532 SCIENCE PROGRESS. 



the oxygen or the water is thrown on to the hydrogen, 

 the amount of which is obtained by difference, and this in 

 turn is to be divided into the directly weighed amount of 

 oxygen in order to determine its equivalent. This is 

 theoretically a very serious, if not a fatal defect in the 

 method. 



This process was first employed by Dulong and Ber- 

 zelius, who made three experiments which were not very 

 concordant but yielded (after correction for the volume of 

 air displaced by the water formed, etc.) the value 15 "So. 4 

 for the atomic weight of oxygen. About twenty years 

 afterwards Dumas (5) published first in the Comptcs Rendus 

 in 1842, and afterwards in the Annates de Chimie in 1843, 

 the results of his classical research. In these papers he 

 criticises the results of Dulong and Berzelius, and points 

 out that the weights of copper oxide, copper and water pro- 

 duced, must be corrected to vacuum standard instead of 

 simply being weighed in air. He goes on to give a descrip- 

 tion of his apparatus and manipulation and gives his weigh- 

 ings from nineteen experiments in tabular form. In these 

 experiments he abstracted from his copper oxide as much 

 as 840*16 1 grams of oxygen and formed 945 "441 grams of 

 water, from which Dumas deduces as the mean of nineteen 

 experiments the ratio of 1 : "12515 which corresponds to 

 O = 15 '98. If we treat the whole series as one huge ex- 

 periment the value becomes 1 : '1253 or O = 15 '96. As 

 pointed out by Dittmar, Dumas although weighing his 

 copper oxide and metallic copper in vacuo seems to have 

 neglected the far more important similar correction ior the 

 water produced, even although he points out (as mentioned 

 above) in the same paper that the results of Berzelius re- 

 quired this correction. This seems almost incredible, but 

 in his tabulated results he gives the weight of the water - 

 collecting apparatus before and after each experiment, and 

 takes the simple difference as the true weight of water 

 produced. One or two other sources of error will be 

 referred to later. 



The next paper (6) in the same number of the Annates 

 de Chimie is by Erdmann and Marchand and is on the same 



