VARIATION AS LIMITED BY THE ASSOCIATION OF CHARACTERS. 597 



that, in the subsequent generations, all sorts of intermediate forms 

 crop up equipoised between the hybrid type and either parent. For 

 instance, if the hybrid type inclined towards the male in colour of ' 

 flower and towards the female in another respect, I find that some 

 individuals in subsequent generations will do just the opposite, as 

 though the law of change indicated a course of variation which would 

 in time fill up every gap between the two extreme forms represented 

 by the species originally crossed.* 



But to this consummation there is the evident bar of the association 

 of different f characters. We constantly find that certain pairs of 

 characters ca-nnot be dissociated from each other, but continually 

 occur together in^ individuals. This association of certain characteristics 

 (so long as it obtains) appears to rule out the possibility of the 

 occurrence of certain conceptually possible intermediate forms. The 

 Antirrhinums give us one instance of this, for among the dwarf 

 self-coloured forms every rogue as to height is also a colour-rogue, 

 whereas those that are typical in stature will probably not produce 

 1 per cent, of colour-rogues. 



If we define a species as " A group of closely related individuals 

 as between whom procreative variation has suffered a temporary and 

 partial arrest," we mean by this definition that we reserve the use of 

 the term " species " to plants which reproduce themselves true from 

 seed with the number of variants limited to (say) 3 per cent. But 

 this, again, may mean no more than that the association of characters 

 has been so extended as to include all the noticeable characteristics 

 of the type. In this sense variation becomes a question of permuta- 

 tions and combinations, and in every " good " species all the characters 

 hang together so that variation is brought to a standstill for the time 

 being, or reduced within very narrow limits. 



By hybridization this fixity is broken up to some extent, or, in a 

 few cases, a new set of associated characters supersede the old associa- 

 tions. Now if the hybrid is crossed with another hybrid, greater 

 variation becomes evident, until in florists' plants we note an ever- 

 growing inconstancy. t All this hypothetical fixity in " good " species 

 is, however, relative and not absolute. There exists a certain per- 

 centage of variants from the type, and I contend that we cannot fore- 

 tell with certainty the trend of such variation. For as the germ-plasm 



* As an example, I hybridized Hippeastrum vittatum with H. •equestris and 

 produced the hybrid H. x Mandevillei, which showed some inclination towards 

 the female. There are eight divergent characters as between the species inter se, 

 which I designate abcdefgh, orABCDEFGH, using small letters if 

 the characters of the second generation resemble those of the female species, 

 and capital letters if resembling the male species. Now the Mendelian assertion 

 of recessive characters would produce variants as they are here written (above) ; 



but I have found none such. The hybrid type produced a b E F g h and 



0 H 



the variants of the second generation such formulae as A b ^ d e f g - ' &c. 



t The word " diverse " might here be understood in the sense in which it has. 

 been used above. So I have not used this word here. 



+ See my Concepts of Monism, "On Species," p. 335, &c. 



