412 G. F. SPRAGUE 



of each family having the highest oil percentage were again grown in progeny 

 rows for continued inbreeding and selection. When the analyses were avail- 

 able the sibling progeny having the lowest average oil percentage was dis- 

 carded. The two selfed ears having the highest oil percentage in the selected 

 sibling were used to propagate the family. This process was continued 

 through five generations. The general procedures used in selection, with the 

 exception of the chemical analyses, are essentially those commonlv employed 

 in the development of inbred lines by the standard method. 



It should be emphasized that the time requirement, number of pollina- 

 tions and analyses, land requirements, and selection differentials were es- 

 sentially the same for the recurrent and the selfing series. The relative 

 efficiencies of the two methods therefore should be directly comparable. 



RECURRENT SERIES 



The results from the recurrent series will be presented first. The material 

 from the 111. High OilXw^xOs420 series has been carried through two cycles 

 after the original selfings. The frequency distributions are shown in Figure 

 26.3. The distribution presented for the original population is a composite 

 for the two backcrossed populations. The solid vertical line represents the 

 population mean and the dotted vertical line the mean of the selected sample. 

 These selected ears were grown in ear-row progenies the following year and 

 all possible intercrosses made by hand. Bulked seed from these intercrosses 

 provided the source material for the next cycle of selfing and selection. The 

 mean of the first cycle population was essentially the same as the mean of 

 the selected parents — the full selective advantage of the parents had been 

 retained. In the second cycle population the mean was further shifted to the 

 right by an amount equal to 2.1 class intervals, but still failed to equal the 

 mean of the selected parents by an amount equal to 1.1 class intervals. 

 The mean of the original population was 7.2 per cent of oil. The mean of 

 the second cycle population was 10.5 with the extreme deviate at 13.5. 



The ranges and standard deviations of these three populations are of some 

 interest in indicating any changes in genetic variability. Considering first 

 the range: in the original population the range was from 4.5 to 10.5, in the 

 first cycle 5.5 to 12.5, and in the second cycle 7.5 to 13.5 — a difference of 6, 

 7, and 6 class inter\^als respectively. The first cycle had the greatest, the 

 original population intermediate, and the second cycle the smallest standard 

 deviations. The fact that the second cycle exhibited the smallest standard 

 deviation may indicate some loss in genetic variability. However 65 per 

 cent of the selective advantage of the parents was retained indicating that a 

 considerable amount of genetic variability exists. 



SELFING SERIES 

 The selfing series presents a strikingly different picture. The results are 

 presented graphically in Figure 26.4. The values plotted for the Si generation 



