DILUTE SOLUTIONS AT THE FREEZING POINT. 357 



ions such as chlorine. Perhaps, therefore, solutions of these salts may contain a larger 

 number of ions than we have assumed ; thus, the ferricyanide may be resolved into 

 potassium cyanide and ferric cyanide, when both K and Fe would behave as 

 kations. The electrically equivalent weight would thus become half that taken 

 alx>ve, the value of m would be doubled, and the equivalent conductivity reduced by 

 one-half. 



The curve giving the ionization of potassium bichromate is unlike any other. 

 Beginning at extreme dilution, it first falls in a manner similar to that shown by curves 

 for sulphates or other salts with divalent acids. As the concentration increases to a 

 value of about 10~ 8 gramme-equivalent per litre, the ionization becomes nearly constant, 

 and remains so till the concentration reaches about 10~ 2 gramme-equivalent, after 

 which it again begins to fall, and almost looks as though it were passing into a new 

 curve, the expression of a different chemical constitution. Perhaps these results 

 may indicate that a change in the actual ions of the salt is brought about by increasing 

 concentration. It is interesting to note that WALDEN* and OsTWALDf have already 

 given evidence to show that bichromates in solution react with the water at moderate 

 concentrations to form a mixture of the normal salt and acid. 



Another point which may be of interest to chemists appears in the case of 

 potassium permanganate. The slant of its curve is very small, and approximates to 

 that for potassium chloride. It is much less than the slope for substances with divalent 

 acid radicles such as sulphates. It is therefore probable that a molecule of potassium 

 permanganate gives two ions, K and MnO 4 ', and not three ions, represented by K', 

 K', arid Mn 2 O 8 ". The chemical structure of the salt in solution should then be 

 represented by KMn0 4 . Mr. GRIFFITHS' freezing-point measurements may be 

 expected to finally settle this question. 



In order to collect the results of this investigation and exhibit them in a convenient 

 form, the following tables have been compiled, showing the most probable values of 

 the ionizations for certain definite concentrations. These results have been obtained 

 from the smoothed curves, and three series of numbers have been tabulated. 



In the first set, Table IX., the series of concentrations is the same as that used by 

 KOHLKAUSCH, viz. : m = 5, 2, 1 X KT*. 



In the second set, Table X., OSTWALD'S series m = 1, , , &c., is taken. 



In both these sets, m is measured in gramme-equivalents of solute per 1000 

 grammes of solution. 



In the third set, Table XI., the concentrations are measured in terms of the number 

 of gramme-molecules of solvent present to 1 gramme-molecule (not gramme-equiva- 

 lent) of solute. The series taken for n is 5, 2, 1 X 10*, the corresponding values of 



m being calculated by taking the molecular weight of water as 18, when in = 



18 x ft 



for potassium chloride, twice that number for barium chloride, &c. 



* 'Zcits. f. physik. Chcm.,' vol. 2, p. 73, 1888. t Ibid., vol. 2, p. 78, 1888. 



