87fi 



^12. We have tried to gain some more data on the thei-mic 

 behavionr of mixtures, whose composition was lying between that 

 of L/\Si(\ and Li J), and more partionhirly, with the aid of a 

 method, which possibly may open an important way in future, to get 

 results in those cases, where one of the com})onents is liighly volatile, 

 or changes in contact with the air. Therefore we shall describe it 

 shortly. The method is this, that crucibles are made tVom the purest 

 stock of platinum, like that used in drawing the wires of the thermo- 

 elements, which crucibles have the 

 shape and size, i-eproduced in fig 9a. 

 rt rt-Rhü A platinumrhodiumwire of athermo- 



element, about 0,6 m.m. thick, is 

 soldered to it at the bottom b}' 

 means of the oxygenflame ; then 

 the crucible is filled with the finely 

 powdered material, with the aid of 

 the small fumiel and hollow stem; 

 after cleaning it very well, a thick 

 platinumwire of the same diameter 

 as the stem is introduced into it, 

 and the crucible is melted off then 

 o j^ at the top. The platinumwire of the 



Fig. 9. used thermoelement, also 0.6 m.m. 



tldck, is soldered then with the oxygenflame at the top of the 

 crucible, and the other wire is bent as shown in the fig. 9/a Both 

 wires are isolated by means of porcelain capillary-tubes, and the 

 whole apparatus then fixed in a MvRQUAHDT-tube and hung in the 

 furnace. The crucible must be in the region of the furnace, where 

 no appreciable temperature-gradient is present. In the immediate neigh- 

 bourhood of the crucible a second thermoelement is fixed in position; 

 then the meltingpoint of the substance can be determined in the 

 usual manner, by reading both thermoelements alternately every 

 half minute. 



8ome experiments with Li„ Si O^ iniinediately proved the method to he 

 completely adapted f or our purposes -. we found the meltingpoint not only 

 as sharp as formerly, but also the undercooling- and solidification- 

 phenomena gave perfectly analogous results, as in the case, where 

 open crucibles were used. The meltingpoint of L^ ^i 0,, determined 

 in this way, -was found at an E.M.F. of 11968 M.V. ; as the cor- 

 rection of the thermoelement however was determined to be 

 —26 M.V., thus the temperature i.e. lies at 11942 M.V. or 1201° C. 

 We now tried first of all, to determine the meltingpoint of pure 



