STUDIES ON OAT BREEDING. 1 3 



upon the size of the mean dn the same sense that a standard 

 deviation is dependent upon the size of the mean. The larger 

 the absolute size of a character the greater are its chances 

 for variation. It would be desirable to determine the stand- 

 ard deviation of each pure line and then divide each deviation 

 by the standard deviation of its line. Thus expressed in terms 

 of the standard deviation these differences would be entirely 

 independent of any differences in the absolute means. How- 

 ever, the number of rows in each line is often entirely too small 

 to determine a reliable standard deviation. 



By the use of these differences it is possible to treat all of the 

 data together. Thus so long as each deviation is measured from 

 the mean of its own pure line we can put together the results 

 from all the lines regardless of the variety from which they 

 came. 



As a first approach to the question of the influence of selec- 

 tion within the pure line, tables have been formed for each line 

 as follows : 



