( cviii ) 



and there — sure enough — were several minute larvae, hanging 

 by long silken threads from the overspreading branches of 

 an " Inga-saman " tree (Fithecolobium saman). They had let 

 themselves down from a height of 30 or 40 feet, and were 

 swinging in the breeze. This part of the mystery was now 

 solved. They hung suspended until the wind drove them 

 against something solid, and there they immediately attached 

 themselves and constructed their cocoons. 



The caterpillar is small, but robust, measuring — when ex- 

 tended — about 6 mm. It is of a dull, pale, translucent 

 green colour, with an irregular crimson band completely 

 encircling each segment. 



The construction of the cocoon is commenced immediately 

 the caterpillar obtains a foothold. The position appears to be 



Cocoon of Epicephala chcdybacnia, A single globule and silken 



dorsal view ; x 3. cord ; x 25. 



a matter of no consequence. The work is completed within 

 two hours, which accounts for my failure to find uncovered 

 larvae. After weaving a thin silken covering, the creature 

 rests for a few moments, and a convulsive movement of the 

 posterior segments is noticeable. Yery soon a globular pellet 

 — apparently composed of dried bubbles — is voided whole. 

 The caterpillar then turns round inside the cocoon, rapidly 

 attaches the globule to the roof of the cocoon by a stout 

 silken cord, bites a small hole close to the point of attachment, 

 and pushes the globule and cord up through this aperture. 

 The rent is then quickly repaired. This is followed by another 

 short pause, the evacuation of a second pellet, and a repetition 

 of the pi-evious performance, the second pellet being placed at 

 the opposite extremity of the cocoon, in consequence of the 

 caterpillar having reversed its position in the cocoon. The 



