288 Messrs. C. H. Burgess and A. Holt, Junr. [Oct. '27, 



melted. The bead was then inserted through the opening D, when on 

 touching the wire it became fused. The current was switched off, and 

 the bead, on solidifying, remained attached to the edge of the wire AA. 

 Air currents were prevented by closing the opening D with a test-tube. 



A slowly increasing current was then passed through the wire till 

 the portion of the bead in contact with it melted, when the weight and 

 bead fell into the test-tube. The moment this occurred, the voltage 

 between the ends of the wire and the amperes passing through it were 

 carefully noted. From the values obtained the resistance was calcu- 

 lated, and thence the temperature. 



The weight hanging from the bead had no decided effect on 

 the melting point. We found that with weights varrying from 

 0*2 5 grammes, the bead dropped off at practically identical tem- 

 peratures. This temperature at which beads of various compounds 

 drop off the wire is extremely well denned, even in the case of 

 substances which become plastic on heating, like glass, so that provided 

 no chemical change occurs during the heating, the method proved 

 rapid and elegant. A certain amount of heat is conducted away by 

 the ends of the thick platinum wires BB to which the thinner one AA 

 is welded, and also by the bead itself. It, therefore, became necessary 

 to calibrate the wire by means of pure salts of known melting points. 



We found this was a matter of some difficulty, as very many of the 

 most ordinary salts were found to decompose to some extent on 

 heating. Sodium chloride, sodium carbonate, potassium iodide, 

 calcium chloride, and strontium chloride all underwent change. Of 

 these salts sodium chloride was very little attacked, but after only a 

 few minutes' fusion we found it had absorbed quite an appreciable 

 amount of oxygen, being converted into the peroxide. Potassium 

 iodide changed in part to iodate, sodium carbonate to the peroxide, 

 and calcium and strontium chlorides to the oxides. 



These changes completely prevented any accurate melting point 

 determinations, so after repeated trials we adopted potassium nitrate, 

 lithium chloride, and potassium chloride as standards. It is true that 

 potassium nitrate decomposes on heating, but as the nitrite (to which 

 it changes) has almost the same melting point, the small amount of 

 change was found to be immaterial. The chlorides we employed 

 seemed to be, on the whole, very stable. 



The melting points given in Carnelley's tables for these three standard 

 substances were assumed as correct; they were used in fixing three 

 points on the temperature-resistance curve. 



Thus, any temperature between the melting points of potassium 

 nitrate and potassium chloride could be obtained by interpolation from 

 the measured resistance, and those above the melting point of potassium 

 chloride by extrapolation. The results did not seem to contain an 

 error larger than 1 per cent., so that we considered the method was 



