Breeding experiments which show that hybridisation and mutation etc. 2^5 



had a predominating influence on the offspring in this culture. In 

 other families the proportion of be plants was usually somewhat larger. 

 In the whole series of F2 families scarcely one individual exactly cor- 

 responded with ruhricalyx in foliage, and very few agreed wholly with 

 grandiflora. 



In such cases the conception of unit-characters breaks down utterly, 

 and any possible system of sjTnbols fails to represent the facts. The 

 leaves on the same stem may pass 

 from smooth to crinkled and back to 

 smooth, also from broad to narrow 

 or narrow to broad. No real classi- 

 fication of the characters is possible, 

 and it becomes absurd to suppose 

 that "factors" exist for such charac- 

 ters as broad leaf-base or amount of 

 crinkling. 



The variation in foliage is much 

 too great to illustrate fully by photo- 

 graphs, but figs. 17 and 18, taken 

 July 23"' represent two plants (Nos. 

 48 II. 7 and 49 I. i) from the F., 

 of grandiflora x ruhricalyx, while fig. 

 19 shows a third plant (No. 49. VII. 

 19) later in the season (Sept. 26). The 

 foliage of these three plants was 

 recorded on Aug. 8''', as respectively 



b c 



- — b • -, p — b.s, and p.s ; that is, in 



the first case the leaves varied from 



half as broad to nearly as broad r.t 



base as ruhricalyx, while the crinkling 



was about one-quarter as great as in Fig. 19. 



ruhricalyx (cf. fig. 17); in the second 



case, the foliage varied from pointed 



at base in the lower stem-leaves to broad in the upper ones, nearly 



free from crinkling (cf. fig. 18). The third plant, which was recorded 



as ps from its earlier stem-leaves, is obviously (cf. fig. 19) producing 



much-crinkled bracts at the end of the season. 



Fig. 20 shows a plant (No. 57. I. 2) from the F2 of ruhricalyx x 

 grandiflora. It was recorded as nearly ps, but there are obvious signs 



Grundiflnra X yuhricalyx, ¥.,■ 

 Foliage p. s. 



