16 



part of my work by the light of Professor Fernald's researches, 

 which were so extremely valuable from the new characters 

 used and which often vindicated my previous generic sepa- 

 rations. So completed, the work is here offered to the student, 

 It embraces a few forms not occurring in New England and, 

 although not so full in its account of the species and groups, 

 ma}' not be without sufficient interest as to warrant its 

 issue as a work of reference. 



The Sphingidae are a highly organized and specialized 

 assemblage of Moths or Night Butterflies, having smooth, 

 hairless larvae and seem to me an outgrowth of a spinose 

 or bristly haired larval group, the posterier spine or caudal 

 horn being developped probably from one or more spines or 

 bristles, the base of which has been prolonged and stiffened 

 by deposits of chitine. In form and structure I believe they 

 are nearest related to the Ccraforfni/piixif a subfamily of 

 Bpmbyridae, and separated from allied Saturnian genera 

 (such as Heii/iJ/'HCft) by myself and Mr. Robinson in 1806. 

 The SmerintMnae are probably descendants of the oldest forms 

 of the SpJiinyidac. On one side they may have given off 

 the higher series. Ckoerocampinae, Macroglossinae, on the 

 other to the lower, SplinHjhirte, Ach<>>-<>nth)<t<>. The anal horn 

 is a simple thin black spinule in Dri/ocainjxi nilicmula, the 

 lowest form of C&ratocampinae, stouter and more developped 

 in the higher An i no fa stiipiift, and becoming, with the entire 

 armature, complex and well developped in the higher genera 

 of the subfamily such as CifJi<>r</i, SpJunf/icfnifpa, Eaclc*. 

 The bristles are gradually lost as the caudal horn became 

 variedly formed. The pupation is similar, beneath the sur- 

 face, without cocoon, an unusual one in the Itombi/ccN or 

 Spinners. Very interesting is the discarding of the larval 

 horn in the later stages in certain genera of Hawk Moths 

 and the substitution therefore of a raised ocelloid spot. That 

 it should occur in two grape-feeding genera TJti/rrns and 

 Philampelus, warrants me in associating the two groups in 

 which it happens. The point of my arrangement in fact 

 lies in considering the ( 1 h<>ci-i>nnni>iii<ir as next to the Ma- 

 and following them by the Sinrri>tfhiit<ir, to which 



