Modification of Unit Characters 105 



realized these possibilities and tested them. Of course 

 if it was a case of complementary factors which had 

 become separated he would have been able to recombine 

 them in later generations by the proper crosses. Recog- 

 nizing the significant crosses and making them he ob- 

 tained no result; the character did not reappear. In 

 the same way breeding tests for ah inhibiting factor 

 which could be separated out in later generations yielded 

 no results. The unit character was gone beyond recov- 

 ery; it had simply dropped out and was lost. 



Such an occurrence may seem surprising at first, but 

 it is what should be expected as an occasional occur- 

 rence. Mendel's law was repubhshed along with De 

 Vries' mutation theory, and as a consequence scientific 

 breeders, interested in Mendel's law, were also on the 

 watch for mutations. De Vries had shown that muta- 

 tions might involve either the appearance of an entirely 

 new character or the dropping out of an old character. 

 It is this situation, therefore, that appeared in Castle's 

 experiments with mice, namely, the abrupt disappear- 

 ance of a unit character is simply the result of mutation, 

 and this involves no violation of Mendel's law. 



The real attack upon MendeUsm involves a different 

 kind of behavior on the part of unit characters. The 

 notable example of this unexpected behavior of unit 

 characters was discovered in connection with Castle's 

 work on hooded rats (2). 



He isolated from his rat populations a certain strain 

 which showed a particular black and white coat pattern. 

 This type was then inbred for a number of generations 

 to insure that it bred true. This fact having been 

 established it was next determined that this black and 



