BY R. J. TILLYARD. 541 



species. This is an atavism, or reversion to type. All such oc- 

 currences need to be carefully noted; and the changes that occur 

 should be carefully compared with known archaic conditions. 

 Comstock (15) has recorded quite a number of instances of this 

 in the Planipennia and Megaloptera, particularly with respect to 

 the retention of the primitive dichotomic forking of R 4 . 5 . In 

 the Lepidoptera, the family Hcpialidae is occasionally subject 

 to atavisms, which are of special interest, owing to the very 

 archaic position of this family within the Order at the present, 

 day. Thus Comstock (15, fig. 335) has recorded a specimen of 

 Pielus labi/rinthicus, in the forewing of which R. is itself forked, 

 and I have myself seen a specimen with a very similar forking. 

 We know from the venation of Belmont ia (Text-fig. 63) that the 

 Permian Paramecoptera had this vein forked. Thus the atavism 

 serves to strengthen our belief that the HepiaHdae are descended 

 from ancestors which possessed the Paramecopterous forking of 

 R St Comstock (15, fig. 337 ) also mentions the famous c pecimen of 

 Sthenopis, in which the hindwing had M and Cu la only partially 

 fused. This is the same condition which we find in the Triassic 

 fossil Arehipanorpa; and there can be no doubt that the an- 

 cestors of the Lepidoptera must at one time have passed through 

 a stage in which the two originally separate veins M 4 and Cm a 

 were only partially fused, as in this atavistic individual of 

 Sthenopis. 



As it is impossible to say definitely that the condition shown 

 in an atavistic individual actually belonged to the Order in ques- 

 tion, seeing that it ma}- just as well be an atavistic reproduction 

 of a character preserved in the ancestral Order only, and not 

 in the Order descended from it, to which the specimen now be- 

 longs, I have not admitted the evidence of these atavisms as 

 sufficient, per se, to govern the character of the Archetype of an 

 Order. Only when their evidence is supported also by evidence 

 under one of the four headings A-D above, have I deemed it 

 permissible to use it in the construction of the Archetype. Thus. 

 in forming the Archetype of the Order Lepidoptera, I have ad- 

 mitted only four branches to Rs, and have kept ^\! 4 fused with 

 Cu, a right to their tips; since there is no evidence, apart from 

 the atavistic specimens themselves, to show that these characters 

 ever really belonged to the true Lepidoptera; whereas there is 

 distinct evidence, in the form of the fossil record, to show that 

 they did belong to Orders ancestral to the Lepidoptera. 



