liv Department oe Expekimi^ntal Plant-Breeding. 



believed that by the continuous selection of the best individuals, 

 that is, the individuals showing a desirable character in the highest 

 degree, this character could be continuously and gradually aug- 

 mented and improved in a cumulative way. While a belief in the 

 mutation theory does not entirely preclude such an interpretation, 

 yet the two theories in general are antagonistic, and if gradual 

 improvement or augmentation of the character could be secured 

 in a few years by the selection of the best, it could not, in general, 

 be claimed that the variations selected in each case were what we 

 should interpret as mutations in the DeVriesian sense. As the 

 claims of DeVries challenge the correctness, therefore, of the old 

 idea of continuous selection, studies on this question become, from 

 a practical as well as a scientific standpoint, of great importance. 

 It is of the highest importance that we should be able to tell prac- 

 tical plant-breeders whether to select the best continuously every 

 year, or to select the best types appearing at a certain time, and 

 then use the energy to purify these types. The two methods are 

 radically different, although the first selections are made in the 

 same way. In order to solve this problem it will certainly recjuire 

 very careful experiments by many investigators extending over 

 several years, and experiments of this nature have been started 

 here with wheat and oats. 



Studies on correlation in breeding zvork. — Both with plants and 

 animals it has frequently been found that different characters are 

 more or less united or correlated in such a way that the inheritance 

 of one character carries with it the other. In some cases, such 

 correlations have proved of great value in guiding selection. In 

 the selection of corn, as an illustration, the fertilization takes place 

 before evidence can be secured as to the productivity of the plant, 

 and in corn breeding work it would be very valuable if any char- 

 acter visible at the time of pollination could be found correlated 

 with high yield. This would enable the experienter to select high- 

 yielding plants at the time of pollination, so that good yielders 

 could be bred together by hand manipulation. In order to secure 

 evidence as to whether such correlation of characters existed, a 

 study was made of about i.ooo individuals of Funk's 90-day corn, 

 all of the qualities which were thought to be important being con- 

 sidered in their relationship to yield. In a number of cases a 

 high degree of correlation was discovered, but it is doubtful whether 

 the percentage of correlation is high enough in any case to be of 

 very great value from a practical standpoint. 



