and Attached Water. 223 



not only lower than the mean of the two, but lower than that of 

 either or any of the constituents — witness the "fusible metals." 

 When metals are alike in chemical affinities, they have little 

 affinity for one another. They mix mainly by diffusion. They 

 dilute one another. If we mix sodium with sodium, we cer- 

 tainly dilute each sodium with the other ; the adhesion which 

 we supply is exactly equal to the cohesion which we replace; 

 the resulting mass is identical with its constituents. But if 

 we mix potassium and sodium together, by diffusion we di- 

 minish cohesion, and the adhesion introduced is less than 

 sufficient to replace it ; from two solids we get a liquid. The 

 similarity of the metals prevents their having any notable 

 relationship with one another. Equally to the purpose is the 

 recollection of the alteration in boiling-point which follows 

 the mixture of liquids which are chemically neutral to one 

 another. Many a month has been wasted in the attempt to 

 get propylic alcohol from fusel oil, owing to the fact that a 

 mixture of amylic and butylic alcohols will commence to boil 

 at a temperature below the boiling-point of butylic alcohol. 

 Moist ether boils below dry ether ; and even a liquid such as 

 iodide of ethyl, which is almost insoluble in water and which 

 dissolves but little of that body, has its initial boiling-point 

 already lower than that of water. The work done by the heat 

 in separating the particles of a body is already effected by the 

 interpenetration of the associated body. Cohesion is dimi- 

 nished and a less heat-tension is required to complete the 

 separation of the more volatile constituent as vapour. 



From another aspect we may consider the neutral body to 

 act as an infinite series of points, and recall the fact that, on 

 evaporating saturated solutions of various salts over a steam- 

 bath, bubbles are frequently formed beneath the salt-pellicle, 

 owing perhaps to the multiplicity of points, at temperatures 

 certainly below 100° C, while the clear solution of the same 

 salt when saturated demands a temperature several degrees 

 above 100° C. for ebullition. The rise in temperature shown 

 on mixing a dry colloid with water is probably connected with 

 the deficit below 100° C. of the temperature required to boil 

 the resulting colloid solution. 



Returning to the separation of ice from colloid solutions, we 

 must conceive that when ice so separates that the solution 

 becomes enriched in regard to the colloid body, the particles 

 of ice in contact with the source of cold become overcooled 

 (that is, fall below 0° C.) on account of the insufficient circula- 

 tion and the imperfect thermal conductivity of ice and of the 

 colloid solution. But when such over-cooled ice is stirred 

 with the colloid solution a fresh portion of ice is formed if 

 the latter is already at 0° C. ; if above 0° C, then some of the 



