towards a Dynamical Theory of Solutions. 25 



derived from the German standard alcohol tables as given 

 in the PhysiTcalisch- Chemische Tabellen of Landolt and 

 Bornstein, 3rd edition, from which source most of the sub- 

 sequent data used in this communication will be derived. 

 It will be noticed that 10 4 (A + p/± 2 /2)/]:> ] p 2 p passes through a 

 point of inflexion, which is actually, or almost, a minimax 

 point near p 1 = 0'5. Its values show a certain symmetry in 

 relation to that at jt? 1 = 0"5 ; for, if this be subtracted from 

 the values for p 1 — 0'9 and 0*1, the remainders are 652 and. 

 735, for ^ = 0*8 and 0*2 the remainders are 324 and 348, and 

 so on. At first this seemed to me to be a fact of perhaps 

 fundamental significance, but we shall see that it is not so. 

 After much study of the data in this 1able and related facts, 

 I have been led to the following explanation of them. An 

 alcohol is added to water it changes more and more of its 

 trihydrol (H 2 0) 3 into dihydrol (H 2 0) 2 , the conversion being 

 practically complete when p 1 = 0'5. Jf this is so, we can 

 calculate the contraction A for jo 1 =0'5 bv means of the data 

 of "The Molecular Constitution of Water" (Phil. Mag. [5 ) 

 1. p. 460, 1900). At 15° C. water is a mixture of 0-333 parts of 

 (H 2 0) 3 of density 0*88 with 0*667 parts of (H 2 0) 2 of density 

 1*09, so that the contraction caused by changing 1 part of 

 water into pure (H 2 0) 2 is 0*333 (1/0*88-1/1*09) = 0-0728. 

 If then J part of alcohol on mixing with J part of water 

 changes it all into (H 2 0) 2 , the contraction must be 0*0364, 

 the corresponding tabulated number for the actual contrac- 

 tion being 0'0410. We shall look more closely into this 

 soon. Consider now the case where the value of p l is nearly 

 1. say 0*99, a little water being dissolved in a great excess 

 of alcohol. We know that when water evaporates into the 

 gaseous state it changes into hydrol H 2 0. It is likely then 

 that when a dilute solution of water is made in a solvent like 

 alcohol, the water is changed for the most part into hydrol. 

 Now in a contribution to the Faraday Society's Discussion 

 on the Constitution of Water (Trans. Far. Soc. vi. 1910), I 

 have shown that water of crystallization is hydrol having a 

 density 1*26 in the liquid state at 0° C. If then 0*01 part 

 of water on mixing with - 99 part of alcohol is changed 

 from density 1*0 to density 1*26, the contraction must be 

 0-01(1 - 1/1-26) = 0-00206 at 0°, the tabulated value being 

 0*00230 at 15°. Thus then we can give a fair account 

 of the phenomena recorded in Table I. by the statement 

 that as alcohol is added to water, it first converts its tri- 

 hydrol into dihydrol, the change being nearly complete 

 when the alcohol is nearly 50 per cent, of the mixture, and 

 then when the alcohol is present in sufficient excess, it causes 



