486 UNIVERSITY OF VIRGINIA PUBLICATIONS 



Two years later A. A. Noyes showed that the amount of free iodine 

 (I2) in solutions of KI far more dilute than any with which we worked is 

 negligible (26). 



Still more recently Osaka showed that the freezing point of a solution 

 of KI was actually raised by the addition of iodine. In fact there was a 

 small rise in the freezing point as found by Le Blanc and Noyes, but they 

 put this down as experimental error. However, Osaka showed that though 

 small it was easily measurable, and further that it was greater for HI than 

 KI. From this he concludes that KI3 and HI3 must be somewhat less 

 dissociated than KI and HI. He does not make quantitative computations 

 because he feels that the data is not sufficiently accurate for this, though 

 undoubtedly sufficient for a qualitative statement as made above (27). 



Work of the writer under the direction of Dr. Edgar. In October, 1912, 

 the writer, under the direction of Dr. Graham Edgar, began a series of 

 experiments having the end in view to determine if possible the rate of 

 diffusion of iodine in solutions of KI of various concentrations. In all 

 cases it was decided to have the solutions of normal strength or some simple 

 multiple thereof. The iodine was assumed to be present as KI3, and 

 the calculated amount of KI was added extra to the solution containing the 

 iodine. That is to say, the concentration of the KI exclusive of that 

 forming triiodide was the same in both solutions (layers). 



The apparatus used was modeled after that of Stefan, utilizing the four 

 layer scheme and making use of Stefan's tables (28) (29). As the values 

 given in these tables required considerable interpolation, a set of curves 

 was drawn on a suitable scale from which the values for X could at once 

 be read. 



The apparatus used by Edgar consisted of a flat-bottomed glass cyUn- 

 der of about 100 cc. capacity, carefully selected to have uniform diameter 

 and almost flat bottom. A mercury bottom could not be used because 

 the mercury would be attacked by the iodine; nor was any hquid known 

 suitable to use for this purpose. Three portions of KI solution of about 

 25 cc. each were first run into the cyhnder by means of a separatory funnel 

 having a thin walled capillary tube for a stem. This capillerry was set 

 Avithin the cut-off stem of the funnel by means of hard parafiine, great 

 care being taken to prevent any paraffine getting on the inside of the capil- 

 lary tube. This tube extended do'ma to within 0.2 to 0.3 mm. of the bottom 

 of the cylinder, thus causing the inflowing liquid to spread out into a thin 

 sheet as it entered the cylinder. If the layer containing the iodine was 

 slowfy and carefully run in there was no mixing of the layers so far as could 

 be detected by the eye. When the last layer had been run in there was 



