On the Molecular Constitution of Water. 



461 



abscissa and p for ordinate, gives a straight line such as XY 

 for an ordinary liquid. But the corresponding graph for 

 water is like the curve CDE in which the ordinate DN re- 

 presents the maximum density at about 4° C. If the actual 

 data are plotted on a large scale, the branch DE looks as if it 



Ear. 1. 



approached asymptotically the straight line FGr. The reason- 

 able course is then to assume that this asymptote represents 

 the behaviour of one of the pure ingredients of water. The 

 slope of this asymptote gives for k the value '001, which is 

 of the same order as k for ordinary liquids, and the asymptote 

 cuts the axis of p in a point which gives for p for the pure 

 ingredient a value about 1*083. 



Gruided by these facts and by theoretical considerations, I 

 sought to obtain an equation for CDE which would give the 

 value of p for the other ingredient, and also a law of the dis- 

 sociation of one ingredient into the other at different tempe- 

 ratures. A preliminary attempt gave a formula which ex- 

 pressed the variation of p with t to within 1 part in 10,000; 

 but as Mendeleeff (Phil. Mag. [5] xxxiii.) had already 

 furnished an empirical formula for the expansion of water 

 correct to 4 parts in 100,000, it seemed best to examine his 

 formula in the light of the foregoing considerations. It is 



i.= l— 



(*-4) 2 



l-9(J4-l + /)(703-51-£)' 

 Phil. Mag. S. 5. Vol. 50. No. 306. Nov. 1900. 



• (1) 



