No. 627] SHORTER ARTICLES AND DISCUSSION -373 



tracted hooded individuals exclusively plus in cluiracter, range 

 from 1.00 to + 3.50, mean + 2.55. The variability has si- 

 multaneously fallen to .66, which is only about one tliird greater 

 than that of the minus race in the first five generations of the 

 selection experiment. (See Castle and Wright.) One family- 

 containing fourteen thrice extracted hooded individuals has a 

 mean grade for the hooded individuals of + 3.05, which is prac- 

 tically identical with the grade of the thrice extracted hooded 

 individuals resulting from the plus crosses (Table I). 



It thus appears that three or at most four crosses with a wild 

 race suffice to obliterate all the racial differences which had been 

 induced by ten generations of selection in the case of the plus 

 race and sixteen generations in the case of the minus race. The 

 plus race was changed almost immediately by a single cross, but 

 the change was small (a fact which misled me until the results 

 of the minus crosses were secured) . The changes with the minus 

 race were so great that they could not be fully secured by less 

 than three or possibly four successive crosses (eight generations 

 of offspring) . The wild race, which we used in our crosses, evi- 

 dently had a residual heredity much more like that of our plus- 

 selected than like that of our minus-selected race. When the 

 hooded gene from either race was introduced by repeated crosses 

 into this residual heredity, the result w;is to pi-oducc- Iioodod 

 races of very similar grade, a little lower in -^radr tliiui ilie plus 

 selected race, but very much higher in ^nulc than ihc inimis 

 selected race. 



It thus becomes clear that the changes wliich had occurred in 

 the hooded character as a result of selection were detachable 

 changes and are probably in nature independently inherited 

 modifying factors. This is a view which Phillips and I gave as 

 one of two possible interpretations of the results which we pub- 

 lished in 1914. Morgan, Muller, MacDowell and others have in- 

 sisted that this was the only reasonable interpretation which 

 could be given, but I have not been satisfied with this conclusion 

 in advance of a really crucial experiment, such as I believe has 

 now been performed. Meanwhile the probability that tlie theory 

 of multiple modifying factors is corrcH't as a general oxphinntion 



