191 
It must be noted that oxygen always has been found in the 
intercellular spaces. 
During the end of the larval state the animal forms a cocoon, 
wbich is always attached to roots or rhizomes of aquatic plants. 
When the cocoon is built the larva gnaws one or two holes 
through the bottom of it into the intercellular spaces of the plant. 
The cocoon is alwavs airfilled. 
*/ 
Perris made two very important observations, namely, that 
the cocoons are filled with air and that the pupa (and the full- 
grown insect in the cocoon as well) cannot live without the air in 
the cocoon being renewed. 
Perris did not know that the air of the cocoon through the 
hole (or holes) is in communication with the intercellular air of 
the plants; he therefore concluded that the air is renewed through 
the wall of the cocoon. The carbon dioxide formed is thought to 
be osmotically exchanged with the oxygen in the water. 
This hypothesis, however, has been given up after v. Si eb o ld 
had seen, that the larva gnaws a hole through the wall of the 
cocoon into the intercellular spaces of the plants. From this time 
it has been taken for granted that the air necessary for the respi¬ 
ration must come from the plants. 
This view is indeed the most probable, but one cannot be 
quite sure whether or not the oxygen dissolved in the water is of 
any importance for the renewal of the air in the cocoon. 
I have therefore endeavoured to make analyses both of the 
air contained in the cocoons and of that in the corresponding 
intercellular spaces, taking care that the air from the intercellular 
spaces is taken exactly from the spot, where the cocoon is 
situated. 
This is of great importance as the composition of the air in 
the roots varies in a considerable degree from the upper end, 
where the oxygen percentage is highest, to the tip of the root, as 
the following analyses show. 
