Dr. E. J. Mills on Elective Attraction, 523 



proportioned to symbolic value. It is further sufficiently appa- 

 rent that (excepting rubidic nitrate) a and 1, increase and dimi- 

 nish in the same general order. 



Assuming the correctness of this law^ and a possible small 

 interference with it for each nitrate^ the occurrence of somewhat 

 wide variations from the mean value of a in the case of plumbic 

 and argentic nitrates, while in the case of nitrates of lower sym- 

 bolic value that variation is very small, meets with an evident 

 and satisfactory explanation. 



(4) The quotients Q represent the weights of nitrates which 

 correspond to the unit of elective attraction. They are therefore 

 strictly equivalent numbers in the legitimate sense of that term. 

 Within the limit of experimental error they constitute an incom- 

 plete arithmetical series, the most probable value of whose first 

 term, as determined by the method of least squares, is 6*258 ; so 

 that Q = m6'258, m being integral. Such a series is already known 

 to chemistry. The following remarks*, containing the first an- 

 nouncement of its existence, are evidently applicable to my own 

 experiments, which were very mainly concerned with the diffu- 

 sion of a solid nitrate in liquid phosphoric oxychloride. " The 

 fact that the relations in diffusion of different substances refer to 

 equal weights of those substances, and not to their atomic 

 weights or equivalents, is one which reaches to the very basis of 

 molecular chemistry. The relation most frequently possessed 

 is that of equality, the relation of all others most easily observed. 

 In liquid diffusion we appear to deal no longer with chemical 

 equivalents or the Daltonian atoms, but with masses even more 

 simply related to each other in weight. Founding still upon 

 the chemical atoms/we may suppose that they can group toge- 

 ther in such numbers as to form new and larger molecules of 

 equal weight for different substances, or, if not of equal weight, of 

 weights ivhich appear to have a simple relation to each other. It 

 is this new class of molecules which appear to play a part in 

 solubility and liquid diffusion, and not the atoms of chemical 

 combination.^' The next step in this direction was taken by 

 Chizyiiskif, who performed a series of remarkable experiments 

 on the fractional precipitation of mixed calcic and magnesic 

 chlorides by means of a phosphate. He found that '^ chemical 

 action is proportional to the product of the chemical masses into 

 their coefficients of affinity,^' and that " equal masses of calcic 

 chloride and magnesic chloride have equal, but oppositely active, 

 coefficients of affinity.^'' Here, then, we have a third group of 

 salts, in which, as chemical evidence shows, the real or dyna- 

 mical " equivalent '^ is an " equal weight.''^ 



* Graham, Phil. Trans. 1850, p. 46. 

 t Ann. Chem. Pharm., Supp. iv. p. 226. 



