KJ^'OW LEDGE OF THE MUTATING OENOTHEEAS. 29 



in tlie amount of anthocyanin, particularly in the sepals, hypantliium and rosette-leaves. 

 Photographs of O. rubrinervis and 0. ruhy^icalyx can, however, he used interchangeahly, 

 for the morphological characters are exactly the same in both. PI. 2. fig. 25 shows a 

 pan of seedlings of 0. ruhricalyx and O. rubrinervis, in which it will be seen that the 

 leaves are narrower, more pointed and less crinkled than in O. Lamarcldana (PL 1. 

 fig. 1). Fig. 26 represents an older rosette, in which the three types of rosette-leaves 

 [of. the rosettes of O. Lamar chiana and O. Imvifolia) are present, though type 2 is 

 inconspicuous in 0. rubrinervis, differing from type 3 only in the more even margin, 

 more rounded tip and less crinkling, but usually not having a broad blade sliarply 

 narrowed to the petiole. The rosette in PI. 2. fig. 26 shows three leaves of type 1, and 

 three of type 2. 



The)*e is considerable variation in the mature rosette of 0. rubrinerms in width of leaf, 

 amount of red developed in the midribs, amount pf crinkling and pubescence, etc., so 

 that certain individuals transgress the boundary line separating them from 0. Lamarck- 

 iana in each of these characters or in several of them together. But the occasional 

 indeterminable rosettes which occur nearly always show decisive characters when the 

 plants come into bloom. The leaves of the mature rosettes also vary in the width of 

 the leaf-tip, and PL 2. fig. 27 shows an individual with exceptionally broad tips. The 

 transgressive variability between 0. Lamarchiana and O. rubrinervis becomes more 

 evident under varying or unfavourable conditions of growth. Thus in 1911, in a culture 

 of O. rubrinervisx 0. Lamar cJciana, P^ 64 rosettes developed, 12 of w^hich were classed 

 as O. Lamarchiana and 52 as O. rubrinervis. These were transplanted too late in the 

 season to become annuals, so they continued to produce rosette-leaves. The leaves 

 appearing towards the end of the season (September 27) on the 0. Lamarchiana rosettes 

 were much narrower and lighter green than normal, so that they were practically indis- 

 tinguishable from 0. rubrinervis leaves. It is evident that local conditions of nutrition 

 have much to do with the fluctuating variability of the Oenotheras, often producing a 

 surprising range of variation. But this is clearly distinct from the germinal variations, 



which are inherited in each generation. 



Comparison of large numbers of rosettes of 0. rubrinervis and 0. Lamarchiana 

 shows the impossibility of discriminating, merely by inspection, between fluctuating and 

 germinal variations. Plants and animals frequently breed true to remarkably small 

 differences, as many recent breeding experiments, particularly the work of Johannsen 

 (1909) with " pure lines " in beans and of Jennings (1911) with Faramecium, show. The 

 only way to determine whether the offspring of a variant individual will regress towards 

 the mean of the race, or will reproduce its peculiarity in all or a portion of its offspring, 

 is by growing its progeny and determining their characters. 



Another example of the manner in which a plant may immediately react to a change 



of environment by prod 



furnished bv a large plant of 0. rubri 



nervis which was removed from the garden to a greenhouse in September, 1907, while 

 in flower. It continued to bloom, hut the weight of the inflorescence soon bent the tall 

 stem into a nearly horizontal position. Growth in this position continued and a series 



