ATOMIC WEIGHTS. 531 



definitely settled what is the true ratio between the atomic 

 weights of these two elements. Until this is finally settled 

 it seems absurd to take hydrogen as the standard even 

 though it may be asked : " What is then your unit ?" The 

 reply to this is, that the unit is as good, and is of the same 

 kind, as a foot or an inch, which are but subdivisions of the 

 length of the standard yard. A point of some practical 

 importance also is, that if we adopt O = 1 6 we have a large 

 number of atomic weights which for everyday use may be 

 taken as whole numbers, but this is not so if we take 

 H = 1. 



We must next consider the state of our knowledge as to 

 the relative values of the atomic weights of hydrogen and 

 oxygen. The determination of this ratio with an accuracy 

 such as the present state of science demands is one of very 

 great practical difficulty, and in spite of the many attempts 

 and because of the various results obtained by different 

 experimenters we are still undecided. 



The methods adopted have been both chemical and 

 physical. The purely chemical methods depend mainly on 

 the determination of— 



( 1 ) The composition of water, 



(2) ,, ,, ammonia and ammonium salts, 



(3) ,, ,, hydrocarbons. 



The first of these is of primary importance, and the 

 others mainly of interest in as far as they corroborate (or 

 the reverse) results obtained by it. The physical method 

 is based on the relative densities of the two gases coupled 

 with a knowledge of the proportions by volume in which 

 they combine. 



The determination of the composition of water gravi- 

 metrically, as is well known, depends mainly on the reduc- 

 tion to the metal of a weighed quantity of copper oxide, 

 which thus gives up a definite quantity of oxygen to form 

 a quantity of water which may be weighed. As Dumas 

 points out in his paper (5) it is a process only calculated 

 to yield good results when carried out with extreme 

 care, because any error in the determination of either 



