Nitrogen Thermometer from Zinc to Palladium. 97 



fixed points commonly in use, as given by various observers, 

 are due, not to differences in the properties of different gases 

 used, nor to differences in pressure, nor to differences between 

 the constant-volume and constant-pressure scales, all of which 

 have been frequently discussed from the theoretical stand- 

 point ; but to systematic errors residing in the apparatus and 

 the methods employed. A large portion of the present work 

 has therefore been devoted to finding out experimentally the 

 effect of variations in all those conditions which might affect 

 the results systematically. 



2. Apparatus. 



In all essential particulars the gas thermometer apparatus is 

 that developed by Day and Clement and already described by 

 them in detail (loc. cit.). It consists of four principal parts : 

 (1) bulb, (2) furnace, (3) furnace jacket, and (4) manometer. 



(1) The Bulb. — A great deal depends upon the material of 

 which the bulb is made. Primarily and obviously, the bulb 

 must be able to hold the expanding gas without absorbing or 

 losing any portion of it throughout the temperature range of 

 the measurements. 



A secondary requirement, the importance of which increases 

 rapidly when high accuracy is sought, is that it shall be possi- 

 ble to use several thermoelements in the furnace with the bulb 

 without their readings being endangered by contamination 

 from the bulb material. As long as this intermediary role of 

 the thermoelement remains a necessary one and alloys of 

 platinum continue to provide the only successful bulb material, 

 the contamination* of the platinum wire of the element by 

 the alloyed iridium or other platinum metal in the bulb w T ill 

 remain a serious consideration in all temperature measurement 

 above 900°. 



Although the platin-iridium bulb served well as a gas con- 

 tainer, the contaminating effect of the iridium upon the ther- 

 moelements made the life of the latter, for measurements of 

 such extreme accuracy, very short. Accordingly, at the close 

 of the first series of experiments, a change was made from the 

 platin-iridium bulb to one of platin-rhodium (80 parts platinum, 

 20 parts rhodium) 160 mm long and 4T mm in diameter. Inas- 

 much as one of the wires of the thermoelement itself contains 

 10 per cent of rhodium to which the platinum wire is always 

 exposed (and which gradually contaminates it, too, although 

 very slowly), it was thought that the substitution of a rhodium 

 alloy in the bulb might serve to retain the necessary qualities 

 of stiffness and regularity of expansion with a minimum of 

 disadvantage in the matter of contamination. These expecta- 



* For a detailed account of the behavior and treatment of contaminated 

 thermoelements, see Walter P. White, Phys. Rev., xxiii, 449, 1906. 



