OF ARTS AND SCIENCES. 155 



determinations was 15.9G07, with a probable error of ±0.0070, the 

 highest value being 16.024, and the lowest 15.892. 



The investigation of Erdmann and Marchand* was far less ex- 

 tended, and some of the precautions taken by Dumas were neglected 

 because deemed unnecessary. No pains seem to have been taken to 

 obtain pure cupric oxide, and the material used in several of the deter- 

 minations was described as " kaufliche Kupfergliihspan," while that 

 used in the others was obtained by igniting cupric nitrate ; and no 

 proof is adduced in either case of the purity of the material employed. 



The results are divided into two groups, and in the experiments of 

 the second group the air was exhausted from the combustion tube 

 before weighing ; but it appears from the paper that the marked dif- 

 ference between the two series of experiments depended rather on the 

 character of the cupric oxide, and on varying conditions used, than on 

 this circumstance. The first series of four results, when averaged, 

 gave the value 15.937, with a probable error of ±0.0138, while the 

 mean of the second series was 16.009, with a probable error of ± 0.0030. 

 The study of the paper, however, does not confirm the expectation 

 that the results of the second series are more trustworthy ; for the 

 closer agreement and smaller probable error appear to be the result of 

 the identity of conditions, which was maintained in this series, but not 

 in the first. Judging from the paper, we should be inclined to place 

 most reliance on the first series, in which the conditions of the experi- 

 ments were varied, rather than on the second, which seems obviously 

 to be influenced by some constant error. 



The results, then, thus far obtained, are as follows: — 



Direct Determinations. 



Dumas (nineteen determinations) 15.9607 ± 0.0070 



Erdmann and Marchand (first four) 15.9369 ± 0.0138 



" " " (second four) 16.0095 ± 0.0030 



Confirmatory Determinations. 



Dumas and Boussingaltf (gas densities) 15.954 ± 0.031 

 Regnault % (gas densities) 15.961 ± 0.0044 



Thomsen§ (not fully described) 15.960 



* Journ. f. Prakt. Chem., 1842, vol. xxvi. p. 461. 

 t Compt. Rend., xii. 1005 ; also Constants of Nature, Part V. p. 6. 

 \ Compt. Rend., xx. 975; also Constants of Nature, Part V. p. 6. 

 § Berichte der deutsch. Chem. Gesell., 1870, p. 928; also Constants of Nature, 

 Part V. p. 8. 



