EFFECTS UPON FISHES OF CHANGES IN SALINITY OF WATER. 71 



If the tish lost ill woio-ht. it would he necessary to conclude that the medium was 

 h^'pertonic." But the limiting membranes, in order to produce this etiect, might be 

 either semipermeable (entirely preventing the passage of salts), or permeable to 

 both water and salts, though to the latter in a lesser degree than to the former. 



If a gain of weight occurred, it might be concluded either (1) that the medium was 

 hypotonic, in which case the membranes might belong to either of the categories 

 mentioned in the preceding paragraph; or (2) that an imbibition of water had taken 

 place, due to a chemical union of some sort. It will be shown that in the case of 

 dead tishes such an imbibition takes place regardless of the osmotic pressure of the 

 medium. 



It nuist be repeated that the behavior of the mem))ranes of a given fish \ aries 

 greatly with the conditions, and that seemingly contradictory results may be obtained 

 with tishes of the same species. An extensive series of experiments which was 

 carried on during the spring of 1905 at the New York Aquarium gave results that 

 were almost wholly in harmony with one another, and certain conclusions were 

 framed which seemed to be supj)orted by nearly all of the facts then known to me. 

 A repetition of some of these experiments during the summer of 1905 gave, in certain 

 cases, ([uite different results, and led to a modification of some of my conclusions. 

 The only differences in the conditions, so far as I am aware, were the temperature of 

 the water and the previous history of the fishes used. The New York experiments, 

 since they form a harmonious series, will be descril)ed first, after which the Woods 

 Hole experiments will be discussed and an attempt made to harmonize some api)arent 

 contradictions. 



The method employed throughout this portion of the work was to weigh the 

 fishes carefully befoi-e and after the change in the density of the medium. This 

 procedure involves numerous difficulties. In the first place, fishes under such treat- 

 ment should not be fed, since the gain or loss of weight through feeding or defecation 

 would confuse the results. Fortunateh', most fishes may be kept unfed for da3's or 

 weeks without any appreciable detriment to their health. In the second place, it is 

 necessary- to determine the normal loss of weight through waste which would occur 

 throughout the period of the experiment. This rate of decrease for the fishes used 

 was in manj' cases determined before the commencement of the experiment; in other 

 cases a conti'ol set was kept ixnder observation. In the third place, and this is all- 

 important, the same conditions must Ijc maintained each time the fishes are weighed. 

 The difference in the amount of moisture adhering to the body surface may make a 

 serious difference in the weight recorded, even 1 or more i)er cent in the case of 

 smaller fishes.'' 



In most of uw experiments the aggregate weight of a number of fishes subjected 

 to the same treatment was taken. They were commonly laid upon a dry towel for 

 one-half minute, then laid upon a soft cloth for one-half minute more, the cloth 

 meanwhile being moved in such a way that the fishes should be gently rolled over 

 the surface. Where the procedure was varied, care was taken that the same condi- 



11 The question of gain or loss of weight through the taking in or passing o>it of matter from the alimentary canal is 

 here left out of account. It will be dealt with later. 



''It was at first thought necessary in each case to clip the tishes into water uf I 

 order that the water adhering should be of the same specific gravity, but this was foi 



