Dr. E. J. Mills on Elective Attraction. 521 



to the double formula of thallous nitrate^ or TP(NO^)^; and 

 Ti^ (NO^)^ cannot be admitted into the series 2, because it con- 

 tains (NO''^)^ instead of (NO^). Since_, then, we can only take 

 into account one half of Tl^(NO^)^ the actual symbolic value^ it 

 becomes necessary to represent the double energy of thallous 

 nitrate by doubling the experimental value of u, which then 

 becomes 8'76. This argument is justified by the natural posi- 

 tion in which it places thallous nitrate — that is, near to argentic 

 and plumbic nitrates. 



(2) The exact significance of a is not difficult to ascertain. 

 AYhen one nitrate surpasses another in the power of fixing chlo- 

 rine per unit of phosphoric oxide, that is an index of superior 

 chemical activity, or, to name the ultimate cause, attraction. 

 When this attractive effect can be specified for several nitrates 

 in a series, thereby valuing the choice of the oxychloride, it is 

 shown that the phenomenon is of that kind which Bergman 

 termed " elective attraction/'' Hence u is the coefficient of elec- 

 tive attraction, or elective coefficient of the nitrates. 



At first sight it might appear that a coefficient of this kind 

 may be evaluated with equal propriety on the basis of phosphoric 

 oxide formed per unit of chlorine fixed. But if we were to 

 adopt this supposition, we should find the following results. 

 The value of the coefficient in the case of lithic nitrate would be 

 0*62, in the case of thallous nitrate O'll. Now, although a de- 

 pends essentially on the mere occurrence of the reaction, yet, in 

 forming this a priori estimate of it, we must be guided by the 

 general account of chemical change effected. A reference to 

 the tabulated results will show that as much as about one half 

 the thallous nitrate became active in one experiment ; in another 

 the whole of the lithic nitrate became active ; and these are the 

 extreme terms of the series. Suppose these (absolute) weights 

 are to each other as 1:2, and compound them (1) with the 

 ratio 8-76 : 1-61 (say, 5 : 1), and (2) with the ratio O'll : 0*62 

 (say, 1:5). "We have in the former case the relation 5 : 2, in 

 the latter I : 10. Bearing in mind the fact for which we have 

 to provide, that any nitrate in the series might become wholly 

 active*, which would make the ratio of absolute weights 1 : 1, 

 it is clear that we must select that construction for u which 

 causes its value, when compounded with the ratio of such abso- 

 lute weights as have been selected, to approach the more nearly 

 to 1:1. Now 5 : 2 is nearer to 1 : 1 than is 1 : 10; and the 

 mode of evaluating a adopted in this memoir is therefore more 

 reasonable than the inverse method. 



In order to render my argument more plain by illustration, I 

 shall recur to the conception of stock as already mentioned. Sup- 

 * Compare Argentic nitrate, exp. III. 



