390 H. G. MAY. 



In the Ba selections the results are very different and not so 

 easily explained. Here also there must have been a downward 

 shift as well as an upward shift. Disregarding the seventh 

 generation for reasons already given, the following differences 

 between the high and low lines are obtained: fi, 22; f2, 40; 

 f"3, 61 ; f4, 65; fs, 79; f6, 95. There is a gradual increase which 

 becomes somewhat smaller as selection proceeds. If it were not 

 for the interfering fluctuations the author's results would be 

 much like those obtained by Zeleny and Mattoon. Even three 

 additional generations of selection failed to produce pure lines. 



In many respects the results obtained in these experiments 

 resemble those obtained by MacDowell in the selection for 

 extra bristles. The average of all the lines is raised or lowered 

 but one can not predict within rather wide limits just what the 

 offspring of any two parents are going to be like. This is clearly 

 shown in Table II. and Fig. 6 in case of the matings difi-2 and 

 difi-3. These matings were made with the same male but 

 with females of 91 and 56 facets respectively. The average for 

 the offspring from the higher female is considerably lower than 

 that for the offspring of the low r er female. In this case the 

 unexpected results can not be explained on the basis that only a 

 part of the character was observed. It is true that the counts 

 w T ere made only on one eye, but the difference between the two 

 eyes falls within definite limits and is very small compared with 

 the differences between the eyes of two flies. A comparable 

 case would be the possibility of less than one extra bristle out- 

 side of the observed rows in the experiments of MacDowell. 

 The large, parallel fluctuations of all the lines seem to indicate 

 that there are environmental factors capable of almost doubling 

 or cutting in half the facet numbers of flies of the same germinal 

 constitution. The unexpected results must be regarded as 

 being due to the fact that the hereditary factors are only a part 

 of the total factor group controlling the facet number. When 

 the environmental factors have been studied it may be possible 

 to control them in such a way as to obtain uniform results. In 

 that case more rigid selection will be possible. 



The crosses made between the high and low lines show no 

 evidence of sex-linkage. The offspring from both sets of matings 



