THE RELATION BETWEEN LIGHT AND PIGMENT-FOEMATION. 549 



weed. The whole stood in strong' diffused light and had a 

 double circulation. 



These experiments give a ]nuch more definite result. In 

 green weed surroundings three of the fish lost in a week 

 tlieir initial gi-eenness, and, together with the remainder/ 

 became entirely brown. Not only so ; out of four, three 

 showed considerable amounts of red pigment, and, as we shall 

 see, contrasted very markedly with the other experimental 

 batches (fig. 3). 



The brown weed experiments gave a curious result. The 

 fish were initially green and so remained, but in the red weed, 

 out of five specimens, two of which had been originally 

 greenish, three were now green and the remaining* two 

 showed only tinges of brown, although at first they had 

 been barred with that colour. 



In this experiment, therefore, green and red weed acted 

 quite differently from each other ; the green light-filter 

 encouraged the brown colour and i-ed pigment, whereas 

 red encouraged green colour and yellow pigment. Brown 

 surroundings resembled red ones in maintaining the green 

 tint. The contrast between this result with transmitted light 

 and the former with reflected light is so striking and puzzling 

 that I, at this point, undertook the experiments (referred to 

 on pp. 552-561) on Hippolyte with a view to clearing up 

 the discrepancy. 



The explanation, I believe, is to be sought in the spectro- 

 scopic analysis of the light transmitted or reflected by thin 

 and by thick masses of the respective weeds. Thus Ulva, 

 two fronds in thickness, transmits red to green, but as the 

 thickness is increased it transmits orange, yellow, and green 

 onl}'. Single fronds of NitophyUum transmit red to 

 green, but several transmit almost pure red with a trace of 

 green. Brown weeds transmit from red to green, the general 

 intensity being low. Taking, therefore, the red as the pui-cst 

 screen, the remarkable feature about it is that along with the 

 greater purity and intensity of the red light there is in the 

 fish submitted to its action a green result due to the combina- 



V()L. 55, PART 3. NEW SBRIKS. 3(3 



