August 24, 1905] 



NA TURE 



411 



It is clear, however, that the labours of Vogt have been 

 precisely in the direction indicated by Teall in the words 

 that I have quoted, " experiment controlled by the modern 

 theory of solution"; and if his opponents are tempted to 

 think that he may have carried the principle too far with 

 insufficient data, they cannot but admire the brilliancy, the 

 persistency, and the ingenuity with which he has applied 

 the newer theories of solution at every turn. 



Heycock and Neville's Work on Alloys. 



I must next refer briefly to another remarkable series of 

 researches which have recently been published. 



The laws which govern the solutions of metals in 

 metals, that is to say alloys, appear to be the same as 

 those which prevail in the case of other solutions ; it is 

 in alloys that the nature of eutectic mixtures has been most 

 fully studied ; and the phase-rule and Roozeboom's deduc- 

 tions from it have been applied with signal success to their 

 investigation. A new impulse has been given to the subject 

 by the work of Heycock and Neville Avhich is summarised 

 in their Bakerian lecture delivered last year upon the 

 copper-tin series of allovs. They have studied the changes 

 which occur during the cooling of an alloy by taking small 

 ingots of the cooling metal and chilling them at certain 

 temperatures ; this arrests the gradual process of cooling 

 and causes all that is liquid at the moment of chilling to 

 become suddenly solid ; it is then possible by polishing and 

 etching the ingot to show the solid crystals set in the con- 

 gealed ground-mass and to study their nature. They have 

 been able to interpret their results by means of Roozeboom's 

 remarkable work on the solidification of mixed crystals 

 published in 1899. For our present purpose it is sufficient 

 to consider these results as applied only to alloys. If a 

 diagram be constructed with the temperatures for ordinates 

 and constitution for abscissce, Roozeboom has shown that 

 two curves may be drawn. The first is the freezing-point 

 curve, or liquidus, giving the temperatures at which an 

 alloy of any composition begins to solidify : this is a broken 

 curve and each section of it represents the temperature of 

 equilibrium between the liquid and a different solid alloy; 

 the breaks represent the temperatures and constitution of 

 the liquid at which one solid ceases to be produced and 

 another begins. The curve is, of course, far more com- 

 plicated than the simple \' of Meyerhoffer, since that re- 

 presents the cooling of a mixture the constituents of which 

 do not form compounds or isomorphous mixtures, whereas 

 the alloys do both. In this respect the alloys resemble a 

 silicate magma which is crystallising as a rock-mass : 

 indeed it will be remembfred that Mendel^eff insists upon 

 the general similarity of silicon compounds to metallic 

 alloys. 



The second curve of Roozeboom is the melting-point 

 curve, or solidus, representing the temperatures at which 

 an alloy of given composition becomes completely solid. 

 Points above the liquidus represent the condition of alloys 

 which are completely liquid ; points below the solidus that 

 of alloys which are completely solid ; points between the 

 two that of cooling alloys which are only partially solid ; 

 and the curves themselves show which solid compounds can 

 be in equilibrium with the liquid and with each other at 

 any temperature. 



The cooling-curves of Roberts-.'Vusten and Stansfield had 

 shown that considerable evolutions of heat may occur in 

 cooling alloys far below the temperature of solidification, 

 indicating that changes are going on in the solid as well as 

 in the liquid condition. Heycock and Neville carry their 

 investigations below the temperature of complete solidifica- 

 tion and study these changes also. 



In the case of the copper-tin series of alloys they find 

 that, according to the temperature and constitution of the 

 liquid, crystals belonging to no less than six different types 

 may separate, namely : — 



a, a solid solution of Cu with less than 9 per cent, of Sn. 



0, a solid solution of Cu with less than 27 per cent, of Sn. 



7, of which the constitution is not known. 



5, which probably has the composition Cu,Sn. 



71, which probably has the composition CUjSn. 



H, which probably has the composition CuSn. 

