46 Prof. J. H. Poynting on 



begin to approach (that is, the horizontal part of the isother- 

 mals would decrease), and that ultimately ice would pass gra- 

 dually into water without any abrupt change of volume (that 

 is, there would be a critical point). Below this critical point 

 ice and water would probably be identical. 



A similar conclusion is arrived at from the latent-heat equa- 

 tion. On the supposition that at the critical point the latent 

 heat vanishes, the temperature given by that equation is 

 — 122 0, 5, with a pressure of over 16,000 atmospheres (Baynes, 

 Thermodynamics,' p. 169). 



It is usually assumed that we must stop the isothermal at 

 the base-line of no pressure. But we know that water can be 

 subjected to a negative pressure; as, for instance, when it rises 

 in a capillary tube in a vacuum, or when it adheres to a baro- 

 meter-tube at a height greater than that of the barometric 

 column. It seems probable that, if perfectly freed from foreign 

 gases, it might even be subjected to a very high negative pres- 

 sure without the particles being torn asunder. So, too, a mix- 

 ture of ice and water might probably be subjected to tension. 

 It seems at least worth while to draw the isothermals for ice 

 and water on such a supposition. 



Prof. J. Thomson's result for the alteration of the melting- 

 point by pressure would hold for at least a short distance above 

 0° when we replace pressure by tension. Assuming it to 

 hold for 4°, we should have to put on a tension of 4-f- '00733 

 atmosphere = 545 atmospheres. But if the expansion of 

 water under a tension equals its compression under an equal 

 pressure, the expansion is about 2T000 F er atmosphere*; so 

 that the volume of the water at 4°, under a tension of 545 

 atmospheres, will be 1*026. The ice, whose volume at 4 C 

 under no pressure would be 1*088, probably will not expand 

 nearly so much under tension. The change of volume on 

 melting will therefore probably be not very far from 



l-088-l-026 = -062, 



against a change at 0° of "087. Then the two branches of the 

 ice-line will converge very considerably for temperatures above 

 0° and with negative pressures. At this rate of convergence 

 the meeting-point is at about 14° C. At higher temperature 

 the ice would pass gradually into water — that is, we should 

 here have another critical point, — the two critical points being 

 at opposite ends of the closed curve which represents the 

 water-ice line. 



* Might not the truth of this supposition be tested by the propagation 

 of sound through the water above a barometric column at a negative 

 pressure ? 



