430 Studies in the Theory of Descent. 



The answer to these questions is not far to find. 

 The wider two stages of a species differ in struc- 

 ture, the less does correlation become possible ; 

 the nearer the two stages are morphologically 

 related, the more powerful does the action of 

 correlation become. It is readily conceivable that 

 the more widely two succeeding stages deviate in 

 structure and mode of life, the less possible does 

 it become for characters to be transferred from 

 one to the other. How is it possible, for example, 

 that a new character in the proboscis or on the 

 wings of a butterfly can be transferred to the 

 caterpillar? If such correlation existed it could 

 only manifest itself by some other part of the 

 caterpillar changing in correspondence with the 

 change of the proboscis or wings of the butterfly. 

 That this is not the case has, in my opinion, been 

 conclusively shown by all the foregoing considera- 

 tions respecting the independent variability of the 

 chief stages of the metamorphosis. 



There are, moreover, an endless number of 

 facts which prove the independence of the indi- 

 vidual stages of development I refer to the mul- 

 titudinous phenomena presented by metamorpho- 

 sis itself. The existence of that form of develop- 

 ment which we designate as metamorphosis is 

 alone sufficient to prove incontestibly that the 

 single stages are able to change independently 

 of one another to a most remarkable extent. 



If we now ask the question : how has- the so- 



