Ross (1949) established a small new family of Annulipalpia with 

 3 species of the genus Xiphocentron Brauer, 1870 (which was placed 

 before in the Hydropsychidae (Ulmer, Genera Insectorum, 1907:176, 

 Figure 217)) and 1 species from south China (X. hwangi Ross). He 

 described later X. sturmi Ross from the tropical forests of Colombia; this 

 species has an interesting larva and pupa and is terrestrial (cf. Sturm, I960). 

 Ross stressed its close relationship "to the ancestral stock of the 

 Psychomyiidae" (Ross, 1949:2). 



The larva of X. m e x i c o (the close relationship of which to the 

 Psychomyiidae is even more marked than in the adult) was recently found 

 in the sources of a cold, slowly running brook in Texas; this close 

 relationship is expressed in such characteristic structures as a labium 

 without palps, elongated into a narrow spinning organ, a trochantin with a 

 large distal lobe, anal legs with claws bent at a right angle and the pupal 

 mandibles elongated into thin filiform distal endings. Edwards (1961 ), who 

 described the larva, placed it in the Psychomyiidae together with Lype 

 150 and Tinodes; Flint was of the same opinion. However, this opinion is 

 contradicted by the important fact that this species has no mental sclerite, 

 its frontoclypeal sclerite has no lateral notch and the claws of the anal 

 legs have no ventral spinules. The larva builds galleries at the bottom of 

 a source which are about 5 cm long and differ from those of Psychomyiidae 

 in that part of them is situated below the water, the other part above water, 

 among the crusts on stones of the littoral. The characteristics of the pre- 

 imaginal stages of X. m e x i c o thus do not invalidate the status of the 

 family Xiphocentronidae. The larvae of X. sturmi build similar structures 

 but outside the water in humid forests (relative humidity of 90—90.9%), 

 among slimy deposits on horizontal branches. The full-grown larvae of 

 X. sturmi have not been described, but the pupal structures of this species 

 are quite characteristic, spindle-shaped with double walls* hanging from 

 a thin, hollow style (Sturm, I960). 



The differences in the descriptions of the larvae and pupae of Xipho- 

 centron show that this small and very ancient genus is complex; 

 possibly, it should be divided into subgenera according to the morphology 

 of the preimaginal stages, their structures and their geographical 

 distribution, as stated by Ross (Ross, 1949:51). 



Ross (1938:135-136; 1944:51— 52, Figures 189-201) enlarged taxonomic 

 units.** He united the Psychomyiidae and Polycentropodidae at first as 

 Polycentropodidae and later as Psychomyiidae, disregarding the marked 

 difference between the larvae of these families, which had already been 

 noticed by Siltala (l907:60€— 60l). Objections to the unification of the 

 foregoing families were made by Mosely (1945:1946), Ulmer (1951:109—110) 

 and Lepneva (1956:10—15) on the basis of the morphology of the larvae. 

 Some American authors accepted the views of Ross. 



Ross and other authors recently considered the former subfamily 

 Glossosomatinae as the family Glossosomatidae; we accept this view. 



Ulmer, in his work on the Sunda Islands, described the subfamily 

 Ecnominae of the family Psychomyiidae as follows: "Vielleicht miissen die 

 Ecnominae einmal als besondere Familie betrachtet werden" (Ulmer, 

 1951:146). 



* Apparently representing cocoon and case. 

 *" See section 5 and the following genera of section 4 of McLachlan (1874—1880:392). 



140 



