K^^OWLEDGE OP THE MUTATIN& OENOTHEEAS. 39 



the most striking case I have observed, of change in leaf- characters during the ontogeny 

 of a rosette. PI. 4. fig. 47 shows the adult rosette of the same type at the age of nine 

 months, the rosettes having been kept constantly under tropical conditions, so that they 

 developed slowly and displayed their characters to the full. In fig. 47 the characteristic 

 type of leaf with pinnatifid basal lobes continues to be produced, and the first type has 

 disappeared. In the 0. grandiflora type of rosette ontogeny there are, then, three 

 leaf- types as in O. LamarcJciana and some of its derivatives, but only types (2) and (3j 

 are represented in figs. 45 and 46. Type (2) in 0. grcmdiflora is represented 

 number of leaves, while in O. Lamarckiana and its relatives types (1) and (2 

 only represented (at least in out-of-door cultures) by three leaves each. The numbei 

 of leaves of each type is probably always variable, depending upon the conditions of 

 development. 



When O. grandijiora is grown under ordinary cultural conditions — i. e., when the seeds 

 are sown in January and the young plants are planted out in May or June — the adult 

 type of rosette-leaf with pinnatifid lobes is usually omitted entirely ; and frequently, as I 

 shall show in other 0. grandijiora races, there is no rosette at all, the young plant 

 forming a stem with internodes almost at the very beginning. This interesting 

 difference from 0, Lamarckiana in physiological behaviour results in the plants 

 developing more rapidly, and frequently blooming earlier. PI. 4. fig. 48 shows such a 



I 



plant of typical O. grandiflora in my cultures from Alabama. The absence of a rosette 

 gives the plant a shrub-like appearance. Such plants were top-heavy, and it was 

 necessary to have them staked in order to prevent their being broken down by the wind. 

 ^i%' 48 also shows the ovate-lanceolate shape of the stem-leaves in 0. grandiflora. They 

 taper at both ends, the lower end narrowing to a petiole, and (except the uppermost 

 bra(its) are not sessile with aurate base, as is usual in 0. Lamarckiana. Such buds are 

 shown, reduced in size, in PL 4. fig. 49. Unlike those of 0. Lamarckiana and all its 

 derivatives, the buds are free from long, muricated hairs, though they usually bear soft, 

 pubescent hairs, but are sometimes quite glabrous. Fig. 49 also shows the range of red 

 colour pattern in these buds. 



0. biennis^ Linn. 



It is w^ell-known that a large number of North American geographical races of 

 O, hiemiis exist. New species, such as O. strigosa and 0. canovirens, are published as 

 segregates from time to time, but many others exist, and my experience iu growing 

 numerous races from wild seeds collected in various parts of the continent has beea 

 that usually each new locality yields a race which is recognizable as distinct by its 

 leaf-characters and habit. One of the most interesting questions in connection with the 

 evolution of the Oenotheras is. How have these races arisen and what relation do tliey 

 bear to the evolution of the group ? The fact that these forms are aU self -pollinating 

 adds to their interest and makes the question of their manner of origin and evolution 

 somewhat different from that of the larger-flowered open-pollinated forms. Por it is 

 clear that in the latter intercrossing of races must frequently take place, so that the 



