No. 497] THE BUREAU OF FISHERIES 323 



from this cause alone. In the display aquaria the fishes 

 had fared pretty well, thanks to the use of perforated 

 pans, which were suspended several feet above each tank. 

 The water was in no case allowed to enter below the 

 surface, as experience had shown that the result would 

 have been disastrous. The deaeration apparatus was 

 first put into use on August 31, and two days later the 

 pans over these aquaria were removed, a full head of 

 water being allowed to enter below the surface in each. 

 This condition was maintained until September 23. Dur- 

 ing this period of twenty-one days no deaths occurred 

 beyond those occasional ones to be expected in a large 

 miscellaneous collection of living fish. No symptoms of 

 the gas disease were noted, either in the presence of sur- 

 face "blebs," or in the occurrence of "pop-eyed" fishes. 

 The collection seemed in particularly good condition, and 

 the absence of the splashing water overhead made it 

 possible to keep the premises much more dry and neat. 



The natural inference from these facts was that the 

 deaerator had fulfilled its requirements perfectly.. But 

 before forming definite conclusions on this point, it was 

 necessary to perform the control experiment of throwing 

 the apparatus out of the circuit, and using the salt water 

 just as delivered by the pump. This was done on the 

 afternoon of September 23. The water supplying the 

 aquaria was allowed to enter below the surface as before. 

 On the following day (September 24) bubbles were noted 

 upon the surface of some of the fishes, either in the mucus 

 or beneath the epidermis. From records made through- 

 out the ensuing twelve days, it appeared that out of 85 

 fishes, belonging to 18 species, which were present at the 

 time the process of deaeration was discontinued, 13, or 

 more than 15 per cent., died within the ensuing six days; 

 while within the first twelve days 15 others, or nearly 

 18 per cent., manifested more or less marked symptoms 

 of the gas disease. Almost exactly one third of the 

 stock had accordingly been killed or obviously injured. 



