BY THE SELECTION OF SOMATIC VARIATIONS. 67 



clones of plants with decreased amounts of red indicates a specific 

 decrease in power to develop red and the results of pedigree culture 

 indicate that continued selection in this direction would give lines 

 with only slight amounts of red or perhaps that are entirely free of 

 all red. 



The metidentical characters of green, red, or yellow are themselves 

 fluctuating not only in expression but in inheritance through cell 

 lineage. They do not appear to be independent. I have never yet 

 obtained a plant by seed or by vegetative propagation that did not 

 possess some degree of red coloration, and as yet no strain has been 

 isolated that was pure for loss of yellow. 



On the whole, vegetative propagation of any new type that arises 

 gives a progeny that exhibits a rather marked degree of constancy with 

 fluctuations and variations about a new mode. Selections for pure 

 green do not give a progeny of pure-green plants, but do give a greater 

 number of green plants than does any other pattern. Selections for 

 decreased red or for increased red likewise give clones with this 

 tendency prevailing. In other words, readjustments of the processes 

 concerned with total production of pigments and their distribution 

 tend strongly to occur in growing-points or to so affect them as to 

 secure a certain degree of permanency. 



The assumptions of de Vries already noted in the introduction seem 

 to apply quite adequately to the behavior of the metidentical char- 

 acters as far as expression in individual cells is concerned, but does 

 not fully explain the phenomena of pattern changes as well as the con- 

 ception of a further influence of intercellular relations, modifications 

 of which may in time affect more or less permanently the expression of 

 metidentical qualities. Any Mendelian conception of pattern factors 

 that are units in heredity is quite inadequate, as is also such a con- 

 ception for even the metidentical characters. 



This analysis of the nature of variegation and the significance of 

 bud variation in Coleus has a direct bearing on the nature of certain 

 other types of variegation. 



One of the most clearly marked types of variegation is that of the 

 infectious chlorosis. The best known cases are those of tobacco 

 (Beijerinck, 1899; Woods, 1899) and Abutilon (Baur, 1904 and 1906; 

 Lindemuth 1897, 1899, 1901, 1905, and 1907). All investigations 

 agree that in these types the variegation is not transmitted to seed 

 progeny. The searching investigations of Beijerinck and Baur lead 

 to the conclusion that a living fluid or virus carries the contagion. In 

 the case of tobacco the infection is readily accomplished by various 

 agencies. In Abutilon grafting is necessary, and by this means the 

 variegation has been transmitted to numerous species of Abutilon and 

 related genera. Similar types of infectious variegation exist in other 

 groups of plants, as Fraxinus, Jasminum, Liburnum, and Ligustrum. 



