STUDIES ON THE GENETICS OF FLOWER-COLOURS, ETC. 
127 
actual number of the two kinds of segregates does not very well agree with 
the expected (expectation on 3:1 basis, 93*75 ±4*8: 31-25 ±4-8). How the 
magenta of such composition has arisen from the white cannot yet be determined, 
and though its origin might be perhaps inferred this is merely a matter of 
conjscture, and it would not be worth while to state here such an inference. 
II. Reversion in the Somatic Cell .—'To this class belongs No. 11 in 
Table YHI. Suppose that the reverse mutation has occurred in a certain 
somatic cell of the composition CCvrbb, at least one cell-generation before 
the reducing division, and suppose further that this cell has got the composi¬ 
tion CCJRrbb in consequence of this process. From the latter call the male 
and the female gametes CItb and Cvb will be derived after one or more 
cell-generations according to different cases. The fertilisation between them 
will give rise to zygotes consisting of 1 CCJRRbb, 2 CCJRrbb and 1 (7(7- 
vrbb ; the production of 20 reds and 5 oranges in No. 11 may be due 
to such a process (expected 18*75±2 and 6*25±2). 
III. The mutations described in I and H are, as already stated, the 
reverse ones, i.e. those from recessive to dominant condition. The mutations in 
the opposite sense, i.e. those from dominant to recessive condition (so-called 
“ loss-mutation ”) have been also observed sometimes, though not frequently, 
to which might belong Nos. 23 and 29 in Table VHI. In No. 29 one 
orange which should be theoretically homozygous, i.e. CCvvbb , has produced 
1 magenta, 118 oranges, and 58 whites. According to our hypothesis a 
certain somatic cell having the genotypic composition CCrrbb has undergone 
a “loss-mutation,” and changed into Ccrrbb', the gametes Crb and cvb 
are derived from it after a number of cell-generations, and their free assort¬ 
ment has given rise to oranges and whites in the approximate ratio 3 : 1 (119 
oranges : 58 whites, theoretically 132*75±5*8 : 44*25±5*8, if we will count 1 
magenta as orange); 1 magenta was produced at the same time, perhaps 
because 1 gamete CRJB has arisen by reversion of b into fr. Thus the 
whole process in No. 29 consists in loss-mutation and reversion combined. 
No. 23 (magenta) was found to segregate in F> into magentas and reds, 
whence we might at once be led to the assumption that that magenta parent 
should have the composition CCJRJRJBb (75 magentas : 42 reds, expected 
87*75±4*7 : 29*25±4*7). Since however this parent has been originally derived 
