VARIABLE PROTECTIVE RESEMBLANCE IN INSECTS 119 



gestions which are not tested because of their apparent 

 probability. It would have been very easy to transfer 

 a freshly formed pupa from one colour to another 

 which is known to produce an opposite effect upon it ; 

 and yet if this simple experiment had been made the 

 theory would have collapsed, for the pupa would have 

 been found to resemble the first colour and not the 

 second. Furthermore, Mr. Wood's suggestion raised 

 the difficulty that chrysalides which had become ex- 

 posed in the course of a dark night would have no 

 opportunity of resembling the surrounding surfaces, 

 for the pupal colours deepen very quickly into their 

 permanent condition. In working at the subject I 

 determined to pay especial attention to such ques- 

 tions. 



Experiments upon the chrysalis of the Peacock 

 Butterfly 



I began work with the common Peacock Butterfly 

 (Vanessa lo), of which the chrysalis appears in two 

 forms, being commonly dark grey (see fig. 28), but 

 more rarely, bright yellowish-green (see fig. 29) : both 

 forms are gilded, especially the latter. The gilding 

 cannot be represented in the woodcuts. Only six 

 caterpillars could be obtained, and these were placed 

 in glass cylinders surrounded by yellowish-green 

 tissue paper. Five of them became chrysalides of the 

 corresponding colour; the sixth was removed imme- 

 diately after the caterpillar skin had been thrown 



