sphingiNjE. 269 



are made up of scattered white points, and he recounts certain 

 observations which tend to show that the purple borders of the 

 stripes in this species have been modified from a dark green border 

 at the sides, but not above, where the latter faintly persists ; in 

 the same manner, the pure white stripes have arisen from lines, 

 like those of the Amorphids, which still remain above. He con- 

 cludes that the beautiful colours of Sphinx ligustri have been acquired 

 very late in the phylogeny. In a later paper (loc. cit., 1885, pp. 281 

 et seq.), Poulton details the ontogeny of S. ligustri, and makes an 

 erroneous guess* (p. 283) that the larva of this species, which has 

 a clothing of short secondary hairs in its 2nd instar, also has them 

 in its 1st, and he later (loc. cit., 1886, p. 141) refers to this as an 

 observed fact. This and other errors of observation tend to discount 

 the value of much otherwise good work. We have already {anted) 

 referred - to the subdorsal line as a characteristic feature of the 

 more generalised Sphingid larvae and as occurring in the early 

 larval stages of many Eumorphid (sens, lat.) larvae that lose it 

 more or less in their later and more highly specialised stages. 

 Weismann points out (Studies in the Theory of Descent, transl. pp. 242, 

 317, 372) the relationship of this line with the more recent and special- 



* In spite of the detail into which Poulton enters in the papers already 

 referred to, the} 7 are defective and even inaccurate from many points of view (due 

 largely to the fact that he had not the advantage of knowing the importance of the 

 position of the primary tubercles which Dyar has since shown), and they must be con- 

 sidered as useful only for the purpose of tracing- the development of the larval 

 markings, and not to be relied upon in the references to detailed structure. His 

 description of the primary tubercles in Sphinx ligustri [Trans. Ent. Soc. Lond., 

 1885, p. 282) accounts for i, ii, iii on abdominal segments 1-7, but leaves out 

 entirely iv, v, vi ; even then he erroneously states that iii is composed of two 

 setae, and he further notes that, on abdominal segments 8-10, the setae are more 

 abundant and without definite arrangement. He also states (p. 283) that he did 

 not examine for "small" hairs in stage 1, but siuce he found them in stage 2 he 

 assumed that ' ; there must be a comparatively thick coating of them" in the 

 1st stage, and then deals with them as if they were present, remarking that 

 " there can be no doubt of their presence," &c, whilst we read yet again of 

 the presence of these hair-bearing- shagreen-dots in the 1st instar of S. ligustri 

 floe, cit., 1885, p. 299), in fact, the larva of this species, and that of -5*. ocellata 

 in the same stage, are described as " hairv larvae with tubercles at the bases of 

 the hairs." As a matter of fact, these small hairs do not appear in the 1st instar of 

 S. ligustri. Similarly we find him uncertain [loc. cit., 1888, p. 520, whether 

 shagreen-dots appear at the end of the 1st instar of Agrius convolvuli, or whether they 

 are merely those of the 2nd instar showing through the transparent skin just before 

 the 1st moult. Further, he did not, until 1884, demonstrate the presence of the well- 

 known bifid hairs in the Amorphids, although he had been publishing his observa- 

 tions on the larvae of this group in [883, and even then (in 1884) did not detect them until 

 the 3rd larval instar, thus showing that the work was done with an altogether too low 

 power to deal with the detail attempted. As late as 1886 floe, cit., 1886, p. 141), 

 a comparison between the rirst stages of Smerinthus [ocellata and populi) and 

 Sphinx ligustri is made, and the error that the newly-hatched larva of S. ligustri 

 is covered with hairs or bristles springing from ordinary shagreen-dots is yet again re- 

 peated, whilst he further states that the dorsal and lateral rows of hairs (= those arising 

 from i, ii, iii) spring from larger shagreen-dots, a most remarkable statement as to the 

 origin of primary larval setae. In his notes on the larva of Agrius convolvuli [loc. 

 cit., 1888, p. 520; he also treats the primary tubercles, i to vi, as shagreen-tubercles, 

 and states that, in the 1st instar, " there are at first no other shagreen-tubercles upon 

 the larva." Altogether, whilst recommending the perusal of these papers for 

 their general account of the development of the Sphingid larval markings, we 

 would suggest that the student should compare carefully all statements relating to 

 details of larval structure with the details under the several species published in this 

 work (antea, iii., pp. 386 et seq.) before accepting the same. 



