Breeding experiments which show that liybridisation and mutation etc. 2^1 



and ruhricalyx. These Fx rosettes were, altogether, extremely like 

 the reciprocal already described, and such differences as existed may 

 perhaps be chiefly attributed to slight differences in the parent races 

 of grandi flora. The larger size of the rosettes in grandiflora x ruhri- 

 calyx can, however, scarcely be attributed to this cause. The rosette- 

 leaves in the latter cross were frequently (fig. 12) more distinctly 

 repand-denticulate than in the reciprocal (fig. 13) or in ruhricalyx 



Fig. 13. Fj rosette, riihyicalyx X grandiflora. 



(fig. 8). The slight differences in the parental grandiflora races preclude 

 our concluding that precisely reciprocal crosses would not give iden- 

 tical morphological results, though in rate of development each cross 

 is decidedly nearer the male parent. Both these crosses showed the 

 same range of types in the Fg. 



As regards the factor R in the Fj of ruhricalyx x grandiflora, of 

 the plants (58) which bloomed, there were 30 i?'s and 28 r's, in almost 



