562 



THE AMERICAN NATURALIST 



[Vol. LV 



\/4.5, or 17.67 per cent.. \vlii<'li is ihc poreentag^e corresponding 

 to 4 factors us <xWvu in Ca^Wc'r. Tahlo IE. In other words, 

 .six factors witli (louiiiimirr pi-odiice as great an increase in the 

 variabilitv ;is four fadofs wouhl produce if none of them 

 exhibited doniiiiaiii-e. 



Tt can 1).' shown m the same way. ])y means of a simple test 

 eas<'. that Castle is entiivly in effoi- in saying- llial "if one factor 



in a rasi> of hlen<liii,t>- iniiefit anee. this will l)e se.'n in the pro- 

 duction of a^\■|!^n<■tl•i<■al oi- mnlt iniodal vai-ialion i)()ly<rons in F, 



casr the faetoi- having' the ureater influence happened to be (h)mi- 

 nani. for in the ahsmcc i»f (h)miiiance each factor enters into 



if wo assume that foui' faeior pairs AABBCCDD differentiate 



