2 AQUATIC COCKROACHES. 
body was in the water, but the tip of the abdomen was never 
submerged, even when all the rest of the body was covered. The 
abdomen moved up and down with a rhythmic action, and 
bubbles of air issued at more or less regular intervals from the 
prothoracic spiracles. These air-bubbles were seen to form 
gradually, to grow larger and larger, and finally to break away 
from the spiracles; about twenty per minute passed through the 
spiracles. Air issued from the mesothoracic spiracles only when 
the insect was violently agitated. From these observations it 
seemed fairly obvious that the terminal abdominal spiracles were 
inspiratory in function, the thoracic spiracles expiratory, and 
that it was necessary, therefore, for the insect to have the tip of 
the abdomen exposed to the air, but that it was a matter of in- 
difference whether the expiratory spiracles were above water or 
below it. In order to settle the question beyond all manner of 
doubt, some specimens were fastened with cotton threads to 
strips of cork; half the number were fastened head downwards, 
the other half head upwards. The cork-strips with the attached 
insects were then immersed in tubes of water. In the case of 
the reversed specimens the water covered the thorax and basal 
segments of the abdomen, but the tip of the abdomen projected 
above the water-level ; the other specimens had the abdomen in 
the water, but the thorax exposed. The results in every case 
proved the inspiratory and expiratory functions of the abdominal 
and thoracic spiracles respectively. The reversed specimens 
endured their constrained position for many hours (twenty-four 
to forty-eight or more), and when released seemed little the worse 
for their experience. On the other hand, the specimens with the 
abdomen immersed in water died in less than twelve hours, some- 
times in less than six. The structure of the thoracic spiracles 
in Cockroaches is quite different from that of the abdominal 
spiracles,* and a difference in function is only to be expected ; 
nevertheless, when repeating these experiments with terrestrial 
Cockroaches, such as Panesthia javanica, | was unable to demon- 
strate satisfactorily the functional differences of their spiracles. 
This failure may be accounted for—in part, at any rate—by the 
fact that this species struggled long and violently when pinioned 
to the cork-strips, and, as they are extremely muscular insects, 
* Miall and Denny, ‘ The Cockroach,’ 1886, pp. 151-155, ff. 85-88. 
