318 Dr. J. H. Long on the 



substances. A short description of Graham's method will not 

 be out of place here. Into a cylindrical glass vessel 87 millims. 

 wide and 152 millims. high was poured 0*7 litre of water ; and 

 under this, by means of a small pipette, 01 litre of salt-solu- 

 tion was carefully brought. The water was displaced upward 

 by the heavier solution ; and the surface separating the two 

 media remained sharply defined. The total length of the 

 liquid column was 127 millims. 



The vessel was now allowed to remain at rest a number of 

 days; and at the end of that time the liquid was carefully drawn 

 off, layer by layer, in sixteen equal portions, of 50 cubic cen- 

 tims. each. The amount of diffused salt was determined in 

 each one of these portions separately, with the exception of 

 the two lower ones, which were analyzed together. 



As an example of these determinations the following may 

 be given: — 



Diffusion of a 10-per-cent. solution of NaCl in pure water 

 for 14 days at 10° C. 



x-r p 1o rnv Grams of NaCl 



No,ofla y er - diffused. 



1 0-104 



2 0-129 



3 0-162 



4 , 0-198 



5 0-267 



6 0-340 



7 0-429 



8. 0-535 



9 0-654 



10 0-766 



11 0-881 



12 0-991 



13 1-096 



14 1-187 



15. 



} 



2-266 



Graham experimented with a large number of substances 

 according to this method, and arranged the results in tables 

 similar to the above. These results Stefan has made the sub- 

 ject of an analytical investigation, whose object was to ascer- 

 tain whether or not they stand in harmony with the theoretical 

 requirements of Fick's law. This he has shown to be not 

 always the case. But in many instances, however, the agree- 

 ment is very close, as illustrated in the following Table, found 

 on page 4 of Stefan's paper : — 



