EFFECTS UPON FISHES OF CHANGES IN SALINITY OF WATER. »i) 



water adhering to the surface should be removed. Thej- were weighed without any 

 preiiminar}- drj-ing, such as was necessar\' when very accurate weight determinations 

 were required. Except in a few cases, the tishes used liad been kept unfed for 8 or 

 more days prior to the experiment. 



The chief difficult}' which beset this method of procedure was the fact that the 

 ti>^hes had to be kept in comparatively small volumes of water. Aeration was hi 

 many cases effected by the use of tanks of compressed air, but, even thus, many of tlic 

 lishes were overcome by asphyxiation before the experiment had pi'oceeded far. In 

 such cases we may" feel sure that the osmotic phenomena were not wholly normal. 

 These cnniplirations, however, ar(> indicated in the records cited below, where due 

 allowanci' i^ made for them. 



My reasons for taking the amount of chlorine as an index of the <(uantily of 

 salts passing from the tish are obvious: (1) The test is an extremely simple one; 

 ('2) the chlorides (sodium chloride and magnesium chloride) together form more than 

 s8 per cent of the entire saline ingredients of sea water." The bromine, which 

 reacts like chlorine to the silver test, may be omitted as too insignilicant in amount. 

 The test employed by me was Mohr's silver nitrate titi'ation method. According to 

 this, a given volume of the solution to be tested is lirst tinted j^ellow bj- the addition 

 of a few drops of a solution of neutral potassiiuu chromate. Silver nitrate solution 

 of a given strengtli is added from a burette until, after all of the chloi'ine lias been 

 lonibined as silver chloride, any excess of silver is free to combine with the chromic 

 acid. At this point an abrupt change of color results, due to the appearance of the 

 red silver chromate. The amount of chlorine present is of course readily coniputt'd 

 from the amount of silver employed. 



For the determinations here I'ecorded I do not claim any very high degree of 

 accurac}'. Nor was this either necessary or possible under the circumstances. In 

 general, the larger the proportions of chlorine the more exact are the tiguies. 

 Where the amount was slight, however, the determination was more diflicult. and it 

 is probable that my figures in such cases are commonly too high. Again, in those 

 cases in which death occurred and decomposition commenced, the presence of \ari(ius 

 organic matters in the water obscured the reaction. In such event the solution was 

 sometimes evaporated and the residue charred. 



Allowance was made in all cases for the quantity of rhlorinc aln'atly present in 

 the fresh water used, this being about 0.003 gram per liter. 

 E.qx'ri merit 81. 



New York, May, I'.tOo.'' Ten F. I,,f,rn,-I!_f,is. of aggregate weight llo.;5 

 grams, taken from water of density 1.0i;j, were put into fresh water. Per- 

 centages of chlorine pas.sed out (i. e., amounts per lOO grams of body weight) 

 were 0.008 during the first hour, 0.030 during first day, 0.041 during -2, days. 

 At the end of this period the fishes were all in seemingly normal condition. 



In this experiment several facts of interest are to be noted: (1) A quite appreci- 

 al)le amount of chlorine passed from the bodies of the fishes during the first hour. 

 (2) The amount passing out during .the entire first day is less than four times the 



n Dittmar (1884), in his hypothetical formula for the "proximate composition" of sea salts, gives the percentage oJ 

 XaCl as 77.7.58, that of MgCIo as 10.878. In whatever combinations it may actually occur, however, the percentage. of 

 chlorine is given by him as 55.292, 



(>This entire series was carried on during May and .June of 1905 at tlie New York .A.iuanum. 



