EDWIN B. POWERS 



311 



extreme variation from the mean is less than 33 per cent of the total 

 variation in the experiments. 



Duplicate oxygen tests could not be made by the methods em- 

 ployed since only 2 quarts of sea water were used in each experiment. 

 However, the accuracy of the method was determined in connection 

 with other work done at the same time. In 88 duplicate oxygen tests, 

 the average deviation from the mean was ± 0.024 cc. per liter. This 

 is more than 27 per cent of the average variation of the oxygen content 

 at end of experiment from the mean curve (Fig. 3). The extreme 



L2 



J. 1.0 

 t. 0.8 



i °-^ 



o 



0.4 

 0.2 



pHGS 6.4 6.6 6.8 7.0 7.2 7.4 7.6 7.8 8.0 8J2 84 8.6 



Fig. 3. The small circles represent the experimental data ot the herring 

 (Clupea pallasii Cuv. and Val.) when five fish were tested in a 2 quart mason 

 jar, and the large circles when ten to twenty-five fish were tested in a 42 liter 

 carboy. The abscissae give the hydrogen ion concentration; and the ordinates 

 the oxygen in cc. per liter at the end of the experiment. 



deviations of the 88 duplicate tests were one 0.094 cc. per liter and 

 four from 0.078 to 0.072 cc. per liter. This, again, is more than 33 per 

 cent of the extreme deviations of the experiments from the mean 

 curve. In other words, from 27 per cent to 33 per cent of the devia- 

 tion of the experiments are probably due to the method employed 

 and not the individual variation of the fish. Thus the average indi- 

 vidual variation of the fish is less than 8 per cent and the extreme is 

 less than 22 per cent of the difference of the total variations of the 

 experiments. Further justifications for the form of the curve will 

 be taken up in connection with the discussion of other experiments. 



