168 PROCEEDINGS OF THE AMERICAN ACADEMY 



by Dumas by an obviously less direct and less accurate method, yet it 

 was still possible that there might be some constant error, and that 

 some cause might yet be found which would raise the maximum by 

 the forty-six thousandths required to give the whole number 16. It 

 was true that the probable error was only about one tenth of this dif- 

 ference ; still, as the materials had been purified, the maximum had 

 constantly risen, and the theoretical limit was in sight. In reviewing 

 the work, it was obvious that the degree of accuracy of the methods 

 used for determining the weights both of the hydrogen and of the 

 water was so great, that no possible error in these values could account 

 for the difference in question. This would imply an error of 1.2 mil- 

 ligrams in the weight of the hydrogen, and of 10.8 milligrams in the 

 weight of the water, and the possible error of a single determination — 

 leaving out of account the reduced probable error of the average 

 value — was far within these limits. If there was a constant error, it 

 must result from the want of purity of the hydrogen gas, and we there- 

 fore determined to try another method for preparing the hydrogen. 



The apparatus next used is represented in Fig. 7 (Plate), and differs 

 from the last only in the generator. Here the generator is a three- 

 necked bottle having a capacity of about two litres, filled to about one 

 eighth of its capacity with a semifluid amalgam of mercury and pure 

 zinc. On this rests dilute hydrochloric acid, containing about twenty 

 per cent of HC1, nearly filling the bottle. Into this acid dips a plati- 

 num electrode, while a straight glass tube passing through the middle 

 neck and dipping under the amalgam gives the means of establishing 

 an electrical connection between the large platinum plate which forms 

 the negative electrode and the amalgam. In addition, a siphon tube 

 for drawing off the acid when saturated with zinc, a funnel tube for 

 introducing a fresh charge, and an exit tube, all well cemented to the 

 several necks of the bottle, complete the generator. When the electri- 

 cal connection is broken, all chemical action ceases, but on connecting 

 by a wire the platinum electrode with the amalgam, a very steady but 

 slow evolution of hydrogen gas takes place, which can be regulated 

 with the greatest nicety by varying the resistance of the connecting 

 wire. On interposing two cells of a Bunsen battery the evolution of 

 gas becomes very rapid. Besides its special use in this connection, 

 the apparatus will be found of great value as giving an absolutely 

 constant source of pure hydrogen whenever required. In charging 

 the generator with acid, the same care was taken to exclude every trace 

 of air as with the previous apparatus, and with hydrogen thus prepared 

 a second series of five consecutive determinations was made, whose 



