THE ACTIOxN OF CHEMICAL AFFINITY. 
217 
case of precipitation will apply mutandis with equal force in tlie case of vola- 
tilization. I am not acquainted with any exceptional instance. 
The testimony of crystallization . — It will sometimes happen that certain quantities 
of AB and CD are mixed in an amount of water which is insufficient to keep in per- 
fect solution AD, should the whole of A combine with the whole of D, although the 
salt itself is a soluble one*. In such a case, if Bergman’s view be correct, either 
no AD will form, however concentrated the solution, or, should double decomposi- 
tion ensue, it will form to the fullest extent possible, and may be expected to crystal- 
lize out at once with something like the rapidity with which precipitation usually 
takes place. If, however, Berthollet’s theory be a true expression of the fact, a 
certain amount of AD will always be fortned, but it may remain dissolved in the liquid, 
although if the whole of A had entered into combination with D it must have sepa- 
rated : yet on concentration AD will make its appearance; and should this, or 
anything else, determine the formation of crystals, or should they ensue on the 
primary mixing, the crystallizable salt is joro tanto put out of the field of action, and 
a redistribution of the acids and bases will take place with further crystallization, 
until an equilibrium is obtained. Now the latter of these deductions describes what 
actually does take place, but there are several circumstances attending crystallization 
from a mixture of salts which are not readily explained, and which I have as yet but 
imperfectly investigated. 
The testimony of diffusion . — Professor Graham has shown'l^ that binary compounds 
vary greatly in the rates at which they diffuse through water. Let it be supposed 
that AB and CD are mixed in one of his diffusion-cells, and that the compounds of 
A diffuse more rapidly than those of C. If no decomposition take place upon mixing, 
the amounts of A and B in the diffusate will be in equivalent proportions ; and so 
will likewise the smaller amounts of C and D. If a complete interchange of acids 
and bases take place, the amount of A in the diffusate will exactly correspond wdtii 
that of D, and in a similar manner C with B. If, however, A and C divide themselves 
between B and D, as the four compounds will be unequally diffusive, it will be very 
improbable that the amount of either A or C in the diffusate should happen just to 
correspond with the amount of either B or D. There is nothing in Pi’ofessor 
Graham’s published researches that will indicate which of these is the case, nor have 
I made any experiments on the subject; but I entertain little doubt that the latter 
result would be arrived at were the matter to be investigated. 
Yet diffusion will never serve as a means of determining numerically the strength 
of the affinities in a mixture; for, supposing the four compounds are actually pro- 
duced, the more diffusive one will speedily pass away from the field of action, which 
* Few if any salts are absolutely insoluble in water, but this will not alfect the reasoning in a previous 
section, for the action of insolubility in a case of crystallization produces the same consequences as in a case 
of amorphous precipitation. 
t Philosophical Transactions, Part I. 1850. 
