Conclusion. 505 



are not in fact what they are called, in the strict sense 

 of the term ; they are only more or less incomplete copies 

 of a type already existing. They may be observed just 

 as well in the later as in the earlier generations : in fact 

 better, because there is little prospect of finding them at 

 first, the number of mutants being so small. It is onlv 

 when the latter can be propagated by an unlimited quan- 

 tity of seed that complete series of transitional forms may 

 be expected. 



These transitions are partly atavistic phenomena, 

 partly instances of ordinary variability ; and partly of 

 transgressive variability. An instance of atavism is af- 

 forded by 0. nanella which may be characterized for the 

 whole of its life by unstalked leaves, but which during 

 a brief period of its youth exhibits the stalked leaves of 

 its ancestors (see Fig. 78 on p. 362). This case is per- 

 fectly analogous to the well-known embryonic forms of 

 many other plants. Atavism is, so to speak, accidentally 

 brought about in 0. laevifolia whose leaves, normally 

 smooth, are sometimes crumpled, either singly or over 

 the whole plant when it is an unhealthy one. Most char- 

 acters exhibit transgressive variability more or less ; but 

 there is always a wide gap between the largest 0. nancUa 

 and the smallest flowering O. Lamarckiana. The leaves 

 of O. gigas exhibit a high degree of transgressive varia- 

 bility; they may be broader and narrower than those of 

 the parent species and sometimes, even, come to re- 

 semble those of ruhrinervis and other narrow leaved 

 forms. Further, the size of the flowers varies directly 

 with the vigor of the plants, in all the species. 



If we deal with one character at a time we can make 

 perfectly continuous series embracing O. Lainarckiaua 

 and all the species which have arisen from it. Vor ex- 



