Table 40 {continued) 83 



PROBABLE VALUES OF THE GENERAL PHYSICAL CONSTANTS 



tion of AgNO s to Ag. Since 0= 16.000, by definition, the sole error is due to 

 that in N. The proportional error is due to that in N0 3 , only about one fourth 

 the probable proportional error in N. The ratio AgN0 3 /Ag can be determined 

 with great precision. The most accurate ratio, by far, is that by Richards and 

 Forbes, 1 yielding also 1.57479. A very elaborate investigation by Honig- 

 schmid, Zintl and Thile, 2 gives again exactly the same ratio. With our adopted 

 value of N and the above value of AgN0 3 /Ag = r, one has N0 3 /(r-i) 

 = (62.0083 ±0.0008)/ (0.57479) = I07.8799± 0.0014. 



The atomic weight of silver can be obtained in many ways. Clarke 3 lists 

 43 methods, yielding a final weighted average of 107.8804. It seems reasonable 

 that at the present time only the AgN0 3 /Ag ratio results need be considered, 

 with a final real error in Ag due merely to that in N. It seems reasonable to 

 adopt 



Ag = 107.880 ±0.001. 



Iodine. — The atomic weight of iodine enters into the discussion of the value 

 of the faraday. Clarke 3 lists eight methods, with a mean of 126.926. This 

 result will bear closer scrutiny. The most accurate is the direct determination 

 of the I/Ag ratio, assuming the atomic weight of silver as known. Among 

 the values of this ratio, 1. 176603, obtained by Baxter, 4 in 1910, is the most 

 reliable. Clarke lists all determinations. Now the four earlier results are all 

 approximately 1.1753, while the later results run much higher. These earlier 

 results probably are vitiated by some systematic error. They are quite self 

 consistent, and so by Clarke are given a high weighting. With the four earlier 

 results eliminated, we have a new weighted average of 1. 176549, in closer 

 agreement with Baxter's 1910 result. This ratio, combined with Ag= 107.880, 

 gives 1 = 126.926, while Baxter's result gives 126.932. Using the revised 

 average value for the I/Ag ratio with Clarke's results for the other seven 

 methods, we obtain a final weighted average of 1 = 126.932, in place of Clarke's 

 value 126.926, and in exact agreement with Baxter's result. Doctor Birge 

 adopts 



I = 126.932 ±0.002. 



In conclusion it is of interest to note that Aston gets 1=126.932, in exact 

 agreement with our adopted value. 



Carbon. — The atomic weight of carbon can be determined directly from oxy- 

 gen. The result of all such determinations, as obtained by Clarke, 3 is 1 2.0000 ± 

 0.00026. This result (written 12.000) was accepted in 1925 by the Inter- 

 national Committee on Atomic Weights/ and has since been used by Baxter. 8 



1 Journ. Amer. Chem. Soc, 29, 808, 1907. 2 Z. anorg. Chem., 163, 65, 1927. 3 Mem. Nat. 

 Acad. Sci., 16, 1920. 4 Journ. Amer. Chem. Soc, 32, 1591, 1910. 5 Journ. Amer. Chem. 

 Soc, 47, 597, 1925. 6 Journ. Amer. Chem. Soc, 50, 603, 1928. 



Smithsonian Tables 



