12 MAINE AGRICULTURAL EXPERIMENT STATION. I915. 



EFFECT OF SELECTION TESTED BY METHOD OF DEVIATIONS. 



In view of the faot that external conditions are able to cause 

 such great fluctuations in the mean yield from year to year it 

 is not possible to compare directly the yields of one year with 

 those of the next. In other words selection in the plus direc- 

 tion might very well have produced a marked effect, yet owing 

 to environmental conditions the yield of these plus selections 

 might be lower than the yield of the pure line in the previous 

 year. 



Further our pure lines have been selected from a number of 

 different varieties. Many of these varieties show tendencies to 

 yield at different rates. Consequently from such heterogeneous 

 material it would not be possible to compare the direct effect 

 of selection upon yield for all of the pure lines. Thus if we 

 should approach the question by the correlation method and 

 correlate the yield of selected plants with the average yield 

 of the resulting rows we mig'ht very well obtain a significaut 

 correlation due entirely to the heterogeneity of the material. 



For these reasons it is desirable to obtain some measure of 

 the yield which is not so greatly affected by the seasonal fluc- 

 tuations. The simplest way of doing this is the method of 

 differences. Thus we may determine the mean yield for a pure 

 line in a given year. We may thdn find the deviation, either plus 

 or minus, of each row from this mean. The sum of all these 

 deviations is of course equal to zero. Similarly we may find 

 the mean of this same pure line in the previous year and the.i- 

 determine the amount and direction of the deviation of the 

 selected plants from this mean. The sum of these latter devia- 

 tions is not necessarily equal to zero. They are the deviations 

 of a few selected plants from a mean determined from all the 

 plants of that line grown in the same year. 



These differences indicate the amount and direction that a 

 given selected plant deviates from the mean of its line and 

 likewise the amount and direction by which its daughter rows 

 deviate from the mean of the same line in the year in which 

 they were grown. Thus, if there is an effect of the selection a 

 plus selection ought, on the average, to result in a row which 

 also deviates in the plus direction. 



These differences are to a large extent independent of the 

 absolute size of the mean. They are, of course, dependent 



