BY THE SELECTION OF SOMATIC VARIATIONS. 59 



Such phenomena of variation appearing in hybridization experiments 

 are usually considered as due to segregation and recombination of 

 hereditary units during the processes of self- or cross-fertilization. 

 Bud variations in vegetatively propagated plants are, of course, inde- 

 pendent of such recombinations. 



That bud variations are generally due to a complete loss during cell 

 division is not substantiated by the results here reported or by the 

 bulk of other experimental work. In the majority of cases the 

 character concerned does not breed true. Mendelian students have 

 interpreted this to mean that such bud variations are produced by a 

 loss of only one factor of a diploid pair, giving heterozygocity. This 

 illustrates the tendency of the Mendelian interpretation to assign the 

 numerous cases of fluctuation in characters to heterozygocity rather 

 than to fluctuations in a factor or to irregular mutational changes 

 spontaneous in the organism. 



In regard to the range of expression in a single plant the laciniate 

 leaf-shape is a more striking character even than the color patterns. 

 It arose, as already noted, in 13 individuals obtained by vegetative 

 propagation, but these were all derived from a few plants of the next 

 preceding generation. This character was inherited through vegeta- 

 tive propagation by all plants grown but one, but the leaves on each 

 individual plant varied from deeply laciniate to fully entire. Plants 

 raised from seed gave all types from extremely laciniate to fully entire, 

 the particular type appearing in the first leaves that developed and 

 remaining quite constant for all leaves developed in the 6 months that 

 the plants have been grown. The special point of interest is that a 

 single individual of the laciniate group passed through a series of 

 fluctuations, giving all grades of leaf-shape from entire to fully laciniate. 

 The range in a single plant is greatly more marked than the differences 

 between the Urtica hybrids (Correns, 1905, 1912), in which the serrated 

 type of leaf was dominant. In the hybrids of the normal and the 

 laciniate types of Chelidonium ma jus (de Vries, 1900) there seems to be 

 no published data regarding the range of variation in the F 2 generation. 

 Of the hybrids between palm-leaf and fern-leaf types of Primula sinen- 

 sis, Bateson (1909) states that " dominance is usually complete/ 7 

 but that he has seen two strains with intermediate leaf-shape. Greg- 

 ory (1911) states that "the palmate character is dominant, though a 

 slight difference can sometimes be recognized between pure and hetero- 

 zygous palmate types." Crosses between an ivy-leaf (a palmate 

 shape with margins crenate) with the fern-leaf gave the normal palmate 

 leaf as an FI hybrid. The F 2 generation exhibited a wide variation, 

 which Gregory groups into 4 classes and assumes that shape of the 

 entire leaf and crenation of the margin are two independent characters. 



On the basis of character of lobing of climax leaves, Shull (1911) 

 distinguishes four biotypes in Bursa bursa-pastoris. The view that 



