260 Variation of Thermal Conductivity of Metals. 



for iron, after reducing their galvanometer-deflections to Cen- 

 tigrade degrees by means of their little interpolation table; but 

 it appears impossible to get any value for X. The form given 



i ° 

 is very nearly ~* 



A simple modification of their method of observing tempe- 

 ratures, devised by Professor G. C. Foster, seems, however, 

 to promise very accurate results ; and it was in view of being 

 thus able to observe temperatures with great accuracy that I 

 set to work at the preceding calculation. 



Suggestions for future Experiment. 



29. In any experiments which may be conducted in vacuo 

 there will probably be some difficulty with the radiating sur- 

 face of the rods, which should be the same for all. Silvering 

 has only one objection; but that is fatal — viz. that it diminishes 

 the radiation-constant so much that the heat flows down the 

 rod almost as if it were a slab; moreover the convection effect 

 of residual air cannot be neglected in comparison with the radi- 

 ation with so great impunity as if the surface were lampblacked. 

 A coating of Brunswick black would probably be perpetu- 

 ally spoiling the vacuum. A close layer of thin paper (after 

 Forbes) seems a very good method, or perhaps a close 

 spiral of very thin cotton thread wound over the rod ; but it 

 cannot be said that the law of cooling in such cases is so well 

 known as it is for a metallic or blackened surface. Probably 

 a coating of stove-blacklead would be the best. 



Any such covering would, of course, necessitate fixed ther- 

 moelectric joints, say very fine wires passed into or through 

 the rod at certain accurately measured and equal intervals. 

 The electromotive force generated at each joint would be mea- 

 sured directly by a compensation arrangement — the arbitrary 

 readings so obtained being reduced to Centigrade tempera- 

 tures afterwards, either by direct experimental comparison, or 

 by calculation from the following formula, 

 t = ^-{(N-T) 2 -ke}*, 

 where N is the neutral point of the two metals which form the 

 joint (the rod itself would do for one), T is the temperature of 

 the cool joint, which may be the same all the time, e the ob- 

 served electromotive force on any arbitrary scale, and k a 

 constant expressing the value of this scale determined once 

 for all by an observation with a known value of t. 



The only satisfactory method of heating is by the boiling of 

 some liquid or the condensing of its vapour. Water vapour 

 is scarcely hot enough, even under pressure; but oil might do. 

 The heat should be conducted into the vacuum by a thicker 



