460 Mr. Poulton's///r^/ier experiments upon 



The effect of the screens upon the backgrounds was 

 easily determined by comparing the effect upon the 

 latter when the former was interposed in the path of the 

 light on its way to the prism, with that of its with- 

 drawal. 



The reflecting power of the backgrounds having been 

 thus determined, a few days later the whole process was 

 repeated, and the second set of observations compared 

 with the first. In most cases they agreed : when they 

 did not, we made a third observation. These determi- 

 nations were made in the laboratory of Balliol College, 

 Oxford. 



I propose to consider the backgrounds in the following 

 order : — (1) Dark surfaces, such as black and brown, 

 reflecting very faintly, but from every part or many parts 

 of the spectrum ; (2) coloured surfaces, chiefly reflecting 

 particular rays : these will be considered in the order of 

 the spectrum from red to blue ; (3) white or bright 

 surfaces (white or metallic), reflecting strongly from the 

 whole of the spectrum. 



{See Table, pages 4G1, 462, 463, 464.) 



These results all combine to prove the validity of the 

 suggestion made in my earlier paper, that rays from the 

 yellow and orange part of the spectrum are effective in 

 dismissing pigment, and favouring green (or bright) 

 larvae and pupre. It seems tolerably certain that it is 

 the yellow and orange rays which, reflected from leaves 

 and shoots, stimulate the larvae and ]Hip8e to become 

 green. It is shown above that if a red background be 

 offered, pupae become dark ; but if an orange surface be 

 substituted, only differing in reflecting an additional 

 narrow strip of the spectrum, but in that strip including 

 orange and yellow rays, both larvae and pupae are 

 strongly influenced in the direction of green, although 

 there is hardly any green in the light which reaches 

 them. I attach less weight to the evidence from yellow 

 backgrounds, because they reflect so much of the spec- 

 trum. But the evidence from the green backgrounds is 

 the strongest of all. If the above argument holds good, 

 artificial greens which are strong in the yellow and 

 orange ought to act like leaves and shoots, while those 

 which are weak in this part, although as greens they 

 may be extremely bright, ought to produce dark larvae 



