30 MAINE AGRICULTURAL EXPERIMENT STATION. I915. 



greater than i ; in one case an index of — 23.5 and in another 

 +48.8. These extreme variations tend to balance each other 

 and no doubt would do so if sufficient rows were grown. But 

 with a small number of rows such a selection may materially 

 change the value of the mean index. There is a priori no rea- 

 son why such small selections should be omitted. For on the 

 selection theory a small variation should result in a much 

 smaller deviation in the daughter row. However, there are 

 very few of these small selections in the present data. 



In order to show the effect of this one small deviation, which 

 is by far the smallest one in these data, we may give the mean 

 index for the 191 1 selection upon the 1912 and 1913 rows when 

 Line 262 is omitted entirely.'"' 



Table 9. 



Mean Index with Line 262 Omitted. 





Rows Grown in 





Mean Index. 



. 





+selections. 



— selections. 



All selections. 



1912.. 

 1913.. 





0.9952 



1.1249 



1.1723 

 0.8910 



1 .0842 

 1 .0043 



Comparing these figures with those in the first two rows of 

 Table 8 it is seen that omitting the effect of this very small 

 selection results on the average in bringing the mean index 

 nearer to i. 



On the whole these indices do not give us any evidence that 

 the selections have had any effect upon the succeeding gener- 

 ations. The values tend to fluctuate about i.o as demanded 

 b}'' Johannsen's theory. It should be pointed out that if regres- 

 sion took place within these pure lines according to the Galton- 

 Pearson theory of ancestral inheritance the value of these 

 indices should approach the value 0.333. There is not the 

 slightest indication that they do this. 



"There were no rows grown from this snuall selection in 1914. 



