304 FRANCIS B. SUMNER. 



broadly, while Garrey says of Fiinduhis heteroclitus : "The in- 

 tegument and gills are therefore impermeable." Garrey is 

 cautious enough, however, not to postulate an absolute imperme- 

 ability, either for Fundulus or for teleosts in general. Greene, 

 though he finds that the osmotic pressure of the blood of the 

 Pacific salmon undergoes a decrease of about 17 J^ per cent, 

 when the fish ascends a river to spawn, is nevertheless doubtful 

 whether osmotic exchanges with the surrounding water are re- 

 sponsible for this decrease. 



In view of my own experiments, however, we are certainly 

 not justified in concluding from the absence of osmotic equi- 

 librium between the fish and its environment that no osmotic 

 interchanges normally occur. On the contrary, abundant experi- 

 ments seem to prove that both water and salts may, under certain 

 conditions, be transmitted in either direction without any harm 

 resulting to the fish. These conditions seem impossible to state 

 in advance for a given case. In general we may say that : 



1 . Measurable changes in weight result only from considerable 

 changes in the surrounding water, but 



2. Not all such changes of density suffice to produce changes of 

 weight, even w/ien the fish is transferred to a medium uhich is 

 knoivn to be strongly Jiypertonic or hypotonic to its own body fluids. 



3. Changes in the salinity of the water may or may not result in 

 changes in the salt content of the body. 



4. Changes in tJie bodily salt content may or may not be accom- 

 panied by changes in ivcig/it. 



5. Neither the changes in zveight nor in salt content are at all 

 proportional to the changes in the density of the external medium. 



It ivould appear that there is normally a tendency on the part 

 of the fish to resist osmotic changes and to maintain the fluids of 

 the body at a definite degree of concentration. Under various con- 

 ditions, however, this resistance is overcome and a certain degree of 

 permeability is established. This is generally a differential per- 

 meability, resulting in osmosis and consequent changes of weight. 

 In such cases, /lozcevcr, the membranes are not strictly semiperme- 

 able, but transmit salts in some measure. Indeed it ivould seem that 

 at times the permeability is indiscriminate, in which case the salts 

 may diffuse freely, but no changes in weight occur. These various 



