3O2 FRANCIS B. SUMNER. 



Brackish water (sp. gr., 1.002) (5 lots of 4 fishes each), 0.134 [or 0.142]' 

 5 days (2 fishes employed) - I 43 



Salt water (1.023) i 



| 10 days (3 fishes employed) 0.151 



It will be seen that the last of these figures is about 78 per 

 cent, greater than the first ! And yet the fishes were all alive 

 and apparently well at the time they were killed for analysis. It 

 will be likewise seen that whichever figure be regarded as the 

 more correct one for the brackish-water fishes, the latter agree 

 much more closely with the salt-water than with the fresh-water 

 individuals (the comparison being of course between the extreme 

 members of the series). It must be added in strict fairness, how- 

 ever, that in two different tests fishes kept for only one day in sea- 

 water gave a much higher percentage of chlorine than those kept 

 for five or ten days. For this apparent anomaly I believe that a 

 satisfactory explanation can be given, but this has been deferred 

 to my longer work. 



A series of figures somewhat similar to the above was obtained 

 from experiments with the white perch. 



Experiments with both F. lictcroclitus and F. majalis agreed in 

 showing a great difference between the effects upon the chlorine 

 content of the body of pure fresh water and water having a certain 

 small percentage of salt. This difference is extremely significant 

 in view of the difference, already mentioned, in their effects upon 

 the life of the fishes. Moreover, it was further found that the 

 average percentage of chlorine contained in the salt-water fishes 

 analyzed was of the same order of magnitude as that of water 

 containing just enough salt to maintain the fishes in health. That 

 such water could not have been even approximately isotonic with 

 the body fluids of these fishes seems evident from the cryoscopic 

 determinations of other investigators. 



4. Careful control experiments excluded the possibility that the 

 water or sails entered or passed from the body through the alimen- 

 tary canal, leaving as t/ie only probable alternative an osmotic 

 exchange through one or more of the external membranes. 



The alimentary canal, and indeed the whole abdominal viscera, 

 together with the washings from the body cavity, were found in 



1 This second figure is the mean which results when one very questionable deter- 

 mination is included. It is inserted for the sake of strict fairness. The other aver- 

 ages are in each case derived from all of the fishes tested. 



