with Boiling Point of Sulphur. 521 



ward in the turbine tube and downward past the bulb. With 

 this arrangement the top of the bulb was cooler. 



In an air bath it had not proved possible to reduce the vari- 

 ation of temperature over the surface of the bulb below 1°, so 

 that the liquid bath, even in this form, was an improvement over 

 the air bath. But still greater uniformity was desired. After 

 some experimenting, which need not be described here, the 

 arrangement shown in the diagram (tig. 1) was adopted. It 

 amounted, briefly, to one bath within another ; that is, the tube 

 containing the stirrer was continued across the bottom of the 

 tank and upward about the bulb, which it fitted with but little 

 clearance (10 mm ) in order to insure the very rapid circulation 

 of a thin layer of liquid past the bulb, while the remainder of 

 the bath remained at an approximately uniform temperature 

 without. With this arrangement no systematic temperature 

 differences greater than the errors of observation of the ther- 

 moelements (0*1°) were observed. 



3. Method of Procedure. 



In this apparatus temperatures were read simultaneously (1) 

 upon the gas thermometer, (2) upon three thermoelements dis- 

 tributed at different points in the bath, — one in a re-entrant tube 

 extending to the center of the bulb, and one each at the top 

 and bottom of the outside wall. 



For the measurements at the benzophenone boiling-point 

 thermoelements of copper-constantan and of platinum-platin- 

 rhodium (Heraeus) were used, for the higher temperatures 

 platinum-platinrhodium only. After a trial of the copper- 

 constantan elements at the next higher temperature (zinc), evi- 

 dence of permanent changes in their readings was obtained 

 which was more than sufficient to offset their increased sensi- 

 tiveness. They were accordingly abandoned in work at the 

 higher temperatures. 



A thermoelement suggested by Geibel,* of gold against an 

 alloy of 60 gold, 40 palladium, was also tried. This gives an 

 electromotive force about equal to that of copper-constantan 

 and over six times that of the platinum-platinrhodium element. 

 But the alloy wire proved so inhomogeneous that the accuracy 

 of the element was much less than that of the platinrhodium, 

 and it was accordingly rejected. 



Through the courteous cooperation of the Bureau of Stand- 

 ards, a sensitive resistance thermometer in charge of Drs. 

 Dickinson and Mueller of the Bureau was placed alongside the 

 bulb during a part of the measurements, and later a similar 

 instrument ingeniously constructed for the purpose by Dr. 



* Zs. anorg. Chem., lxix, 88-46, 1910. 



