74 E. CARLETON MACDOWELL 



averages, 14 a little lower, and 8a much lower. In the second 

 group of families in Fs (fig. 3C and table 8C) female averages 

 from parental sums 7, 8, 14a, and 14b are similar, whereas 

 those from sums 9 and 11 are a whole bristle lower. The male 

 averages for parental sums 7, 8 and 14b are similar while those 

 from 11 and 14a are lower, and from sum 9 a whole bristle lower. 



In Ft is a set of cousins consisting of four families (table 8D 

 and figure 3D) . In this group the low parental sum has the low 

 male and female averages, whereas sums 11 and 14a and 14b 

 are much aUke. In Fio there is a group of cousins in four families 

 (table 8E and fig. 3E). The female average for parental sum 

 12b is considerably larger than that for the sum 10, yet the 

 female average for the sum 12a is about equal to that for the 

 sum 10. Parental sum 11 has the lowest male and female aver- 

 ages. In Fu a group of four families (table 8 F and fig. 3F) 

 shows averages for parental sum 11 that are above averages 

 for 10 and 12b. Parental sum 12a has averages much below 

 any of the other famihes. This study of averages must be 

 followed by careful observation of the figures referred to above, 

 which show the frequency distributions of each of the families 

 that has been considered. These curves have been plotted so 

 as to enclose in each polygon similar areas, to facilitate com- 

 parisons. The actual numbers of flies are given both in the tables 

 and the figures. 



Further data bearing on the question of the relation between 

 the grades of parents and children is to be found in the series 

 of curves representing inbred Unes through single matings (table 

 11 and fig. 6). One case must be especially emphasized. The 

 last generation. No. 216, in the series beginning with No. 18 

 (fig. 6 A) was produced by flies with no e?rtra bristles. This 

 generation which follows seven selections has a higher distribu- 

 tion than the preceding ones which were from high grade parents. 

 There are no normals from these normal parents, and no females 

 with less than two extra bristles. This extreme case is in ac- 

 cord with all that has been found in regard to the independence 

 of the parental and fihal grades. This is further borne out by 

 these int)red lines in the comparison of parents and children in 

 successive generations. 



