( 440 ) 



this it wonlil ajipear that the steruite iii'/iiriformis had preserved the more ancestral 

 shape, and that the asymmetrical one was the younger development. We think just 

 the reverse is the case. For in H. venata, as iu Cephonodesjanus, we find the tenth 

 sternite divided into two ])rocesses, curving laterad at tip in reiinta, as in Sesia, etc. 

 This i-cmita sternite, representing the essential features of the divided sternite of 

 the allied genera from Sesia down to Madorijx and Pachi/Ua, is a more ancestral 

 type than the apparently simple sternite of the other Uacmorrlmgia. As ernafa 

 agrees strnctnrally with //(ci/'ormis, etc., except in the tenth abdominal segment and 

 tlie iienis-sheath, tjiere must l)e a cnnnection between the divided and the simple 

 sternite. We have shown under Cep/w/wi/rs that the apparently simple tenth 

 tergite of kylas, etc., is the result of the reduction of the left lobe, which remains 

 vestigial at the base of the long right lobe. On looking over the great number of 

 dissections oi Ilaemorrhagia, we find that there is in many cases, at the base of the 

 ventral process on the right side, a jiiece of chitin separated from the process by the 

 vestige of a groove (PI. XLIII. f. 25, r) ; this piece is doubtless the rest of 

 the right process of the sternite. We regard, therefore, the single asymmetrical 

 process of tlit/shi', staudingeri, etc., homologous with the asymmetrical left jirocess 

 of the divided sternite of ceiiata, and think that the nearly symmetrical process of 

 fuciformis is a derivation from an asymmetrical one. 



The difJ'erence between the right and left clasper is obvious in all species, but 

 more marked iu some than iu others (PL LI. f 17—21 ; LII. f 4. .5). The claspers 

 arc! most similar in di/linis (PI. LI. f. 21). The left harpe (//,/) is always smaller 

 than the right one ; it has never a prominent process, while the right harpe is in 

 many species produced into a more or less clubbed process, generally armed at the 

 end with spines or clothed with long bristles. We re])eat, it is the left harpe which 

 is tile more reduced, while it is the right i)rocess of the tenth sternite and tlie left 

 Jirocess of the tenth tergite which is obliterated in most Haemorrhagia and some 

 Cephonodes. This is remarkable, for it is not difficult to perceive that it should 

 have been the left process of the tenth sternite which disappeared, instead of the 

 rigiit one, if the oblitei-ation of the left process of tlie tenth tergite and of the left 

 har]ie of the ninth segment was due cnily to an inherent tendency in these segments 

 to become reduced on the left side. As in the case of the tenth segment it is the 

 left side of the dorsal jilate and the right side of the ventral plate whicii atrophy, 

 it is clear that there must be some other reason for this peculiar development. 



Comparing tiie most primitive tenth segment, as preserved in //. venata, with 

 the most sjjeciulised one, as found in Cephonodes leucogast.er, we see that at the 

 highest degree of specialisation attained, the double forceps, which moves vertically, 

 is rejilaced by a single forceps moving liori/.ontally (PI. XLllI. f. 1.1. IU). At (irst 

 sight it appears to be great waste to drop the right ventral process and to move the 

 left one towards the right side and npwanls, instead of drojiping the left one and 

 bringing the nearer right one in the position in which the sternal process is situated 

 in Icucoguster. But the species oi Ilaemorrhmjia and several Cephonodes show that 

 the right ventral process is lost without the movement of the lei't one towards the 

 right side having begun. Therefore it is obvious that the absence of the process is 

 not dependent on the twisting of the segment to form a horizontal forceps, and that 

 the complete twisting of the segment, as observed in C. leucogaster, took place 

 probably after the loss of the right ventral process. 



Besides the reduction of the bifurcate sternite to one with a single process, we 

 observe anotiier line of development, illustrated by Ce/dionodes janiis and hylai. 



