STUDIES ON OAT BREEDING. 



25 



Table 7. 



Shozmng the Effect of Two or Three Successive Selections in 

 the Same Direction. 





1913 Rows. 



1914 Rows. 





+ + 

 Selection. 



Pelection. 



+ + + 

 Selectior. 



£ election. 





+ 



— 



+ 



— 



+ 





+ 



— 



Number of rows 



Total deviations 



Average deviation per row 



28 



26.93 

 .962 



39 



28.75 

 .737 



24 



16.21 

 .675 



41 



30.45 

 .743 



17 



20.01 

 1.177 



23 



30.07 

 1.307 



17 



19.60 

 1.153 



21 

 36.66 



1.746 



From this table it is noted : 



1. That with two successive plus selections (1913 rows; 

 there is no effedt of the selection. There is a slight excess of 

 the total minus deviations, i. e. in the direction opposite to the 

 selection. 



2. With two successive minus selections (1913 rows) there 

 is a substantial excess of the total minus deviations. The total 

 minus deviations are almost twice the total plus deviations. 



3. With three successive selections in the same direction 

 (1914 rows) there is an excess of the minus deviations in each 

 case. Even where the selection has been plus in each year the 

 total minus deviation is one-third greater than the plus devia- 

 tions. Here there is a substantial excess in the opposite direc- 

 tion' to the selection. 



4. With three successive minus selections there is a marked 

 excess in the direction of the selection!. 



Judging from these figures it would appear that successive 

 plus selections have resulted in rows which on the average were 

 below the means of their lines. On the other hand successive 

 minus selections have shown in each case very marked excess 

 in the direction of the selection. 



The results so far obtained are to a large extent contradictory. 

 It has been seen that, considering the effect of each selection on 

 the years following it, there is some evidence oi a positive 

 effect on the year immediately following. If this effect were 

 really due to the selection, then successive selections in the same 

 direction ought to produce a still more pronounced effect. From 

 table 7 it is seen that this prediction is realized in the case of the 



