1 68 RESEARCHES UPON ATOMIC WEIGHTS. 



more no precautions are mentioned either for preventing access of water to the 

 material before weighing or for preventing the reduction of the silver salt by 

 the phosphorous acid formed in the decomposition of the trichloride with water. 

 Recalculated on the basis of the atomic weight of silver as 107.88, his five analy- 

 ses give results which vary between 30.99 and 31.08. The average is 31.03. 



Van der Platts ^ made two determinations by each of three different methods. 

 He obtained the values 30.90 and 30.97 by the precipitation of silver from silver 

 sulphate solution with phosphorus. His results from the analysis of silver 

 phosphate were 31.08 and 30.95. He gives no details of the method of prepar- 

 ing and analyzing this substance, merely making the statement, "it is difficult 

 to be sure of the purity of this salt." Finally, by the combustion of yellow 

 phosphorus in oxygen he obtained the results 30.99 and 30.96. The very meagre 

 descriptions of these experiments preclude criticism. 



Using Leduc's data for the densities and compressibilities of phosphine and 

 oxygen, Daniel Berthelot ^ has calculated, by the method of limiting densities, 

 the molecular weight of phosphine to be 34.00 and the atomic weight of phos- 

 phorus to be 30.98. 



Very recently Gazarian^ has obtained a considerably lower value for the 

 molecular weight of phospliine, 33.93. This value was calculated from the ex- 

 perimentally determined weight of the standard liter by the four methods of 

 molecular volumes (Leduc), limiting densities (Berthelot), critical constants 

 (Guye), and "indirect" limiting densities (Berthelot). The different methods 

 give essentially identical results, except in the case of the direct method of lim- 

 iting densities. By the latter method a value 0.06 unit higher is obtained, but 

 Gazarian rejects the result on the basis of inaccurate knowledge of the compres- 

 sibiHty of phosphine. It is highly desirable to obtain more certain knowledge 

 of the compressibility of phosphine, since the method of limiting densities is the 

 most rehable of all the methods for applying the correction to the densities made 

 necessary by deviations from the laws of a perfect gas. 



The other methods are burdened with arbitrary assumptions a'nd empirical 

 constants, and moreover Baume '' has shown that both the method of molecular 

 volimies and the method of critical constants give correct results only with gases 



for which the ratio — ^ is nearly i, whereas for phosphine this ratio is 1,26. 



If the molecular weight of phosphine be assumed to be 33.93, the atomic 

 weight of phosphorus is 30.91. IntheUghtof this low result it is unfortunate 

 that Gazarian prepared phosphine by only one method, and that he did not 

 determine the purity of the gas, i. e., by absorption. Gazarian used the method 

 of Matignon and Trannoy,^ which consists in heating calcium phosphate and 

 aluminum together until they react, and then treating the product of this re- 

 action without further purification with water in a gas generator. Matignon 



» C. R., 100, 52 (1885). « C. R., 126, 141S (1898). 



• Jour, de Chim. Phys., 7, 337 (1909). * Jour, de Chim. Phys., 6, 76 and 86 (1908). 



^ C. R., 148, 167 (1909). 



