Molecular Constitution of Water i 481 



The terms on the right are now all known, as from Cohen's 

 .curves we can estimate with considerable accuracy drj/rjdf at 

 0° when df is 200, 400, and 600 atmos respectively, namely 

 (305, 224, and 192) x 10 -9 , and the other data are contained 

 in the previous sections. Thus in the three cases we get for 

 the left side of (35) the values (55, 56, and 56) x 10~ 7 . The 

 greater part of these values must be due to p2 v 2$Vzldf, so 

 neglecting ]h v idvi/df we find 



-^§=- 00034; (36) 



.that is to say, that an increase of pressure of 100 atmos causes 

 the viscosity of trihydrol at 0° to increase by 3'4 per cent. 

 For ethyl oxide and benzene at 20°, Warburg and Sachs 

 {Wied. Ann. xx.) found a corresponding percentage increase 

 of 7*3 and 9'3, and Cohen for turpentine found 15, and these 

 are all more compressible liquids than trihydrol. In regard 

 to viscosity, trihydrol behaves like a normal liquid, and the 

 exceptional change of the viscosity of water under pressure 

 is due to the dissociation caused by the pressure, which 

 replaces some of the highly viscous trihydrol by the less 

 viscous dihydrol. Cohen's curves show that at about 30° 

 the viscosity of water must be almost independent of pressure 

 up to 1000 atmos ; at this temperature the normal effect and 

 the dissociation effect neutralise one another. 



The preceding treatment of water's viscosity shows that 

 the viscosity of aqueous solutions must be highly complicated 

 on account of the action of the solute in altering the pro- 

 portions of trihydrol and dihydrol in the solution. It is 

 known that the viscosity of some aqueous solutions of solids 

 is less than that of water itself. The reason for this sur- 

 prising old fact is now apparent ; the solute converts enough 

 of the viscous trihydrol into the less viscous dihydrol to more 

 than compensate for the increase of viscosity which its own 

 presence imparts. 



7. Dielectric Capacity. 



Of the dielectric capacity of water and ice we have already 

 made some study in section 2, for electric fields alternating 

 with the frequency of light, since n 2 stands for K, the di- 

 electric capacity. But for K in more slowly alternating 

 fields and in a steady field the experimental results are in 

 apparent conflict. Thus Heerwagen (Wied. Ann. xlix.) and 

 Drude (ibid, lix.) find a steady, almost linear diminution of i 



