RICHARDS AND ROWE. — SPECIFIC HEATS OF LIQUIDS. 183 



concerning the heat capacities of carbonate solutions suggest however 

 that the amounts of carbonate to be found in hydroxides of the highest 

 grade would have a scarcely appreciable effect upon the specific heat. 

 For example, the specific heat of a solution of sodium hydroxide with 

 100 mols of water as determined by Thomsen is 0.968 while that of 

 sodium carbonate of an equivalent concentration is given severally by 

 Thomsen as 0.95S and by Marignac ^ as 0.9695. Taking Thomsen's 

 two values as the more strictly comparable, it is evident that each 

 percent of carbonate may be supposed to lower the specific heat by 

 one unit in the fourth decimal place. The highest grade of sodium 

 hydroxide to be found in commerce, as employed in our experiments, 

 contained from 1 percent to 2 percent of carbonate; hence solutions 

 made directly from this material would show a heat capacity 0.01 

 percent to 0.02 percent lower than a solution of pure hydroxide. 

 Because, however, such assumptions introduce uncertainty; and 

 furthermore, because 0.02 percent is a perceptible quantity, it was 

 obviously best to secure specimens of really pure substances. 



Our material was prepared by first adding to a rather concentrated 

 solution of the purest material of commerce an excess of pure barium 

 hydroxide in hot aqueous solution. After the settling of the precipi- 

 tated barium carbonate, the clear, supernatant liquid was siphoned 

 off in an atmosphere free from carbon dioxide. A series of gravi- 

 metric determinations of the excess of barium present in the solution 

 were then made, and the theoretical amount of very pure sulphuric 

 acid was added. After the settling of the precipitate, the supernatant 

 liquid was again siphoned off with the usual precautions, and made 

 up to the proper strength by the addition of freshly distilled water, 

 which had been cooled in an atmosphere free from carbon dioxide. 

 Such alkaline solutions were always kept in paraffin-lined bottles — 

 a procedure which sufficiently preserved their initial purity, as is 

 evidenced in the accompanying analyses. 



For example, 101.20 grams of solution were found by titration to be 

 equivalent to 9.8976 grams of a certain solution of hydrochloric acid. 

 On the other hand, 12.9746 grams of this latter solution were found to 

 yield 10.3197 grams of silver chloride. Hence, 100.000 grams of the 

 alkaline solution must be equiA'alent to 7.8723 grams of silver chloride 

 or must contain 2.171 grams of dry sodium hydroxide. The solution 

 contained therefore 2.171 percent of alkali. Other analyses are tabu- 

 lated below. 



The end point was essentially the same with methyl orange as with 



6 Marignac, Arch. Sci. Nat., 55, 113 (1876). 



