320 Dr. J. H. Long on the 



of the substance employed. Their accuracy has also been 

 questioned more than once*. 



The importance of the subject seemed to make another in- 

 vestigation necessary ; and this, at the suggestion of Professor 

 Lothar Meyer, I have undertaken. The method described in 

 the following pages was likewise proposed by him; and part of 

 the experiments were conducted in his laboratory at Tubingen. 

 Another part of the work was performed in the physical labo- 

 ratory of Professor 0. E. Meyer in Breslau. I would take 

 this opportunity of expressing to both these gentlemen my 

 sincere appreciation of the many kindnesses shown me by them 

 during all the time spent in their laboratories. 



The New Method. 



There are several serious objections to the methods of Gra- 

 ham, Beilstein, and Marignac. In applying them, one is un- 

 able to follow the progress of the diffusion ; that is, one cannot 

 tell with any accuracy how much salt diffuses from hour to 

 hour, or from day to day even, nor when the condition of con- 

 stancy has been reached and how long it lasts. 



These methods give only the average velocity of diffusion 

 for the duration of the experiments. During the first days it 

 may be relatively large, and during the last relatively small. 

 Another objection to the methods is, that in them the diffusion 

 takes place into a medium of constantly increasing density. 

 This increase, being different for different substances, must 

 exert in each case a different disturbing force, whose amount 

 we have no means of determining. Hence it seemed desirable 

 that a method should be employed in which these and other 

 sources of error would be avoided. Such a method is the one 

 proposed by Professor Meyer ; and its chief features are 



(1) that it allows a determination of the rate of diffusion from 

 hour to hour, or even at shorter intervals if necessary, and 



(2) that the diffusion takes place into a medium whose concen- 

 tration is always zero. 



How these ends are attained will be best understood from 

 a consideration of the diagram and its explanation. 



Fig. 1 (Plate VII.) represents the profile of the apparatus 

 in natural size, m n op is a beaker-glass into which fits the 

 bent tube ar stb, whose internal diameter is about one and a 

 half millimetre. At ef this tube is joined to the larger tube 

 efg h, whose internal diameter is fifteen millims. This latter 

 tube is open below ; and the appearance of the whole as seen 

 from above is represented in plan in fig. 2. 



The tube arst and the beaker-glass are filled to the level 

 * For instance, see Voit, loc. cit. 



