128 SCIENCE PROGRESS 



nitrate, i.e. for mixtures of three components, the two salts 

 and water. 



No doubt the same principles prevail in the solidification of 

 metals, alloys, and rocks, although the problem is here compli- 

 cated by the formation of compounds and isomorphous mixtures 

 (solid solutions) ; but whatever substance solidifies, it can only 

 crystallise from a solution which is supersaturated with regard 

 to it, and unless this is made to take place by inoculation, 

 the solution must first become sufficiently supersaturated to 

 crystallise spontaneously. 



The freezing-point curves terminate at M and m in the 

 melting points of the pure substances A and B ; and the super- 

 solubility curves terminate at S and s, the temperatures of 

 spontaneous crystallisation of A and B. We have determined 

 these for salol and betol. We have also made experiments with 

 water, and have found that pure water contained in sealed tubes 

 crystallises by shaking at — 1*9° C. ; and even if heavy substances 

 be enclosed with the water in the tubes so as to produce 

 friction, we have been unable to make it crystallise at a higher 

 temperature than — o"4 C. 



If the crystallisation of a drop of any solution be watched 

 under the microscope, remarkable changes in the process will 

 generally be noticed so soon as the drop passes from the 

 metastable to the labile condition. Experiments made in my 

 laboratory by M. Chevalier have shown that, in a drop of labile 

 solution, potash-alum crystallises as a rectangular network of 

 fine needles, and not in the ordinary octahedra. This observa- 

 tion accords with the general experience that when crystals 

 grow very rapidly they usually grow as needles. I believe this 

 to be due to the fact that when a crystal has started in a labile 

 solution the liquid in immediate contact with it is at once 

 reduced to the metastable condition both by loss of solute and 

 by rise of temperature, so that the growth only takes place 

 within the sheath of less concentrated liquid with which the 

 crystal envelopes and protects itself, and therefore proceeds 

 quite slowly. But if any point of the crystal has begun to 

 advance so rapidly that it projects through this sheath before 

 the zone of weaker solution has time to close round it, then 

 it may continue to advance rapidly through the strongly super- 

 saturated labile solution ; and the concentration currents which 

 play most energetically upon the end of a crystal will ensure 



