426 
DR. F. P. BURT AND DE. E. C. EDGAR ON 
Comparing our result with those of Morley and Noyes, we have :— 
Morley .... 1*00762, 
Noyes. .... 1*00787, 
or, as estimated by Clarke from the same data, 
1*00783, 
Burt and Edgar . 1*00772. 
Our value, therefore, is very nearly the arithmetic mean of the other two. 
In 1914, a paper was published by Germany ( 14 ) containing a series of measurements 
of the density of oxygen. 
As a mean of fifteen experiments he obtained the value 1*42906 for the weight of 
a normal litre. The gas was made by heating potassium permanganate, and, after 
passing over potash and phosphoric oxide, was liquefied and fractionated. This last 
precaution entitles his results to special consideration. Collecting the results obtained 
by Morley, Rayleigh, and Germany, we have 
Morley (41 experiments). . . . 1*42900 
Ra\ t leigh (16 experiments) . . . 1*42904 
Germany (15 experiments) . . . 1*42906. 
The arithmetic mean is 1*42903, but Germany, attaching rather greater weight to 
his own result, partly because his oxygen was liquefied and fractionated, and partly 
because of special precautions taken in measuring the pressure, proposes the figure 
1*42905 as the most probable value on existing data. Further confirmation of the 
higher value is afforded by the work of Scheuer( 15 ), who has recently obtained a 
result identical with Germany’s. The substitution of this value (1*42905) for 
Morley’s in the density ratio alters our atomic weight to 1*00769. It seems unlikely 
that any probable future alteration in the density ratio would bring our value as low 
as Morley’s or as high as Noyes’, and we may conclude that the true value of the 
atomic weight of hydrogen lies very close to 
1*0077. 
Part of the expense of this work was defrayed by a Royal Society Grant which we 
desire to acknowledge here. We are indebted to Prof. Dixon for his continued 
interestin the research and for a number of helpful suggestions. 
