210 



THE AMERICAN NATURALIST [Vol. XLVIII 



forms, referred to as types 3 and 4, the latter being ca- 

 pable of further subdivision. Neither of these was de- 

 scribed by Shull. At least one difference between rhom- 

 boidea and simplex, on the one hand, and types 3 and 4, on 

 the other, could be noted at once, i. e., the relative width 

 of the leaf. As has been shown above, the former have 

 their first leaves twice as long as broad, the latter three 

 times as long as broad. The idea suggested itself that 

 there might exist a factor which determined these charac- 

 ters. Since the original parent belonged to type 4, the 

 narrow character of the earlier leaves must be dominant 

 over the broad character. Also, since the original parent 

 produced both " narrow' ' and "broad" types, it must 

 have been heterozygotic for this character. Using (N) 

 to indicate the gene, we get for the zygotic construction 

 of the parent plant aaBbNn. 



aBN I aBn abN abn \ 

 abN | abN abN abN I 



bn J aBN aBn abN | abn 



Since self-fertilization is the rule in Capsella, it was an 

 easy matter to test the validity of the theory. A form 

 aaBbNn, one with unelongated primary lobes, sinuses 

 reaching the midrib and with early leaves of a "narrow" 

 type should yield, on self-fertilization, the following com- 

 binations: l.bbnn (square 16), a plant of which, accord- 

 ing to our definition, the earlier roset leaves should be 



