Thermochemical Constants. 61 



2FeS0 4 aq.,0, S0 8 aq. . . . = 2x18,400 



2Cr0 3 aq.,3S0 3 aq = 2x18,442 



Mn 2 7 aq.,2S0 3 aq = 4x18,162 



Mn 2 7 aq.,5H 2 2 aq.,2S0 3 aq. . =10x18,924 

 Mn 2 7 aq., 5H 2 2 aq., 2H 2 C1 2 aq. = 10 x 18,214 



The " perfect analogy " of these various reactions, even 

 when we confine ourselves to the members of each separate 

 group, is certainly of a somewhat questionable character, and, 

 in my opinion, cannot be regarded as existing in even the 

 most superficial sense ; and where an analogy does exist, as in 

 the formation of the six sulphates, we might expect the heat to 

 be the same in every case, but can attach no meaning what- 

 ever to its being a multiple of some constant; nor can we 

 understand why certain other perfectly analogous reactions 

 should not exhibit a similar relationship if this constant had 

 any true signification. 



The variation, also, in this constant is somewhat large, 

 amounting to 7 per cent, or x x 1,226, where the value of x 

 may be as high as 18, giving a maximum error of 22,068, a 

 quantity larger than the constant itself. 



That this constant is of an entirely fictitious nature may be 

 shown most palpably by taking any other number and tabu- 

 lating all the thermal values which are multiples of it. The 

 number 15,000 was taken in this way, and all the quantities 

 which are multiples up to 10 of it, within an error of + 2*3 

 per cent, only, are given in the accompanying table. They 

 were selected from the data referring to the non-metallic and 

 metallic combinations tabulated by Thomsen in vols. ii. and 

 iii., and though they amount to nearly 120, they could no 

 doubt be considerably increased by a more thorough and ex- 

 tended search. Out of these it is evident that we could form 

 many tables of " analogous compounds," the heat of forma- 

 tion of which is some simple multiple of our " constant " 

 15,000, and a similar investigation would show that any 

 number would act just as well as a " common constant of 

 affinity/' 



The whole question, therefore, of these constants disappears, 

 and resolves itself into the chances of coincidences occurring 

 in a certain number of numbers. 



