166 



Variation of Thermal Conductivity of Rocks. [May 30, 



temperature in the stone by observing the current through the 

 galvanometer, due to differences of temperature between any two of 

 the three junctions B, M, T placed in circuit with it. 



10. We made many experiments to test what appliances might be 

 necessary to secure against gain or loss of heat by the stone across 

 its vertical faces, and found that Meselguhr, loosely packed round 

 the columns and contained by a metal case surrounding them at a 

 distance of 2 cm. or 3 cm., prevented any appreciable disturbance due 

 to this cause. This allowed us to feel sure that the thermal flux lines 

 through the stone were very approximately parallel straight lines on 

 all sides of the central line BMT. 



11. t The thermometer which we used was one of Cassella's (No. 

 64,168) with Kew certificate (No. 48,471) for temperature from to 

 100, and for equality in volume of the divisions above 100. We 

 standardised it by comparison with the constant volume air 

 thermometer* of Dr. Bottomley with the following result. This is 

 satisfactory as showing that when the zero error is corrected the 

 greatest error of the mercury thermometer, which is at 211 Q G., is 

 only 0-3. 



12. Each experiment on the slate and granite columns lasted about 

 two hours from the first application of heat and cold ; and we 

 generally found that after the first hour we could keep the tempera- 

 tures of the three junctions very nearly constant. Choosing a time 

 of best constancy in our experiments on each of the two substances, 

 slate and granite, we found the following results : 



Slate : flux lines parallel to cleavage. 



wT = 50 0> 2 C. 

 = 123-3. 

 = 202'3. 



Phil. Mag.,' August, 1888, and ' Edinb. Eoy. Soc. Proc.,' January 6, 1888. 



