138 DISCOVERY REPORTS 



least interference. Experiments on the determination of oxygen in dilute dinoflagellate cultures were 

 made at the Marine Biological Association's Laboratory at Plymouth before the cruise. Those demon- 

 strated that the principal interference took place through absorption of iodine by the dinoflagellates. 

 Samples titrated immediately after acidification and liberation of the iodine showed a higher oxygen 

 content than those titrated 2 hr. after acidification. In this period the mean absorption of iodine 

 was equivalent to 0-4 ml. of oxygen per litre at 15 C. in 3 hr. No difference was observed in samples 

 stored in alkaline-precipitated condition for periods up to 10 hr. It was decided, therefore, to store 

 samples in the latter condition. 



Allee and Oesting (1934) found that quantities of nitrite of 071 mg. atoms N/m. 3 and over were 

 sufficient to affect the Winkler method. The interference occurs through the nitrite's liberating iodine 

 from the potassium iodide and giving a greater quantity of iodine than would normally be released 

 through the acidification of the manganic hydroxide precipitate in the presence of potassium iodide. 

 The error is, of course, to overestimate the amount of oxygen present. 071 mg. atoms N/m. 3 is quite 

 a high figure for the sea, but as no nitrite estimations were made the extent of any error likely to arise 

 was unknown. The modified Winkler method used to deal with hydrogen sulphide, however, also 

 removes any interference due to nitrite. 



Some of the sediments on the sea-bed in the region of Walvis Bay are populated by sulphate- 

 reducing bacteria, whose activity results in the production of hydrogen sulphide. This gas, liberated 

 into the overlying water, appears to coexist, in a dynamic state, with the dissolved oxygen (Durov and 

 Turzhova, 1947). To obtain a true estimation of the oxygen content at a particular time and position 

 it was necessary, therefore, to put an end to the oxidation-reduction reaction as soon as possible after 

 sampling. Although Alsterberg's method (1926) would have been preferable we tried to accomplish 

 this by using the Rideal-Stewart modification of Winkler's method, which could conveniently be 

 carried out with the chemicals available on board. The method consists of a preliminary oxidation, 

 carried out by treating the acidified water sample with potassium permanganate. When the reduction 

 is complete any remaining permanganate is destroyed by the addition of potassium oxalate, after which 

 the normal Winkler method is continued. 



Parallel determinations on samples from Walvis Bay with both the Rideal-Stewart and unmodified 

 Winkler methods gave the results shown in Table 1. 



Table 1. Dissolved oxygen content of water-samples from Walvis Bay 



Sample no. 



1 

 2 



3 

 4 

 5 

 6 



7 

 8 



Oxygen determinations on water-samples taken in Walvis Bay using both the Rideal-Stewart and unmodified Winkler 

 methods. The samples have been arranged in order of concentration, and they were all analysed at the same intervals after 

 collection. 



A further test was carried out on four similar samples with the unmodified Winkler method. Two 

 of the samples were analysed immediately after collection, while two were kept for 4 hr. before 

 analysis. The first two showed respectively 041 and 0-39 cc. 2 /l. more than the second two. If this 



