THE PUPA OF LIBYTHEA CELTIS. 285 



nose-spine. The front of the pupa of L. celtis is very much of the type of 

 those of Pararf/e egeria and Avf/ynnis arjlaia; a transverse, rather sharp 

 ridge runs between the eyes, and just outside the glazed eye this ends in 

 rather a sharp point, whilst there is a waved rise interior to this, with a 

 depression in the middle line, so that there exists clearly the outer double 

 spine that is so marked in the Vanessids, &c., whilst the inner pair are 

 also indicated. At the angulated spine in front of the eye, the ridge of 

 the nose-spine is continuous across the eyes and antenna with the 

 wing ridge that continues to the spine on the 2nd abdominal dorsum. 

 The whole line of this ridge, therefore, forms a complete circle round 

 the pupa, in some it is a little waved, in others it is so straight that it 

 all lies in one plane. This ridge is the most characteristic item in the 

 pupal facies. Another marked peculiarity is the curving forwards of the 

 abdominal segments, in some little degree by attitude, but chiefly by 

 the narrowness of the fronts of abdominal segments 5, 6 and 7, so 

 that, in some cases, possibly assisted by a little contraction by drying 

 (though the pup« are alive), the suspensory button on the anal seg- 

 ments almost touches the ends of the appendages on the 4th abdominal 

 segment. The 8th and 9th segments are practically evanescent 

 ventrally, but fairly well-developed dorsally. The total result is that 

 the cremastral spine projects ventrally or even slightly forwards instead 

 of posteriorly. The actual cremastral surface, with its armament of 

 abundant minute reddish hooks, has the appearance of being on the 

 back of the 10th segment instead of at its end, owing to the dorsal 

 surface of that segment being extremely short, unlike its condition in 

 Vanessa, Pieris, &c. This whole structure is very like that of Avfiynnis 

 aglaia, where the ventral sides of the abdominal segments are shortened 

 so as to bring the cremaster to the front of the pupa, and where the 

 hooks are similarly apparently dorsal by the shortening of the dorsum 

 of 10. The cremastral surface in L. celtis is more like that in the 

 Vanessids than in any other related subfamily, being narrow and 

 straight in an antero-posterior line. The similarly shaped surface in 

 Pontia is transverse. The general surface of the pupa is free from any 

 hairs and neither in this nor any other respect does it suggest any 

 Lyc^nid character or affinity. 



The pupa is 13mm. long, of which 3'4mm. only are occupied by 

 the abdominal segments 5-10. Viewed laterally the smallness of these 

 segments is seen to be quite as great as this measurement shows. The 

 depth of the pupa (back to front) being 6mm. at the hump on 2nd 

 thoracic, little over 3mm. at 5th abdominal. The dorsal spine on 2nd 

 abdominal is nearly as high as that on 2nd thoracic, at least there is a 

 difference of less than 1mm. in the thickness of the pup^e at these two 

 points. The " waist " at 1st abdominal segment is made very marked 

 by the following spine, the depth of the depression is quite 1mm. A 

 ridge or keel runs dorsally the whole length of the pupa from the 1st 

 thoracic to the last segment, and it is this that forms the spine of the 

 2nd thoracic ; on the 3rd thoracic and the 1st abdominal it is a slight 

 ridge on a flat surface, and is so also on the the 2nd abdominal to the 

 spine, behind the spine it quickly subsides, and on the later segments 

 is rather a white line than a ridge. 



The sx)ine of the 2nd abdominal is formed by the dorsal ridge, meeting 

 the ridge, that is so far as I know peculiar to lAbytheinae, which starts 

 from the ridge of the nose-spine, appears on the eye outside the glazed 



