and Dispersion of Light by the Alums. 



165 



Substance. 



Refraction-equivalent. 



Calculated. 



Observed, and reduced 

 to A. 



Soret. 



Topsoe and 

 Christiansen. 



Ammonium Aluminium alum 



Sodium ,, „ 



Methylamine „ ,, 



Potassium ,, ,, 



Rubidium ,, „ 



Caesium „ ,, 



Ammonium Chromium alum 



Potassium „ „ 



Rubidium „ „ 



Caesium ,, ,, 



Ammonium Iron alum 



2520 

 239-3 

 267-2 

 2451 

 2540 

 268-0 

 2640 

 2571 

 266-0 

 280-0 

 270-6 

 263-7 

 272-6 

 286-6 



252-2 



238-5 

 267-7 

 2468 

 253-7 

 262-3 

 265-9 

 261-2 

 266-7 

 275-5 

 269-1 

 2650 

 273-2 

 276-0 



268-6 

 261-4 



Potassium ,, ,, 



Rubidium „ ,, 







The agreement between the results calculated and found is 

 as near as might be expected, except in the case of caesium. 

 It confirms the general law, as the variations only in one 

 instance amount to as much as 1 per cent., and are some- 

 times plus and sometimes minus. There is little doubt that 

 the caesium, in one set of observations or the other, was im- 

 pure : an old determination of mine, from the chloride, would 

 give figures lower than Soret's. 



Do these data afford us the means of determining the 

 refraction-equivalents of the elements with more exactness 

 than heretofore ? It is evident that in the series of aluminium 

 alums the metal having the smallest refraction-equivalent is 

 sodium, and that the rest follow in the order — potassium, 

 ammonium, rubidium, methylamine, caesium, and thallium ; 

 the same order is preserved in the chromium and iron alums 

 as far as they extend. This agrees with the order previously 

 determined both by myself and Kanonnikoff ; but when we 

 look more closely into the matter it is evident that the figures 

 are not very exact. Thus, in the aluminium series the differ- 

 ence between 2NH 4 and 2K is 5*42 ; but in the chromium 

 series it is only 4*72, and in the iron series 4*12, or 7*19 ac- 

 cording to Topsoe. The values of potassium deduced from 

 this would vary considerably : assuming the value of NH 4 to 

 be 11-1, it might be either 8'4, 8*7, 9*0, or 7*5 ; which are 

 wider differences than between my old estimation, 8*1, and 

 Kanonnikoff 's recent independent determination, 7*75. This 

 will not be wondered at when it is remembered that all expe- 



