396 LEPIDOPTERA CHAP. 



Fam. 21. Arbelidae. Closely allied to Cossidae, but with- 

 out frenulum, and with less complex wing-nervures. A small 

 family believed to be similar to Cossidae in the life-history. 

 The tropical African Arbelidae are considered by Karsch to be 

 a distinct family, Hollandiidae. 



Fam. 22. Chrysopolomidae. This family has been estab- 

 lished by Dr. Aurivillius l for an African genus, allied in wing- 

 nervuration to Cossidae ; the Insects are like Lasiocampidae. 



Fam. 23. Hepialidae (Ghost- and Svrift-Moths). Moths of 

 very diverse size, some gigantic ; wings not fitting together well 

 at the bases ; without a frenulum ; no proboscis ; the scales 

 imperfect ; the nervures complex. The Hepialidae are extremely 

 isolated amongst the Lepidoptera ; indeed, they have really no 

 allies ; the conclusion that they are connected with the Micro- 

 pti'rygidae being certainly erroneous. Although but small in 

 numbers only about 150 species being known they exhibit a 

 remarkable variety in size and colour. Many are small obscure 

 moths, while others are of gigantic size six or seven inches 

 across the wings and are amongst the most remarkably coloured 

 of existing Insects. The great Charagia of Australia, with 

 colours of green and rose, bearing white spots, are remarkable. 

 The South African Leto venus is of large size, and has an 

 astonishing supply of glittering metallic splashes on the wings, 

 making a barbaric but effective display. The South Australian 

 Zelotypia staceyi, of enormous size, is also a handsome ninth: 

 but the majority of species of the family are adorned only in 

 the feeblest manner. 



Very little is known as to the larvae ; they are either sub- 

 terranean, feeding on roots, or they live in the wood of trees and 

 shrubs. They are nearly bare, and are apparently the lowest type 

 of Bombycid larva. At the same time, it would appear there is 

 considerable variety amongst them. Packard says 2 the young 

 larva of Jfi-ju'/tfi/s mustelines has the arrangement of setae that is 

 normal in Tineidae. The larva of H. hurnali seems to be a very 

 simple form, but H. liectus shows a considerable amount of 

 divergence from it. They probably live for several years; the 

 larva of //. (irgenteo-iii<-nl<it us in North America lives for three 

 years, ;it first eating the roots of Alder and then entering the 



1 Ent. Ti-dsTcr. xvi. isnr,. ),. n r>. 



- On lurvnr (.]' Hrjiialidau, J. JW: York c nf. Soc. iii. 1895, p. 69, Plates III. IV. 



