334 NATURAL HISTORY OF AQUATIC INSECTS CH. 



much expanded at the end. Side-pieces, correspond- 

 ing to the labial palps, arc attached, and there is 

 commonly a pair of spines or claws, which secure the 

 struggling victim. The details of the structure vary 

 according to the species. When the larva is at rest 

 the apparatus is folded up, the broad joint being 

 spread over the front of the mouth, while the arm is 

 bent backwards between the fore-legs. In Libellulid 

 larvae the side-pieces can be brought together in the 

 middle line like the jaws of a rat-trap, which they 

 further resemble in their toothed edges. No more 

 murderous weapon exists among animals than this. 



We must next consider how the submerged larva 

 obtains its supply of air. In the broad-bodied 

 Libellulid larvae, a pair of large spiracles can easily 

 be seen on the dorsal surface behind the head (be- 

 tween the narrow prothorax and the mesothorax, i.e. 

 between the fore and middle segments of the thorax). 

 In the larvae of Agrionidae the same spiracles are 

 hidden, but can be made out by careful search. 

 Another pair of thoracic spiracles can be discovered 

 by dissection. These openings have been usually 

 supposed to be completely closed during the aquatic 

 stage. Reaumur found that oiling the spiracles did 

 not destroy life, as it does in ordinary air-breathing 

 Insects, and this fact has often been quoted as 

 decisive proof that the spiracles are not realh' open 

 in the Dragon-fly larvae. Reaumur himself draws no 

 such conclusion, but allows that the oil possibly did 

 not adhere to the wet surface, or that the spiracles 

 closed and shut out the oil. He might have given a 

 third explanation as an alternative, viz., that the larvae 



