Nitrogen Thermometer from Zinc to Palladium. 99 



nace lining, affords uniform temperature distribution over a 

 length of 20 cm in the center of the tube for a range of tem- 

 perature from 300° to 1550°, and no one temperature is more 

 difficult to regulate than another. This arrangement contains 

 a limitation, however, of considerably greater magnitude than 

 was at first suspected. The ends of the bulb face the com- 

 paratively cold ends of the furnace tube and radiate a sufficient 

 quantity of heat toward these cold ends to reduce the tem- 

 perature of the end surfaces of the bulb some 6 or 8° below 

 the mean temperature of the cylindrical surface. 



In so far as this may appear to be a rather obvious condition 

 to be overlooked, it may be remarked parenthetically that it 

 is a common practice of physicists to assume that the tempera- 

 ture is constant over a radial cross-section near the center of a 

 cylindrical furnace which is reasonably long in comparison 

 with its diameter. With this in mind, the probability is even 

 greater that a metallic conductor only 4 cm in diameter (the end 

 surface of the bulb) perpendicular to the axis in such a furnace 

 will have a uniform temperature between its center and peri- 

 phery. The fact is that neither of these assumptions is justi- 

 fied, even in furnaces as long as twenty times the diameter. 

 This was shown in a number of actual measurements made 

 under varied conditions, differences of several tenths of a 

 degree being found as low as 300°, and of several degrees at 

 1000° and higher. 



This situation demonstrates the futility of depending upon 

 metallic conductivity (of platinum) to equalize a steep tem- 

 perature gradient, and again emphasizes the fact, if further 

 emphasis is necessary, that the air bath, or, more explicitly, 

 the temperature distribution within the heating chamber, is 

 the most uncertain factor remaining in gas thermometry. 



On account of difficulties in manipulation and accidental 

 leakage into the thermoelement system, we preferred not to 

 introduce additional heating coils into the furnace tube, and 

 accordingly undertook to stop the loss of heat by inserting 

 thin, platinum - covered diaframs opposite the ends of the 

 bulb. The situation was still further safeguarded, in exchang- 

 ing the platin-iridium for the platin-rhodium bulb by adding 

 a reentrant tube at the lower end of the bulb, to enable us to 

 measure the actual temperature prevailing at its center as well 

 as over the surface. In this way, we thought to obtain a more 

 representative integral of the surface temperature and a com- 

 petent comparison of this integral with the temperature 

 actually prevailing at the center of the bulb. 



(3) The Furnace Jacket. — The furnace jacket was water- 

 cooled and could be closed air-tight around the furnace and 

 bulb together, so that the pressure could be maintained the 

 same within and without the bulb to avoid deformation. 



