Crucibles 1 and 2 having been brought to constant weight against their tare, were ready for use. The piece of cadmium was weighed and placed in 1. An excess of pure nitric acid was added and a gentle heat applied until all the metal had dissolved. This required from twenty to forty hours. A sand-bath was constructed by placing a large porcelain crucible in an iron crucible and filling the intervening space with sand. The pair of crucibles (1 and 2) was placed in the porcelain crucible and the contents evaporated to dryness by warming very carefully at first and gradually increasing the temperature. The pair of crucibles was then transferred to a bath constructed as the above where iron filings took the place of sand. This was heated by a single burner until the nitrate was all decomposed when a triple burner was added and finally two for six or eight hours. This was not sufficient to effect complete decomposition. When cold, the pair of crucibles was placed in the nickel crucible as represented in fig. 3 and sharply heated over a blast-lamp for several hours. This completed the decomposition of the nitrate and the removal of the last traces of oxides of nitrogen. During the blasting the lid on crucible 3 was raised a little to one side to allow free access of air. The nickel crucible was forced tightly into a hole cut in the center of an asbestos board about ten inches in diameter, to prevent any reducing gases from the lamp entering the crucibles while hot. This was the same arrangement as was used by Partridge[4]. [4] Amer. Journ. Science XL, 379. It was found that the final decomposition of the nitrate could not be effected in a muffle furnace as with zinc, since at very high temperatures cadmium oxide attacked the porcelain with great energy and injured the crucibles. The decomposition of the nitrate was shown to be complete not by constant weight alone, but by testing for oxides of nitrogen with starch paste rendered extremely sensitive with potassium iodide. That the test should be reliable, Morse and Burton have pointed out that all the reagents used must be free from oxidizing agents. The presence of iodate in the iodide is especially to be avoided. This was removed by boiling the solution with zinc amalgam. Air was removed from all the solutions by boiling. When the starch-potassium-iodide solution had been prepared as sensitive as possible, a portion of it was treated with a little hydrochloric acid, to determine if any iodine was liberated. If no coloration was observed the cadmium oxide was added. It dissolved in the hydrochloric acid and if any oxides of nitrogen were present they would have revealed themselves by the liberation of iodine and a blue coloration of the starch paste. In no one of the ten determinations was the slightest coloration detected.