National Bureau of Standards BurntRefractory, sample no. 77, consistently gave lower results than theoretical. Fluoride could not be detected in this sample, eliminating one possible explanation for the oxygen deficiency. This sample was later held at 500°C. for 3 hours in pure oxygen to determine if it might have been in a lower state of oxidation. After this treatment the oxygen analysis averaged 49.0% as compared with 46.4% in the original sample.This oxidation, with the data in Table V, has illustrated the ability of the method to descriminate between oxidized and reduced glasses.