August 24, 1905] 



NA TURE 



407 



deposits are of enormous extent, more than looo feet 

 thick, and consist of fairly well-defined layers of various 

 sulphates and chlorides of sodium, magnesium, and 

 potassium, and their double salts and hydrates. It has 

 long been supposed that the minerals have been derived 

 from the evaporation of sea-water which contains in solu- 

 tion the chlorides of sodium, magnesium, and potassium, 

 with sulphate of magnesium and small quantities of 

 calcium salts ; and the general sequence of the minerals 

 is that of their solubility ; the less soluble sodium chloride 

 crystallised out first and is at the bottom, while the very 

 soluble magnesium chlorides, having been the last to 

 crystallise, occupy the top of the series. But the problem 

 is by no means so simple as to be one of mere solubility 

 in water ,■ the rock salt itself persists through the whole 

 series, and some of the associations are difficult to explain. 



As is well known, the modern theories of solution mainly 

 rest upon the behaviour of dilute solutions from which 

 the principles of electrolytic dissociation have been de- 

 duced ; but in the case of the concentrated solutions from 

 which dissolved substances actually crystallise, very little 

 is really known about the liquid itself. A great deal is 

 known, however, about its equilibrium with the solids 

 that separate from it, and the general laws of this equi- 

 librium are e.xpressed by the phase-rule deduced from 

 mathematical considerations by Willard Gibbs, which 

 states how many mechanically separable constituents can 

 coe.xist under varying conditions of equilibrium in a system 

 containing a definite number of chemical components. 



A solution saturated with a given substance is one which 

 is in equilibrium with that substance when the latter is 

 in contact with it in the solid form ; the phase-rule in- 

 dicates the number of solids which must be in contact 

 with a given solution ; the only difficulty in practice is to 

 determine the nature of the double salts or distinct hydrates 

 that may be formed. 



By means of a series of experiments upon the solu- 

 bilities of these salts, either singly or in the presence of 

 one another, in order to determine the composition of 

 solutions saturated simultaneously with two or more sub- 

 stances, it is possible to obtain a graphic representation 

 of all possible solutions containing the salts present in 

 sea-water. From this the course of crystallisation of any 

 particular solution, for example sea-water, can be pre- 

 dicted. 



The general sequence thus theoretically predicted is as 

 follows : — (i) Rock salt ; (2) Rock s.«ilt with the magnesium 

 sulphate, epsomite ; (3) Rock salt with the double sulphate 

 of potassium and magnesium, leonite ; (4) Rock salt with 

 leonite and the potassium chloride, kainite ; (5) Rock salt, 

 the magnesium sulphate kieserite, and the double chloride 

 of potassium and magnesium, carnallite ; (6) Rock salt, 

 kieserite, carnallite, and the magnesium chloride, 

 bischoffite. This last combination will persist until all 

 the water is evaporated. This is found to be the general 

 sequence, not only of the salts obtained on evaporating 

 sea-water at 25°, but also of the Stassfurt deposits. 



Up to this point the results have been summarised by 

 Dr. E. F. .■\rmstrong in a report presented to the British 

 -Association in 1901. .Since that date the research has been 

 prosecuted actively by van 't Hoff and his pupils, and now 

 the conditions of equilibrium at 25° have been mapped out, 

 not only for the above compounds, but also for the minerals 

 thenarciite, glaserite, astrakanite, and reichardtite, which 

 occur in these deposits. The whole process of crystal- 

 lisation of the solution, from which no fewer than twelve 

 different salts have separated, can, therefore, be predicted, 

 and their sequence and associations can be traced through 

 numerous stages, beginning with the separation of rock 

 salt and ending with a mixture of rock salt, kieserite, 

 carnallite, and bischoffite. 



In reconstructing the history of these deposits account 

 must also be taken of the varying vapour pressures of the 

 solutions which are saturated with the different com- 

 pounds, as this really determines which particular com- 

 pounds are stable, so that the matter is bv no means so 

 simple as might appear from this brief sketch. It is 

 further necessary, in order to bring the process within 

 reach of calculation, to assume that each deposit is re- 

 moved from contact with the mother liquor after it has 

 crystallised out ; but fortunately this is practically what 

 has happened in the Stassfurt deposits, for each layer is 



more or less separated from the succeeding one by an 

 intervening layer of clayey material. 



