482 Mr. W. Sutherland on the 



water with temperature from 0°to 80°; while Thwing (Zeitsch.. 

 f. Phys. Chem. xiv.), with rapidly alternating oscillations 

 (period not given), finds the capacity rise to a sharp maximum 

 at 4°. In his measurements for different mixtures of alcohol 

 and water at a fixed temperature, instead of a steady curve 

 representing the variation of K with composition, he gets a 

 curve containing cusps at points which correspond to the 

 formulae C 2 H 6 + 6H 2 0, C 2 H 6 + 3H 2 0, and C 2 H 6 + H 3 0; 

 and similarly for mixtures of water with propyl alcohol, methyl 

 alcohol, glycerine, and acetic acid, he gets cusps at points of 

 definite molecular proportions. Drude (Zeitsch. f. Phys* 

 Chem. xxiii.), on the contrary, gets a steady curve for K for 

 water-alcohol mixtures running the same course as Thwing's 

 with the cusps smoothed out. It looks, therefore, as if 

 Thwing had used a period of vibration which was particularly 

 sensitive to arrangements of electrons in regular order corre- 

 sponding to the molecular proportions in his mixtures. His 

 sharp cusp for water at 4° is indicated by the following 

 excerpts from his data : — 



Temp 0° 2-2° 4° 6° 7° 



K 79-46 80-84 85-2 80-84 79-4 



Otherwise his observations for water make K run with 

 temperature a course very similar to that found by Drude 

 and Heerwagen. If his cusp for water has the same signifi- 

 cance as his cusps for other mixtures, we should have to take it 

 as implying for water at 4° the composition (H 2 0) 3 + 3(H 2 0) 2 , 

 that is, *333 trihydrol mixed with "667 dihydrol, while our 

 formula (9) gives "363 with *637. An alteration in our 

 somewhat arbitrarily chosen density of liquid trihydrol ('88) 

 at 0° would change these proportions to '333 and '667 ; but 

 until Thwing's remarkable observations have been confirmed 

 by further special experiments, I should not feel quite war- 

 ranted in making such changes in the constants of this paper 

 as would make the composition of our standard mixture at 4° 

 to be (H 2 0) 3 + 3(H 2 0) 2 . That ice near 0° shows K of the 

 order 78 when tested by electric oscillations of period about 

 10 -1 to 10 -2 second, and of the order 2 for periods of 10 -6 

 and less, is a fact of prime importance in the relation of 

 molecules to electrons. Dewar and Fleming's measurements 

 of K at low temperatures bring out also the high promise of 

 K for giving an insight into molecular architecture. But for 

 the present the experimental data for water and ice hardly 

 allow of more elaborate investigation than the above. 



