262 Dr. C. R. A. Wright on the Determination of 



formula as being probably more accurate than tbose made by 

 Bosscha's alleged corrected formula. 



32. A number of observations, the details of which are 

 given in Part II., show that, admitting the accuracy of Lati- 

 mer Clark's valuation of the E.M.F. of his standard cell(Proc. 

 Roy. Soc. 1872, vol. xx. p. 444) as being 1-457 x 10 8 C.G-.S. 

 units at 15°, the value of e in the expression #=E — H%J 

 (§ 11) is 1-5003 x 10 8 C.G.S. units when water is electrolyzed 

 at a temperature of 15°-20°, the value of J being taken as 

 42 x 10 6 — the probable error of the mean result being ± 0*0048, 

 or ±0*32 per cent. Owing, however, to certain causes de- 

 tailed in Part II. (§ 39), this value is slightly too low — this 

 result being esentially due to the impracticability of collecting 

 every trace of gas evolved, so that the value of H (the heat 

 evolved per gramme-equivalent of water decomposed) is 

 slightly overvalued, and hence e slightly undervalued. 



From the combustion experiments of previous investiga- 

 tors, summarized in the preceding eight paragraphs, it results 

 that the value of e in the expression e = H^J (§ 11) is 

 1*5038 x 10 8 (since IF = 34,100), J being, as before, taken as 

 42xl0 6 . These numbers, 1-5003 x 10 8 and 1-5038 x 10 8 , 

 accord so closely as to show that (so far as these experiments 

 can decide) the value of J cannot be far from 42 x 10 6 . Ap- 

 plying the formula 



HP 



J= (H + H') X 



(§ 24) to the experimental data given in Part II., the average 

 value deduced for J is 41*96 x 10 6 , this value for J giving the 

 following values for e : — 



From formula e= E-H % J . . e= 1*5023 x 10 8 

 „ v e=W x J • • • 6= 1-5024 xlO 8 . 



Since, however, the mean value of H is slightly overvalued, 

 it results that the value of J finally deduced is slightly above 

 41*96 x 10 6 , with a probable error of less than ±0*4 per cent., 

 i. e. of less than ±0*016 x 10 6 , due to the electrolytic experi- 

 ments, the probable error in the determination of H / not being 

 taken into account. 



33. It is evident that, considering the nature of the data 

 from which this conclusion is drawn, not much dependence 

 can be placed on the final value of J deduced except as an 

 approximation. It is noteworthy, however, that this value of 

 J very closely agrees with that deduced by Joule in 1867 

 B. A. Reports, 1867), by determining the heat producible 

 in a wire of known resistance by a known, current, and that 



