174 PROCEEDINGS OP THE AMERICAN ACADEMY 



On examining the table, it will be noticed that the mean of the 

 determination by the electrolytic method is the mean of all the deter- 

 minations combined, and that the probable error of the total average 

 is only about one fourth as great as the error of the nineteen deter- 

 minations of Dumas, which are incomparably the best that have hitherto 

 been made. 



It does not now seem possible to escape from the conclusion, that 

 the proportions in which the purest hydrogen that can be made com- 

 bines with oxygen to form water are those of 2 to 15.953, with a 

 possible error far within the T ^ 7 of a single unit. 



The question, of course, still remains, Is the hydrogen thus prepared 

 the typical hydrogen element? But this is the same question which 

 must arise in regard to any one of the elementary substances ; and all 

 that we can say is, that the evidence in regard to the purity of the 

 hydrogen we have used is as good as any that can be adduced in re- 

 gard to any one of the elementary substances whose atomic weight has 

 been most accurately determined. The question as regards Prout's 

 hypothesis narrows itself now to this one point ; and here we must be 

 content to leave it until further investigation has given us more 

 knowledge in regard to the nature of elementary substances. 



The writer at first planned to carry out the investigation on a much 

 larger scale, and for the purpose had blown a globe similar to that 

 represented by Fig. 1, but of five times the capacity, and counterpoised 

 it by the same general method. This globe held twenty-five litres 

 (somewhat over two grams of hydrogen gas), or five times as much as 

 the globe actually used ; but the difficulties of carrying out the deter- 

 minations on this scale led him to reduce the scale of the determinations 

 to that actually adopted ; and in the view of the results finally reached, 

 it is evident that no appreciable advantage would have been gained 

 from the enormous expenditure of time and labor which the process 

 on a large scale involves. Assuming that the difficulties of preparing 

 pure hydrogen gas on' that scale could have been overcome, it would 

 have required from five to seven hours to fill the globe, and four or 

 five days continuously to complete the combustion. 



Moreover, after many trials, the writer could not procure a globe 

 that would stand the requisite pressure weighing less than two and 

 one half kilograms, and with this weight and volume it was not 

 possible, with the best balance he could command, to distinguish 

 half a milligram with as much accuracy as he could one tenth of a 

 milligram with the smaller apparatus, while a vastly longer time was 

 required to reach equilibrium. A great deal of time was spent in 



