540 Prof. H. L. Callendar on a 



The objections commonly urged against the constant- 

 pressure type of gas-thermometer are that it deviates from 

 the theoretical scale, and that the accuracy of reading dimi- 

 nishes as the temperature rises. The first objection is pro- 

 bably groundless in the case of hydrogen and helium. For 

 other gases the correction is certainly larger, but also less 

 uncertain for the constant-pressure than it is for the constant- 

 volume gas-thermometer. The second objection is purely 

 imaginary. Although it is not possible to read the pressure 

 difference, or the expansion, to the same fraction of a degree 

 at 1000° C. as at 0° or 100° C, the relative accuracy as com- 

 pared with the fundamental interval diminishes very little, 

 and is many times greater at high temperatures than is 

 actually required, in consequence of uncertainties due to 

 other sources of error. With a bulb of 100 c.c. capacity and 

 a balance weighing to 1 mgm. on 400 gm., it is possible to 

 read to one part in 360,000 on the fundamental interval 

 (as the mass of mercury displaced would exceed 360 grms.), 

 which is a higher order of accuracy than can be attained by 

 pressure measurements. At 1000° C. 1 mgm. would corre- 

 spond to l/230th part of a degree, or a relative accuracy of 

 1 in 230,000 on the interval, which is more than ample. 

 With a constant-volume thermometer, assuming one hundredth 

 of a millimetre as the limit of accuracy of the measurement 

 and correction of the mercury columns, with an initial pres- 

 sure of one metre, it would be possible to read to one part in 

 36,000 only on the fundamental interval ; and it would be of 

 no advantage, even if it were possible, to be able to read to a 

 higher relative order of accuracy on the larger differences of 

 pi-essure. For work at 1000° C. it would be necessary to 

 adopt a much lower initial pressure, such as 273 mm., thus 

 reducing still further the limit of accuracy on the funda- 

 mental interval. 



From other points of view the constant-pressure thermo« 

 meter possesses certain practical advantages. (1) Since the 

 pressure is constant there is no strain of the bulb at high 

 temperatures, and the uncertain pressure correction is avoided. 

 (2) By the use of a standard atmosphere, consisting of a bulb 

 containing gas at constant pressure in melting ice, it is pos- 

 sible to avoid all reference to mercury columns, whether 

 barometer or manometer. The observations are much 

 simplified, and the uncertain temperature correction of the 

 mercury columns is avoided. (3) The operation of measuring 

 the expansion by weighing the mercury displaced is susceptible 

 of greater accuracy than the measurement of large differences 

 of pressure. There is the further advantage that the weighing 



