122 SCIENCE PROGRESS 



causes crystallisation to begin, and the same effect may often 

 be produced by shaking. 



Break thou deep vase of chilling tears 

 That grief hath shaken into frost, 



are well-known lines from In Memoriam that refer to this 

 remarkable effect. 



It is curious that no systematic observations have hitherto 

 been made upon the behaviour of solutions which are con- 

 tinuously shaken while cooling. 



In the course of some experiments upon the refractive 

 indices of cooling supersaturated solutions of various salts 

 made by Miss Florence Isaac and myself in the Mineralogical 

 Laboratory at Oxford we noticed that when such a solution 

 is stirred in an open vessel, a thin shower of crystals generally 

 appears at or about the saturation temperature ; but that after 

 the temperature has fallen about io° below this point there 

 ensues a dense shower of crystals, which is generally so thick 

 as to make the liquid almost opaque. The sudden weakening 

 of the solution at the lower temperature is accompanied by a 

 sudden fall of the refractive index, which attains its maximum 

 value at this temperature. 



On the first day, when we experimented with a solution 

 which we had not previously used, the earlier and thinner 

 showers of crystals did not appear, and the later and denser 

 shower might be somewhat retarded. But in subsequent 

 experiments the thin shower (accompanied by a slight diminu- 

 tion in the rate of increase of the refractive index) occurred 

 at exactly the saturation temperature, and the dense shower 

 (accompanied by an actual diminution of the refractive index) 

 occurred at a perfectly definite lower temperature. 



We infer that the first shower takes place because some fine 

 crystalline particles of the dissolved substance get into the 

 solution with the dust of the air, and continue to grow, and 

 that the second shower is due to a different cause. 



If this be so, then the same solution enclosed in a sealed 

 glass tube should not yield the first shower. 



We have made this experiment with numbers of aqueous 

 solutions of different substances and of different strengths, and 

 have always found that in a sealed tube the first shower is 

 prevented, and that the liquid cannot be made to crystallise 



