the colours oj certain Lepidoptera. 395 



These pupae were wonderfully uniform and transitional, 

 so that their classidcation was a matter of great diffi- 

 culty. It will be noted that there are only two out of the 

 whole number which were other than intermediate [some 

 form of (3) ] varieties. There was remarkably little gold 

 on the pupae, the classification of the (3)s being entirely 

 dependent on the amount and depth of the pigments 

 present. 



In order to test the results of this experiment still 

 further, a different mode of comparison was adopted. 

 Neglecting the pupae fixed to the glass, all the others in 

 Kows I. and 11. were arranged according to the parts of 

 the body which were exposed to black or gold. They 

 fell into four classes : — 



A. — 3 pupae with head well in gold. 



B. - 6 „ „ black. 



C. — 3 ,, just in gold. 



D. — 2 ,, „ black. 



But there was no tendency for the anterior part of 

 pupae A. and C. to be any lighter than the same part of 

 B. and D. respectively. 



Those fixed on Eow III. were then similarly compared. 

 They fell into two classes : — 



E. — 10 pupae with head in black. 



F. — 1 pupa ,, gold. 



The last was certainly lighter than any of E., but it 

 was light altogether, and not specially about the head 

 or anterior part. 



It is thus clear that the colours did not produce 

 localized effects. The anterior or posterior end of a 

 pupa was often specially light, but this was quite 

 irrespective of the colour of the band against which it 

 had rested. 



These comparisons were carried out with the greatest 

 care, the pui)ae being not only arranged side by side on a 

 sheet of white paper, with the light falling on the same 

 side of each of them (for this plan is always adopted in 

 my comparison of pupae), but the sheet was gradually 

 turned round to permit their equal illumination on all 

 parts of the body. 



Although the results are entirely negative, thus con- 

 firming my earlier experiments in 1886, it is clear that 



2 g2 



