78 BULLETIN Of THE BrREAU OF FISHERIES. 



water. It iimv l)<> ()l)jected, therefore, that any osmotic change occurring in this fi>li 

 M'hen pliiccil iti licsh water niu.st be classed as pathological. I do not think that llii- 

 conclusion is a necessary one, for the following reason: Both the gain and suhsi 

 (juent loss in weight occur long before any ill efi'ects are manifested, and if the tislicv 

 are removed in time to their normal medium no harm will have resulted. In utiv 

 case such objections do not apply to most of the other e.xperiments with this specir-. 

 nor do they apply to any of the experiments with the salmon, nor probabh' to eitlni 

 of those with the white perch. 



A few words are likewise necessary in reply to uuotlicr possible criticisui of my 

 interpretiitions. It maj' be asked, may not the increase or decrease in weight li" 

 due to water being taken in or passed out of the alimentary canal ^^ Have \m' 

 necessarily to do with osmotic phenomena at all? While the former possibilitj' has 

 not been absolutely excluded, and could not be excluded, save by ligaturing })otli 

 ends of the alimentary canal, I have very strong reasons for rejecting it. In tin' 

 first place, there is strong a priori improbability that the effects of osmotic action 

 should be so closely paralleled in any other way. Why should a tish from salt watn 

 swallow fresh or brackish water, and why should a tish from fresh or brackish water 

 pass out a part of its intestinal contents when placed in salt water? And \vli\ 

 should this latter initial loss be followed by a relative gain, as shown by experiment- 

 53 and 5.5 i 



Yet more detiuite evidence on this question is furnished by certain experiment ~ 

 described in the next section. It is there shown that salts pass from the l)ody of a 

 salt-water fish when placed in fresh water. It is likewise shown that the entiir 

 contents of the alimentarj' canal (indeed the entire viscera of the lish) are not 

 adequate to furnish as nuich salt as leaves the body in the course of a few honis. 

 It seems probable that the water enters or leaves the body along the same path as do 

 the salts, that path being through one or more of the bounding membranes. Other 

 experiments show that in some fishes, at least, the gills are the organs chietly 

 concerned. 



As before stated, the results of certain experiments at Woods Hole failed to 

 conform to those previously obtained. This is not true, however, of all these later 

 experiments, although as a whole the results are certainly not as clear-cut and 

 unvarying as in the New York series. The only really serious difficulties to br 

 encountered relate to the behavior of Fundnlus majalis. Description of these 

 experiments >vill accordingly be deferred until after the others have been considered. 

 Eight experiments with 7^. /ti-tiroc/ifus are includ(>d in this later series. 



Exferhnents 57-6'4. 



Woods Hole, July and August, litOo. Out of 4 experiments in which fishes 



were transferred to fresh water, an average gain of about 1 per cent (in i to 1 



da}') was shown in 3 cases; in the fourth case, however, a loss of 1.5 per cent 



is recorded. 



In 2 experiments in wiiich water of density l.dol was used, a gain of about 



1 per cent in each case was likewise recorded. 



Of 7 lots which were put into water of density 1.009, 3 showed a gain (in 1 



case as high as 2 per cent), and 3 a loss, while 1 remained stationary. These 



'1 This criticism has been made by Bottazzi and Enriques (1901) ol the weight determinationsof ^pIj/siobyQiiinton (I'.i 



