520 Reports and Proceedings — British Association. 



solutions, a good deal may be learnt about tbeir bebaviour. Having 

 complained tbat we need experiments in tbis field, I may perbaps be 

 pardoned if I allude to some unpublisbed experiments of my own 

 wbicb relate to tbe general bebaviour of crystallising liquids, and 

 appear to me to explain two difficult problems in petrography. 

 To such experiments tbe objection of Doelter does not apply. 



The Metastable and Labile Conditions. 



When a solution of any salt such as alum or sodium nitrate is 

 allowed to crystallise at a uniform temperature tbe crystals will only 

 grow so long as tbe solution is supersatui-ated ; a crystal growing in 

 the supersaturated solution will continue to do so until a condition 

 of equilibrium is attained. If tbe solution be kept at rest and 

 maintained at a constant temperature, tbe crystal will continue to 

 concentrate tbe liquid around itself and to withdraw solid material, 

 until by diffusion of the impoverished liquid tbe whole mass is 

 ultimately reduced to saturation, equilibrium is established, and the 

 crystal ceases to grow ; but most saturated solutions are so viscous 

 tbat a very long time is required before this point is reached. 

 Prolonged and vigorous stirring is required if tbe supersaturation is 

 to be completely relieved within, say, a day ; without stirring, weeks 

 may be required. 



Further, it may be possible, as is well known, to keep a super- 

 saturated solution in a sealed tube for years without change ; and it 

 is also possible to start crystallisation in such a liquid by dropping 

 into it a crystal of the dissolved substance, or of one isomorphous 

 with it, or sometimes by shaking it. 



But it is, perhaps, not generally known that supersaturated 

 solutions are of two sorts. 



In 1897 Ostwald published some experiments upon supercooled 

 liquids and supersaturated solutions, which were carried out with 

 the oV>ject of showing how extraordinarily minute are the quantities 

 of solid material capable of starting crystallisation in sucli liquids, 

 but at the same time that they have a limit of size. He called 

 attention to the radical difference which probably exists between the 

 state of a saturated solution which cannot crystallise spontaneously 

 and that of the more strongly supersaturated solution which can 

 do so. 



Tbe former is one in which crystallisation can either take place 

 spontaneously or can be induced by stirring or shaking, or a variety 

 of causes : this Ostwald calls the labile state. The latter is one 

 in which crystallisation can only take place if a solid crystal of the 

 dissolved substance, or a fragment of one, is brought into contact 

 with the liquid : this be calls the metastable state. It is highly 

 probable that no amount of stirring or shaking, or introduction of 

 foreign substances, can make the metastable liquid crystallise. 



Until recently no attenq^t to ascertain the exact limit between the 

 metastable and labile states, or even to establish the existence of 

 such a limit, bad lieen successful, and practically no attention has 

 been paid to the difference between them. Tamman, who measures 

 the velocity of crystallisation by counting the number of the centres 



