380 Messrs. Wright and Thompson on the Determination of 



values most consistent with the bulk of evidence at present 

 extant the various E.M.F. determinations described in the 

 previous portions of these researches, it will practically suffice 

 to multiply them by '987, i. e. to subtract 1*3 per cent, of 

 their value, whether the figures be direct determinations of 

 E.M.F. or deductions from calorimetric experiments. 



It may further be noticed that when the value '987 x 10 9 

 for the B.A. unit is taken, in consequence that of Clark's cell 

 becomes '987 X 1*457 = 1*4:38 volt; so that the mean value of 

 the E.M.F. corresponding to the work done in electrolysing 

 water, as directly determined in Part II. § 46, becomes 

 1-4808 x 10 8 C.G.S. units, taking J=41*555 x 10 6 instead of 

 42 x 10 6 , as in Part II. The mean heat-valuation arrived at 

 § 31, Part I., from the results of various experimenters em- 

 ploying the calorimeter, being 34100 gramme-degrees, the 

 product 34100 x JF becomes 34100 x 4355 = 1*4850, which 

 value exceeds 1*4808 by only *29 per cent., a quantity falling 

 within the limits of probable error in the experiments (±'32 

 per cent.). Inasmuch as from various causes a slight syste- 

 matic error of defect necessarily attended the direct determi- 

 nations, the final conclusion is warranted that these direct 

 determinations are sensibly in accordance with the two propo- 

 sitions, firstly that the B.A. unit is actually only -987 x 10 8 ; 

 and, secondly, that Clark's cell has an E.M.F. of -987 x 1*457 

 = 1*438 volt. 



159. The experiments described in § § 153 and 154 lead to 

 the conclusion that a Daniell cell containing solutions of zinc 

 and copper sulphate of equal strengths has an almost invari- 

 able E.M.F. whatever the strengths of the solutions when the 

 plates are of amalgamated zinc and electro-copper; but rises 

 in E.M.F. if the plates are either amalgamated zinc and amal- 

 gamated copper, electro-zinc and electro-copper, or, a fortiori, 

 electro-zinc and amalgamated copper. It seemed desirable to 

 substantiate these deductions more rigorously. Accordingly 

 two pairs of solutions of zinc and copper sulphate were care* 

 fully prepared, such that one pair had the composition 

 2*069 MS0 4 100 H,0, and the other the composition -0937 

 MSO^lOOHjO. Cells were then set up with these fluids, 

 opposed to one another, and series of readings taken. It was 

 found that the numbers thus obtained exhibited much greater 

 divergences and fluctuations than those similarly obtained 

 with cells where the molecular strengths of the solutions were 

 not equal in each case respectively. In the latter case the cur- 

 rent always passed in one direction, and did not exhibit any very 

 marked alteration in magnitude during the first 15 to 30 minutes 

 after setting up (during which period the observations above 



