BY THE SELECTION OF SOMATIC VARIATIONS. 73 



are inherited as such. The intercellular relations involving amounts 

 and distribution of pigments are widely and suddenly disorganized 

 and readjusted during the processes concerned with seed formation. 

 On the other hand, the metidentical characters green, yellow, and red 

 are quite uniformly transmitted in both sorts of cell-divisions.' 



The behavior of the laciniate character is significant in this con- 

 nection. After this character made its appearance in a plant it was a 

 constant feature in the development of all plants (but one) derived 

 by vegetative propagation. The seed progeny, however, exhibited 

 wide differences, ranging from deeply laciniate through all degrees or 

 grades to the entire type of leaf. Furthermore, the laciniate leaf 

 appeared in all seed progenies thus far grown, even when derived from 

 plants of a Hne in which the character had never appeared. The proc- 

 esses of reduction and self-fertihzation in a Coleus plant seem to bring 

 out latent possibilities for various developments of leaf-shapes. 



It remains to be seen if any of the types appearing in seed progeny 

 are more constant in vegetative or in seed propagation than are the 

 similar types that develop by bud variation. Already several bud 

 variations have appeared in plants of seed progeny, indicating vegeta- 

 tive changes in the processes concerned with pigment and pattern 

 formation. The colors involved in the variegation of Coleus represent 

 every type of coloration (green, yellow, white, and red or blue) con- 

 cerned with the variegation and coloration of plants. There is no 

 evidence that the essential nature of these characters in Coleus differs 

 from that of the characters concerned with variegation and pigmenta- 

 tion in corn (Emerson, 1914) , in Mirabilis (Correns, 1909) , in Melandrium 

 (Shull, 1914), in Pelargonium (Baur, 1909), in Antirrhinum (Baur, 1910). 



The explanations here given regarding the spontaneous variability 

 of the characters concerned in the development of pigments and of the 

 changes in intercellular and intertissue relations influencing develop- 

 ment of color patterns in Coleus apply equally well to such cases as 

 those just noted. The evidence indicates that the characters in ques- 

 tion, and most especially the pattern characters, are not represented by 

 units or factors, unless these assumed factors are to be considered in a 

 general sense as temporary conditions descriptive of tj'pes of develop- 

 ment and not as particular localized units of germ-plasm, which is the 

 conception that gave the Mendelian interpretation its definiteness and 

 simplicity. 



The knowledge of the nature and the heredity of color character- 

 istics will be advanced more by studies of the natural variability of 

 the characters involved and by chemical and phj'sical investigations of 

 the processes concerned in the formation and distribution of such sub- 

 stances as melanin, flavone, and phenol compounds rather than by 

 further elaboration of complicated formulae involving multiple factors 

 that attempt to explain fluctuations, inherited variations, and cases 

 of increased variability that appear in hybrid seed progenies. 



