BY THE SELECTION OF SOMATIC VARIATIONS. 25 



As a whole, there was a rather large proportion of the plants that 

 remained within the type as classified, although there was hardly a 

 plant grown during the winter that did not become somewhat greener 

 during that period. 68 plants fluctuated in a marked degree; 47 of 

 them were almost entirely green during the winter, but were again 

 quite uniformly yellow spotted in summer, although some of these 

 remained much greener in summer. None of the latter, however, could 

 be considered as of the pattern green-red blotched. 



13 plants fluctuated irregularly, giving mixed patterns, mostly of 

 green-yellow-red blotched, green-yellow spotted-red blotched, and green-red 

 blotched, all more or less intermingled among the various branches and 

 on the same branch. These were not used as parent plants, but doubt- 

 less by selection it would be possible to obtain a marked degree of 

 constancy for the irregular and mixed patterns, with, also, production 

 of plants that would be uniform for various types. 



The most uniform and marked of the fluctuations was the case 

 of 8 plants of clone 14 which gradually became more yellow during 

 the winter of 1912-13, until they were quite typical green-yellow-red 

 blotched', 5 of these were used as parents of the plants of clone 14 already 

 reported with table 3. The change in pattern arose as a gradual 

 increase of yellow from various degrees of a yellow spotted condition to 

 a well-defined yellow bordered pattern that was quite uniform for the 

 entire plant, and which when tested in progeny was subsequently quite 

 as constant as cases which arose by sudden variations. An analysis of 

 the pedigrees of these 8 cases shows that all of these descended from 

 only 3 of the 7 plants grown from cuttings of the original branch 14. 

 This phenomenon of the appearance of the same variation in different 

 plants that were derived from the same more remote ancestor is common 

 and constitutes what we may call duplicate-reversions or variations. 



Besides the fluctuating variations in regard to green and yellow, 

 there were numerous cases of fluctuation in the red-blotched condition 

 both of the epidermis and of the subepidermal tissues, giving extremes 

 of very finely red-blotched or coarsely blotched. No selections were 

 made to secure types of the red-blotched condition. None of the 

 plants fluctuated to a no-blotched or to a solid-red pattern. 



The behavior of the 79 plants of this pattern in clone 13 is especially 

 interesting. They constitute a test for this pattern obtained by the 

 selection of gradual and accumulated fluctuation. The progeny of 

 5 plants grown in 3 generations, subjected to the same sort of selection 

 as the other clones, showed the highest degree of constancy obtained in 

 the clones of this pattern. 



Seven plants gave, during the winter, a marked increase of green, ac- 

 companied by the production of cut and laciniate leaves (fig. 7), the 

 appearance and constancy of which are quite fully discussed later. 



