60 NATURAL SELECTION FROM 



ing new has entered into any combination. But 

 when the Aster boundary is crossed, and still 

 more when the boundaries of Compositse and 

 then of Angiosperms are crossed, absolutely new 

 characters are met on every hand ; and still the 

 phylogenetic connections seem convincing. For 

 example, perhaps no plant morphologist doubts in 

 these days that at least some of the Gymnosperms 

 are phylogenetically related to ancient ferns. The 

 distinguishing mark between the two groups is 

 the absence of seeds in the one and their presence 

 in the other. The seed and all that goes with it 

 is a new character, and how selection could have 

 originated it, is a question at whose answer even 

 scientific imagination balks. It is evident that 

 the ovules of Gymnosperms are related by de- 

 scent to the sporangia of ferns in some way, but 

 so extensive a change does not seem to come 

 within the possibilities of Natural Selection. We 

 have relatively primitive ovules, but they are 

 enormously different from fern sporangia; and 

 we can imagine how selection may have trans- 

 formed these ovules into those of more advanced 

 type, for this is only manipulating a structure 

 already in existence and is adding nothing new. 

 The leaves of sporophytes, the vascular system, 

 the root system are further illustrations of the 

 same kind. Absolutely unrepresented in the 

 lower groups, they are new, complex, and fully 

 functioning structures when we meet them first. 

 Of course lost records and an inconceivable lapse 



