254 



a few instances when conditions of ice or vegetation necessi- 

 tated a temporary modification of the direction. 



Collections subsequent to May 20, 1896, were made by 

 means of the plankton pump, a known volume of water being 

 strained through the silk net. The water strained was taken 

 in such a way as to represent a vertical column of equal di- 

 mensions from bottom to top. This was accomplished by low- 

 ering the inlet of the hose from the intake of the pump to the 

 bottom, or as near it as we could go without fouling the water 

 by disturbing the unstable deposits, and raising it to the sur- 

 face during the progress of the pumping. To secure a perfect 

 column it is necessary to begin raising the hose with the first 

 strokes that deliver water to the net, and to arrive at the sur- 

 face long enough before the required amount is pumped to 

 allow surface water to reach the net. With a fixed hose 

 length and a known capacity of pump, this is easily deter- 

 mined by experiment. The volume strained varied with the 

 contents of the water. As a rule, one fourth of a cubic meter 

 was strained. When plankton was scanty and silt light the 

 quantity was doubled, and occasionally in excessive plankton 

 or unusual silt but half this volume was strained. Variations 

 of minor importance in the methods here noted will be men- 

 tioned in connection with the discussion of the plankton at 

 the several stations. These variations are such as were ne- 

 cessitated by difficulty of access with collecting apparatus, or 

 by the exigencies of flood, ice, and current 



PRESEKVATION AND MEASUREMENT. 



During the first three years the plankton was killed and 

 preserved in strong alcohol. In subsequent years 70 per cent, 

 alcohol to which formalin had been added to the grade of 2 

 per cent, was used, and proved to be a better preservative than 

 the strong alcohol. 



The quantity of plankton present in the catch was deter- 

 mined by compression in a Purdy centrifuge for two minutes 



