402 ShuHes in the Theory of Descent. 



doptera, but is found in Diptera, and especially in 

 Hymenoptera in every degree. The larvae of 

 ichneumons which live in other insects, require 

 (not always, but in most instances) that the female 

 imago should possess a sharp ovipositor, so that 

 in this case also the structure and mode of life of 

 the larva influences the perfect insect. This does 

 not depend, however, on inherent laws of growth 

 (correlation), but on the action of external 

 influences, to which the organism endeavours to 

 adapt itself by natural selection. 



I will now let the facts speak for themselves. 



It is shown by those species in which only one 

 stage is di- or polymorphic that not every change 

 in the one stage entails a corresponding change 

 in the other. Thus, in all seasonally dimorphic 

 species we find that the caterpillars of butterflies 

 which are often widely different in the colour 

 and marking of their successive generations are 

 absolutely identical. On the other hand, many 

 species can be adduced of which the larvae are 

 dimorphic whilst the imagines occur only in one 

 form (compare the first and second essays in this 

 volume). 



There are however facts which directly prove 

 that any one stage can change independently of 

 the others ; I refer to the circumstance that any 

 one stage may become independently variable 

 that the property of greater variability or of 

 greater constancy by no means always occurs in an 



