( 169 ) 



The copulatory armature is on the whole comiilioated in this snlifaiuily, the 

 species differing often very remarkably from one another. The greater iiroj)ortiou 

 of the species are slow-flying insects which do not wander mucb, and that accounts 

 largely for the astonishingly great differences presented by the geographical races 

 of several species (compare I'seiiilnclanis postica, Poli/j/li/c/uis (rilineatus). 



The early stages of the Ambiilicinae are interesting on account of several 

 peculiar points in their structure. The larvae are generally said to be characterised 

 by a triangular head and a granulose skin. These characters are indeed found in 

 the European and American species, but not in all the African and Indian ones. 

 So far as the larvae are known, and so far as the descriptions and figures are 

 reliable, there is no smooth-skinned larva among tlie Amhulicinae, all being 

 granulose or spiuose. The granules are prolonged to short spines in Coequosa, to 

 longer and stronger spines in ]!liailinojmm,s.m\ to long dentate ones in LopJwstethus. 

 Of these larvae Coeqiiom has a triangular head, while it is rounded in the other two. 

 A rounded head is also found in the full-grown larvae of Duphnusa. and Clams 

 (and perhaps in other genera). Now the question arises whether the bulky- 

 headed larva of these genera, which are not all closely allied with one another, 

 has preserved the ancestral head-form of the Ambulicine larva, or whether the 

 triangular head is the more generalised one, from which the round head of (.'lam's, 

 etc., originated. The first stage is not known of any of these larvae, unfortunately, 

 but For.saj-eth figures the yonnger larva of Clanis {Trans. Ent. Soc. Lorul. 1884. 

 [I. 303. t. 1.5) as having a triangnlar head, and states that the rounded bulky head 

 is acquired at a later stage. From this one must conclude that in the large-headed 

 larva the triangnlar head-form is lost iu consequence of a lateral expansion of the 

 head-case. But this does not necessarily mean that the ancestral larva of the 

 Amljidicinae had a triangnlar head. Considering that the first stage of Sphinx 

 ocrllata (and also of the Acherontiine genus Lapara) presents a rounded head, 

 which assumes later on the well-known triangular form, and that the head of 

 PoUjptijchus grcii/i is jjroduced in the earlier stages into a long process, and assumes 

 an obtusely triangular firm in the last stage, it seems to be probable that the 

 triangular head is a derivation from a rounded one, and has developed again in 

 some instances into an enlarged rounded head. Thus the caterpillars of Lophostethus, 

 Ihpknusa, Rhadinopasa and Clanis would be later forms than the acrocephalic 

 larvae of other Amhulicinae. If this is true, the spines of Rkadinopa.m and 

 Lophostethim would also ajipear to be exaggerated developments of the setiferous 

 granules of Mannnlia and others, and would not represent an ancestral feature pre- 

 served from the common ancestor of the Sphingidae and Saturniidae. It is 

 necessary to study the first stages of the Sphinyidae. more closely and of more 

 species than has hitherto been done. We have almost entirely to depend on 

 descrijitions and figures, which mostly fail in giving the essential points. The 

 conclusions based on such scanty and not always reliable data are not convincing. 



The horn is long and curves gently upwards iu the generalised forms ; it is 

 occasionally lost {Coequosa). 



The chrysalis of the Amhulicinae is as a rule rather stumpy at the frontal end ; 

 in many forms the frontal part bears two tubercles. The sheath of the tongue never 

 projects, as in the greater proportion of the Acherontiinae ; it reaches either to the 

 end of the wing-cases (rarely, Compsogene only ?), or is shortened. A coraj)aratively 

 long tongnc-case is retained in some species which have a strongly reduced tongue 

 {.\limics liliae, for instance). This fact, which is corroborated by the preservation 



