40 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 54 



Hence, on the oxygen scale, H= 1.00772 ±.00001. 

 If we reject the seven highest values under the first heading, and omit 

 Morley's defective hydrogen series under the second, we get — 



By syntheses of water = 15.8760, ± .00017, or H = 1.00781, ± .00001 



By gaseous densities O = 15.8769, ± .00058, or H = 1.00775, ± .000035 



General mean O = 15.8762, ± .00016, or H = 1.00779, ± .00001 



The two component values of the last mean are remarkably concordant, 

 differing by only one part in 17640. For practical purposes the last 

 decimal of the hydrogen value may be rounded off, giving 



H = 1.0078, ± .00001 



as the atomic weight under consideration. The actual uncertainty of 

 this value, however, is greater than the so-called " probable error." The 

 latter, it must be borne in mind, is a mathematical expression which 

 should not be used in a colloquial sense. For computations of this 

 kind the probable error is essentially a coefficient of concordance, which 

 merely indicates the relative value or weight assignable to a given scries 

 of observations in comparison or combination with otliers. 



THE NITROGEN-OXYGEN RATIO. 



The direct ratio between nitrogen and oxygen has been determined by 

 analyses of nitrous and nitric oxides, and by measurements of gaseous 

 densities. The different methods may be considered in regular order. 



The exact analysis of nitrous oxide, with reference to the atomic 

 weight of nitrogen, was effected by Guye and Bogdan.^ The gas itself 

 was condensed in carefully purified charcoal, and so weighed ; it was 

 then passed slowly through a tube containing a spiral of iron wire, which 

 was heated to redness by an electric current. The iron was oxidized, and 

 its gain in weight gave the amount of oxygen in the lSr.,0. The results 

 obtained were as follows : 



.0033 



For the complete gravimetric analysis of nitric oxide we have the 

 elaborate data furnished by E. W. Gray.^ The gas was weighed, and 



iCompt. Rend., 138, 1494. 1904. Journ. Chim. Phys., 3, 537. 1905. 

 ^Journ. Chem. Soc, 87, 1601. 1905. 



