AND THE RELATION OF SPECIFIC HEAT TO ATOMIC WKICIIT. 243 



REGNAOLT'S copper on analysis gave " only traces of foreign matters." Korr used 

 " commercial copper wires." 



The iron used in my first experiment was prepared by dissolving fine iron wire in 

 hydrochloric acid, boiling the solution with nitric acid, and precipitating with excess 

 of ammonia. The completely washed precipitate was dried and heated strongly in a 

 platinum dish, then introduced into a glass tube and heated to redness in a stream of 

 hydrogen made by dissolving aluminium foil in solution of caustic potash, and purified 

 by passing through a moderately strong solution of potassium permanganate. The 

 spongy metal was compressed into short rods in a steel cylinder, and then melted in 

 an oxyhydrogen flame upon a support of lime. The buttons in solidifying gave off 

 a considerable quantity of gas, presumably hydrogen. They were then rolled into 

 moderately thin strips. 



A sample analysed by solution in cupric ammonium chloride and combustion 

 indicated O'Ol per cent, of carbon. This must be derived from atmospheric dust, as 

 every precaution had been taken to exclude carbon compounds from the gases used 

 in the preparation. 



Iron in this state is apparently unaltered by contact with pure steam at 100, but 

 a small amount of rusting occurred during the subsequent cooling in the presence of 

 atmospheric air. 



The results were not as uniform as could have been desired, the two former 

 experiments yielding results appreciably higher than the two latter. This may 

 {Kxssibly be due to the presence of occluded hydrogen, which was gradually oxidised 

 or expelled ; at any rate no other explanation presents itself, as the experiments 

 passed off successfully. 



Specific Heat of Iron nearly pure and after fusion and rolling. 

 S.G. 775 at 18/18. 



Specific heat. 

 11022 

 11037 

 10946 

 10926 



Arithmetical mean . . -10983 



The nature and amount of impurities present in a metal affect the value of the 

 specific heat seriously. This is well known, and an instance is afforded by the results 

 quoted at the beginning of this paper relating to impure cobalt. REGNAULT showed 

 that the specific heat of a specimen of white cast iron, containing perhaps 6 per eent. 

 of impurities, was '12728, while that of malleable iron was '11380. There are no 

 data for estimating the effect of carbon by itself upon the specific heat of a metal, 



2 I 2 



