21-4 REVIEWS. 



— not to infuse the character of the male parent, for he fer- 

 tilized the progeny with the pollen of the female parent, and 

 thus early bred out the other blood, but to induce variation, 

 which, once initiated in the internal disorder consequent upon 

 the crossing-, was apt to proceed, or might be led on by selec- 

 tion, to great lengths, according to Vilmorin. The variations 

 in question, being mainly such as are sought in floriculture, 

 may not have passed the line laid down by Naudin, or actually 

 have introduced new features. But such plants would surely 

 have no exemption from the ordinary liability to variation. 

 If other plants vary, in the sense of producing something new, 

 so may these. 



This brings us to another inference which Naudin draws. 

 Having observed that his hybrids in their manifold variation 

 exhibited nothing which was not derivable from their im- 

 mediate ancestry, he directly (and in our opinion too confi- 

 dently) concludes that all variation is atavism, — that when 

 real variations are set up in ordinary species, this is not an 

 origination but a reversion, a breaking out of some old ances- 

 tral character, a particular and long deferred instance of this 

 variation desordoti?iee, which would thus appear to be the 

 only kind of variation. This view has been presented before, 

 but not, perhaps, so broadly. Adducing some theoretical 

 considerations in its favor — to which we may revert — and 

 some sound reasons against the view that variation is caused 

 by external influences, he declares it " infinitely more prob- 

 able that variation of species properly so called is due to 

 ancestral influences rather than to accidental actions." We 

 might think so if these two categories were exhaustive, and 

 external conditions must be supposed to act immediately, as 

 the cause rather than the occasion of variation. But the sup- 

 position that " accidental actions," whatever they may be, and 

 external influences of every sort do not produce but educe and 

 conduct variation — which is our idea of what natural selec- 

 tion means — avoids the force of Naudin's arguments. 



Moreover, Naudin's view, regarded as an hypothesis for 

 explaining variation, leaves the problem just where it finds it. 

 To explain the occurrence of present and actual variations, 



