196 



ME. E. B. POULTON ON THE EXTEENAL 



size in the larva. The identification of the terminal spine (x') with the anal flap of the 

 larva was published by Mr. W. H. Jackson in ' Forms of Animal Life ' (1888, p. 153). I 

 also find from my notes that I had independently arrived at the same conclusion. 

 Important confirmation is afforded by the previoiisly-described relation of certain 

 larval structures and colours to those of the pupa. When a spine is absent, and the 

 terminal part of the pupa is rounded, the part above the anus nevertheless corresponds 

 to the larval anal flap (see Plate XX. fig. 27, Plate XXI. fig. 12, &c). The rounded 

 cushion-like structure (x) represents the left anal clasper of the larva, and bears a 

 similar relation to the anus and dorsal part (x'). The proof of this identification is 

 chiefly found in the previously described exceptional individuals in which this part 

 retains the characters of the anal claspers (see Plate XX. figs. 8, 9, 10). 



An examination of Plates XX. and XXI. will show that the condition represented in 

 woodcut 7 is typical among pupse. There are, however, certain species in which the 



Fig. 3. 



./vm 



Fig .4. 



Fig. 5 



Fig .6 



Fig .7. 



Fig 



Fig 



arrangement is somewhat different. Woodcut 8 represents diagrammatically one of these 

 exceptions. An example is found in the pupa of Aglia taa (Plate XXI. fig. 16). The base 

 of the terminal spine is somewhat constricted off from the rest of the tenth abdominal ; 

 the latter is divided into an upper and lower part by an oblique furrow. In certain 

 Geometrge the same conditions are still more strongly marked ; they are diagrammati- 

 cally represented in woodcut 9. Melanippe fluctuate/, is a good example of this condition 

 of the tenth abdominal (see Plate XXI. figs. 21, 22, 23), while Amphidasis betularia is 

 ' transitional from this to the normal condition shown in woodcut 7 (see Plate XXI. 

 fig. 20). 



Por a long time I believed that the tenth abdominal is in reality composed of two 

 segments arranged, as in woodcut 7, one over the other. The arrangement shown in 

 woodcut 9 is then explained by the supposition that the two component segments have 

 assumed a more normal mode of succession, the anal segment (x in woodcut 7) becoming 



