3 
within the membranous walls of the air-sacs, and the various 
definite prolongations of these. This latter was the description 
given two centuries ago by Harvey and Perrault, and more re- 
cently by Colas, Guillot and Sappey; and its correctness does 
not admit of a doubt. An interchange of gases may, indeed, 
take place between the air in the air-cells and the blood in the 
' vessels of their walls: to a very trifling extent however; for the 
blood-vessels of these delicate membranes are very minute, and 
sparsely scattered. We cannot therefore but believe that the 
extensive, elaborate and (as regards the class of birds) ubiquit- 
ous system of air-cells must have some far more important office 
than that of conveying oxygen to these minute vessels, The 
question is, What is this office ? 
Bearing in mind that the air-sacs occur in all birds; that 
their number and position is very uniform; that they always 
have very free communications with the lungs; that the greater 
part of the air inspired passes into them, and is again expelled 
in expiration, we can scarcely doubt that they are a necessary 
part of the respiratory apparatus of birds. 
The respiration of birds, even when in repose, has been shewn 
to be much more active than that of mammals. But in order 
that birds may be equal to the enormous exertion required of 
them for sustaining themselves in the air for considerable periods 
of time, very ample provision must be made for respiration. If 
therefore the lungs were constructed after the mammalian type, 
they would require to be very large, and powerful muscles must 
be provided for the respiratory movements. But this would add 
unduly to the weight of the body. ‘The lungs therefore are 
small, very porous, and light; and yet nevertheless they are 
more efficient respiratory instruments than mammalian lungs of 
greater weight and volume. This we know from experimental 
comparisons of the quantity of oxygen absorbed, and of carbonic 
acid evolved by birds and mammals, for each pound weight of 
their bodies, in equal times. The increased efficiency is due to 
2—2 
