11 



Fo. Moreover, when Fn is still being filled, F^ can be 

 replacée! by a third flask F3. In this way the water may 

 be caught in a consécutive séries of flasks, or, if neces- 

 .sary, also with regular intervais. After loosening the flasks 

 they were immediately supplied with the reagents and 

 then shut air-tight with a glass stopper. 



c. Détermination of the velocity of the watevcucrent. 

 We showed already in § 1, that overflow of fluid, from 



the flasks, can be coUected in the measuring cylinder C, 

 which enables us to measure the velocity of the watercur- 

 rent. By opening the valves entirely, about IQ à 80 cM^. 

 water per minute streamed through our apparatus. Generally 

 we measured the mean assimilation-rate during a period 

 of 4Vv à 5 minutes. In this space of time about 350 cM^. 

 streamed through the apparatus, when the valves were 

 opened entirely. Because of this we hâve ever since 

 expressed the velocity of the watercurrent by measuring 

 the time necessary for collecting 350 cM^. water, also 

 when the valves were not openend entirely. 



d. Détermination of the amount of oxygen. 



After having fixed the amount of COo in a sample of 

 the water from the réservoir B, which had not flowed 

 along the shoots, the (very small) amount of oxygen in 

 a similar sample had to be determined. After this the 

 water was guided along the Helodea-shoots by turning 

 the valves, and then the amount of dissolved oxygen had 

 to be determined from the water samples successively 

 collected. 



The amount of oxygen was determined according to 

 the method of Winkler^), titrating with a N/lOO 

 solution of NaoSo03 the quantity of jodium équivalent to 

 the oxygen dissolved in the water. Therefore in our 



^) See: Tiemann — Gartner: Handb. d. Unters u. Beurt. d. Wàsser. 

 S. 308. Braunschweig 1895. 



