RATE OF DIFFUSION OF IODINE IN KI 489 



there is no check in the two layer system there was no advantage in re- 

 moving tlie second layer. (Titration being very much more accurate than 

 the other factors involved.) Some of the original iodine solution was 

 alwaj's placed in the bath along with the C3^1inders, and a pipette full was 

 removed at the same time that the samples were removed from the cylin- 

 ders. A small porcelain dish was used to receive the iodine solution as 

 it came drop by drop from the delivery tube (a, fig. 1). As soon as a 

 sample was removed it was transferred to a small glass stoppered flask, a 

 pipette full of the original iodine solution being also put into a similar 

 flask. When all four samples had been removed, the iodine in the flasks 

 was titrated with sodium thiosulphate (approximately n/75). Starch 

 paste was used as an indicator. Thiosulphate as weak as this will change 

 strength from day to day; but this introduces no error, as a fresh sample 

 of the original iodine solution is titrated every time a determination is 

 made. The equation used to determine k is: 



r' 2.30 ,__ 8 Vo 



(The derivation of this formula will be given later.) Here h is the diffusion 

 constant, t the time in days, Fo the total amount of iodine in the cylinder, 

 Fi, the amount of iodine in the top layer after time t has elapsed, and r 

 is the length of the total liquid column, i.e., both laj^ers. It will be noted 

 that the quantity F, occurs to the first power in both denominator and 

 numerator of our fraction (and there only); hence, it does not matter in 

 what unit V is measured. This is of great practical value, for it allows us 

 to be indifferent to the exact strength of thiosulphate used. 



In a few cases, instead of placing the cylinder in the thermostat it was 

 placed in a Dewar flask, being securely held in place by a well fitted cork. 

 This was possible only in siunmer, and at such times as the temperature 

 of the room was nearly at 25° and fairly constant. The flask was filled 

 with water at as nearly room temperature as possible, the empty cylinder 

 and thermometer inserted, and the whole left for a day or so, in order that 

 the system might be in thermal equilibrium. Then the experiment was 

 carried out as in the other cases. The temperature recorded for the 

 experiment was the mean observed (4 times a day) during the run. Con- 

 stants thus obtained were reduced to 25° by means of the formula given 

 on a previous page (16). 



It was only rarely that conditions permitted of the use of this method, 

 and as it offered no special advantages over the use of the thermostat, it 

 was only used a few times. The one interesting thing learned from this 



