248 Mr. H. L. Callendar. [Dec. JO, 



tare," * I described in detail a somewhat elaborate form of air ther- 

 mometer with which it was found possible to attain an accuracy of 

 ihe order of 0'01 C. I have since succeeded in overcoming some of 

 the difficulties encountered in that investigation, and in evolving on 

 similar lines a form of instrument which is capable of a much higher 

 order of accuracy, and which has the farther advantage that both the 

 observations and the calculations are immensely simplified. 



The standard instrument for measuring temperature selected by 

 Regnault in his classical researches was the constant-volume air 

 thermometer. In my earlier experiments I employed air thermo- 

 meters of this type, but modified them by the introduction of a 

 sulphuric acid gauge of small bore between the thermometric bulb 

 and the mercury manometer. I was thus enabled to reduce the 

 correction to be applied for the small volume of air which was not 

 exposed to the temperature to be measured, and at the same time to 

 observe small variations of temperature with greater accuracy. 



The constant-volume type of air thermometer, however, has 

 several disadvantages. The degree of accuracy attainable depends 

 primarily on the exact measurement of pressure by means of a 

 mercury manometer. The observations involved are slow and 

 laborious, and it is difficult, unless the temperature is absolutely 

 steady, to secure an accuracy of the order of a tenth of a degree C. 

 At high temperatures this method has the further disadvantage that 

 the bulb is exposed to variations of pressure, the effect of which in 

 altering its volume cannot be accurately estimated. 



For these and other reasons I soon abandoned the constant- volume 

 air thermometer in favour of the constant-pressure type. The bulb 

 may thus be entirely freed from stress at high temperatures, and the 

 mercury manometer may be dispensed with. The auxiliary reservoir 

 containing mercury into which the air is allowed to dilate may be 

 kept permanently in melting ice, and the volume representing the 

 dilatation of the air may be determined by weighing the mercury 

 displaced. This observation may be made at leisure, and admits of 

 very considerable accuracy. 



With the form of instrument described in the previous paper, it 

 was still necessary to read the barometer. These readings were 

 found to be by far the greatest source of uncertainty. I was so much 

 impressed with this in the course of some experiments on the boiling- 

 point of sulphurf that I determined to construct an instrument 

 which should be altogether independent of the measurement of 

 mercury columns. 



If the pressure of the air enclosed, instead of being adjusted to 

 equality with that of the atmosphere, be adjusted always to the same 

 * ' Phil. Trans.,' A, 1887. 

 f ' Phil. Trans.,' A, 1891, p. 130. 



