HIGH TEMPERATURE MEASUREMENT 501 



these conditions may be roughly estimated from a knowledge 

 of the number of degrees in the thermometer stem, which are 

 not at bulb temperature, the average temperature of the ex- 

 posed portion, and the apparent coefficient of expansion of 

 mercury in the particular kind of glass used over the range 

 in question. An approximate value for the latter at moderate 

 temperatures is •00016. The determination of the " stem 

 correction " as it is called is, however, at best very uncertain, 

 and if the extent of exposed stem be great and the temperature 

 high, the correction may reach 30° C. Thus, for example, if the 

 bulb temperature be 300° and the amount of stem exposed be 

 200° at an average temperature of 50° C, the correction would 

 amount to 9° C. It will thus be seen that if a mercury ther- 

 mometer be intended to be used only partly immersed, either 

 it should be graduated to read correctly with some definite 

 amount of exposed stem or a correction must be applied. 



During the past few months experiments have been com- 

 menced with a view to the manufacture in England on a large 

 scale of mercury thermometers in which fused silica is substi- 

 tuted for glass, the difficulties in the way of the construction 

 of capillary tubes of reasonably uniform cross-section having 

 been largely overcome. Owing to the well-known excellent 

 properties of silica-glass at moderate temperatures these ther- 

 mometers, if produced sufficiently cheapl}^, should present 

 notable advantages. 



Our knowledge of high temperatures has been greatly ex- 

 tended during the past ten years by a number of researches, 

 among which may be mentioned those of Holborn and his 

 associates working at the Reichsanstalt at Charlottenburg, of 

 Day and his co-workers at the Geophysical Laboratory at 

 Washington, also the work of the American Bureau of Standards 

 at Washington and the National Physical Laboratory at Ted- 

 dington. These and other workers have carried gas-ther- 

 mometer measurements to much higher temperatures and b}'' 

 perfecting various forms of electric furnace to obtain the 

 necessary uniformity of heating, have succeeded in reaching 

 comparative agreement ^at temperatures up to i,ooo°C., the 

 upper limit of the experiments thus far made being i,soo°C. 



Although the gas-thermometer, as representing as nearly as 

 possible the absolute scale, has been accepted by general con- 

 sent as the ultimate standard of temperature measurement, to 



