II FLIES WITH AQUATIC LARV^ 141 



bottom, and provided with bunches of fine branching 

 filaments, which expose a considerable surface, and 

 are well adapted to the abstraction of oxygen from 

 the water, especially when waved to and fro by the 

 contractions of the body. The pupae of the surface- 

 forms on the other hand have respiratory trumpets, 

 which are well suited for drawing-in gaseous air. 

 Such tubes would obviously be quite useless at a 

 depth of even one inch from the surface of the water. 

 It seems most likely that the surface-form is the more 

 primitive, and that the bottom-form is a special 

 adaptation of this. The primitive Tipulid, we may 

 be pretty sure, had neither respiratory tubules nor 

 haemoglobin in the larval stage. It is equally unlikely 

 that it possessed respiratory filaments in the pupal 

 stage. The majority of aquatic Diptera (Culex, 

 Tanypus, Corethra, &c.) still retain with slight modi- 

 fications what we suppose to be the primitive form of 

 pupal respiratory trumpet. The pupa of Simulium, 

 on the other hand, which is immersed and fixed 

 beneath the surface of the water, has bunches of 

 filaments like Chironomus plumosus. 



It is interesting to see in these Insects, how natural 

 selection can act upon the earlier stages of the life of an 

 Insect without materially affecting the structure of the 

 adult. The naturalist who should attempt to classify 

 aquatic Diptera by the larva and pupa, would place 

 Chironomus plumosus and Chironomus motitator in 

 different families, yet the flies seem to differ in nothing 

 of greater importance than the length of the first 

 joint of the fore tarsus. Other instances of the special 

 modification of provisional organs brought about in 



