422 Dr. J. H. Long on the 



stant volume of liquid — 10, 20, 40, and 80 grains of salt in 

 1000 watergrain measures of the solution." In Graham's 

 first paper*, the percentages referred to parts by weight of 

 the solution. These values I have reduced to grams per litre, 

 and the amounts of diffused salt likewise to grams. Then 

 constructing a curve from these data, I have taken out the 

 value corresponding to the amount of salt used per litre in my 

 experiments. For instance, I used an n solution of BaCl 2 — • 

 that is, a solution of 104 grams per litre, or, corresponding to 

 the above, a solution of 10*4 per cent. On the curve I have 

 found the amount which should be diffused from a solution of 

 this concentration in the Graham method and compared this 

 with the value found by my own. These amounts depend on 

 the dimensions of the diffusion-vessels ; that is, in the general 

 formula established by Fick, 



But I may be taken as equal to V, as the distances between 

 the levels of concentration-zero and those of the strong solu- 

 tions are about the same in both cases. Then 



q : q' : : S : W, 

 or 



in which q is the section of the neck of Graham's cells, q r a 

 section equal to those of the cells used by me, S the amount 

 of salt diffused in a given time in a Graham cell, and W finally 

 the amount which should be diffused in a Graham cell of the 

 same dimensions as those of the cells used by me. This latter 

 is the value sought. S, q, and q f are known ; and if we con- 

 sider the lower section of the neck of Graham's cells as deter- 

 mining the amounts diffused, we may write 



S'= 7r 7' 2 5 ' S =0-226 S. 

 7rl5*24 2 



If S represents the amount of salt diffused in one day taken 

 as the unit of time, no further reduction is necessary ; if not, 

 it simply remains to divide by the number of days of the ex- 

 periment. This gives, of course, only the mean value for the 

 rate of diffusion during the whole time ; but that is the best 

 which can be done under the circumstances. 



In this way I have reduced Graham's results ; and the values 

 will be found below side by side with my own. No reduction 



* Phil. Trans. 1850, p. 1. 



