On the Practical Measurements of Temperature. 



233 



cut off and disposed in whatever way is most suitable to the require- 

 ments of the particular experiment. 



Comparisons were also made between the platinum pyrometer 

 and the air thermometer at constant mass and volume up to nearly 

 600° C. By enclosing the fine platinum spiral inside the bulb of the 

 air thermometer itself, errors of unequal heating were to a great 

 extent avoided. A modified form of air thermometer was devised for 

 the purpose, in which the air is confined at constant volume by sul- 

 phuric acid in an auxiliary gauge which is connected to the mercury 

 manometer by vnlcanised rubber tube. For the high temperature 

 observations the bulb was made of a combustion glass tubing which 

 does not soften till 700° C. ; the linear expansion of this kind of glass 

 was determined by a very accurate method, in which the mean 

 temperature of the glass tube is given by a platinum wire extending 

 down its axis. It was found that if kept for any time at about 

 500° C, or higher, the glass bulb is liable to capricious changes of 

 volume amounting to about 0*1 per cent, or more; by a slight and 

 convenient modification of the sulphuric acid gauge such changes 

 may be determined on the principle of the volumometer and allowed 

 for. By avoiding the permanent change the linear expansion could 



be represented by the formula 1 = 1 + "00000680 t + 27 . 10~ 9 t\ with 



a mean error of 0'2 per cent, per single observation. Among other 

 sources of error affecting the air thermometer at these temperatures 

 we may mention surface condensation. It seems certain that the 

 surface air film varies with the temperature, and when completely 

 removed by exhaustion at a high temperature does not recover its 

 normal state for some time, so that the zero pressure may decrease 

 upwards of a millimetre of mercury in a week. The investigation of 

 this is, however, still incomplete. 



The results of all the observations with three different instruments 

 are well represented by the formula 



R=e^ ' ' * (e) 



where t is temperature Centigrade by air thermometer, and the values 

 of the constants are 



oc=0 -0034259, (3=0 '0015290. 



Between 0° and 200° C. the curve (e) is very nearly a straight line, 

 R=l + 0-003460*, as it has a point of inflexion at about 80° C. The 

 mean deviation of an observation between 400° and 600° from this 

 curve is about 2'5° C, but it would be rash to use the formula (e) 

 outside the range covered by the experiments : it is for all reasons far 



