264 Dr. C. E. A. Wright on the Determination of 



megalergs respectively when Regnault's formula for the spe- 

 cific heat of water (§ 25) is taken ; whence the mean value 

 of J is 41*555 megalergs from the two sets of friction 

 experiments (1850 and 1878), and 42*10 by the electric-cur- 

 rent method (1867) — the latter consequently being - = 



1*0131 times the former, or 1*31 per cent, in excess. The 



value for J deduced above from the writer's experiments (viz. 



41*96 

 41*96 megalergs) is ... ^rr =1*0098 times that got from 



Joule's friction experiments, showing an excess of 0*98 per 

 cent. Hence, on the whole, the writer's numbers and Joule's 

 1867 values concord much more exactly than might have been 

 anticipated, each showing an excess over Joule's water-friction 

 value (1850 and 1878) of about 1 per cent., Pegnault's 

 formula being employed for reducing the observed rise of 

 temperature to gramme-degrees at 0°-l° C. 



35. Reasons have already been given (§ 32, videniso Part II. 

 § 39) for supposing that the value 1*5003 found for the E.M.F. 

 representing the affinity for oxygen and hydrogen in water is 

 slightly understated. The effect of an error of this kind would 

 be to raise the value of J deduced by the formula 



and hence to make the excess of the value of J thus deduced 

 over Joule's water-friction value to be somewhat more than the 

 1 per cent, or thereabouts just calculated. It is worthy of 

 notice that there is also reason for supposing that Joule's 1867 

 value is understated. Experiments are described in Part II. 

 the result of which is to show that an unobserved source of 

 error existed in Joule's 1867 experiments, the effect of which 

 was to diminish the value of J thence deduced by an amount 

 probably equal to some tenths per cent. ; so that on the whole 

 Joule's 1867 value, when corrected, is probably from 1*5 to 

 2*0 per cent, higher than his water-friction value (1850 and 

 1878). 



This difference between the two values is precisely that 

 which would subsist did an error to an equal amount exist in 

 the B.A. resistance-unit valuation : i. e. if the B.A. unit were 

 1*015 to 1*020 earth-quadrant per second instead of being 

 exactly 1 earth-quadrant per second, the value of J deduced 

 from Joule's 1867 experiments would be 1*015 to 1*020 times 

 the true value ; for it is calculated by the formula 



T C 2 R* " 

 J=- H -, 



