NA TURE 



August 24, 1905 



graphical problem, that of the differentiation of magmas 

 into various rock types ; for in this connection very few 

 experiments have been made, and practically none of recent 

 date. Observations of the facts as they present themselves 

 in the field accumulate every day ; almost every important 

 petrographical region is being studied with the particular 

 object of determining the mutual relations of its rock 

 masses and the factors which have contributed to their 

 differentiation. They have been ably discussed by Becke, 

 Brogger, Becker, Cole, Harker, Iddings, Judd, Lacroix, 

 L^vy, Pirsson, Rosenbusch, Teall, Washington. Zirkel, 

 and many others ; appeal has been made to the action of 

 gravity, of temperature differences, of diffusion, of electric 

 currents, of fractional crystallisation, of re-fusion, of 

 chemically combined water, of absorption of the country 

 rock ; but with the exception of a single case, observed 

 in the glassworks of Targowek, in which the top of a 

 molten glass was found to contain less lime and more 

 silica than the bottom, and some observations by Doelter 

 upon boron-glass, there is scarcely a single experiment 

 upon silicates which really bears directly on the question. 

 That artificial glasses are far from homogeneous is known to 

 glass-workers and to makers of lenses, but there is nothing 

 comparable with the splitting of a magma into two or 

 three distinct liquids which solidify as different rocks. 



It is in the case of laccolites that the problem ought to 

 present itself in the simplest form, for w'e may regard 

 them as basins of igneous rock which have been practically 

 imprisoned within solid walls and have crystallised in situ. 

 There can, I think, be no doubt that differentiation has 

 generally taken place even in such basins, that the margins 

 have often a different mineralogical and chemical const'- 

 tion from the more central portion, and that the differences 

 are greater than can be accounted for by solution of the 

 enclosing rock, and are often of a chemical nature which 

 cannot be so explained. 



The various theories that have been propounded fall 

 into two distinct classes — those which seek the cause in 

 the separation of solid material from the liquid, so thai 

 when the latter subsequently crystallised it constituted a 

 different rock from the former ; and those theories which 

 assume that different liquids have separated from each 

 other and then solidified as different rocks. 



The first conception satisfactorily explains the manner 

 in which the least soluble minerals are concentrated at the 

 bottom or margin of an igneous mass, for they naturally 

 crystallise first where the mass is coolest, or where contact 

 with other crystals may have occurred ; or even if thev 

 have been precipitated as a cloud throughout the magma 

 they must be carried about by convection currents and 

 ultimately sink together unless the magma be very viscous. 

 Most geologists will probably agree with the conclusions 

 of Vogt that some of the most important deposits of 

 metals, metallic oxides, and sulphides have been produced 

 by magmatic differentiation from deep-seated magmas 

 which now constitute basic rocks associated with them. 

 But this does not explain how the mass which has crvstal- 

 lised out may be not a mineral but a rock. 



The actual observations on crystallising solutions do not 

 amount to much ; it is quite clear from laboratory experi- 

 ments that crystals do grow by means of convection 

 currents, which produce a flow of stronger solulion towards 

 the crystal and of weaker and warmer solution upwards 

 and away from the crystal. The concentration currents 

 can easily be seen in any ordinary aqueous solution as 

 streaks in the liquid. .Again, that there might be a slight 

 difference in the concentration of the upper and lower, 

 or of the warmer and cooler parts of a solution has also 

 been shown. That a very considerable difference in con- 

 centration can be produced by centrifugal action was 

 proved only last year by the experiments of Calcar and 

 de Bruyn, in which solutions contained in rapidly rotating 

 vessels became more concentrated in the portions furthest 

 from the axis of rotation. 



Schweig has recently suggested that the crystals which 

 fall to the bottom of a rock-magma may be unstable com- 

 pounds, which' re-dissolve when the pressure is relieved, 

 and so give rise to an underlying magma of different 

 chemical constitution. 