It may be possible even to go a stage further and obtain 

 a clue to the actual temperatures that prevailed, for two 

 minerals, langbeinite and loweite, are absent from the 

 theoretical model made by van 't Hoff to represent what 

 must happen during evaporation at 25° ; and this indicates 

 that while the deposits crystallised the temperature really 

 rose higher than 25°, probably as high as 43° ; in fact, 

 after the conditions of equilibrium have been worked out, 

 the appearance or disappearance of certain minerals can 

 be used as a sort of geological thermometer, capable of 

 indicating the limits within which the temperature can 

 have varied. 



The whole investigation is a splendid example of experi- 

 mental research devoted to a particular problem and 

 directed by a well-established theory ; the chemist in his 

 laboratory has now succeeded in tracing the changes that 

 took place ages ago in the bed of a land-locked sea as it 

 laid dov/n its contents and finally became a dry basin, 

 although he is not able to reproduce the original conditions 

 or to work for the long periods which Nature had at hei* 

 disposal. Without the logical consideration of the con- 

 ditions necessary for equilibrium, countless experiments 

 might be made upon these salts, and an immense amount 

 of speculation might have been devoted to their possible 

 reactions in the liquid state, about which we know so 

 little, instead of to their equilibrium when solidifying, 

 about which we know so much more. 



SoDie Petrographical Problems. 



The other geological problems which I have mentioned 

 have also been beyond the reach of actual experiment, for 

 it is hopeless to attain the immense pressures and high 

 temperatures or the enormous time that may have been 

 required for the growth of natural minerals in rocks and 

 veins ; and so when difficulties are encountered there is a 

 tendency to " explain " them (if the word may be so mis- 

 used) by reference to the mysterious effect of conditions 

 which cannot be brought directly within the reach of 

 experiment. 



I cannot help thinking that this has to some extent 

 occurred in the discussion of the petrographical problems 

 which I propose to consider next. There are two great 

 liquid reservoirs from which minerals have crystallised — 

 the sea, with its dissolved salts, and the subterranean 

 baths of molten silicates, from which the igneous rocks 

 have been derived. It is true that in the sea two of the 

 constituents, water and sodium chloride, largely pre- 

 dominated over the others ; but, after all, both sea and 

 lava are liquids subject to the same physical and chemical 

 laws. 



An admirable summary of the evolution of petrographical 

 ideas was given in the Presidential Address to the Geo- 

 logical Society of London in 1901 by Or. Teall, who dealt 

 both with the consolidation of rocks from molten magmas 

 and their differentiation into species. It is not, therefore, 

 necessary for me to consider anything but recent work 

 which has been done during the last four years, and the 

 earlier controversies may be left out of account. 



Among the many problems relating to the mineral and 

 chemical constitution of rocks which have yet to be solved, 

 two, and those perhaps the most important, should lend 

 themselves most readily to experimental treatment. The 

 first is the problem of rock differentiation ; why does a 

 magma, even one which has presumably crystallised in situ, 

 separate itself into zones, or layers, or streaks of different 

 constitution? And the second is the problem of mineral 

 differentiation : why does a granite magma, for example, 

 crystallise as a mixture of the particular minerals mica, 

 felspar, and quartz, and why is the least fusible mineral 

 the last to crystallise? 



It will scarcely be possible for me to deal in this Address 

 with more than the second of the two problems, but it 

 will be apparent from the somewhat parallel case of the 

 salt deposits that the mere order and manner of crystal- 

 lisation of a mass of molten silicates must be a sufficiently 

 complex problem to exhaust our attention for the present. 

 Magmatic Differentiation. 



If we are to consider only recent experiments which 

 have a bearing upon the problems of rock magmas, it is 

 not necessary to say much about the first great petro- 



NO. T869, VOL. 72] 



