264 



The coefficient has been computed in each case on the ba- 

 sis of measurements by the gravity method, by the centrifuge, 

 and by enumeration of all the larger and quantitatively more 

 important constituents of the catch. An examination of the 

 table will indicate that the relation and direction of the dif- 

 ferences of the various coefficients do not materially differ by 

 the three methods. The results by the enumeration method 

 give the largest coefficient — probably as a result of the elimina- 

 tion of the silt factor in some instances, and possibly by reason 

 of the large margin of error involved in the method. 



It is evident from the table that an average of a number 

 of catches, not only by the net but also with the pump, should 

 be used if empirical coefficients are to be established with ac- 

 curacy. It is probable that the low coefficients seen in a few 

 instances result from insufficient pump catches, or from some 

 error in paralleling the catches. Since the coefficient problem 

 was eliminated in our later work by the use of the pump, fur- 

 ther efforts to establish empirical coefficients were abandoned 

 for lack of time to carrv on more elaborate tests. 



Three alternatives were thus before us. First, to adopt 

 the coefficient computed according to Hensen's formula, and 

 use this one factor, 1.32, for all catches irrespective of the age 

 of the net and of seasonal, local, quantitative, and qualitative 

 differences in the catch. This method Apstein ('96) and other 

 European planktologists have adopted. Reighard ( '94), Ward 

 ('95), and Juday ('97) have also followed this plan, but in each 

 case they were dealing only with catches taken in midsummer 

 from the same or similar bodies of water, and the resulting 

 error thus introduced was much less than would result from 

 the adoption of a uniform coefficient for our varied catches. 

 Furthermore, we had the evidence of the probable extent of 

 this errror which the pumping method afforded. 



A second alternative was to ignore the coefficient question 

 entirely ; but this involves even greater distortion of the prob- 

 able seasonal and local fluctuations in the plankton. 



A third method, and the one finally adopted, was that of 



