The fact that sulphide of cadmium was always too light after the first hour’s heating in hydrogen sulphide proves that it must have contained some oxide of cadmium even after this heating. Oxide of cadmium is readily absorbed by the glaze on porcelain, and some error must have been introduced in this way because it would not be converted into sulphide after forming a silicate. The effect of this would be to give a low result for the atomic weight of cadmium. To get some idea of the magnitude of this error, the sulphide was poured out of the boats used in the first and second determinations. They were then warmed with nitric acid for a few moments, washed in water, and heated over the blast lamp for a few minutes. The boats used as tares were treated in exactly the same manner. On weighing, the boats in which the oxalate in determinations I and II had been decomposed, were found to be 1.12 milligrammes and .82 milligrammes heavier respectively than at the beginning of the determinations. This would only introduce an error of .03 of a unit in the atomic weight on account of the small difference in weight between these amounts of oxide and equivalent amounts of sulphide. The boats were warmed, as above mentioned, with nitric acid to remove any adhering sulphide. This might have decomposed some cadmium silicate at the same time, and the error due to cadmium oxide thus be found smaller than it really is. The following experiment was made in the hope of avoiding the formation of cadmium silicate. The glaze was removed from the inside of a porcelain boat by hydrofluoric acid followed by a thorough scouring with sand and water. The boat was then heated in the flame of a blast lamp for several minutes, tared against another boat which was not treated with hydrofluoric acid. Both were heated to redness in a current of hydrogen sulphide for an hour, cooled, weighed, and then heated in hydrogen sulphide for another hour, and weighed again. The boat gained 1.7 milligrammes during this second heating, showing that a boat whose glaze has been removed by hydrofluoric acid could not be used in this method. Throughout this work, great care was taken to exclude the oxygen of the air from the cadmium sulphide, while hot. The current of hydrogen sulphide in which the cadmium sulphide is heated must not be too slow, otherwise the sulphur in the dissociated gas will diffuse to the colder parts of the tube and condense, the residual gas becoming very rich in hydrogen. The hydrogen will then reduce some of the sulphide to metal, causing loss by volatilization. One determination was lost in this way, over two milligrammes of the sulphide being sublimed out, and it could easily be detected on the side of the tube. It is believed that the cause of the variations in the four determinations made by this method, is due to imperfect dehydration of the oxalate. It did not seem advisable to continue this part of the work any farther; therefore the