selected for obtaining a strain with a high index; F 6 3, formula 

 i .66:1, was taken for obtaining a strain with a low index. F 6 5 



