RELATIONS TO OTHER SCIENCES 253 



vaporizing this to reach the lowest temperature yet attained, that of 

 the solidification of hydrogen, 252. 5, or 20.5 above absolute zero. 

 Thus the usable scale of temperature has been considerably lengthened. 



Less fortunate than Sir James Dewar, we have not succeeded, in 

 a long series of experiments made by use of the electric furnace, in 

 determining exactly the extreme limit of temperature reached. As 

 the outcome of some delicate experiments, M. Violle has assigned as 

 the boiling-point of carbon the temperature of 3500. But as we shall 

 prove further on, the temperature of the arc increases with the in- 

 tensity of the current, and the measurement of these high tempera- 

 tures requires further investigation. In order to fix the conditions 

 of our experiments, we carefully stated the voltage and amperage of 

 the current and the duration of each experiment. The diameter 

 of the electrodes and the capacity of the furnace had been deter- 

 mined beforehand, and remained constant. . 



At the very start, we found that at the temperature of our electric 

 furnace, the metallic oxides, hitherto regarded as irreducible, were 

 easily decomposed. Also reactions which were only partial at the 

 highest temperatures of ordinary furnaces became total here. A large 

 number of compound substances were dissociated at these high 

 temperatures, and on the other hand, new series of combinations, 

 definite and crystallized, were obtained. We thus prepared unknown 

 compounds of great stability, such as the carbides, borides, and sili- 

 cides. Most of these new binary compounds can also be partly or 

 wholly broken up by still further increasing the intensity of the 

 current and with it the temperature. 



Some of these carbides will furnish us a very definite scale of dis- 

 sociation. We also meet here, in the neighborhood of 3000, the 

 same general laws which govern the decomposition of substances 

 by heat at lower temperatures. Moreover, the boiling of a mixture 

 of copper and lead, of tin and lead, or of copper and tin, presents the 

 same peculiarities between 2000 and 3000 as does a mixture of 

 water and ether, of water and alcohol, or of water and formic acid, at 

 much lower temperatures. The laws of the fractional distillation of 

 two liquids apply therefore to the boiling of metals at very high 

 temperatures. 



In using our electric furnace, we operate in a reducing atmosphere, 

 and if a strong enough current is employed we get very quickly a 

 constant temperature, which is the boiling-point of quicklime. If, 

 on the contrary, the substance to be studied is put very close to the 

 arc, that is to say, very close to the gaseous conductor composed of 

 carbon vapor which unites the electrodes, the temperature rises 

 with the intensity of the current. A chemical reaction proves this. 

 With a current of 100 amperes and 50 volts, the reduction of titanic 

 acid by the carbon gives an oxide of an indigo blue color. With 