Harker, also, some time ago, suggested the existence 

 of horizontal layers of different liquid magmas above each 



NO. 1869, VOL. 72] 



other, thus attempting to explain the presence of quartz 

 in basic rocks as due to the crystals which had sunk 

 into the basic magma from a more acid magma floating 

 upon it. 



The second theory, that of liquid differentiation, regards 

 such layers as actually produced by the spontaneous 

 division of a magma into two liquids of different com- 

 position, and if it be tenable seems more capable of ex- 

 plaining the geological facts. 



The experiments bearing on the subject are well know'n, 

 and have been quoted by Backstrom and Teall ; mixtures 

 of phenol and water, or of aniline and water, which form 

 a homogeneous solution above a certain temperature, may 

 below that temperature (which is a sort of critical point 

 of the solution) divide into two solutions, one consisting 

 of phenol in excess of water, the other of water in excess 

 of phenol ; and these two solutions are not miscible, but 

 separate into two distinct layers. 



Many pairs of substances have now been found to exhibit 

 this incomplete miscibility, which varies with the tempera- 

 ture and may at certain temperatures become complete ; 

 among them are some of the metals such as zinc, lead, 

 bismuth, and silver. 



If rock-magmas can really behave in this way, there is 

 no difficulty in explaining their differentiation ; but ex- 

 periments upon fused silicates have not disclosed anything 

 of the sort, though they are made far below the critical 

 temperature. 



The case of nicotine and water, which has recently been 

 described by Hudson, is remarkable and suggestive : above 

 a temperature of 205° a mi.xture in equal proportions is 

 a clear liquid ; at 205° it divides into a saturated solution 

 of nicotine in water floating on a saturated solulion of 

 water in nicotine ; at 90° these two layers change places ; 

 at 64° they mix again and the liquid becomes once more 

 homogeneous. 



It is, of course, possible that fused silicates at experi- 

 mental temperatures correspond to nicotine and water 

 below 64°, and that rock-magmas correspond to the same 

 mixture at higher temperatures. 



In discussing the reasons why in laccolites of the .Square 

 Butte type the margin should be more basic, and in lacco- 

 lites of the Magnet Cove type more acid than the centre, 

 Washington regards the magma as a mutual solution of 

 an alumo-alkaline substance with a ferro-magnesian sub- 

 stance ; whichever of these is in excess may be regarded 

 as solvent, and crystallises first, for example, either the 

 syenite or the shonkinite. In a laccolite where no 

 differentiation has taken place, as in the Henry Mountains 

 type, he supposes the mixture to be eutectic or such that 

 they crystallise together. Pirsson, in a paper recently 

 published upon the " Highwood Mountain Laccolites of 

 Montana," while attributing a greater part in the process 

 to the action of convection currents, also regards the ferro- 

 magnesian minerals, taken together, as constituting the 

 solvent and crystallising first as shonkinite. 



In fact, stated quite baldly, these latest views tend to 

 a compromise between the two theories which I have just 

 mentioned. They regard the splitting of the magma as 

 produced by a fractional crystallisation, only now the mass 

 which crystallises is not a mineral but a rock ; in other 

 words, they assume that rocks may be dissolved in each 

 other, and may crystallise from each other as though they 

 were minerals. 



In this matter of magmatic differentiation, then, there 

 has been during the last few years a large accumulation 

 of geological evidence, a little new speculation, but prac- 

 tically no new experimental work, and scarcely any 

 progress. 



^[incra} Differentiation and Eateelies. 



Let us pass to the second petrographical problem, that 

 of mineral differentiation, the nature and order of the 

 minerals which crystallise when a cooling magma becomes 

 a solid rock mass. 



It has been laid down by Rosenbusch, and is accepted 

 as a general rule (in spite of many exceptions), that the 

 order in which the various minerals crystallise is one of 

 increasing acidity, ores and oxides and so-called accessory 

 minerals first, then those minerals which are comparatively 

 poor in silica, then those which are richer, and finally, if 



