( 535 ) 



XXIV. — On the Thermal Conductivity of Iron, Copper, and German Silver. By 

 A. Crichton Mitchell, Esq. Communicated, with an Introduction, by 

 Professor Tait. (Plates XXXIV., XXXV.) 



(Read 4th July 1887. Revised 24th October 1887.) 



[Introduction. 



Shortly after I read to the Society my paper on " Thermal and Electric 

 Conductivity" (Trans. R. S. E., 1878), in which I stated that the results 

 were " by no means final, even so far as my own work is concerned," I was 

 requested by Sir Wyville Thomson to undertake the examination of the 

 "Pressure Errors of the 'Challenger' Thermometers." This investigation led to 

 another on the "Compression of Sea- Water," and allied subjects, which is not yet 

 finished. Meanwhile, though I had prepared everything for my promised 

 repetition of the experiments on Thermal Conductivity, the bars formerly 

 used having been nickelised, &c, I found that it would be impossible for me 

 to carry out the investigation. I therefore asked Mr Mitchell, who, as 

 Neil-Arnott Scholar, had already done good and careful work on Thermal 

 Conductivity in my Laboratory, to repeat the experiments under the altered 

 conditions. I put at his disposal all the apparatus which was employed in the 

 former research. The Government Grant Committee allowed a sum for the 

 payment of a computer to reduce the results, and the observations were at 

 once commenced. The results are now laid before the Society, and are 

 probably as good as the method and the thermometers employed can 

 furnish. 



As regards the method, one grand defect is the uncertainty as to the 

 relative amounts of surface loss of heat in the two parts of the experiment. 

 The nickelising has, to a very great extent at least, removed the part of this 

 uncertainty which was clue to oxidation of the bars ; but there remains 

 another part, not at all easy to reckon and allow for, which depends on the 

 fact that each thermometer in the long bar is maintained for hours in a nearly 

 constant state of graduated temperature throughout its stem, while the 

 corresponding state of that in the short bar not only varies rapidly as the 

 cooling proceeds, but probably always materially differs from it. No attempt 

 has been made to correct the results so far as this cause of error (which is 



VOL. XXXIII. part ii. 4 K 



