328 MORRIS M. WELLS. 



measurable difference in gas concentrations, as compared with 

 samples taken at the exit tube of bottle A, and it is not probable 

 that the salt content was appreciably affected by the fishes. 



The method of analysis that has been used throughout is a 

 slight modification of the standard methods of the Public Health 

 Association. 1 In all cases, the collections of samples were made 

 in narrow-necked bottles, through which the water was allowed 

 to run for several minutes, with the delivery tube reaching to the 

 bottom of the bottle. The bottles were kept corked as much as 

 possible until after all the chemicals had been added. These 

 precautions were essential, for in the case of low oxygen a sample 

 may show 2-3 c.c. per liter, if collected and titrated in an open 

 250 c.c. graduated cylinder, while if tested as above, it may be 

 found to contain but from .I-.I5 c.c. per liter. For the carbon 

 dioxide tests, a special bottle was devised. With this bottle, 

 the cork was removed only to run in the sodium carbonate in 

 titrating. This was always done immediately after the collection 

 was made. 



The water flowed through the bottles at the rate of from 

 500 to 600 c.c. per minute. This changed the water in the 

 larger experimental bottle, once in about 15 min. and in the 

 smaller once in 6 min. In both experimental bottles, the 

 introducing tube led to the bottom, while the exit tube reached 

 just to the under side of the cork. Finally, in order that the 

 temperature of the water might not vary in the experimental 

 bottles, a stream of cold tap water was kept running over each 

 throughout the experiment. The number of fishes used in an 

 experiment varied with their size. A smaller number was 

 usually placed in the smaller bottle. The number in any single 

 experiment varied from 2 to 10 and about 160 were killed, in all. 

 About 25 experiments were performed; the shortest occupied 

 from two to three hours and the longest five to seven days. 



The fishes used in the experiments were taken in one of the 

 small creeks near Chicago and were for the most part experi- 

 mented upon before they had been in the laboratory aquaria for 

 more than two or three days. In no case had they been in 

 captivity for longer than three weeks. In the laboratory they 



1 See Jour. Infectious Diseases, '05, supplement No. i. 



