284 NATURAL HISTORY OF AQUATIC INSECTS 



plctcly cast at the last moult, when the spiracles open 

 for the first time and allow air to pass directly into the 

 tracheal tubes. In Perla, however, the tracheal gills 

 persist, though in a greatly reduced form, after the 

 spiracles have opened. In the winged fly the gills 

 can be discovered with a lens in exactly the same 

 places where they occur in the larva. This fact, 

 which was established by Palmen, was of great use 

 to him in refuting a view which had been extensively 

 adopted — viz., that spiracles are derived from pre- 

 viously existing tracheal gills. It was supposed that 

 the casting of the gill in all cases opened a way into 

 the tracheal system, which was previously closed. In 

 the winged Perla, however, minute tracheal gills, pro- 

 ject from every thoracic spiracle. They are probably 

 quite useless. The chief theoretical interest of Pal- 

 men's arguments is that they dispose of a theory 

 which implied that Insects were primitively aquatic, 

 and became adapted to aerial respiration by the con- 

 version of their gills into air-slits. On the contrary, 

 it is on all grounds probable that Insects were primi- 

 tively terrestrial, and that the various adaptations to 

 aquatic conditions are secondary. Where Insects 

 possess gills, these are not primitive organs pos- 

 sessed by the more typieal Insects, but relatively late 

 acquisitions, developed as it were casually, and with a 

 marked relation to the peculiar needs of the family 

 or small group in question. No organs of an Insect arc 

 more variable in position, number and .structure than 

 the larval gills, and this is one proof that the}- were 

 not inherited from a common ancestor, but acquired 

 by many small groups independently of one another. 



