AN'I) THK KKLATION <>F SI'KCIFK- HKAT TO ATOMIC WKICHT. 



237 



A portion of the compressed but unmelted metal placed in a j>orcelaiii crucible and 

 covered with glass was heated till the latter was melted and the crucible much 

 softmed. The metal in this case contracted visibly but was not melted, and it was 

 preserved separately for a further attempt to determine the specific heat with the 

 object of observing the difference, if any, between the melted and the unmelted 

 metal. 



Series II. 



The density of the fused cobalt was determined by weighing in benzene, the 

 density of which was specially determined. Two experiments gave 87171 and 

 87191 respectively, or a mean of 87181 for the density of the metal at 21/4. 



This cobalt was afterwards found to leave a minute trace of black substance on 

 dissolution in nitric acid. 



A fresh and larger supply of the metal was therefore made as follows: The 

 purpureo-chloride was first prepared and precipitated from aqueous solution by adding 

 strong hydrochloric acid, the precipitate being well drained, re-dissolved in boiling 

 water, and again precipitated by hydrochloric acid. The salt was then heated in a 

 platinum dish till converted into the anhydrous chloride, which was then transformed 

 into sulphate, and finally into ammonio-sulphate by addition of pure ammonium 

 sulphate. The double salt was crystallised from water in small crystals, which were 

 well drained and washed. From a concentrated solution of this salt in water the 

 metal was deposited by electrolysis, using as anode the cobalt of the preceding series 

 of experiments. 



After fusion in oxy hydrogen flame the metal gave the following values for the 

 specific heat : 



