154 PROCEEDINGS OF THE AMERICAN ACADEMY 



and pressure not only do not exactly obey Mariotte's law, but the 

 deviations from the law in these two cases are in the opposite direc- 

 tions, oxygen gas being condensed by increasing pressure more, and 

 hydrogen gas less, than the law requires. Hence theory would not 

 lead us to expect that the ratio of the densities of these gases at the 

 standard conditions would be the exact ratio of their atomic weights. 

 But obviously it may be that this inference from the molecular theory 

 is not legitimate, or it may be that the effect of the imperfect aeri- 

 form condition would not perceptibly influence the apparent atomic 

 ratio ; and hence the confirmation afforded by Regnault's results is of 

 value. 



In the same category we must class also the determination of the 

 atomic weight of oxygen made by Thomsen of Copenhagen, wbo 

 weighed the amount of water obtained by burning a measured volume 

 of hydrogen gas. Here the. reduction of the volume to weight involved 

 a knowledge of values and conditions which could not be known with 

 the greatest accuracy, and unfortunately the details of the experiments 

 have not been published. 



Again, we should class simply as confirmatory results deduced in- 

 directly, and involving the values of other atomic weights, however 

 accurately these subsidiary values may be supposed to be known ; 

 such, for example, as Stas's determination of the amount of chlorine 

 in amnionic chloride. 



Turning now to the actual direct determinations of the combining 

 proportions of oxygen and hydrogen, there are only two which are of 

 any present value. Of these by far the most important is the classical 

 investigation of Dumas, " Researches on the Composition of Water." * 

 This is one of the most memorable investigations in the history of 

 chemistry, and its general principles are known to every student of the 

 science. An indefinite amount of hydrogen was burnt by means of 

 cupric oxide ; the amount of oxygen consumed was determined by 

 the loss of weight of the combustion tube, and the amount of water 

 formed was collected and weighed directly. The experiments were on 

 a very large scale, the amount of water produced varying from 15 to 

 70 grams. The greatest care was taken to insure the purity of the 

 materials used, every known experimental means was employed to 

 secure accuracy, and all necessary corrections were applied to the 

 results. Estimated on the system at present in use, the value of the 

 atomic weight of oxygen obtained by Dumas as the mean of nineteen 



* Ann. de Chim. et de Physique, 3d ser., viii. 189 (1842). 



