Studies of the distribution of O-2 in air will inform us about critical problems in the global carbon cycle which are not readily accessed by other measurements, including the rate of seasonal net production in the oceans on a hemispheric scale, the rate at which the oceans are taking up anthropogenic CO2, and the net rate of change of the continental biomass. In this paper, we outline a method for measuring the O-1/N-2 ratio of air to a standard error of +/-6 per meg (+/-0.006 parts per thousand) for a sample analyzed in quadruplicate, corresponding to +/- 1.2 ppmV O-2 in air out of 210,000. The method involves measuring the ratio of O-16(2) to (NN)-N-15-N-14 by isotope ratio mass spectrometry. Potential and actual problems with this method include fractionation as sample and reference gases are introduced to the mass spectrometer, mass spectrometric nonlinearity, effects of imbalance of sample and reference ion currents on the measured isotopic ratio, isobaric interferences at masses 28 and 29 due to the formation of CO+ from CO2 in the source, and zero enrichments. We discuss the magnitude of errors introduced by these factors and outline the relevant corrections. The ultimate mass spectrometric uncertainty is about +/-2 per meg (+/-0.4 ppmV) for a 1 h instrumental analysis. Overall precision is currently limited by fractionation as sample and reference gases are introduced into the mass spectrometer. A considerable improvement in precision may be possible.
