410 



THE AMERICAN NATURALIST 



[Vol. LV 



orations, but eventually settles down to a value not far 

 from 1.00. This last \-alue ma\' therefore be taken as 

 the normal dimori)hism in a honiozyo-ous low bar popu- 

 Icition. 



In tlie high selection line there is a simihu- ra])id de- 

 cline, in this case from i').Vl in the ])a rental generation to 

 1.45 in the third seh'etion uviicral ion. 'lliere is then a 

 pronounccMl iiiciTa>c and Hnclnalion l>etween 2.00 and 

 4.00 from Ihc tiflh to Ihc t wcnl \ -t hird generation. After 

 a second general decliiu' the \ alue remains for most of 



While the genei-al course of the curves is the same in 

 the two cases, the low selection line is the more consistent 

 and readies t]u> lower level. 



The i)rol)al)le exi)lanation of the decrease in dimor- 

 phism in l)oth seh'ction lines is to be sought in the fact 

 that some of the accessory factors affecting facet number 

 are sex-linked. The unselected population is a mixed one 

 and accordingly has a considerable degree of hetero- 

 zygosis for these factors in tlu> femah^s. One of the 

 results of selection I'oi- facet nunihei- with iid)reedinu- is 

 a decrease in this hetero/y-^osis. Ira\iiiu- the femah'S 

 homozygous for low facet factors in the h>w line and for 

 high facet factors in the hiuii line. As lenu" as the popu- 

 lation is honio/>i:Dn> il is prohahic thai the degree of 

 dinioi phi-m d(»e> nut clianue to an\ ureal e\1('nt because 

 there i. nu diMnrh; nee of tie fact.»rial proportions. An 

 inciv.-i.e (M. one Mde i. ac<-on,panhd l.\ a propoil ioiiate 

 iiicrea^e nn the other. The ^ame uonhl he tian- a> far 

 a^ the helerozxoou^ fenial- ai-e eoncenu'd if the d..nil- 

 iian<-e of low and lii-li factor> were alike. If. however, 

 the factor- t'oi- low facet nnniher have a hiuher dominance 



that -eleetl.m as p.-a.-tised will pnulnce a decrease in 



po])ulation. When thoe lietei-o/.\ u-ote.> ai'e eliminated 



