56 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 54 



Hence CO = 28.0046, and C = 12.0046; ±.0008. 



This work of Stas ^vas also criticised by Scott/ in connection with the 

 determinations by the first method. The process employed is subject to 

 several possible errors, two of them being especially serious. First, the 

 carbon monoxide may have contained hydrogen or hydrocarbons. Sec- 

 ondly, the copper oxide, which was prepared by calcining copper nitrate, 

 almost certainly contained occluded nitrogen. The value found for C, 

 however, is probably not very far from the truth, and it is not unlikely 

 that errors in opposite directions tended to compensate one another. 



For the density of carbon monoxide there are available determinations 

 by Leduc ^ and Eayleigh.^ Leduc used a globe which had a capacity of 

 2.9440 grammes of air. Filled with CO it held the following weights, 

 giving the accompanying densities: 



Weight CO. Density. 



2.8470 .96705 



2.8468 .96698 



2.8469 .96702 



.96702, ± .000015 



This density, combined with Leduc's determination of the density of 

 oxygen, 1.10514, ±.000032, gives the crude ratio — 

 0, : CO :: 32: 28.0007, ± .0010 



Eayleigh's determinations may be stated in the following form : A 

 globe which held 2.62760, ±.00004 grammes of oxygen, held of carbon 

 monoxide^ — - 



2.29843 



2.29852 



2.29854 



Mean, 2.29850, ± .000024 



Corrected for the compression of the globe when empty this becomes 

 2.29906, ±.000024. From these data the crude value for CO is 27.9989, 

 ±.0012. Combining this with Leduc's determination, the general mean 

 becomes — 



CO = 28.0000, ± .00077 



Eayleigh, it must be observed, prepared his three samples of carbonic 

 oxide by three distinct methods, and the concordance in their weights 

 gives strong assurance of their purity. 



^ Loc. tit. See also Richards, Anier. Chem. Journ., 20, 701. 1898. 

 - Compt. Rend., 115, 1072. 1893. 

 3 Proc. Roy. Soc, 62, 204. 1897. 



