Species crosses in Rats. 119 



tamer and quieter. But the g:roup must have possessed some potential 

 variability in this respect, for in the pure fieldrats no effect of a number 

 of cagebred generations was seen. 



Conclusions and generalizations. 



The main value of the experiments described lies in demonstrating 

 that new characters can originate by crossing, in a way which closely 

 simulates mutation. If, to account for the production of such novelties 

 we can choose between an explanation requiring only a difference in 

 genotype between two phenotypically identical forms, a natural ex- 

 planation therefore, and an inexplicable phenomenon, the spontaneous 

 loss or rreation, or change of a gene, mutation, we prefer the natural 

 explanation. 



Had we held to the purely unnecessary assumption that the pro- 

 duction of a new, recessive character necessarily proved "mutation", 

 we would nevei' have found the real reason of the production of these 

 novelties. We demonstrated, that such nova, originating as double- 

 recessives, aabb from a rross between AAbb and aaBB are different 

 in only one gene from either parent species. This of course make these 

 cases doubly difficult of analysis. 



Is it possible that the classical "unitcharacters"' in mice originated 

 in this way? On first sight this seems impossible, because all these 

 novelties, albinism, dilution, waltzing, etc. behave as single recessives. 



If in these rats novelties arose in a way, which can be explained 

 but which at first sight looks like mutation, the {question arises whether 

 in other similar cases, especially in Drosophila, the origin of similar 

 novelties should not be recombination of genes rather than mutation. 



Just how easy is it to mistake a case of the production of a donble 

 recessive novelty for mutation. The possibility of mistaking the one 

 for the other is indeed veiy great, if one is not actually expecting the 

 process to happen. In the first place, unless one can make breeding- 

 tests with the ancestors of novelties, which we can do in our mice, but 

 which is wholly excluded in shoitlived animals like Drosophila, the only 

 proof that a novelty arose as a doublerecessive rather than as a single- 

 recessive mutant lies in the numerical ratio in which it originated. A 

 doublerecessive novum is liorn among sixteen brothers and sisters with 

 the normal character. A real mutant, differing from its ancestors in 

 one gene, is born as a novelty in a proportion of one in four. The 



