box, and sort them as soon after as possible, as there are many 

 species that will devour the others in a very short time. The 

 larvae of the Cuspidates do not roll themselves into a ring when 

 they fall, and they may be known also by their having from one to 

 four small humps on their bodies, and by their having but little 

 hair; they are, besides, very active when touched. Any that 

 present these characteristics, it will generally be safest to place by 

 themselves, for the greater number of them are almost sure to prove 

 cannibals, which will attack and devour both larvae and pupae of 

 other insects, and in a single day a batch of them will often destroy 

 the results of a whole season's collecting. They do not eat the 

 whole body of any larva they may attack ; but will crawl about the 

 cage, and fasten at the back of the head of any larva they may 

 meet with and keep their hold there until their victim falls dead to 

 the bottom of the cage, when they renew the attack on another. 

 In this way I have seen one of these larvae destroy seven others in 

 the course of about half an hour. At that time I had only one 

 breeding cage ; but now, profiting by experience, I have three 

 made of wood and muslin for the more peaceful kinds, and one 

 made of glass and zinc, perforated at the ends, specially for the 

 accomodation of larvae that will not live amicably with the other 

 kinds. Geometer larvae especially suffer from the attacks of these 

 Cuspidates, as they are very delicate, and a slight grasp will suffice 

 to kill them. The Geometer pupae are delicate likewise, and I 

 believe that they are a favourite prey of the larvae specially under 

 notice. Whenever it is possible, the better plan in collecting is to 

 keep the larvae in separate boxes from the first, and if they are 

 taken up in small boxes, these can be placed open in the breeding- 

 cages, and the larvae will soon crawl out by themselves, and so 

 save the risk of injury liable to be caused by having to handle 

 them. Lepidopterous larvae that have been much handled fre- 

 quently develop crippled or otherwise imperfect imagos, with the 

 wings curled up like dried leaves, so that no one could set them 

 in a satisfactory manner. In searching for the larvae of the 

 Cuspidates search the trees carefully when the leaves are falling, 

 looking out especially for small clusters of withered leaves. On 



