296 Messrs. Wright and Thompson on the Determination of 



It is evident that, having regard to the impossibility of 

 completely eliminating errors from difference of character of 

 plate-surface, these numbers coincide sufficiently closely to 

 show that whether no current at all circulates (electrometer), 

 or a small current not too great to set up any notable " pola- 

 rization," sensibly the same numbers result whether the observa- 

 tions be made by carefully valuing separate cells and subtracting 

 the values from one another ; by opposing two cells and measur- 

 ing the current due to the difference in their E.M.FSs ; or by 

 means of diffusion- cells (representing the terminal liquids of 

 two opposed cells of Raoult's form, when directly connected). 

 The same conclusions also follow from the observations detailed 

 in § 154 made with copper- sulphate solution and electro- 

 copper plates. 



2. Electro-zinc Plates. 



Observations made with Diffusion-cells only : probable error 

 in no case exceeding ±*75 millivolt. 





w=3-005 



^=1-016 



»=-'10l2 



w=5-584 

 3-005 

 1-016 



11-2 



20-0 



8-8 



30-5 

 19-3 

 105 



3. Bright Zinc Plates. 



Observations with Diffusion -cells only: very much less con- 

 cordance between the results of these experiments than with 

 other kinds of zinc plates, and much more alteration (and of 

 more irregular character) on allowing cells to stand a few 

 minutes. 



Probable error about + 1*5 millivolt. 





n= 3-005 



rc=-10!2 



w=5-584 

 =3-005 



100 



296 

 196 



The following table represents these three sets of values 

 reduced by interpolation to a common origin (m = 5'0) and 

 common abscissae: the curves respectively marked 1, 2, and 3 

 in fig. 1, Plate VI., indicate the three sets of reduced values 

 thus obtained. 



