404 Prof. H. BufF 071 the Thermal Conductivity 



lims. below the metal plate. The two exterior ends of the 

 wires, near each other but well insulated, are connected by 

 means of copper wires with an astatic galvanometer. Both 

 are surrounded by a thick covering of cotton-wool to protect 

 them against sudden alterations of temperature. 



The substitution of a thermoelectric pile for the mercurial 

 thermometer has in the first place, like the metal plates of the 

 brass cylinder, the object of diminishing the influence of ra- 

 diation ; but, besides that, it is more sensitive to thermal infl-u- 

 ences than the latter, on account of the adjustment of the sol- 

 dered junction. 



In the first experiments a strip of silver was placed above 

 the soldered junction as a screen to protect it from radiation ; 

 but in later experiments, on finding this precaution super- 

 fluous, it was removed. Magnus had found that the advan- 

 tage he had expected to derive from the use of the screen w^as 

 more than counterbalanced by the screen itself becoming warm 

 and communicating heat to the sides of the vessel. This I 

 found to be the case in my experiments also. 



The whole apparatus is surrounded by a wider glass cylinder, 

 the ground edge of which fits onto the plate of the air-pump. 

 The space betw^een this outer cylinder and the apparatus can 

 therefore be filled with cold water. I, however, like Magnus, 

 did not at first use the cooling arrangement. 



The very first experiments disclosed a source of error, which 

 had to be eliminated. The hot water in the brass cylinder not 

 only heated the junction of the thermoelectric pile and the air 

 surrounding it, but also, very probably to a higher degree, 

 the sides of the glass cylinder, both by conduction and radia- 

 tion. The increase in temperature was perceptible even to 

 the touch, and extended to below^ the level of the soldered 

 junction. 



It is thus evident that a current of gas ascends along the 

 sides of the glass, and hence the experiments were useless, as 

 far as the question of the conductivity of the gases is concerned. 



Magnus had certainly noticed this influence of the heated 

 sides of the glass cylinder ; but he probably failed to appreciate 

 its importance, inasmuch as by his method of observation the 

 sides of the cylinder were not accessible *. 



I had hoped to eliminate this error by abbreviating the du- 

 ration of the experiment. Magnus maintained during the 

 whole experiment the temperature of the heating-vessel at the 

 boiling-point by the introduction of steam. As the mercury 

 thermometer required from 20 to 40 minutes to reach its 

 maximum, it is evident that the heat had ample time to extend 

 through the whole apparatus. 



♦ Compare the drawing of his apparatus in Phil. Mag. voh xxii. pi. i. 



