METHODS OF FLANKTON RESEARCH. 527 



pass through the filter, so that even the method with 

 pump and filter fails to give the smallest forms and also 

 the bacteria. 



This loss is easily seen by an investigation of the 

 filtrates from a hardened paper filter, which reveals the 

 fact that as much as 26 per cent, of the Gymnodineae 

 and the same per cent, of naked Chrysomonads can pass 

 through ; and a much greater percentage of bacteria 

 would be found to have done so. 



Of even greater importance than this loss is the fact 

 that many softened fragile forms are killed by the 

 filtration, and Monads, Amoebiclae, and small Grymnodinae 

 will be absent for that reason from the filter catches. More- 

 over, it is very difficult to recognise most of these forms 

 when fixed and preserved, and, therefore, for these forms 

 alone it is necessary to use other apparatus which does 

 not require any filtering mechanism and which will allow 

 of the organisms being studied in the living condition. 



The apparatus consists of (1) a means for procuring 

 samples of water from different depths, and (2) a centri- 

 fuge. By this method small water samples can be taken 

 and examined from the various parts of a water column 

 down to the greatest depths in the ocean, and from these, 

 by interpolation, the average number or volume of 

 organisms present in the complete vertical column can be 

 calculated. 



A " Kriimmel " water bottle (fig. 1) is the most 

 satisfactory for the purpose of obtaining water samples. 

 It is already made sufficiently large to bring up three 

 litres of water from any depth required, which is sufficient 

 to allow of a portion being used for the determination 

 of the hydrographical conditions — an absolute necessity 

 in plankton work. The water bottle is lowered open and 

 closed at the required depth by a falling weight, sent 



