No. 600] PIEBALD BATS AND MULTIPLE FACTOBS 



725 



swered by the statement that "no slowing up is observ- 

 able in the rate of change of the racial character under 

 selection either plus or minus " (Castle, :16, p. 96). This 

 is assuredly a very vital point in the contention that mul- 

 tiple factors will not explain the results. For, if the rate 

 of advance has not fallen off, and if, during seventeen 

 generations, each selection has been as effective as the 

 preceding one, it certainly would look as though this prog- 

 ress were due to constantly varying germ plasm, and not 

 to the sorting out of certain groups of factors. Were a 

 sorting out of factors going on, each advance would re- 

 strict the possibilities for further advances, so that in a 

 series of selections the rate of advance would decline. 



In Castle's " Heredity," page 122, Fig. 41 are shown 

 the curves of the averages of the first eight generations 

 of the plus and minus races. These curves begin with 

 the average of the offspring that appeared after the first 

 selection. From this point on, the advance shown by the 

 curves is gradual. But should not the advance resulting 

 from the first selection be recorded? The average of the 

 first selected generation was not the point of departure. 

 To show the advance resulting from the first selection, 

 the first point of the curve must give the average of the 

 hooded race before the first selection. Unquestionably 

 the difference between the average of the unselected race 

 and the first selected generation was an advance due to 

 selection, yet this advance is apparently ignored in the 

 statement quoted above, as well as in the figure cited. 

 The first selection resulted in a very much greater ad- 

 vance than any other single selection in the whole series. 

 It took the ten subsequent selections to separate the means 

 of the two races as far as the first selection separated 

 them. If each selection had produced a like advance, the 

 eleventh generation of selection should find the averages 

 of the two races eleven times as far apart as they were 

 after the first selection instead of twice as far apart. 

 Failure to consider the advance due to the first selection 

 has concealed one of the most striking features of the 



