130 THE COLOUES OF ANIMALS 



satisfactory results in spite of the slight susceptibilil^y 

 of stage iii. was to employ large numbers of larvse, 

 and to pay careful attention to minute differences of 

 pupal colour as well as to the time during which the 

 conflicting colours had been applied. 



The experiments were conducted in two ways. 

 In the first the larvae were induced to suspend them- 

 selves from sheets of clear glass, by placing them in 

 wide shallow glass boxes so that the ascent to the 

 glass roof was easily accomplished. As soon as sus- 

 pension (stage iii.) had taken place, each larva was 

 covered with a cardboard tube divided into two 

 chambers by a horizontal partition which was fixed 

 rather below the middle. There was a central hole 

 in the partition just large enough to admit the body 

 of the larva. The tube was fixed to the glass sheet 

 with glue; the upper chamber was lined with one 

 colour, e.g. gilt, and the lower chamber with the 

 opposite colour, e.g. black, with which the outside of 

 the cylinder was also covered, in case the larva should 

 stretch its head beyond the lower edge. The parti- 

 tion was fixed at such a height that the larval head 

 and rather less than half of the total surface of skin 

 were contained in the lower chamber, while rather 

 more than half of the skin surface was contained in 

 the upper chamber. The arrangement is shown in 

 section in fig. 30. 



The second method of conducting the conflicting 

 colour experiments was superior in the more equal illu- 



