158 PROCEEDINGS OP THE AMERICAN ACADEMY 



that the ratio of O to H is not constant, but varies with the state of 

 division and of saturation of the oxide, the duration of contact of the 

 water formed with the oxide, and with the temperature, from 7.95 to 

 8.15. The latter value is obtained with a saturated and divided oxide 

 filling the tube, the former with oxide in lumps filling the tube for a 

 space of 25 cm. "With a larger empty space the ratio has fallen to 

 7.90. When the synthesis of water is effected by weighing the hydro- 

 gen consumed (as by dissolving a known weight of zinc in HC1) and 

 the water formed, the ratio differs according to the contents of the 

 combustion tubes. If it contains granular CuO of a length of 89 cm. 

 heated to redness, the ratio O : H is 7.96 to 7.98 to 1 ; at a low tem- 

 perature, 7.90 to 7.93 ; if the CuO is replaced by PbCr0 4 , from 7.89 

 to 7.93." 



In addition to all this, impurities in the oxide of copper might have 

 a serious influence on the result. As before said, Erdmann and 

 Marchand speak of using " kaufliche Kupfergliihspan " ; but in our 

 work we could find no commercial cupric oxide which did not contain 

 a marked amount of arsenic. "We examined a number of specimens 

 coming from the best German and American dealers, and there was 

 not a single instance in which we did not find arsenic, and even when 

 the material was marked " purissimum." In some cases the amount 

 of arsenic was so great, that, after successive reductions and oxidations, 

 abundant crystals of arsenious oxide collected at the exit end of the 

 combustion tube. It is unnecessary to add that the hydrogen used 

 was free from all such impurity. 



For our own experiments, of which the results are given below, not 

 only was the oxide of copper prepared from absolutely pure electro- 

 lytic copper, but also, as will be shown, the combustion tube was left 

 at the end of the determination as it was at first, and the same tube 

 was used for a number of experiments. 



Apparatus for weighing Hydrogen. 



In entering on a new investigation of the oxygen and hydrogen 

 ratio, it was evident at the outset that no advantage was to be gained 

 by multiplying determinations by the old method. The only hope of 

 improvement lay in finding some method of weighing the hydrogen 

 with sufficient accuracy ; and it was essential to determine this weight 

 to within one ten-thousandth, or at least one five-thousandth, of its 

 value. 



A gas can only be weighed by enclosing it in a glass globe, or some 

 similar receiver, and hydrogen is so exceedingly light that its total 



