Breeding experiments which show that hybridisation and mutation etc. 26^ 



Table VIII (contd.) 



(G X R) xG=^ 

 (R xG) xG = 



(R XG) X R = 



(G X R) X (R xG) = 

 (RXG) x(G X R) = 



A fairly uniform type, intermediate between grandiflora x 

 riibricalyx and pure grandiflora, both in foliage and in 

 time of blooming. Cults. 73, 74, 78. 



Much more variable, both in foliage and time of blooming; 

 but taken as a population, distinctly intermediate between 

 riibricalyx x grandiflora, and pure grandiflora. Cults. 67, 

 68, 69, 71. 



Foliage for the most part intermediate between rubricalyx x 

 grandiflora and rubricalyx. The same is probably true of 

 the time of blooming, which is not, however, recorded. 

 Cult. 70. 



Foliage characters very variable though nearer grandiflora; 

 time of blooming intermediate between (G x R) X R and 

 (G X R) Fo, i. e., late. Cults. 75, 76, 77. (See table VI.) 



Foliage characters variable, but much nearer rubricalyx than 

 grandiflora. Time of blooming early, about same as 

 (R X G) X R. Cult. 72. 



A comparison of tables VII and VIII shows that the results of 

 the two series of crosses agree more or less in Fi, though the bienhis- 

 muricata crosses are more strongly patroclinous. Wlien the F^ families 

 of the grandiflora-ruhricaly X crosses are crossed back with the parents 

 however, a distinct difference appears. For while, e. g., (B x M ) x B 

 is found to give pure biennis, (G x R ) x G, produced a type which 

 is again intermediate between grandiflora x rubricalyx and pure grandi- 

 flora both as regards the foliage characters and the time of blooming. 

 There is decidedly not segregation in the population as a whole, but 

 blending and fractionating of characters. A new blended type is pro- 

 duced, which did not exist previous to the cross (G x R) x G. The 

 same is true of (Rx G)x R and (Rx G) x G, though these furnish 

 some evidence of variability or segregation in foliage characters. Cer- 

 tainly fR >. G) X R does not give pure R but a new type which is 

 for the most part intermediate between i? x G and R. Also (R x G ) 

 X G gives not R x G but a type midway between R x G and G. Again, 

 the double reciprocal crosses in the grandiflora-rubri calyx series show 

 no tendency to produce in one case only pure grandiflora and in the 

 other rubricalyx, but there is rather a widely variable series of forms 

 just as in the F2 of the original crosses. 



Goldschmidt (1912) has obtained some evidence indicating that 

 the bicnnis-muricala crosses may exhibit merogony, the male nucleus in 

 each case entering the cytoplasm of the egg cell and developing an 

 embryo while the egg nucleus degenerates. If this result is confirmed 

 t will explain why the double-reciprocal crosses revert to the "outside" 



