KNOWLEDGE OF THE MUTATJXG OEjS'OTHEEAS. 27 



and that, on tlie wholei tlie former species, such as 0. bieimis and 0. muricata, are more 

 successful and widespread than the latter, such as 0. grandiflora and 0. Jlooheriy does 

 not favour the view that occasional crossing is a necessity or even an advanta 

 this genus. 



Schouten (1908) in several cultures of 0. laevifolia has found that it yielded various 

 mutant forms, including 0. laemfolia salicifolia, 0. scintillans, O. laevifolia nanella, and 



o 



O. rubrinervis. 



0. hi'evistylis 



This form also was discovered by DeYries in Hilversum, hut has never appeared in 

 cultures of 0. Lamarckiana. Its almost completely sterile ovaries render its propagation 

 possible only by using its pollen to cross with another form, so that it must clearly have 

 originated as a mutant, very probably in the locality where it w^as found. The charac- 

 ters of this form have not been accurately shown in previous accounts. PI. 2. fig. 21 

 shows how distinct are the young seedlings from 0. Lamarcldana (PL 1. fig, 1), 0. rubri- 

 nervis (PI. 2. fig. 25), or O. laemfolia (fig. 15), the blades being characteristically broader 

 than in any except 0. Lamarcldana, from which the leaves differ in having longer 



petioles and (at this time) very little crinkling. PI. 2. fig. 22 is a characteristic mature 

 rosette, showing some leaves broad-pointed, but many having obtuse, blunt tips such as 

 never occur in. the mature rosette of O. Lamarcldana (cf. PI. 1. fig. 5). There is always 

 this wide range of variation in the tips of the leaves of each rosette. This is shown in 

 PL 2. fig. 23, with w^hich PL 1. fig. 6 of O. Lamarcldana should be compared. It will 

 be seen that the ranges of variability overlap, though O. Lamarcldana never bears a 

 leaf with tip as blunt as the broadest O. brevistylis, and O. hrevistylis never bears a leaf 

 with tip as narrow as the narrowest 0. Lamarcldana. Hence, again as in 0. laevifolia, 

 though their ranges of variation overlap there is no doubt whatever that these two forms 

 are perfectly distinct. If there were any doubt regarding the distinctness of the rosette 

 stage it could be set at rest when the plants mature, for not only has O. hrevistylis 



sterile ovaries but its style is extremely short so that the stigma barely projects above 

 the calyx-tube, and the free tips of the sepals are also very short, as shown by PI. 2. 

 fig. 24. In addition, the terminal rosettes of the branches are more compact than in 

 O. Lamarcldana, the bracts being shorter, with broader and less pointed tips. I found 

 no difficulty in identifying the rosettes of 0. bf^evisfylis when mixed with O. Lamarck- 

 iana and was able to confirm my determination in each case when the plants bloomed, 

 by opening a bud and discovering the very short style. It would be impossible to obtain 

 more conclusive proof that a range of variation which is apparently continuous is in 

 reality not so, for in the ontogenyjDf one series of individuals we pass from the rosette 

 sta^-e which yields a continuous series of leaf-shapes and in which there is some difficulty 

 in recognizing the two types, to the flowering stage in which all the flowers of each 

 individual are recognizable at a glance as belonging to one type or the other. This race 

 also furnishes a striking instance of what occurs in all these Oenothera forms, namely 

 the " holding together " or correlated variability of characters affecting all parts of the 



plant, the rosette and stem-leaves as well as the flowers and fruits. One must adopt the 



E 2 



