474 Mr. W. Sutherland on the 



The average tension per square centim. of the film /=a/D. 

 Above we have found /= 5380 atmos and 



D = 12-6xl0- 9 centim. 



We can now calculate p x in equation (21), that is the 

 amount of 1 in one gramme of the surface-film. For if we 

 divide the surface-tensions of water at different temperatures 

 by 12' 6 x 10 -9 , we get the average tensions per unit area in the 

 surface-film as given in the table below. But if we divide 

 the values of <p\ y derived from those of p 2 in Table I., by the 

 values of dpjdf at the same temperatures from Table VII., 

 we get the tension required to convert all of 1 in the body of 

 the water into trihydrol in the surface-film. The difference 

 between the former tension and the latter, expressed as a frac- 

 tion of the latter, gives that fraction of p x which has not been 

 converted into trihydrol, so that if we multiply this fraction 

 by p x for the body of the liquid, we get p x for the surface- 

 film as required in equation (21), and as given in the next 

 table along with the derived values of u x . 



Table X. 



t , 60° 80° 100° 



a 6425 6077 57-11 dynes 



fr+dpjdf 6210 7160 8420 atmos 



/ 5120 4840 4550 atmos 



lO 3 ^ in surface... 131 248 360 



ai 61-28 57-48 53-48 dynes 



^(18^)1 429-3 397-6 365-2 



^(36/Pi) 1 681-4 631-0 579-6 



The values of da^lS/p^/dt between 60° and 80° and 

 between 80° and 100° are 1-585. and 1-62, mean 1-60 ; while 

 for da^&lp^ljdt the values are 2*52 and 2'57, mean 2*54. 

 These are about equally different from 2°121 for an average 

 liquid ; but we can more satisfactorily test whether 18 or 36 

 is the molecular mass of our ingredient 1 by assuming that 

 the critical temperature of water 368° C. is that of 1, and 

 therefore «i(18/ p-^ and a 1 (36/p 1 )* must both vanish at about 

 368°; so that on dividing their values tabulated above by 

 308, 288, and 268 respectively, we should get much more 

 reliable values of the appropriate coefficients. In this way 

 for « 1 (36//3 1 )* we get the temperature-coefficients 2*21, 2*19, 

 and 2*16, mean 2'19, whereas the mean for the similar co- 



