OP ARTS AND SCIENCES. 153 



of most of the atomic weights, when given, as is usual in recent text- 

 books, on the basis of the hydrogen unit. 



When as the result of his investigation on the atomic weight of anti- 

 mony there was presented to the writer the ratios of whole numbers 

 as shown in the first of the above proportions, with the single excep- 

 tion of the atomic weight of hydrogen, the question was at once sug- 

 gested : Is the ratio of the atomic weights of oxygen and hydrogen in 

 fact that of 16 : 1.0025, as the general average of all trustworthy 

 determinations hitherto made seems to indicate, or was there some 

 constant error lurking in these results which caused the very slight 

 variation from 16 to 1 required by the theory? In looking at the 

 proportion thus displayed, it seemed as if the variation from the theory 

 must be apparent, and he determined to ferret out the hidden error if 

 possible. This investigation was undertaken in the autumn of 1883, 

 but owing to the condition of the writer's sight the work has been 

 greatly delayed. 



No one can study the record of the investigations by which the ratio 

 of the weights of the oxygen and hydrogen atoms have been deter- 

 mined, without receiving the impression that they are by no means 

 decisive in regard to the theory we are discussing, and it is also equally 

 evident that this ratio, if it could be fixed beyond doubt, would be a 

 crucial test of the theory. 



Previous Work. 



The methods by which the atomic weights of oxygen and hydrogen 

 have been determined may be divided into two classes ; first, the direct 

 method of determining the ratio in which the proportions of oxygen 

 and hydrogen uniting to form water were actually weighed ; secondly, 

 the confirmatory method, to whose results small weight could be given 

 independent of the first. 



Among confirmatory methods we must unquestionably class the 

 classical determinations by Regnault of the density of oxygen and 

 hydrogen gases under normal conditions at Paris ; that is, iu so far as 

 these determinations bear on the question of the ratio of the atomic 

 weights. 



According to the molecular theory the ratio of the densities of oxy- 

 gen and hydrogen gases could only be the ratio of their molecular or 

 of their atomic weights when both materials were in the condition of 

 perfect gases, of which condition the test would be an exact conformity 

 to Mariotte's law. Now, as Regnault himself elegantly demonstrated, 

 oxygen and hydrogen gases at the standard conditions of temperature 



