458 



THE AMERICAN NATURALIST [Vol. LV 



tcr of dominance without linka{?e, sucli distributions would be 

 expeeted to show right-hand skewness. But they do not con- 

 In regard to the second objection, that of recombination of all 

 favorable factors, Collins has given a large number of figures to 

 show what was already well known, that with a large number of 

 factors the cliances for recombination are remote with the small 

 progenies grown in experimental plots. It was not intended to 

 maintain that pedigree cultures were adequate to show that such 

 a recombination could not be made. The point in mind, if not 

 clearly expressed, was that natural selection in isolated popula- 

 tions oi cultivated plants had not brought about any noticeable 

 approach to stability. In the hills of New England maize has 

 been grown for long periods of time in isolated fields. Some 

 varieties hav«^ probably been grown for at least fifty years with- 

 out admixture. Yet these varieties when self-fertilized show as 

 ra{)id a reduction in growth as other varietire which arc lately 

 the product of extensive hybridization. 



There is an enormous difference in the possibilities for imme- 

 diate recombination with and without linkage. To illustrate: 

 with twenty independent factors the chance for the bringing 

 together of all dominants in a homozygous state in one genera- 

 tion is tlH'otvtically one in 4"-^". With the same twenty factors 

 dist ril.iiitMl by twos in ten different chromosomes, each being 

 separat.Ml l)y wu units of crossing-over, the chance for recombi- 

 nation is tliiMiivtinilly one in 20''\ This is a difference in total 

 luuiihi is so v;i-T ;is to be almost inconceivable. Working over 

 loni;- p. iio(U n!' time, linkage may not be a hindrance to recom- 

 bination, as ta.-toi's one*' brought together tend to stay together 

 as fifiiily as ttu'v once resisted separation. Many cross-fertilized 

 species in^ the wihl whose age is^^ measured in geological periods 



isolate.! for a considerable time show no iioticeable approach to- 

 ward this coiiditif)!!. It seems reasonable to suppose that the 

 arrannviiK-Mt of factoivs in the chromosomes has something to do 

 witli this stati' of affairs. Therefore, until the chromosome 

 llirofy ot" hfifdity was developed, there was considerable plausi- 



