62 POPULAR SCIENCE MONTHLY. 



pure white which breed pure white, and some pure gray mice which 

 breed pure gray, besides some impure forms like those of the first gen- 

 eration, which breed both colors. In other cases the new form if bred 

 to the parent stock gives rise to offspring which even in the first gen- 

 eration are like the new form. Thus it appears that when the ancon 

 ram was bred to an ordinary ewe the offspring were like the father. 

 Similar cases are recorded for the merino sheep, for the japanned 

 peacock, and for other new types. It appears here that the new type 

 dominates in some or in all of the offspring. Darwin had in mind 

 cases like these when he rejected the view that discontinuous variations 

 have furnished the material for natural selection.* 



There are, however, still other kinds of sudden and definite varia- 

 tions, and these, not the former, are the kind which are mainly re- 

 sponsible for new species. New elementary species arise not through a 

 loss of some character of the old form (although new species may have 

 amongst their new characters some that are due to loss), but by the 

 appearance of a new character or characters which follow a different 

 law of inheritance. When such elementary species are crossed with 

 the parent type, a new type may arise, which is usually intermediate 

 in character between the parents, but in some cases may be more like 

 one than the other. Similar results follow when the new elementary 

 species are crossed with each other. Thus the results are comparable 

 to those that occur when Linnsean species are crossed, although in the 

 case when forms belonging to two widely different Linnaean species are 

 combined the results may be so complicated that it is beyond our power 

 to give a satisfactory analysis of the results at present, f 



The term mutation has been used for the sudden appearance of new 

 variations belonging to any one of the preceding cases. Whether it 

 may not prove necessary to use different terms for the different kinds 

 of discontinuous variations in so far as they follow different laws of 

 inheritance remains for the future to decide. We might, perhaps, use 

 the term mutation (and mutant) only where new elementary species 

 are formed; retrogression where a character is lost (elementary varie- 

 ties being formed, which follow the Mendelian law of inheritance, as 



* It is not clear from the records whether this class follows the Mendelian 

 rule of inheritance, with the new form as the dominant in the first generation. 

 From the statements that are ordinarily made, it might appear that even in 

 the first generation some of the offspring are like one parent and some like 

 the other, which, if true, would make these cases different from the first class 

 given that follow Mendel's law of inheritance. 



f In my book on ' Evolution and Adaptation,' T pointed out that while a 

 single step or imitation could not he halved, yet if a number of steps were taken 

 in the same direction the number might then be halved. This may be true 

 but is unnecessarily hypothetical, and, as pointed out here in the text, the ap- 

 parent difficulty that Darwin met with can now be explained in other ways. 



