32 



Indiana University Studies 



One point of interest concerning Castle's work may be 

 briefly discussed inasmuch as I have made an experiment 

 which bears directly upon the same question. Castle found 

 that when he crossed either of his selection lines to the wild, 

 the extracted hooded animals obtained in Fo showed a regres- 

 sion toward the mean of the hooded race before selection 

 was begun. As this fact suggested the presence of multiple 

 modifying factors, he outcrossed the extracted F/s again to 

 the wild, and obtained twice extracted hooded rats in the 

 second Fo generation. Castle believed he should obtain a 

 further regression if modifying factors are the cause of the 

 regression in the first extracted rats. However, he did not 

 find a further regression, but found that the movement of 

 the mean, mode, and range of variation was in the reverse 

 direction. From these results Castle concludes, ''So the 

 hypothesis of modifying factors to account for the regression 

 and for the progressive changes observed under selection 

 becomes untenable." 



I have repeated this same experiment with my flies, where 

 I have demonstrated the presence of multiple factors. Instead 

 of stopping with twice extracted flies, the outcross has been 

 made five times. The results are given in Table X. This 

 table shows the number of flies in each generation with the 

 normal bristle number (four), the number of flies with extra 

 bristles, the percentage of flies with the normal bristle num- 

 ber, the mean bristle number of all flies, and the mean bristle 

 number of the extra bristled flies. These data are given for 

 the F/s as well as the Fo's. In each case it was the female 

 which was outcrossed to the wild, and in each case this female 

 had extra bristles. In the first outcross females with nine, 

 ten, and eleven bristles were used. In the others the females 

 had either seven, eight, or nine bristles. The parents used in 

 the first outcross were taken from the twenty-second, twenty- 

 third, and twenty-seventh generation of the selection line. 

 The mean bristle number of these three generations is 8.41. 

 The mean of the first Fo's is 6.07, This is a marked regres- 

 sion toward the normal. In the second F.'s as in Castle's 

 rats, there is a slight swing in the opposite direction, the 

 mean reaching 6.11. This difference, however, is hardly 

 enough to be of any marked significance, especially when the 

 percentage of normal flies increased from 47.59 in the first 

 F. to 50.17 in the second. In the third F/s there is a further 



