( 689 ) 



altered portion, the two portions of the connodal solid- vapour are 

 again drawn (unfortunately the altered connection is omitted in the 

 figure). At m lies the maximum density of water under its own 

 vapour pressure. The preliminary model made by us serves to show, 

 that the most notable properties of H,, will be correctly given if 

 we consider the ice ridge as an ordinary solid ridge on the surface, and 

 the liquid ridge as changed by the last mentioned process owing to the 

 association. 



We hope shortly to obtain models which will exhibit the trans- 

 formation of the different modilications or allotropic states as well as 

 the peculiarities of the expansion phenomena. In this way we hope 

 to contribute to a somewhat better insight, at least to a better 

 survey, of the thermodynamical character of the different substances. 

 However in addition to the questions relating to association there 

 are others, which have reference to mixtures, saturated solutions 

 etc., and which would necessitate a long investigation. We hence 

 think that the results now obtained from a sufficiently independent 

 whole to be published. 



We have thus in the above made the meaning of our work 

 clear and can now proceed to describe the models somewhat more 

 fully and to show that the experimental data can really be obtained 

 on these surfaces. 



I. The solid-liquid plait on the Gibbs' surface. 

 (Representation of the continuity of the solid and gaseous states). 



The specific heat of the imaginary substance in the gaseous state 

 has been assumed to be equal to half that of liquid carbon dioxide: 

 the specific heat in the solid state as equal to that which is found 

 by employing Neumann's law. Also the substance obeys the van der 

 Waals equation of state for carbon dioxide in the liqLiid and gaseous 

 states. It appeared useful to give for this model (pi. II, fig. J) front 

 and side elevations and plan (pi. Ill fig. 1, 2, 3, 4) Avith the lines 

 •>/ = const., i" =: const., 7'= const, (dashed lines) /}==: const, (dotted 

 lines), for which the drawings are sufficiently explanatory. The con- 

 nodal fluid (gas)-solid is drawn on this model and the coexisting 

 states joined by steel wires. 



The principal difference between our representation and Tammann's 

 is very clearly seen on comparing the plan (pi. Ill fig. 2) with his 

 figure (Drude's Ann. 3 p. 190). 



This model is also useful for a comparison of our idea with the 

 well known scheme given by Maxwell (Theory of Heat, p. 207) 



