120 CARRITT [CHAP. 5 



values for saturation from an atmosphere with a pressure of 760 mm Hg, actual 

 variations in atmospheric pressures over the sea surface may produce departures 

 of oxygen concentration from the standard atmosphere values as high as 0.4 ml 

 2 /l. It will be noted below that this effect can be at least partially determined 

 by measurement of the concentration of dissolved nitrogen or argon, which 

 undergo little or no chemical change when in solution. 



Procedures that involve standard solutions of potassium dichromate, 

 potassium iodate or potassium biniodate as the means of standardization for 

 the Winkler titration are in common use. Thompson and Robinson (1939) 

 recommended the use of potassium biniodate in preference to the other two 

 substances after a study of the accuracy and precision attainable with each 

 standard. Jacobsen, Robinson and Thompson (1950) adopted this recom- 

 mendation and gave detailed instructions for the entire Winkler procedure. 



Nevertheless Foyn (in litt.) has objected to the use of potassium biniodate on 

 the grounds that the variability in the composition of solid potassium biniodate 

 leads to measurable variation in the normality of the thiosulfate being 

 standardized. At least one commonly used procedure (U.S. Hydrog. Off. Pub. 

 No. 607) recommends the use of standard potassium dichromate. 



Armstrong (1959, unpublished MS) and Carritt (1959, unpublished MS) 

 showed that air oxidation of iodide to iodine during standardization procedures 

 can lead to low values for thiosulfate normality and so to low oxygen values 

 in the Winkler titration. The oxidation is a photochemical reaction that is 

 favored by high concentrations of iodide and of acid. The rate of reaction of 

 dichromate with iodide is increased at high acid and iodide concentrations, and 

 in some procedures increased speed has been sought by increasing the con- 

 centrations of these substances to the point where the photochemical air 

 oxidation becomes a significant source of error. 



Both iodate and biniodate react rapidly at acid and iodide concentrations 

 where the photochemical oxidation of iodide is insignificant. 



The use of an oxygen solution of known concentration obviously will eliminate 

 many of the problems encountered with the standards named above. A simple 

 saturation device has been constructed and tested and found to be an adequate 

 solution to the problem (Carritt, 1959, unpublished MS). The device consists of 

 a three-liter flask, which in use contains about a liter of water, arranged to be 

 slowly rotated by a motor-drive in a constant temperature water bath. The 

 speed of rotation is such that no cavitation occurs, yet a large area of con- 

 tinually renewed liquid film comes in contact with the atmosphere. Tests using 

 distilled water, for which the saturation concentration-temperature relation 

 appears to be accurately known, show that excellent agreement between 

 experimental and tabulated values can be obtained provided that temperature 

 is controlled or known to be constant to ± 0.05°C and that atmospheric pressure 

 has changed less than 1 mm Hg during equilibration. Tabulated saturation 

 values must be corrected for the difference between actual barometric pressure 

 and the standard 760 mm Hg atmospheric pressure. One hour of equilibration 

 produced saturated water starting with water 66% saturated. 



