12 THE ATOMIC WEIGHTS. 



.014 



. .0036 



The effect of discussin.sj these two series separately is somewhat start- 

 ling. It gives to the four experiments in Erdmann and Marchand's 

 second group a weight vastly greater than their other four and Dumas' 

 nineteen taken together. For so great a superiority as this there is no 

 adequate reason ; and it is highly probable that it is due almost entirely 

 to fortunate coincidences, rather than to greater accuracy of work. We 

 will, therefore, treat Erdmann and Marchand's experiments as one series, 

 giving all equal weight, the mean now becoming = 15.975, d= .0113. 

 If we take the sum of the eight experiments, 483.137 grammes water 

 and 429.352 grammes oxygen, and compute from these figures, then 

 = 15.966. 



It would be easy to point out the sources of error in the foregoing sets 

 of determinations, but it is hardly worth while to do so in detail. A few 

 leading suggestions are enough for present purposes. First, there is an 

 insignificant error due to the occlusion of hydrogen by metallic copper, 

 rendering the a23parent weight of the latter a trifle too high. Secondly, 

 as shown by Dittmar and Henderson, hydrogen dried by passage through 

 sulphuric acid becomes perceptibly contaminated with sulphur dioxide. 

 In the third place, Morley * has found that hydrogen prepared from zinc 

 always contains carbon compounds not removable by absorption and 

 washing. Erdmann and Marchand themselves note that their zinc con- 

 tained traces of carbon. Finall}^, copper oxide, especially when pre- 

 pared by the ignition of the nitrate, is very apt to contain gaseous impuri- 

 ties, and particularly occluded nitrogen.t Any or all of these sources of 

 error .may have vitiated the three investigations so far considered, but it 

 would be useless to speculate as to the extent of their influence. They 



* Anier. Cheni. Journ., 12, 469. 1890. 



t See Richards' work cited in the chapter on copper. 



