Lerts anp Braxe—Estimation of Oxygen in Solution. 455 
difference being only 0°28 cc. of oxygen per litre of water, 
showing that the method gives good results with ordinary 
drinking waters.” 
With Thames water at Richmond the greatest difference was 
0-98 e.c. per litre, the least 0:07, and the mean 0°56, while with. 
Thames water opposite the southern sewage out-fall, worse results 
were obtained, the mean difference being 0°63 c.c. 
* Thresh’s method,! in the hands of its inventor, appears to have. 
given exceedingly accurate results. The test experiments were’ 
chiefly made with distilled water saturated with air at 10°, 15°,. 
20°, and 25°C., respectively, and the results compared with the 
gasometric determinations made by Roscoe and Lunt at the same 
temperatures. The greatest mean difference between the two sets: 
of figures amounted to 0-09 c.c. of oxygen per litre of water, and 
the least to 0°02. Thresh makes the following remark :—“ The 
results, however, can be made to vary, the extreme limit being 
0-5 mgs. (0°31 c.c.) of oxygen per litre of water, using 250 c.c. 
for the estimation.” 
One of the authors of this Paper has employed this method on 
several occasions, but although every care was taken, and all the 
precautions mentioned in Thresh’s paper carefully observed, the 
results were not very satisfactory, and in the case of sea-water 
polluted with sewage they were definitely unsatisfactory. With 
distilled water, a variation of 0-4 c.c. of oxygen per litre from the 
true amount was found several times; while in the case of the. 
polluted sea-water, some of the samples, when kept in stoppered 
bottles filled to the brim and immersed in water, actually showed. 
a gain in dissolved oxygen when kept for twelve days, amounting 
in one of the samples to more than 0°6 ¢.c. per litre, although 
there cannot be the slightest doubt that a decrease had really 
occurred. As an important technical question was involved in 
these determinations, and interests of considerable magnitude 
were at stake, the determinations were practically valueless. 
In the two volumetric processes mentioned above, and appa- 
rently indeed in all similar methods for determining dissolved 
oxygen, the titration is performed in a vessel through which a 
current of hydrogen or other inert gas is made to pass, which not 
1 Chem. Soc. Journ. ‘Trans. 57 [1890], p. 185. 
