FURTHER NOTES ON METRIOPTERA ROESELII. 39 



capture, was lemon-green. By the time one of them died, six 

 days later, the border had become decidedly yellowish ; in the 

 case of the other male, which lived two weeks longer, practically 

 all trace of green disappeared before death. In all instances 

 the transition from green to yellowish began at the posterior 

 margin of the border, and proceeded from behind forwards. 



As throwing some light upon the direction taken by colour- 

 development during the process of growth, I may mention that 

 the female nymph obtained at Heme Bay on August 28th was 

 very green indeed, and the green colour on the prothoracic lobe 

 was not only present on the border, but invaded a considerable 

 area of the lobe itself ; moreover, the black in the same region 

 was not at all intense or clearly marked off from the green 

 portion. The venter is another region of the body that is subject 

 to considerable variation in respect of colour. In some specimens 

 it is light brown, while in other examples it is golden or light 

 yellow. 



All the living specimens which I have had under observation 

 in captivity were enclosed in a large dry fish- globe kept indoors, 

 and it is probable that, if they had been constantly exposed to 

 the light as in a state of nature, the assumption of the fully adult 

 coloration would have been more rapid and more complete. 

 Again, it is not unlikely that the greenness of so many of the 

 individuals taken in 1912 may have been due to the sunless 

 weather which prevailed towards the close of the summer, and 

 also to the circumstance that, during the same period, grass and 

 other vegetation was kept particularly green by the constant 

 rains. 



Like Mr. South's Essex male of 1911, all the specimens 

 which were kept alive were fed upon fresh grass, from which 

 they ate readily until it became at all dry. In the evening 

 of September 30th it was noticed that the last female then 

 remaining was moribund or even already dead, but it was not 

 removed at that time. The next morning it was unquestionably 

 dead, and the femora of both hind legs had been partially eaten 

 away, no doubt by the male which was still surviving. When 

 at last the female was removed, it was seen that all the tarsi of 

 the fore and mid-legs had been nibbled away. Similarly, the 

 tarsi of both mid-legs had been eaten by other individuals in the 

 case of a male found dead in the fish-globe on September 28th. 

 That these insects do not mutilate themselves in their last 

 moments is shown by the fact that such individuals as ended 

 their lives in solitude suffered no damage of this kind. It is 

 worthy of note, however, that all the specimens which died in 

 captivity, whether kept with others or not, had their antennae 

 more or less broken. I have previously recorded instances of 

 cannibalistic feeding on the part of M. brachyptera, and it now 

 appears that the same habit is shared by M. roeselii also. 



