296 Mr. Poulton's not*'* in 1886 



fig. 6, nat. size). The larva appears to depend entirely 

 upon tactile stimuli for the direction in which to move 

 its terrifying full-face, and towards which to eject the 

 irritant acid secretion. Visual sensations appear to 

 play no part as guides in the assumption of the de- 

 fensive attitude. The large and full-fed larva which is 

 represented in fig. 6 was only sensitive to tactile im- 

 pressions, and the slightest touch upon either side or 

 upon the back was always followed by a corresponding 

 movement of the anterior part of the larva, and the 

 attitude thus taken up on each occasion was maintained 

 for some considerable time, although instantly altered 

 when another part of the body was touched. 



In the same paper (loc. cit.) I described a complex 

 form of prothoracic "gland" which was everted by 

 applying pressure to the larva of Dicranura furcula. I 

 have now found that a similar " gland " is present in 

 D. rinula; and I wish to correct an error in the 

 description of this structure given in my last paper. 

 The account was written from the memory of an 

 observation made many months before in Switzerland, 

 and it is wrong in describing the everted gland as 

 consisting of six diverging processes. In reality two 

 lateral processes— or everted gland-tubes — are present 

 upon each side, while a median pouch represents the par- 

 tially everted sac, in which the irritant secretion is stored. 

 All are coloured green from the green blood which is 

 forced into them, on eversion. These structures are 

 most easily everted in D. vinida at the beginning of the 

 last larval stage, a time when all the defensive structures 

 appear to be at their highest state of functional activity. 

 At this time the caudal filaments are large and brightly 

 coloured, and are extremely sensitive ; while later in the 

 stage they decline in importance, and generally cease to 

 be capable of eversion. The larva does not seem to be 

 able to evert its prothoracic gland voluntarily, but very 

 slight pressure is at this very sensitive period sufficient 

 to cause eversion. It appears probable that this 

 structure, starting as an ordinary gland, became volun- 

 tarily eversible like the ventral glands of Croesus and the 

 dorsal glands of the larvae of Liparida described in the 

 next section, and that the power of eversion has been very 

 recently lost, as the larva has acquired the remarkable 

 power of ejecting the intensely irritant secretion to a 



