332 Prof. Spencer U. Pickering on Delicate 



and the total range of the instrument only 3°*4 C, giving 

 166 millim. to eactf degree* C; one estimation figure ("05 

 millim.) would therefore represent *0003° G. 



In using this thermometer it is first suspended in the calori- 

 meter containing water at a temperature a certain known 

 degree above that at which the experiments are to be made ; 

 a very small flame is then applied to a point just below the 

 upper chamber (which in such instruments must be of con- 

 siderable size), by which means the column of mercury is 

 broken off at the part thus heated, and the separated portion 

 remains in the upper chamber, the tube being much too fine 

 to permit of it being shaken down again. After cooling it is 

 put into a bath at the required temperature, and compared 

 with some standard instrument. In this way the registration 

 of the thermometer can be adjusted to any required tempera- 

 ture within *05° C. without the least difficulty ; but greater 

 accuracy could be obtained if necessary. 



This thermometer was employed in a large number of ex- 

 periments, and the general results, as will be shown below, 

 proved to be satisfactory. But it possessed one serious im- 

 perfection, or, more properly speaking, its delicacy brought 

 into light an imperfection which must be inherent, although 

 it has hitherto been unnoticed in all mercurial thermometers. 



As is well known, the bulb of a thermometer does not ex- 

 pand regularly when heated ;"en effet," to quote Berthelot's 

 words, "la marche de ^instrument est intermittent, c'est-a- 

 dire que le niveau n'est souvent atteint qu'apres une serie 

 de petites oscillations " *. To obtain a state of equilibrium the 

 thermometer must be tapped on the top w^ith the finger, or 

 some such means adopted to overcome the inertia of the bulb 

 before each reading. The larger the bulb and the finer the 

 bore of the tube, the more apparent does this inertia become : 

 with instruments such as the less delicate ones here mentioned 

 a dozen or less taps are sufficient to overcome it, but with 

 No. 62839 it was found that continuous tapping for two or 



* Mec. Chim. i. p. 178. It is curious to note that even while Berthelot 

 is warning his readers against this source of error he appears to have 

 fallen into it himself. On p. 163 he gives the results of a comparison of 

 two thermometers placed in a bath of water, the temperature of which 

 was rising, and read simultaneously at intervals of ten minutes; from this 

 and other closely concordant comparisons he finds 6*702 degrees of the one 

 instrument are equivalent to 14*827 of the other j but, on looking more 

 closely at each reading, it will be found that 



(1) a rise of 0*9 in the one corresponds to a rise of 1*1 in the other. 



(2) „ 0*7 „ „ „ 2-2 



(3) n 06 » v if I' 5 w 



(4) „ 1-1 „ „ „ 0*9 



(») ^ „ 1*2 „ „ ,, 1*1 „ 



Vastly discordant numbers. 