 Both and y are unstable at ordinary temperatures. The 

 compound 5 crystallises out of or 7 while they are already 

 in the solid state, when the temperature falls sufficiently. 



NO. 1869, VOL. 72] 



A glance through the loi photographs of chilled and 

 etched ingots which accompany Heycock and Neville's paper 

 on this series of alloys shows how impossible it would be 

 from the final coinposition of the solid alloy to ascertain 

 the various stages through which it has passed during 

 cooling; as the authors remark, it is of the nature of a 

 palimpsest. For example, the alloy, containing 14 atoms 

 of tin to 86 of copper, consists at 800° of o crystals in a 

 ground-mass which probably contains j3 ; it solidifies at 

 about 775° ; at 675° there are only 3 crystals ; at 600° 

 there are" a and $ crystals, but here a has crystallised out 

 of e after it became solid ; at 530° there is a much larger 

 proportion of a ; at 470° there are a crystals immersed in 

 a mixture of a and 5 into which the residual $ has broken 

 up on cooling. 



If the course of events is so coinplex in an alloy of 

 only two metals, how much more difficult must it be ^ to 

 decipher in the case of a mass of complicated silicates which 

 are even more prone to form isomorphous mixtures, such 

 as we have in a solid rock, not to mention the additional 

 presence of aluminates, oxides, and sulphides. And yet 

 geologists are accustomed to speculate freely about the 

 crystallisation of rock constituents from the magma without 

 taking account of anything save the final stage. 



I cannot help thinking that the experimental methcd 

 of Heycock and Neville will have to be applied to the 

 study of slags and fused silicates if we are to trace success- 

 fully the evolution of rock species. The value of their 

 work to geologists is not only that the results are skilfully 

 interpreted by the light of m'odern physical chemistry, but 

 primarily that it is experimental work upon actual crystal- 

 lising materials. 



Supersaturated Solutions. 



I do not myself see how we can do otherwise than apply 

 to the study of rock-magmas all that can be learnt from 

 physical chemists concerning the behaviour of solutions, 

 for though we cannot attain in laboratory experiments the 

 high temperatures and great pressures at which rocks may 

 have crystallised, there is no reason to believe that these 

 introduce more than a difference of degree. The principles 

 of equilibrium between the various crystallising components 

 probably remain the same, whatever may be the tempera- 

 tures and pressures at which they have solidified. 



It must at the sanie time be confessed that most of the 

 experiments upon which the modern theory of solutions 

 has been built up have been conducted upon dilute solutions, 

 whereas the problems of crystalline growth are concerned, 

 not with dilute nor even with saturated solutions, but only 

 with solutions which are supersaturated. There is some 

 force in the objection of Doelter that the results of such 

 experiments may not be directly applicable to crystallising 

 slags. 



For example, as I have already mentioned, doubt has 

 . been expressed in the case of silicate magmas, whether the 

 substances in solution are the minerals about to crystallise 

 or only their constituents ; whether viscosity and super- 

 saturation may not invert the theoretical order of their 

 appearance ; whether we are to take into account possible 

 dissociation of the molecules or not ; whether the presence 

 of a common ion in these ininerals is a factor which de- 

 termines their mutual solubility. In fact, very little is 

 known about the actual condition of the materials in a 

 strong solution, although I do not know that there is any 

 evidence available which forbids us to regard a solution 

 about to crystallise as a mixture of liquids one of which 

 is about to pass into the solid state. 



But if little is known about the nature of strong and 

 supersaturated solutions, a good deal may be learnt about 

 their behaviour. Having complained that we need experi- 

 ments in this field, I may perhaps be pardoned if I allude 

 to some unpublished experiments of my own which relate 

 to the general behaviour of crystallising liquids, and appear 

 to me to explain two difficult problems in petrography. To 

 such experiments the 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 

 the crystals will only grow so long as the solution is 

 supersaturated ; a crystal growing in the supersaturated 

 solution will continue to do so until a condition of equi- 



