OP ARTS AND SCIENCES. 185 



The total weight of the hydrogen burnt in the sixteen determina- 

 tions as observed was 6.7029 grams. Add to this sixteen times the 

 correction, or 16 X 0.00138 = 0.0317, and we obtain 6.7346 grams 

 for the corrected weight. The total weight of the water formed was 

 60.1687 grams. Hence we find by difference for the total weight of 

 oxygen consumed in the combustions 53.4341 grams ; and the cor- 

 rected atomic weight of oxygen is 2(53.4341 -f- 6.7346) = 15.869. 



The probable error of this result is no greater than that of the 

 "Total average" given on page 173; for the value of the constant 

 correction must be certainly known within the one fiftieth of a milli- 

 gram. It is true that there are several variable elements which enter 

 into the determination of this value, but they can all be estimated 

 with far greater accuracy than the conditions of our problem require. 

 We may therefore write as the present result of our work, H : O = 

 1.000 : 15.869. 



Atomic weight of oxygen, 15.869 ± 0.0017. 



If we compare this result with that of Dumas, as before, on page 

 175, we have for the complete analysis of water, — 



Percentage of oxygen after Dumas .... 88.864 ± 0.0044 

 Percentage of hydrogen after final result . . 11.193 ± 0.0011 



100.057 



It would now appear that the close agreement before shown was a 

 mere coincidence, and that there must have been a small constant error 

 either in our own process or in that of Dumas. Where the error lies 

 further investigation can alone determine ; for although, after a careful 

 revision of our work, we can discover no flaw, no one can be confident 

 that such a constant error as has already appeared may not hereafter 

 be found, — and certainty can only be secured after repeated confirma- 

 tions by essentially different methods. While, therefore, we feel bound 

 to acknowledge without delay the cause of constant error which Lord 

 Rayleigh has pointed out, we give our corrected result as subject to 

 further revision. It has been suggested by Lord Rayleigh, in a " Pre- 

 liminary Notice " of his work on the relative densities of hydrogen 

 and oxygen, of which advance sheets have been received while writ- 

 ing this note, that in our combustions the hydrogen may have been 

 imperfectly burnt, especially as towards the last of the combustion it 

 must have been greatly diluted (but with air). We have no decisive 

 evidence on this point ; but the whole course of our combustions as- 



