248 Mr. J. Y. Buchanan [May 8, 



pass from the water to the ice ; similarly, when ice is being* formed in 

 a mixture of ice and water, immersed in a freezing-bath, the tempera- 

 ture of the water must be a little lower than that of the ice produced. 

 Therefore, the observed freezing temperature of water, and melting 

 temperature of ice must always be different, if quite exactly deter- 

 mined. But it is found that, by supplying heat at as low^ a tempera- 

 ture as possible in the melting experiment, and removing it at as higli 

 a temperature as possible in the freezing experiment, the observed 

 melting and freezing temperatures of the substance H^O approach 

 each other more and more closely. Therefore, as a matter of experi- 

 ment alone, it is legitimate to conclude that the limiting values of 

 these temperatures are identical. It is called 0" on the thermometric 

 scales of Reaumur and Celsius, and 32" on that of Fahrenheit. 



This temperature was for long held to be invariable ; indeed, it is 

 little more than half a century since it was established that it is 

 lowered by increase of pressure and raised by relief of the same, the 

 quantitative proportion being that of 1" C. to 135 atmospheres. 



In the above definition of the freezing and melting temperatures 

 of H^O, that substance is specified as being present partly in tlie solid 

 and partly in the liquid state. The temperature at which ice 

 begins to take form in water which is cooled when in contact 

 only with itself, or with a solid other than ice, has not been 

 determined, and is in fact uncertain. The moment the smallest 

 particle of ice is present, the water has the opportunity of passing, as 

 a liquid, into itself as a solid : but not till then. 



Evidence of the uncertainty which exists regarding the tempera- 

 ture at which ice begins to form in water, when it is cooled in contact 

 only with a solid other than ice, is furnished by the wet-bulb thermo- 

 meter when it is being prepared for use at temperatures below 0" C, 

 by freezing on it the quantity of water which is supported, against 

 gravity, by the perfectly clean Ijulb. When this is rotated in air of 

 - 1()' to -20'C. ice never begins to form until the temperature of 

 the bulb of the thermometer has fallen to - 2" or - 3° C, and rarely 

 before it has fallen to - 4° C. In many cases I have observed it fall 

 to temperatures as low^ as -8° or -9"C.; and in such cases, when 

 freezing begins, the whole drop of water is frozen without its being 

 able, by the liberation of latent heat alone, to raise the temperature of 

 the bulb of the thermometer to 0° G. 



In our definition of the freezing and melting temperature of 

 H2O, no substance is specified except H^O. But HoO, that is, 

 absolutely pure water, rarely, if ever, occurs in nature. Therefore 

 our definition is not directly applicable to water as it occurs in 

 nature. It has been found by experiment that when, in a mixture 

 of ice and water, the water contains ever so little foreign, especially 

 saline, matter in solution, the temperature at which it freezes, and 

 that at which pure ice melts in it, is lowered. It is therefore 

 probable that in Nature, ice never melts and water never 



