308 henry b. ward: 



was true if Cyanophyceae were present, as these are not 

 thoroughly precipitated owing to low specific gravity. The 

 material is also matted together, preventing equal distribution 

 on the slide if numerical estimation is to be employed. Kofoid 

 (97a) emphasizes the selective error of the centrifuge on living 

 plankton. 



In microscopical water analysis for technical purposes the 

 Sedgwick-Rafter method almost universally employed has been 

 subject to modification in detail by Jackson (96, 98), while 

 Whipple (96) has analyzed most clearly the various errors of 

 the method and the value of each. The same author has also 

 (97) planned a simple form of apparatus for water analysis. 

 Leeds gives a valuable discussion and ' summary of these 

 methods. 



In the filtration of plankton organisms Kofoid (97a) found 

 that the sand filter retained only 40 to 65 per cent of the num- 

 ber of organisms present and .advocated as more satisfactory and 

 precise the Berkefeld filter. Reighard (98) objects to the con- 

 tamination of the plankton resulting from the use of the latter, 

 and Jackson (98) considers that the slow rate of filtration 

 makes its use entirely unpractical. 



For the manipulations connected with the enumeration of 

 individuals in plankton hauls various minor pieces of apparatus 

 have been suggested ; only the more important need be noticed 

 here. Whipple (94b) advised the employment of an ocular with 

 a field suitably ruled, and Zacharias (96a) introduced an ocular 

 of large field with an iris diaphragm. As the enumeration of 

 organisms recorded without reference to size and character is 

 extremely misleading, Whipple (94b) proposed a standard unit 

 of size, 20x20 microns, as a means of correcting the error. 

 Tables for common organisms and an ocular with ruled field 

 assist in the computation. Comparison of lines platted to show 

 the numerical and areal values of the organisms in a haul with 

 the albuminoid ammonia curve for the same demonstrate the 

 much closer correspondence of the areal estimation with the 

 amount of organic substance present. By the use of logarith- 



