The Isolation of Tricotylous Intermediate Races. 419 



it is desirable, in connection with what was said in the 

 first part of this volume, to consider what we should 

 expect to occur in the isolation of such intermediate races 

 as may happen to be present. The intermediate race does 

 not differ from the ordinary types or half races in any 

 visible characters, but only in the frequency of tricotylous 

 specimens. If the tricotylous individuals are selected for 

 culture from crops raised from bought seed, it is by no 

 means certain that these and these only belong to the 

 intermediate race, if indeed such exists at all. For in 

 addition to this, the half race is almost always present, 

 and this, as we know, also contains some tricotyls. On 

 the other hand the intermediate race always produces 

 atavists and usually in considerable numbers. Provided 

 therefore that the crop contains an intermediate race, 

 some of the dicotyls and some of the tricotyls would be- 

 long to it ; but the prospect of obtaining it from the latter 

 w^ill obviously be greater than from the former. For this 

 reason I select, whenever possible, tricotyls only for trans- 

 plantation. All that then remains to be done is to save 

 their seeds separately from each plant and to determine 

 their hereditary values in the harvest. If any of these 

 are especially high they belong to the intermediate race, 

 and all the res! are thrown away inasmuch as they in- 

 clude the half race and the doubtful cases. Ordinarily 

 the race is by this time perfectly pure and can be improved 

 by selection on the ordinary lines. As a rule, the means 

 of the curves describing my intermediate races, lie be- 

 tween 50 and 60%, and can be brought as high as cSO to 

 90% by selection in occasional individuals. 



In contradistinction to the unsuccessful experiments 

 described in the previous section there can be no doubt 

 in these successful cases that we are not dealing with the 



