326 Dr. J. H. Long on the 



the solution, which at the time being was rather concentrated, 

 from the tube i. The amount of this displacement was much 

 smaller than was expected. Instead of sinking to ef or lower, 

 the blue layer remained stationary at c d; and the surface sepa- 

 rating this from the solution was very sharply defined and 

 perfectly reflecting. This remained the case during the day, 

 and, left to itself several days longer, was but little changed ; 

 a sharply-defined surface was still to be seen. 



This allowed of but one explanation, viz. that with a solution 

 of this concentration the position of equilibrium between it and 

 the water was such that their surface of separation could only 

 be found quite high in the tube i. In such a case, not only 

 diffusion but also a mechanical carrying-over of salt was ob- 

 served. An JSTaCl solution of much smaller concentration 

 (1*07) was next taken, and quite a different result obtained. 

 At first the sharply-marked surface was seen as before, but 

 lower in the tube, and in the position ef (fig. 1). Left until 

 the next day the surface of demarcation was still very plain, 

 but its reflecting-power was gone ; and on the third day the 

 surface itself had become indistinct, disturbed by the diffusion 

 from below. The experiment was repeated several times with 

 different solutions, and always with the same result ; from 

 which it was seen that in general for solutions of sp. gr. 1 # 10 

 and over, and for velocities of the flowing water not exceeding 

 70 cubic centims. per hour, the position of equilibrium of the 

 liquids in the tubes was such that their surface of separation 

 was above the line ef, and the progress of the diffusion 

 remained for a great length of time obscured more or less by 

 an excess of mechanically removed salt. 



On the contrary, if the solutions be weaker, the surface of 

 separation is found at the outset in the position efoY there- 

 abouts. A real diffusion soon commences in i, and gradually 

 extends itself to g h, below which point the concentration can 

 decrease but very slowly, as explained above. It is now easy 

 to account for the anomalous action of the 3nKCl solution. 

 This has a density of 1*13 — that is, somewhat greater than 

 that which was experimentally found to stand at the level ef. 

 While in the case of solutions of medium density the diffusion 

 may be considered as taking place between g h and ef, in this 

 case it may be considered as taking place between the same 

 lower level and an upper level c d, which is above ef, and 

 more directly in the path of the flowing water. 



One would expect in this case that more particles would be 

 carried over than are really diffused; and this the results show. 



By means of the blue water the influence of velocity can be 

 nicely illustrated. The experiments undertaken to decide this 



