ItA.TE OF DIFFUSION OF IODINE IN KI 507 



For our case this would mean that the rate of diffusion would approach 

 nearer and nearer to that of the I3 ion as the concentration of the KI 

 increased. As the migration velocity for I3 is certainly less than that for 

 K we should expect the rate of diffusion to fall off with increasing con- 

 centration of KI. This it does not do. (According to Bredig the migra- 

 tion nmnber for I3, is 44.2, according to Crotogino it is 60, that for K is well 

 kno-UTi to be about 65.3 (39) (40) (41). The above forces us to conclude 

 that other and more potent factors are worldng in the opposite direction. 



Discussion of Plate I. A careful review of the factors just discussed 

 leads us to the consideration of the only two independent variables (or 

 variables not knoion to be simply related) having important and traceable 

 effect on the diffusion. These are the fluidity and" the density of the solu- 

 tion. The effect of the fluidity is as one would theoretically expect it 

 to be, at least up to about 2 normal; but it is not clear why the rate of dif- 

 fusion should increase with the density, though it certainlj^ does as is shown 

 b}^ the curves on Plate I. 



The simplest assumption that can be made as to the effects of viscosity, 

 and density, respectively, is that the rate of diffusion varies directly as 

 the density and fluidity. This assumption however can not be even 

 approximately true, for it would necessitate a very much more rapid 

 increase in the rate of diffusion than is shown by experiment. However, 

 the writer has deduced an empirical expression, by means of the three curves 

 in question, which gives a very satisfactory numerical relation between 

 the diff'usion constant and the fluidity and density.. It is, 



50 (K - 1.02) = 100 (/^ - 1) - 20 (p - 1) 

 or 



K = 2i?'-|-0.4p- 1.42 



K calculated according to this equation gives the dotted curve for diffusion 

 on Plate I. The difference between the diffusion constant thus calculated 

 and that actually found by experiment is given in the last column of table 

 VIII. The maximum variation is 2.8 per cent. This is hardly greater 

 than the probable error, and the average variation is certainly within the 

 probable error of experimentation. While it is not to be expected that 

 this equation expresses the exact relation, it is probable that it is a fairly 

 close approximation. It is of intereest to note that the curve thus calcu- 

 lated shows a maximum at a concentration of about 3.5 normal — a point 

 at which we have no experimental data — and falls to very nearly the 

 observed value at 4.5 normal, the next point at which we have an experi- 

 mental value for the diffusion. The writer hopes at some time in the near 



