358 REPOKT— 1886. 



of a masterly memoir by Prof. F. Kohlrauscli on the same subject throws it into 

 the shade ; for there can he no doubt that while the ground covered by both is 

 similar, the Kohlrausch memoir is greatly superior, both in the experiments made 

 and in the discussion upon them.' 



A brief abstract of this portion of Dr. Arrhenius's paper is all, therefore, that is now 

 necessary ; and if my criticisms on parts of it appear in any case caustic, I must 

 express my regret to the author for the adverse opinion, and trust that my appre- 

 ciation of a great deal in the second part will compensate for it to some extent. 

 It sometimes seems as if the author allowed himself occasionally to indulge in an 

 exploded type of reasoning, wherein, by manipulation of imaginary data, a confu- 

 sion is produced, out of which emerge several laws more or less in agreement 

 with experience, which are thenceforth labelled and referred to as theoretical 

 deductions It may be, however, that the italicised and numbered statements 

 throughout the paper are not intended for strict statements of deduced law, but 

 are merely summaries of more or less probable truth. In that case it is their form 

 only which is misleading, and one would judge them by a different standard. I 

 proceed to give an account of the contents. 



§§ 1-8 are devoted to an account of the experimental method employed, and 

 to a justification of it. 



The method consists in the use of a differential galvanometer, one of whose 

 branches consists of ordinary wire, while the other contains a current alternator 

 in the shape of a wheel turned by hand, with alternate bars on its periphery rubbed 

 by two springs (Edlund's ' depolariser '). The alternating current from this in- 

 strument is led to a switch, which is able at pleasure to throw into circuit either 

 the experimental tube of liquid or a wire resistance box. 



A battery is arranged so that its current divides fairly equally between the two 

 branches of the galvanometer ; and one portion of the current, after being rendered 

 alternating by the above arrangement, is diverted by the switch either to the 

 electrolyte or to the adjustable resistance-box. The resistances of the liquid 

 and of the box are considered equal when the deflexion of the galvanometer is in- 

 dependent of the position of the switch. 



I must confess to surprise that this can be a satisfactory mode of measuring 

 resistance : self-induction and electro-chemical capacity being so mixed up in it,^ 

 and the so-called depolariser being a very unsatisfactory instrument. Apparently, 

 however, results fairly comparable with each other can be obtained. 



§ 9. Calculation from the Experimental Data. — First, the conductivity of the 

 original water used as solvent is subtracted fi-om that of the solution, and the re- 

 sult claimed to be the conductivity of the salt alone. For this ride, as an empirical 

 process justified ^ posterioi-i, there may be something (not much) to be said. But 

 as a corollary from a ' law ' subsequently proved in this paper (§ 15, law 2) it 

 is unreliable. Probably not much liarm is done by using this rule in the special 

 cases considered ; at anj^ rate, Kohlrausch does the same thing : but it is well to 

 notice that the rule k = k^ + k.-,, for the conductivity of a mixture, if stated as a 

 general and a priori law, assumes two things : 



1st. That the law of divided circuit must hold whenever two conductors 

 are intermingled ; which is untrue.^ 



' See § 6 of a letter by Dr. Ai-rhenius printed on p. 386. 

 - See § 1 of a letter by Dr. Arrhenius printed on p. 384. 



' Dr. Ai-rhenius in his letter, page 38.5, naturally and justifiably objects to this 

 statement as dogmatic. The reason it is here put so briefly is tliat the point was 

 referred to in my last year's paper, see Aberdeen Eeport, top of page 728. Observe, 

 I only object to the assumption that the law of divided circuit must be true : it mai/ 

 be true in some cases, but a possibility cannot be made the basis of a deduced law. 

 In order that the combined resistance of two solutions when mixed may be the semi- 

 harmonic mean of their separate resistances when alone, the following conditions 

 are necessary and suflScient : 



(1) The solutions must not affect each other's conductivity in any way ; the 

 fact of mixture must not increase dissociation or change viscosity, 

 for instance. 



