90 BULLETIN OF THE BUREAU 



One source of error in this method is th 

 through volatilization. This loss is prol>:i 

 error was likel^y to l)e a nearly constant on 

 unaffected. 



Since it was thought possible that differences in the bodily salt content of the 

 tishes used might depend upon differences in the water contained in their alimentary 

 canals, certain tests were made with fishes from which the latter had been removed. 

 The results of such ex|)erinients were such that this procedure was not thought to 

 be worth while. 



Let me add that 1 have presented here the figures for all of the analyses made 

 l)y me, no discordant results being suppressed. 

 E.r per In lent 102." 



{a) Five F. lietevoditm from a lot which had been kept unfed for some weeks 

 in water of density 1.023 were tested together. The alimentary canals, livers, 

 and gall bladders were removed, and the fishes were weighed after the remo\ al 

 of these parts; thus the percentage of chlorine given represents the proportion 

 found in the eviscerated bodies. The percentage was here 0.167. 



(i) Five fishes from same lot were kept in fresh water for 1 da}'. The 

 subsequent treatment was the same. Percentage of chlorine 0.129. 



((?) Five fishes (from another lot) were kept in 1.001 water'' for 1 day. 

 Alimentar}' canals, etc., removed as above. Percentage of chlorine 0.123. 



{(I) Eight fishes (same lot as last) kept in 1.002 water (see table on p. !h;i 

 for 1 day. Percentage of chlorine 0.159. 

 It would thus appear that in fresh water about 23 per cent of the total chloride^ 

 of the body passed out during a single day. Let it lie recalled that in experiment <s(i, 

 in which the fishes used had been kept in water of the same density as those here 

 emplo3'ed, it was found that the loss of chlorine was O.OiS gram per 100 grams of 

 body weight. Assuming the original percentage of chlorine to have been the same 

 as in the present experiment (0.167), the fishes in experiment 86 lost about 25 per 

 cent of their chlorine. Thus the results reached by these two methods are found to 

 be in striking agreement. It would ))e rash, however, tt) expect such a close 

 correspondence in every case. 



The effect of the 1.001 water was practically the same as that of fresh water, the 

 difference of 0.006 per cent in the anah^ses probably having no significance. On the 

 other hand it is a fact of the highest interest that fishes in the 1.002 water lost little 

 if any of their chlorides. It would thus appear that the degree of dilution which 

 may be endured with impunity by this fish lies somewhere between these two limits. 



E.rp.rn,„l,tl(ii. 



Fiiiiilii] IIS iiiiijiiUs was here used^ two fishes being taken for each analysis. 

 The digestive tracts, etc., were not removed. 



(a) Fishes from .salt water (1.023). Percentage of chlorine 0.179. 



(J)) This lot, originally from salt water, had spent 2 days in water of density 

 1.001 (approximately). Percentage of chlorine 0.178. 



" All of this series were done at Woods Hole during August, 1905. 

 '' The water here used gave a salmometer reading (corrected) of about 1.001. 

 tills reading, however, showed that it contained only 3 per cent as much chlorine a 



