markings and attitudes of lepidopterous larv(e. 291 



larva eats part of another empty shell lying close to the 

 one from which it has just emerged. In one case I saw 

 a newly-hatched larva attack a thread of the gauze to 

 which the eggs were attached, but it soon left this and 

 ate its shell. The process of gnawing through the shell 

 can be watched with a lens. Very slow progress is 

 made as long as the larva is completely within, and is 

 biting at the concave surface of the shell. When a 

 small hole has been made one mandible is thrust out- 

 side, and the shell is very quickly eaten away from the 

 edge of the aperture, until the larva can emerge. The 

 young larva is of a beautiful yellowish-green colour 

 upon the body, while the horn is red. (The length is 

 5'5 mm. when extended in walking ; the horn 2 mm. 

 long, and thus much shorter than that of *S'. Ugustri). 

 The horn is bifid, like that of S. Ugustri (but to a less 

 extent), and terminates in two bristles, while the rest of 

 its surface and that of the body is densely covered with 

 short whitish hairs. There were never any indications 

 that the horn could be moved. The head is thinly 

 covered with similar hairs. The larva spins a web in 

 the same manner as S. Ugustri. (It should be men- 

 tioned that these webs are very slight, and only detected 

 by careful watching. This may explain the failure of 

 other observers in finding them). The ocelli are black 

 and very distinct on the light green head. The oblique 

 stripes and subdorsal can be just made out with difficulty 

 in a newly-hatched larva, but there is no darkening of 

 the ground colour in front of the stripes, although this 

 appears at the end of the stage. The usual attitude on 

 the leaf was as in S. Ugustri, and with the same pro- 

 tective significance. At the end of the stage the regular 

 markings are very distinct. Even at this time the sub- 

 dorsal is more distinct anteriorly than in the rest of its 

 extent. I was very much astonished to find a very 

 small proportion of the larvre with the head of the typical 

 SmerintJuis shape, while the others possessed the more 

 generalised round shape. I could at first hardly believe 

 the accuracy of the observation, but it was subsequently 

 confirmed, and there can be no doubt that this is an 

 instance of the passage backwards of a character in the 

 ontogeny actually taking place before us. The shagreen 

 dots are very numerous, and each hair springs from the 

 summit of a dot. The markings are caused by the dots 

 becoming arranged in a linear series, without the ground 



