i.] SUDDEN VARIATION 67 



that of teeth: it may be for the same sort of reason that 

 white cats with blue eyes are deaf. In these different 

 examples the " correlative' ' changes are only solidary 

 changes (not to mention the fact that they are really 

 lesions, namely, diminutions or suppressions, and not 

 additions, which makes a great difference). But when we 

 speak of "correlative" changes occurring suddenly in 

 the different parts of the eye, we use the word in an entirely 

 new sense: this time there is a whole set of changes not 

 only simultaneous, not only bound together by community 

 of origin, but so coordinated that the organ keeps on per- 

 forming the same simple function, and even performs it 

 better. That a change in the germ, which influences the 

 formation of the retina, may affect at the same time also 

 the formation of the cornea, the iris, the lens, the visual 

 centres, etc., I admit, if necessary, although they are forma- 

 tions that differ much more from one another in their 

 original nature than do probably hair and teeth. But 

 that all these simultaneous changes should occur in such 

 a way as to improve or even merely maintain vision, this 

 is what, in the hypothesis of sudden variation, I cannot 

 admit, unless a mysterious principle is to come in, whose 

 duty it is to watch over the interest of the function. But 

 this would be to give up the idea of "accidental" variation. 

 In reality, these two senses of the word "correlation" are 

 often interchanged in the mind of the biologist, just like 

 the two senses of the word "adaptation." And the con- 

 fusion is almost legitimate in botany, that science in which 

 the theory of the formation of species by sudden variation 

 rests on the firmest experimental basis. In vegetables, 

 function is far less narrowly bound to form than in animals. 

 Even profound morphological differences, such as a change 

 in the form of leaves, have no appreciable influence on 

 the exercise of function, and so do not require a whole 



