336 PHYSIOLOGICAL DEVELOPMENT. 



balancing itself by slight undulations of its fins, shows hoi 

 great an economy results from an internal float, to fishes 

 which seek their food in mid-water or at the surface. Hence 

 the habit of swallowing air having been initiated in the way 

 described, we see why natural selection will, in certain fishes, 

 aid modifications of the alimentary canal favouring its 

 lodgment — modifications constituting air-sacs. In 



the second place, while from air thus lodged in air-sacs thus 

 developed, the advantage will be that of notation only if the 

 air is infrequently changed or never changed, the advantage 

 will be that of supplementary respiration if the air-sacs are 

 from time to time partially emptied and refilled. The re- 

 quirements of the animal will determine which of the two 

 functions predominates. Let us glance at the different sets 

 of conditions under which these divergent modifications may 

 be expected to arise. 



The respiratory development is not likely to take place in 

 fishes that inhabit seas or rivers in which the supply of 

 aerated water never fails : there is no obvious reason why 

 the established branchial respiration should be replaced by a 

 pulmonic respiration. Indeed, if a fish's branchial respiration 

 is adequate to its needs, a loss would result from the effort of 

 coming to the surface for air; especially during those first 

 stages of pulmonic development when the extra aeration 

 achieved was but small. Hence in fishes so circumstanced, 

 the air-chambers arising in the way described would naturally 

 become specialized mainly or wholly into floats. Their con- 

 tained air being infrequently changed, no advantage would 

 arise from the development of vascular plexuses over theii 

 surfaces ; nothing would be gained by keeping open the com- 

 munication between them and the alimentary canal; and 

 there might thus eventually result closed chambers the 

 gaseous contents of which, instead of being obtained from 

 without, were secreted from their walls, as gases often are 

 from mucous membranes. Contrariwise, aquatic 



vertebrates in which the swallowing of air-bubbles, becoming 



