1 92 Seciregation of a Quantitative Character in Beans 



somewhat diffuse, but still intermediate between E and M. For further 

 analysis were chosen two very short types, one from row 9, the other 

 from row 36 ; two very long types fi'om row 31, and a number of^ypical 

 forms between these limits, in all 18 rows. The two shortest types gave 

 rows 49 (from 9) and 62 (from 36) in 1917; the two longest, rows 56 

 and 57. In 1917 rows 49, 56 and 57 are indistinguishable and the 

 group of descendants is in this case very strikingly homogeneous and 

 again intermediate between E and M. To this there is however one 

 exception. The other of the very short forms, row 62, gives an isolated, 

 very short, group immediately distinguishable from the intermediate 

 group itself, and at once suggesting an X group. Continued in 1918 

 the 62 group, of which the three shortest and one longest plants were 

 sown, proved as exj)ected to give an X group, identical with the X forms 

 segregated from other groups. Three extremely long types were also 

 continued in view of the possibility of segregation taking place in that 

 direction also. The result was, as before, negative. 



Thus, this group, with the exception of the one case in which an 

 X group is segregated out, behaves like the 6 group, remaining inter- 

 mediate between E and M. It seems legitimate to assume that the X 

 was produced by splitting in the same way as was found typical in the 

 other groups; and so, to sum up the results, the hybrids between E 

 and M either split into M, an entirely new type shorter and broader 

 than E, and an intermediate type that again splits ; or they do not split 

 at all. The non-segregating type arises twice, independently, and in 

 different years. 



As to the significance of these results, first it is evident that the 

 theory that mutations are due to the loss of a factor or factors cannot 

 be applied here. From the moment the X type is established, the 

 theory of loss of factors is insufficient. No system of recombination of 

 factors, cumulative or otherwise, could explain the appearance of a new 

 type lying outside the limits of the original forms. It is necessary to 

 regard the difference between E and M, not as a case of presence and 

 absence of a factor or factors, but as a case of positive difference. This 

 does not necessitate the assumption that anything has been added. 

 The simplest view seems to be that a factor has been modified in 

 the original pure line, to give the mutation. Why not more than one 

 factor ? Because the fact that, as already indicated, the typical splitting 

 into X, intermediates, and M, gives a ratio of non-ilfs to Ms of 3 : 1, a 

 result confirmed in 1916 when the total number of rows splitting was 

 13, giving a total of 176 non-ilf plants; 63 Ms points to there being 



