With the modified method of computation discussed in detail on page 88
it is seen that such exceedingly exact analyses of oxygen as were
formerly made are unnecessary, and further calculation is consequently
very simple if we know the percentage of nitrogen to within a fraction
of 1 per cent. We have used a Haldane gas-analysis apparatus for
analyzing the oxygen, although the construction of the apparatus is such
that this presents some little difficulty. It is necessary, for
example, to accurately measure about 16 cubic centimeters of pure
nitrogen, pass it into the potassium pyrogallate pipette, and then
(having taken a definite sample of oxygen) gradually absorb the oxygen
in the potassium pyrogallate and measure subsequently the accumulated
nitrogen. The analysis is tedious and not particularly satisfactory.
Having checked the manufacturer's analysis of a number of cylinders of
oxygen and invariably found them to agree with our results, we are at
present using the manufacturer's guaranteed analysis. If there was a
very considerable error in the gas analysis, amounting even to 1 per
cent, the results during short experiments would hardly be affected.