94 REPORTS ON INVEISTIGATIONS AND PROJECTS. 



ing, but they must be tested repeatedly and many of them followed up 

 for several years before they are reported upon fully. It may be well to 

 consider in a general way the subjects we are investigating. 



Inheritance of Characteristics. — One kind of organism differs from 

 another, and the newly evolved species differs from its progenitor, in one 

 or more characteristics. We have not reached the point where new char- 

 acteristics can be produced at will, but we can learn the laws of inherit- 

 ance of characteristics, and such knowledge will enable us to predict how 

 a new variation will behave when crossed with the parental species and to 

 classify variations according to their origin and behavior in heredity. 



We find that when two varieties that differ in some characteristic are 

 crossed it frequently happens that one only of the two forms will reappear 

 in the offspring (viz, the dominant characteristic, of Mendel), and it will 

 be little modified by the presence, in that offspring, of the germ of the 

 opposite characteristic. This is in accordance with the theory that most 

 characteristics are, or may be resolved into, elementary units. Similarly, 

 when a variety that has some new feature not possessed by the ancestor 

 is crossed with that ancestor the offspring usually have the character fully 

 developed (dominant). If these offspring are crossed together the char- 

 acter is absent in a small proportion only, on the average one-quarter, of 

 their offspring. The consequence of this law is that a newly evolved 

 characteristic is not at once swamped by intercrossing, as has often been 

 assumed, but may even spread at the expense of the parental type. Thus 

 the new species is nurtured in its infancy. 



Unit Characteristics. — The fact that in crossing varieties their dissimilar 

 characters do not blend is important, since it supports the theory that such 

 characters first appear as they now are, fully formed. It indicates that 

 since evolution has advanced by the addition of new characteristics it had 

 advanced by steps or jumps. A new species has not gradually arisen from 

 an old one, but suddenly, by mutation. Our breeding experience, conse- 

 quently, supports, in part, the mutation theory of de Vries. The practical 

 consequence of this theory is that it is more important to look for sports 

 or new combinations of characteristics than to work by selection. 



On the other hand, it is becoming clear that unit characters are not 

 immutable ; for, in hybridizing, the dominant form often shows traces of 

 its antagonist ; and if the dominant form is repeatedly infected by con- 

 tinued hybridization it may become much changed. 



R6/e of Selection. — Breeders attribute their success in improving races 

 chiefly to the selection, for breeding stock, of the best-appearing individ- 

 uals. Recently, a school has arisen which maintains that, within narrow 

 limits, selection is impotent. Our results are giving us an exact insight 

 into the truth of the matter and justify to a certain extent the breeder's 

 operations while offering a new explanation for them. 



