studies on oat breeding. 35 



Discussion. 



The statistical analysis of these data has given but very little 

 evidence of the positive effect of selection within these pure 

 lines. The one or two instances in which there is some doubt, 

 may be discussed briefly. The first instance is that shown in 

 Table 7 in which the total deviations of the daughter roiws show 

 a positive excess in the year following the selection. There is 

 some question as to whether these excesses in the direction ol 

 the selection are not to be regarded as coincidences of random 

 sampling. This is rendered more probable because the later 

 ' analysis by means of the indices show very little evidence for 

 such an effect. 



However, if these excesses are to be regarded as significant 

 it is very probable that they are due to physiological rather than 

 genetic causes. It is quite conceivable that the grain from the 

 more vigorous heavy yielding plants differs from 'that of the 

 smaller plants in either the amount or quality of its food 

 material. It was thought that possibly some evidence on this 

 point could be obtained by determining the correlation between 

 the yield and the weight per 100 grains of individual plants. 

 This correlation was determined for several pure lines but in 

 each case it was significantly zero, even showing a slight nega- 

 tive value in some instances. Thus it appears that the size of 

 the grain is not the determining factor in these data at least. 

 Whether the grain from the large and small plants vary in the 

 .character of their nutrient material is a question that we have 

 not studied. 



It was pointed out in connection with Tables 7, 10 and 11 that 

 with successive minus selections there was a positive effect of 

 the selection. This was shown both by a study of the devia- 

 tions from the mean and of the indices. On the other hand, 

 successive plus selections appeared to show a negative effect o^t 

 the selections. Whether any significance is to be attached to 

 these slight differences or not cannot be decided by the present 

 data. 



On the whole these results certainly show that selection within 

 these pure lines of oats has been without any marked effect 

 upon the progeny so far as the characters studied are concerned. 

 The results based upon the study of the indices agrees as cloisely 

 as could be expected with Johannsen's theory. 



