430 



The Journal of Heredity 



from which they sprang, which is at a 

 distinctly hij^hcr level than that of the 

 wild stock and at a very much hij^hcr 

 level than that of the Irish. With the 

 residual heredity of the minus selection 

 series they would doubtless look like 

 ordinary hooded rats. There remains 

 one possible different interpretation. 

 Mutants may differ from their hooded 

 ancestors by a factor which is wholly 

 indeijcndent of that ])resent in wild rats. 

 Symbolically, hooded rats may be 

 ShShmm. mutants ShShMM, while wild 

 rats are SSmm. Castle has, however, 

 virtually eliminated this possibility by 

 crossinj^f wild with mutant and raisinji; 

 an F2. On the hypothesis of inde- 

 pendence 6% hooded rats should ai^pear 

 in F2 but among forty-six individuals 

 none such a]Ji)eared. 



Thus genetic variations of the hooding 

 factor and ones independent of it are 

 both present to give a basis fof selection. 

 The origin of the mutants in the course 

 of the experiment, of course, demon- 

 strates that there is no absolute stability 

 of the germ plasm. Further, where 

 there is one variation which is of great 

 enough effect to be recognizable at once, 

 many more smaller should perhaps be 

 expected. Whether the selection ac- 

 tually hks progressed largely through 

 fa\'orable combinations of factors or 

 through the occurrence of small muta- 

 tions is not easy to answer. MacDowell 

 has cited the decrease in the standard 

 deviation as evidence that each selection 

 series is becoming more nearly homozy- 

 gous. It is, however, not certain that a 

 grade near the extremes of the series 

 means exactly the same thing as near 

 the middle. The fact that the correla- 

 tion between parent and offsi:)ring 

 shows actually a slight increase as 

 selection i)rogresses indicates that the 

 ratio of genetic to somatic variation 

 has not suffered and argues for the 

 frequent occurrence of new variations. 

 These may, of course, be largely inde- 



pendent of the hooding factor in inherit- 

 ance. It should be added that the large 

 amount of somatic variation and the 

 absence of close inbreeding make it 

 highl>' improbable that the parent- 

 oJTs])ring correlation can be due to a 

 splitting up of each selection series into 

 non-interbreeding groups, even though 

 there is assortative mating. 



vSumming up, it appears most prob- 

 able to the writer that genetic variations 

 are occurring sufficiently often to give a 

 basis for selection to an indefinite 

 extent. Most of such variations seem 

 to be inherited independently of the 

 hooding factor, but at least in one case 

 it has been demonstrated exhaustively 

 that the hooding factor itself has 

 varied to a new level in the direction of 

 its allelomorph in wild rats. 



Conditions are doubtless more or less 

 similar in other ])iebald animals. In 

 most cases, however, the variability 

 due to irregularity in develo])ment is 

 greater than in the rat. This is notably 

 the case in the guinea-pig in which the 

 pattern remains exceedingly \'ariable 

 after generations of the closest inbreed- 

 ing. Yet in a stock breeding at random 

 the correlation between parent and 

 offs]jring is about the same as in rats. 

 Perhaps the irregularity in develoi^ment 

 argues a genetic instability from which 

 even the germ cells are not exem])t. 



Finally, under any interpretation. 

 Castle's selection exi^eriment demon- 

 strates the efficacy of Darwinian selec- 

 tion. It is true that one large mutation 

 occurred with effects perhaps as large 

 by itself as the entire ])lus selection 

 series, but where such a variation gives 

 one new level, selection has ])roduced a 

 continuous series of stable levels. This 

 would give selection of small variations 

 a more imjDortant place in evolution and 

 animal husbandry where it is nice 

 adjustments of one character to another 

 or to the environment that count. 



