ATOMIC WEIGHTS. 541 



liquid is almost exactly neutral, when it is again weighed. 

 The amount of acid or ammonia in excess is now titrated 

 by means of dilute alkali or acid standard solutions, when 

 the weights of hydrochloric acid and of ammonia equivalent 

 to one another are obtained. One experiment will make 

 the method plain. 



Weight of hydrochloric acid . . 5*0191 



Added ,, ,, . . "0172 



Total hydrochloric acid . . . 5*0363 



Total ammonia .... 2*3523 



H CI 5*0363 



N T TJ = ~ = 2-I4IO. 



N H 3 2-3523 



The mean of all his experiments is 2*14087 + 0-00009. 

 This when corrected for air displacement becomes 2*13934 

 ± 0*00009. ^ we ta ^e the values for N and CI (O = 16) of 



Stas van der Plaats Thomsen 



CI = 35*457 35"4565 35"4494 



N = 14-044 or 14-055 H'0519 i4'0396 



then it follows — 



H = -9989 or -9945 -9959 -9992 



The uncertainty as to the exact atomic weight of nitro- 

 gen has thus a marked influence on the value deduced for 

 H. The values obtained by his recalculation are maintained 

 by Thomsen to give a truer expression to the ratios found 

 by experiment, as existing between the compounds used by 

 Stas, than the atomic weights of either van der Plaats or of 

 Stas himself. It must be noted, however, that if we adopt 

 Thomsen's value for nitrogen, Stas' value for hydrogen is 

 thrown quite the other way,. and instead of being H « '9992 

 we saw that it must be between 1*0104 and 1*0107. 



One point about Thomsen's method is this: "Are the 

 gases really dry, i.e., absolutely free from aqueous vapour?' 

 The mere fact of the same ratio being obtained, whether 

 four grams or twelve grams or twenty grams of hydro- 

 chloric acid are used in an experiment, only proves that 

 the quality of the gas was the same throughout, but not that 

 it was either pure or dry. Unless we have also in one or 



