378 MR T. C. BAILLIE ON THE 



from the disturbing effects of the burner on the one side and the escaping vapour on 

 the other. The arrangement is just that of a modified reflux condenser. 



The thermometers were suspended with their bulbs in the mercury of the heater 

 (at H in fig. 2), and the temperature of the heater was gradually raised to about 

 100° C, when the regulator was adjusted to act. The thermometers were left there 

 under these conditions from morning till evening. As readings were always taken in 

 the evenings, while the heater was set working in the mornings, the thermometers were 

 never read until they had been at the same temperature for several hours. It was there- 

 fore thought necessary to keep the thermometers the same length of time at 100° C. 

 before testing them at that temperature, so as to allow the glass to take on the same 

 set that it had in the bar at the same temperature. It is possible that if an ordinary 

 mercury thermometer is kept for hours at some temperature before it is read, 

 its reading at the same temperature on some other occasion will only be the same after 

 it has remained at that temperature for some hours. 



The burner D was lit, and after the testing apparatus had been at 100° C. for some 

 time, one of the thermometers was taken out of the mercury heater and quickly put into 

 tube A. After the first two or three occasions, it was found easy to do this so dexter- 

 ously that the reading on the thermometer did not fall more than 2° in the interval. 

 After it had been in A for some time, during which the reading was constant, it was 

 rapidly transferred to B, and by and by to C. The thermometers were hung up verti- 

 cally by means of a plumb line, and the readings taken with the telescope. It was 

 found that when E was too full of water, even when it was just over half-full, the read- 

 ings in A, B, and C were not alike. When that was the case, the thermometer was left 

 in one of the tubes until enough of the water had evaporated. The barometer was read 

 sometime during the test and the true temperature of the bulb of the thermometer found 

 from Regnault's tables. The difference between the observed reading on the ther- 

 mometer and the temperature of the water vapour gave the whole correction at that 

 temperature, the graduation correction and the stem exposure correction being thus 

 lumped together. The same thing was gone through for each of the thermometers. 



The same sort of process was repeated with the same apparatus, after the water bad 

 been dried out and naphthalene put in its place. Pure naphthalene was used, and as 

 the boiling point of pure naphthalene has been determined on the air thermometer scale 

 by Crafts, and has been found to be very constant, it is as satisfactory a " fixed " point 

 on the scale of temperatures as one can wish for. The total correction of each of the 

 t hermometers was thus found at the temperature of the boiling point of naphthalene. 

 The graduation corrections on the Kew thermometers used were known to be small, and 

 hence it was only to be expected that an expression of the form a + bt? would represent 

 the correction. This expression suited the values of the corrections found for all the 

 thermometers except one to within a fifth of a degree, but the value of b was not the 

 same in all cases, as it varied from -00008 to -000115. Curiously enough, b was smaller 

 for those thermometers graduated up to 300° C. than for those which could not read 



