492 



Messrs. Stewart and Loewy on the true 



[June 17, 



mometers was precisely the same as that used in our previous experiments 

 and described in the account we gave of them to the Royal Society. The 

 formula employed for deducing the most probable mean temperature for 

 each experiment was as before, using the same symbols : — 



.(ai>.f±g)+*f±g) .... 



n + n +n 



As explained in the paper mentioned above, we had then no time at our 

 disposal to swing the pendulums used at extremes of temperature, and 

 hence we availed ourselves of the elaborate series of experiments on the 

 temperature-correction of pendulums made by General Sabine (vide Phil. 

 Trans. 1830, p. 251), adopting the mean of his entire results, viz. 0*435 

 vibrations per diem, as correction for 1° of Fahrenheit's scale for our 

 reductions. 



In our present investigation we thought it indispensable to obtain the 

 utmost accuracy, by ascertaining the temperature-correction for each pen- 

 dulum intended to be used by an independent series of experiments in an 

 artificially heated room, the natural annual range of temperature in the 

 Kew pendulum-room being insufficient for our purposes. The arrangement 

 consisted in erecting an iron stove in the vicinity of the pendulum- 

 apparatus, and carrying a long pipe through the whole height of the 

 room. By several preliminary trials, it was soon found that up to about 

 80° the temperature could be maintained constant for several hours, but 

 that the difficulties increased with the rise of the temperature, and be- 

 came almost unsurmountable when the temperature was above 100°. 

 Besides the maintenance of a pretty equable temperature during the dura- 

 tion of an experiment, another difficulty arose. The pillar of masonry 

 which carries the apparatus on one side, and the wall of the room on the 

 other, prevented us giving to the heating-apparatus such a lateral po- 

 sition as to bring the bar which carried the thermometers and the pen- 

 dulum in equal proximity to it. The stove had to be placed in front, 

 somewhat to the left of the apparatus, and hence the brass bar which 

 carried the thermometers was nearer to the source of heat than the pendu- 

 lum. In order to arrive at the most exact measurement of the real 

 "temperature of the pendulum, two additional thermometers were sus- 

 pended behind it, at about the same distance from it as those in front, 

 and all four were read during the experiments. 



Seeing from the preliminary trials that an approximately equal distri- 

 bution of temperature throughout the apparatus could only be relied 

 upon up to about 70°, and that after that point the differences in the 

 readings of the thermometers, behind and in front, increased to an extra- 

 ordinary degree, we decided upon making two different classes of ex- 

 periments, viz. one set confined to temperatures of about 70°, and 

 another comprising higher temperatures ; and we further, during the pro- 



