THERMAL CONDUCTIVITY. 25 



William Thomson, with the object of arriving at a true 

 value of the conductivity of these different substances. 

 The result was that the conductivity is obtained for these 

 different kinds of rock with, perhaps, greater accuracy 

 than the conductivity of any other substance which is 

 here, and the results are as follows : Of trap rock the 

 conductivity is '00415, that is to say, if we had a depth of 

 trap rock one centimeter thick and kept the opposite faces 

 of it at a difference of temperature of 1, then the flux of 

 heat through every square centimeter in that block of rock 

 would be in a second of time sufficient to raise the tem- 

 perature of a gramme of water through '00415 C. For 

 sand, the number is '00262 ; and for sandstone, '01068. 



In the matter of metals the most important observations 

 which have been made are those made on this bar, which 

 were made at Edinburgh by the late Principal Forbes. 

 This bar, which is made of as pure iron as it is possible to 

 get, was mounted on two pivots so as to be free to radiate 

 in all directions. A crucible is attached to one end, into 

 which solder was put and kept continually melted, and it 

 was found by continually adding some of the solid metal, 

 keeping some unmelted always in the crucible, the tempera- 

 ture could be maintained with great accuracy for a good 

 many hours. Then thermometers were placed in holes in 

 the upper surface, and the temperature recorded. The 

 method of reduction was one completely unfailing, and did 

 not involve any theory whatever, and thus these results 

 became a test of the accuracy of the assumptions that 

 Fourier made in his theory. I will try and explain briefly 

 the general method employed in reducing the observations. 

 There was another bar of exactly the same material as this, 

 a short bar, which was heated up to an intensely high 

 temperature and then allowed gradually to cool. A ther- 

 mometer was placed in this bar exactly as in the long bar, 

 and the temperature as it cooled was continually observed, 

 and then if results were jotted down upon a curve or in a 

 table we should be able to tell what was the loss of heat 

 by radiation and convection in a minute of time at any 

 given temperature from direct measurement. Then sup- 

 pose we take a base line and mark it in divisions to 

 represent differences of temperature as we pass along, say, 

 10, 20, and 30 degrees above the temperature of the air ; 



