about 100 ¿tg. of carbon and were measured with a precision of ±11.4%. All samples weighed between 250 and 400 mg. Included in the above errors are those inherent in detection of the carbon dioxide produced. One major source of error is caused by uncertainty in the C02 background rate over the required 15-minute collection time. Normally in bench-top operation a sample can be analyzed in 5 to 6 minutes, but in this case the furnace is 25 feet from the analyzer. These errors can be reduced, particularly for bench-top application, by burning larger samples and minimizing the length of the gas line between the furnace and analyzer. However, it is considered sufficient proof that this furnace is directly applicable for the combustion of steels and refractory metals for carbon or sulfur analysis.
ACKNOWLEDGMENTThe author thanks Frank F. Felber, Jr., for his guidance in the project, and W. E. Hatch and R. A. Norton for their help in the mechanical development. Work performed for the U. S. Atomic Energy Commission under contract No. AT(30-l)-2789.