and Standardizing of Mercurial Thermometers. 183 



where no sudden alteration in the bore sufficient to affect the 

 calibration-curve can be detected. It will often, though not 

 always, be found that the obstacle maybe passed by wrapping 

 round th^ tube a rag soaked in ether which is evaporating 

 rapidly. This same method of cooling is of great service in 

 help'ng the mercury to become detached from the, bulb, when 

 it is required to run it into the upper chamber of the in- 

 strument. 



The thermometers here mentioned were intended for calori- 

 metric purposes, and therefore an accurate knowledge of the 

 exact temperatures which they registered was not of so much 

 importance as a knowledge of the value of any given interval 

 on them, and that they should all be strictly concordant 

 among themselves. Instead, therefore, of being standardized 

 by a direct comparison with the natural standard 56117, they 

 were all compared with 55084, which embraced the ranges of 

 all of them, and which had previously been standardized by 

 means of the natural standard itself. Inasmuch as the scale 

 of 55084 was three and a half times as open as that of the 

 natural standard, a proportionate gain in concordance of the 

 instruments compared with it was in this manner obtained. 



The comparison was effected by suspending the two instru- 

 ments in a large bath of water which had been brought to the 

 required temperature, the surrounding atmosphere being like- 

 wise heated as nearly as possible (within o, 2 C.) to this same 

 temperature. After thorough agitation of the water, the 

 upper ends of the thermometers were tapped, and the height 

 of the columns in each read by means of a lens ; immediately 

 afterwards they were read again, reversing the order in which 

 they were taken. After tapping them again, they were again 

 read twice over as before : the water was then stirred, and a 

 second series of four readings of each instrument obtained in 

 the same way. The operation was repeated a second time; and 

 thus the readings at the points of comparison consisted in 

 reality of twelve separate readings of each instrument. The 

 points of comparison were the highest and lowest available 

 ones in the thermometer being standardized. During every 

 comparison the temperature of the bath was falling appreciably, 

 though very slightly. 



In the case of the two most delicate instruments, 62839 and 

 63616, three and two such comparisons were made respec- 

 tively, using different portions of the standard 55084 for each, 

 i. e. at different temperatures ; this being effected by removing 

 by heat some of the mercury in the thermometers to be stan- 

 dardized into the upper chamber. In the three comparisons 

 of 62839, the lowest point in the scale was thus made to 



2 



