Thermal Conductivity with Temperature. 



521 



The temperatures of the rod were read thermoelectrically, 

 and stated in terms of galvanometer-deflections ; and though 

 a little Table is given showing what these deflections experi- 

 mentally mean in Centigrade excess of temperature, the num- 

 bers actually used in their calculations are the deflections 

 themselves, which are only roughly proportional to the tem- 

 perature excesses. The following empirical relation between 

 S (the deflection) and 6 (ih.Q excess of temperature) is deduced 

 from their little comparison Table, 







^H-tM 



10- 



100 



and this I have used to obtain the second column of the fol- 

 lowing Table : — 



Iron (Wiedemann and Franz). 



[Excess of tern- 

 Galvanometer! perature in 

 deflections. I Centigrade 

 8. degrees. 



lo ^85o^)LzT lo gyK 



= log/(0). 





Ojjfej) 



■ 2 cosh n%. 



230 



50-8 



153£ 



34 5 



100£ 



229 



Q7h 



15-8 



42 



10-1 



257 



6-3 



132 



3-5 



1-5637 

 1-4142 



1-2487 



10965 



•9086 



•7079 



•4558 



1495 £ 

 1575 

 1557 

 1638 

 1712 

 •1846 



2-094' 

 2-168 

 2-068 

 2-172 

 2147 



a The mean of the first four numbers in this column is -1566 ; therefore 



ju£ = -3606. 

 b The mean of these numbers is 2- 130 ; therefore 



\il€h V(2 cosh ,4 -2) = -3606. 



The third column contains the logarithm of the quantity 

 which ought to go in geometrical progression. The fourth 

 column shows that this law of progress is moderately true for 

 the first four numbers, but that the numbers which ought 

 to be constant exhibit a decided increase towards the cooler 

 end of the bar, probably because the bar was so short that 

 the flow of heat along it extended through the point where 

 0=0 — which is contrary to our hypothesis (§§ 32 and 17) that 



— shall vanish with 6. The distance f was 2 *6 centimetres. 

 dx " 



The next Table shows their results for silver tabulated in 



the same way, and on the assumption that m-=. +1000. 



Phil Mag. S. 5. No. 52. Suppl. Vol. 8. 2 N 



