1893.] On the Composition of Water by Volume. 133 



The mean of all the experiments in the variable volume apparatus, 

 and in which potassium chlorate was the chief source of oxygen, is 



2-000903 + 0-00004 impurity equally distributed in both gases, 

 1*99925 +0*00005 „ assumed to be all in hydrogen, 



in 5 series of 19 experiments in all. 



The mean of all the experiments in the constant volume apparatus, 

 in which silver oxide was used as the source of the oxygen, and sodium 

 and steam either directly, or after absorption in palladiu u ior the 

 hydrogen, is 



2-002435 + 0-00006 impurity in both gases equally, 

 or 2-002431 + 0-00006 „ hydrogen alone. 



This is the mean of 53 experiments in 5 series. 

 If 6 experiments be rejected we get the value 



2-002466 + 000003 



as the result of 47 experiments in 5 series, and any impurity makes 

 no difference, whether it be assumed all in the hydrogen or equally 

 distributed in both gases. The most probable value, however, is 

 2-00245. 



This value, combined with the value 15'882 for the ratio of the 

 densities found by Lord Rayleigh, gives for the atomic weight of 

 oxygen 



15862 



Dittmar and Henderson's value is. . . . 15-866 

 Cooke and Richards' „ .... 15*869 



Leduc* found for the ratio of the volumes, by taking the density of 

 electrolytic gas from strong potassium hydrate solution and his own 

 values for the densities of hydrogen and oxygen, 



2-0037, 



and for the ratio of densities 35-905, 

 giving for the atomic weight of oxygen, 



15-876. 



Morley's experimentst are objected to on the ground that his appa- 

 ratus is too complicated, his measuring vessel far too wide for accu- 

 rate measurement and to its being used also as the explosion tube, 

 the transferring of his gases from one mercury trough to another, and 



* ' Comptes Eendus,' vol. 115, p. 313. 



f ' Amer. Journ. Science,' vol. 41, Ser. 3, pp. 220, 276. 



