98 A. L. Day and H. B. Sosman — 



tions have been completely realized. Although the rhodium 

 alloy is less rigid at temperatures of 1000° and beyond thau 

 the iridium alloy, and requires more careful adjustment for 

 equal pressure within and without, no sagging of the bulb 

 walls or deformation from the gas pressure has appeared up 

 to 1550°. Meanwhile, the contamination of the thermoele- 

 ments in the presence of the rhodium alloy is now reduced in 

 magnitude about 80 per cent for a given temperature and time 

 of exposure. 



(2) The Furnace. — The common practice of recent observ- 

 ers (Callendar, Harker, Holborn and Day, Jacquerod and 

 Perrot, Day and Clement) has been to use a cylindrical bulb 

 in which the length was three or four times the diameter, 

 enclosed in a concentric furnace tube (air bath) heated by the 

 electrical resistance of a coil of wire wound upon or within 

 it. To this bulb the heat is applied radially over its cylindri- 

 cal surface, but no heat is supplied at the ends. The furnace 

 tube itself and the winding of the coils have been changed at 

 different times and in a variety of ways in order to vary the 

 distribution of the heat supply. The arrangement used in 

 most of our experiments consisted of one main coil of platinum 

 wire l*2 mm in diameter, wound on the inside of a refractory 

 magnesia tube 36 cm long and 2 ctn thick. As has been our 

 habit for some years, the windings near the ends of the coil 

 are somewhat closer together than those at the middle, but 

 this device is not of itself sufficient* to compensate for dif- 

 ferences of temperature along the bulb at all temperatures. 

 In a particular case a favorable arrangement will provide an 

 almost perfect temperature distribution at 500°, but will over- 

 compensate the ends at 1000° sufficiently to spoil the meas- 

 urements. The conductivity of the bulb metal is wholly 

 inadequate to help out this overcompensation by conveying 

 surplus heat from the ends to the middle of the bulb. On the 

 other hand, a change in the winding which will correct the 

 overcompensation at 1000° provides insufficient compensation 

 at 500°. The arrangement which has become usual with us 

 is therefore to wind the coil somewhat more closely at the ends 

 than in the middle, with the idea of providing partial com- 

 pensation for the inevitable heat losses at the ends of the fur- 

 nace in this way, and in addition, to insert supplementary coils 

 of smaller wire in the ends of the furnace tube in order to 

 provide a small, independently regulated heat supply which 

 can be superposed upon that of the main coil and give the 

 desired uniformity at any temperature likely to be employed. 

 A furnace tube arranged in this way, except for accidental 

 variations, caused, for example, by the flaking off of the fur- 

 *Day and Clement, loe. cit., p. 411. 



