Atomic Weight of Potassium, history
In connexion with the atomic weight of sodium, reference has been made to the general methods employed by Berzelius, by Stas, and by Marignac in determining the atomic weights of sodium, potassium, silver, chlorine, bromine, and iodine. At this point it will suffice to summarize the results obtained by these methods, and those derived from the work of modern investigators.
There are numerous early determinations of the composition of potassium chlorate, giving the ratio KClO_{3}: KCl = 100: x, where x = ranged between 60.791 ± 0.0009 and 60.9487 ± 0.0011.In Stas's first series of experiments the chlorate was decomposed by heat; in the second, it was decomposed by hydrochloric acid. From the foregoing results Clarke has computed the weighted mean
KClO_{3}: KCl = 100: 60.846 ± 0.00038 (A)
Early analyses of potassium bromate by Marignac and of potassium iodate by Millon gave
KBrO_{3}: KBr = 100: 71.3245 ± 0.0207 (B) KIO_{3}: KI = 100: 77.527 ± 0.005 (C) The weighted mean is Ag: KBr = 100: 110.3459 ± 0.0019 (F)
Five analyses of potassium iodide by Marignac in 1843 gave
Ag: KI = 100: 153.800 (G)
From the foregoing seven mean results lettered from A to G the atomic weight of potassium can be induced. For this purpose the antecedent data calculated by Stas from his experiments have been chosen, (O = 16), Cl = 35.457, Br = 79.952, I = 126.850, Ag = 107.930. The values obtained for the atomic weight of potassium are:
(A) 39.136; (B) 39.438; (C) 38.739; (D) 39.138; (E) 39.113; (F) 39.144; (G) 39.038;
The values (B) and (C), derived from analyses of potassium bromate and iodate, are obviously worthless. The results (A), (D), and (F) depend mainly on the careful work of Stas, and their arithmetic mean is 39.139. The results (E) and (G) are based on Marignac's work on potassium chloride and iodide, and approximate reasonably to those of Stas. Excluding (B) and (C), the arithmetic mean of the results is
K = 39.113.
From his own experiments Stas induced the value K = 39.142, and recalculation in terms of O = 16 of the value accepted as the atomic weight of potassium for many years gives 39.14 or 39.15.
The work of modern investigators, particularly that of Richards and his coadjutors at Harvard, has proved Stas's value for the atomic weight of silver to be too high, and his relative values for silver and chlorine to be erroneous. These inaccuracies vitiate the preceding calculations; but there would be no advantage in deriving them from more accurate antecedent data, as modern research has disclosed appreciable errors in even the most painstaking of the earlier work, and furnished more reliable ratios for inducing the atomic weight of potassium. In the subjoined account the calculations have been made with the atomic weights O = 16.000, Ag = 107.880, Cl = 35.457, Br = 79.916, N = 14.008.
In connexion with their work on caesium, Richards and Archibald in 1903 made two analyses of potassium chloride, the results being
AgCl: KCl = 100: 52.0215, whence K = 39.109; and Ag: KCl = 100: 69.1145, whence K = 39.104. In the following year Archibald published the results of four additional experiments: AgCl: KCl = 100: 52.024, whence K = 39.113; and Ag: KCl = 100: 69.114, whence K = 39.103.
Richards and Archibald also made three other experiments, potassium nitrate being heated with silica, and the loss in weight determined:
N_{2}O_{5}: K_{2}O = 100: 87.232,
whence K = 39.112.
An account of the elaborate research of Richards and Stahler on potassium chloride appeared in 1907. The salt employed was prepared by the action of hydrogen chloride on potassium nitrate purified by repeated recrystallization with special precautions. After several crystallizations, the chloride was fused in a current of nitrogen. The silver was obtained by reducing the nitrate with ammonium formate. After electrolytic purification, it was fused in a limeboat in a current of hydrogen, and finally in vacuo. In the analyses the silver required to precipitate the chlorine from a known weight of potassium chloride was determined, and also the weight of silver chloride produced:
5 experiments. AgCl: KCl = 100: 52.0118 ± 0.0004, whence K = 39.095; 7 experiments. Ag: KCl = 100: 69.1073 ± 0.0004, whence K = 39.096.
A parallel research on potassium bromide by Richards and Mueller gave results in good agreement with those of Richards and Stahler:
4 experiments. AgBr: KBr = 100: 63.3727 ± 0.0003, whence K = 39.096; 11 experiments. Ag: KBr = 100: 110.3190 ± 0.0004, whence K = 39.096. A very careful redetermination of the ratio KClO_{3}: KCl by Stahler and Meyer in 1911 gave KClO_{3}: Kcl = 164.382: 100, or KCl: 3O = 100: 64.382, whence K = 39.098.
Neglecting the work of Richards and Archibald, which lacks the experimental accuracy of the later investigations, five modern values for the atomic weight of potassium fall within the limits of 39.095 and 39.098, and indicate the great probability of the value
K = 39.096.
The current table of the International Committee on Atomic Weights gives
K = 39.10. 


