High Temperatures. 189 



because the clay was too porous to hold the melted metal in all 

 cases (molten Cu for instance penetrated the wall and short- 

 circuited a section of the coil), and second, the change from one 

 crucible to another which happened nearly every day was much 

 more convenient with the coil wound upon a separate tube. 

 This oven consumed about 600 watts at 1000° 



Some caution was necessary in heating up to the highest 

 melting point (copper), otherwise the oven coil (l mra in diameter) 

 burned out — its renewal is a simple matter however. 



For the temperature measurement thermo-elements were 

 used which were cut from the same wire as T 2 and which have 

 shown themselves to be perfectly identical with it in their 

 results. Temperatures above 500° are derived from the normal 

 curve (Table XVI) and below from the measurements in the 

 nitre bath. 



In each determination the thermo-electric force was observed 

 from minute to minute and afterward plotted as a function of 

 the time (mentioned hereafter as the "time curve"). A 

 change of 10 microvolts could be compensated directly on the 

 compensation apparatus by inserting 0*1 ohm. Smaller differ- 

 ences were interpolated from the galvanometer deflection, one 

 scale division corresponding to from 0*5 to 1"4 microvolts 

 according to the resistance in the branch circuit. 



The accuracy of the crucible method is only limited by the 

 accuracy with which it is possible to measure with the thermo- 

 element, in regard to which more will be said further on. The 

 wire method is less exact because the interruption of the cir- 

 cuit which indicates the melting point is not independent of 

 any slight tension which may exist in the wires of the thermo- 

 element, particularly the platin-rhodium wire — a factor which 

 varies from one observation to another. 



The highest temperatures observed for a particular metal by 

 this method may therefore be expected to come nearest the 

 truth. The inserted wire may even at times be seen to break 

 just before the real melting takes place. 



Gold. — The melting point of gold was determined by the 

 wire method only. Two different specimens of gold had been 

 placed at our disposal., one (I) from the Gold und Silber Schei- 

 deanstalt in Frankfurt-am-Main, the other (II) specially pre- 

 pared and purified in the Eeichsanstalt. No difference could 

 be detected in their melting temperature. 



The observations on the first two days (following table) were 

 made in the gas thermometer oven, the others in the small 

 oven used for the determinations by the crucible method (fig. 3) 

 in which an empty crucible had been placed. 



The arrangement of the thermo-element in the latter case 

 was the same as above described for the crucible method with 

 the omission of the porcelain protecting tube. The heating 



