Temperatures above the Boiling-point of Mercury. 431 



voir. In fact the air of the reservoir at 500° is two and a half 

 times less dense than that in the capillary tube possessing the 

 temperature of the surrounding space. The weight of the air 

 contained in the reservoir diminishes, therefore, — 



1. By the amount of air contained in the 500 millims. of 

 capillary tube originally filled with mercury at zero, but now at 

 500° tilled with air ; this amount is equal to about the y^r 

 part of that contained in the reservoir at 500°. 



2. By the difference in the quantity of air contained in the 

 anterior portion of the tube originally at zero, but now at 500°. 

 The pressure, and consequently the condensation of this air, 

 being now tripled, and the length of the one portion of the tube 

 being nearly equal to that of the other (that in which the mer- 

 cury has been replaced by air), it will be seen that the increment 

 in the amount of air contained in the first portion of the tube is 

 very nearly equivalent to two-thirds of the quantity replacing the 

 mercury in the second portion of the same tube. 



The above quantities taken together represent a total diminu- 

 tion of one sixtieth in the amount of air contained in the reser- 

 voir, resulting from the increased temperature. 



This diminution will evidently vary with the capillarity of the 

 tube. The foregoing calculations are made for a tube ^ of a 

 inillim. in internal diameter. In a tube ■£§ of a millim. in dia- 

 meter it would amount only to g-fo- ; whilst if the bore were ^ a 

 millim. it would be increased to xV- 



We will now trace the consequences of this variation in the 

 amount of air contained in the reservoir. In the first place, it 

 produces another source of diminution in the length of each de- 

 gree, in addition to that caused by the expansion of the glass 

 envelope. But, after all, this is only a theoretical error, since 

 the length of the degrees is empirically determined by a curve 

 and by means of certain fixed points. 



Still the expulsion of a small quantity of air from the reser- 

 voir gives rise, theoretically speaking, to a more considerable 

 source of error, although, as we shall show, it may practically 

 be disregarded. This source of error arises from the ordinary 

 variations in atmospheric temperature. 



In the different experiments the air contained in the capillary 

 portion does not invariably possess the same temperature, and 

 cousequently its volume changes independently of the variations 

 caused by the temperature of the space in which the reservoir is 

 placed. Then in what does our instrument differ from an ordi- 

 nary air- thermometer ? 



In the latter instrument, the air it contains is divided into 

 two portions, each of considerable volume; the temperature of 

 the portion beyond the reservoir must therefore be determined 



