TRANSFORMATIONS OF BUFFALO GNATS. 19 
thickenings of the cuticle known as the spiracles. The spiracles do 
not seem to be functional as a means of respiration. 
Respiration apparently takes place by what are known as the 
rectal gills (PL I, fig. 2, rg), which are extensions of the rectal wall 
and are formed of the same epithelial layer. These extensions are 
primarily 3-branched, and are either simple or bear numerous lobes. 
Minute tracheolae penetrate these gills, joining the main trunks at the 
base of the rectum. The gills are also filled with blood, are retractile 
into the rectum, but are usually extended in running water. Accord- 
ing to Headlee, they function both as blood gills and as tracheal gills, 
as the presence of blood in them is undoubted, and since the tracheae 
in them are very minute. The tracheal system can be seen very dis- 
tinctly in living larvae. Several larvae, still quite active, were placed 
in a shallow dish of water and examined under the binocular by the 
aid of strong sunlight against a black background. As the larvae 
slowly died the tracheae, owing to the air collected in them, showed 
a bright silvery appearance, so that even the very small ramifications 
and branches could be clearly distinguished. The larvae seem to be 
in a comatose condition an hour or so before death, and the silvery 
appearance of the tracheae remains for about three hours after death, 
so that there is a considerable period in which the tracheae may be 
studied before the air is driven out. In larvae preserved in alcohol or 
mounted in balsam the air, of course, is driven out of the tracheae, 
which becomes almost transparent, so that it is very difficult to trace 
them in this condition. The tracheae become greatly ramified toward 
the caudal end of the body and seem to penetrate the wall of the 
alimentary tract and to extend into the rectal gills. 
Larvae have not been found to survive longer than forty-eight 
hours in still water, and on being transferred from flowing water will 
turn back constantly in an effort to clean the rectal gills with the 
mouth parts. Larvae placed in still water and only barely covered 
will survive longer than those covered to a depth of 3 inches or more. 
That the lack of oxygen is responsible for the death of the larvae 
transferred from flowing to still water is further evidenced by the 
fact that the larvae will survive longer in a tightly corked bottle at 
a temperature of 32° F. than at a temperature of 75° F., because the 
amount of oxygen that can be held in solution by the water varies 
inversely as the temperature. Larvae seem to thrive equally well in 
running water, whether the temperature is 32° or 75° F. Low tem- 
perature seems to retard but not to injure development, and a rising 
temperature up to 75° F. hastens development. 
The structure of the rectal gills and the number of lobes afford 
excellent systematic and specific characters, as may be seen from 
the accompanying figures (PI. IV, figs. 5-9). The variation in indi- 
viduals seems to be bounded by well-defined limitations. 
