40 Mr. W. Sutherland on Weak Electrolytes and 



for eacli value of p u so we can calculate r 2 , that is (n 2 2 — l)/ 

 (n 2 2 -\-2)p 2 for the changing liquid 2, water, the change in 

 the constitution of the water showing itself by small changes 

 in the value of r 2 '. For alcohol and water the data are 

 those of Hess at 15° C. for the line F of the spectrum. 

 From the density 0*80889 at 15° 0. of Hess's alcohol it 

 appears that it was a 95 per cent, spirit by weight. This 

 fact will not interfere with our method of inquiry, which is 

 the main thing at the present stage. With alcohol (95 per 

 cent.) as liquid 1 and water as liquid 2 the results are : 



lQBp, 20751 40890 59993 79865 



10 5 r 2 ' 20854 20783 20738 20751 20879 



If these data are represented on a graph they show a 

 minimum of r 2 near ^ = 0*5, perhaps at 0*46, having a 

 value 0-20733. Since at 15° C. water is two-thirds (H 2 0) 2 

 and one-third (H 2 0) 8 , and at 0° C. r' = 0*209G8 for the D 

 line of the spectrum with (H 2 0) 3 and 0*20434 for (H 2 0) 2 , we 

 find by proportion that at 15° C. with the F line r' = 0*21214 

 for (H 2 0) 3 and 0-20674 for (H 2 0) 2 . Thus the decrease in 

 r 2 with increasing p 1 from pi = to 0'5 means that as the 

 alcohol is increased it converts more and more trihydrol 

 into dihydrol, as we found to be the case in studying con- 

 traction. If the conversion were complete and uncomplicated 

 by any other change when p 1 = 0'4z6 and the mixture is 

 (C 2 H 5 OH) 2 with 3(H 2 0) 2 . then the minimum found for r 2 

 ought to be 0-20674 for (H 2 0) 2 instead of 0*20733 as esti- 

 mated from the graph of the above data. On the whole 

 refraction confirms the conclusion from contraction, that up 

 to jt9i = 0-5 or 0-46 the chief action of alcohol on water is to 

 change all its trihydrol into dihydrol. The refraction data 

 would make it appear that the change is never complete, 

 but the increase of r 2 f when jo 1 >0'5 shows another action 

 taking place, and perhaps part of this action occurs before 

 jo 1 = 0*5 and so masks part of the change in r 2 due to change 

 of (H 2 0) 3 into (H 2 0) 2 . This other action we identified in the 

 study of contraction as the change of dihydrol into hydrol. 

 If this is correct, it appears that r' for hydrol is greater than 

 for dihydrol. This raises a temporary difficulty, because in 

 my contribution to the Faraday Society discussion I found 

 that hydrol in the form of water of crystallization has a 

 much smaller r' than dihydrol or trihydrol. The last little 

 table makes it plain that here we have to do with hydrol 

 possessing a larger r than dihydrol. A graph of the data 

 shows that for an infinitely dilute solution of water in 

 alcohol r 2 ' has a limiting value like 0'2107 or 0'2110. 



