284 Messrs. Wright and Thompson on the Determination of 



same limits of probable error) were found to be attainable 

 with a notably smaller number of observations, provided 

 that certain precautions were observed, detailed more fully 

 later on (§§ 149, 150). It was soon found, however, that 

 the observance of these precautions sufficed to permit of the 

 method of opposition being used so as to obtain directly the 

 difference in E.M.F. between two cells alike in all respects, 

 save that the zinc (or copper) sulphate solution used in one 

 was stronger than that in the other. The values thus ob- 

 tained led to sensibly the same mean curves as those obtained 

 with a far greater amount of labour by means of the electro- 

 meter, and, notwithstanding the diminution in the number 

 of observations, the range of probable error was sensibly 

 diminished. 



One of the essential precautions being the avoidance of dimi- 

 nution in E.M.F. in consequence of so-called "polarization," 

 by using currents of only minute density, it at once suggested 

 itself that under these conditions in all probability the differ- 

 ence of potential set up between (for example) two plates of 

 zinc immersed in zinc- sulphate solutions of different strengths 

 allowed to intermix by interdiffusion (or what may be con- 

 veniently called a diffusion-cell) would vary little, if at all, 

 from that set up between the same two plates immersed in 

 the same solutions respectively when these solutions are not 

 directly connected so as to interdiffuse, but are severally con- 

 nected (by any of the means used in two-fluid cells) with 

 solutions of copper sulphate of identical strengths, in which are 

 immersed copper plates of the same surface-character connected 

 together by a wire : i. e. that subsisting between the terminal 

 plates of two opposed Daniell cells containing the same copper- 

 sulphate solution, but zinc solutions differing from each other 

 in strength and respectively identical with those used in the 

 diffusion-cell, when the copper plates are united together and 

 the zinc plates are the terminals. On trying the experiment 

 this was found to be the case, sensibly identical mean curves 

 being obtained for given variations in solution-strength and for 

 given kinds of plate-surfaces, ichether the method of opposition 

 was used, or that of diffusion- cells. 



148. Whichever method of observation was employed, the 

 following general laws were always verified with a high degree 

 of exactitude whenever sufficient observations were taken to get 

 average results with only a small probable error. All the ob- 

 servations were carried out in every instance at a temperature 

 within a very few degrees of 18° C, almost invariably between 

 15° and 20°. 



(1) In any two-fluid cell containing solutions of two me- 



