TABLE 234. 



AIR AND MERCURY THERMOMETERS. 



Rowland has shown (Proc. Am. Acad. Sci. vol. 15) that, when o and 100 are chosen for fixed points, the relation 

 between the readings of the air and the mercury in glass thermometers can be very nearly expressed by an equation 



oftheform t=r-at(ia -t)(6-t), 



where t is the reading of the air thermometer and T that of the mercury one, a and b being constants. The smaller 

 a is, the more nearly will the thermometers agree at all points, and there will be absolute agreement for t o or 

 loo or />. 



Regnault found that a mercury thermometer of ordinary glass gave too high a reading between o and 100, and too 

 low a reading between ioo j and about 245. As to some other thermometers experimented on by Regnault, 

 little is recorded of their performance between o and 100, but all of them gave too high readings above i<xr', 

 indicating that below loo 1 - the mercury thermometer probably reads too low. Regnault states this to be the 

 case for a thermometer of Choisy le.Roi crystal glass, and puts the maximum error at from one tenth to two tenths 

 of a degree. Regnault's comparisons of the air and mercury thermometers and a comparison by Recknagel of a 

 mercury thermometer of common glass with the air thermometer are compared with the above formula by Rowland. 

 The tables are interesting as showing approximately the error to be expected in the use of a mercury thermom- 

 eter and the magnitude of the constants a and b for different glasses. They are given in the following Table. 

 Regnault's results above 100 C. compared with the formula /rr T ai(ioo t)(6t), give for the constants a 

 and b the following values : 



Cristal de Choisy Le Roi . r= 0.00000032, 0. 



Verre ordinaire . . . a = 0.00000034, ^ = 245. 



Verre vert .... a 0.000000095, ^ 270.* 



Verre de Suede . . . a = 0.000000 14, 6=10. 



Common glass (Recknagel) a 0.00000033, ^ = 290. 



SMITHSONIAN TABLES. 



* Misprinted f +] 270 in Rowland's paper. 

 228 



