220 Mr. E. H. Griffiths. [Jan. 17, 



If we assume with Regnault that L is a linear function of 6 we 

 can deduce the values at and 100, and we get 59673 and 536'63 

 respectively. 



Hence Table Y. 



Values of L. 



0. 100. 



Regnault ................ 536'60 



Dieterici ................. 596-73 



Griffiths (extrapolated) ---- 596'73 536*63 



I think that I am justified in calling this agreement remarkable,. 

 and there is evidence that it is not a mere coincidence. 



Winkelmann, in his analysis of Regnault's work, states that the 

 following formula gives the results of Regnault's experiments with 

 greater accuracy than Regnault's own formula : 



L = 589-5 - 0-2972 o - 0'0032147 0+ 0-000008147 3 .... (W), 



Now if we assume my value of dLjdO we get 



" Total heat" = 596'73 + 0'39900 ........ .... (G 2 ) r 



whereas Regnault's formula is, 



Total heat = 606'5-f 0*305 ................. (R). 



The following table gives in column III all Regnault's experimental 

 results (R e ) below 100, except those (below 63) by his other mode 

 of experiment, which I have given reasons for rejecting. Column YII 

 shows the difference between R e and the value given by formula (R). 

 Column VIII gives R e (W), and column IX R e -(G 2 ). 



