IGNEOUS ROCK-MAGMAS AS SOLUTIONS 245 



minerals, which usually crystallise at an early stage, even when 

 they are present in very small amount. It is by no means clear, 

 however, that these minerals — apatite, spinel, zircon, and the 

 like — are freely soluble in the ordinary silicate-minerals, as these 

 latter are in one another. If there be only a limited miscibility, 

 diminishing with falling temperature, the very high melting- 

 points of these accessory minerals will ensure their early 

 crystallisation without reference to their amount. 



Rosenbusch's first rule, imposing a fixed order of crystallisa- 

 tion, appears at first sight equally irreconcilable with the 

 behaviour of solutions. It is to be remarked, however, that 

 the rocks in which such rule partially holds good, still with 

 important exceptions, are in general rocks containing a con- 

 siderable number of constituent minerals, and must therefore 

 be compared, not with simple, but with very complex solutions. 

 If we examine the less common rocks composed, e.g., of two 

 essential minerals, we find no fixed order of crystallisation, but 

 one depending directly upon the relative proportions of the 

 constituents. On Allival and the neighbouring hills, in the Isle 

 of Rum, there is an extensive development of plutonic rocks, 

 consisting essentially of olivine and a basic felspar, near 

 anorthite in composition. They include every connecting link 

 between a pure olivine-rock and a pure felspar-rock, and in a 

 large part of them there is no other mineral except a few 

 scattered crystals of chromite. In the varieties richer in oliyine 

 that mineral has invariably been the first to crystallise (excluding 

 the chromite), and in varieties richer in felspar the reverse 

 relation is invariably found. Where the two minerals are in 

 certain definite proportions, they have crystallised simultaneously 

 one or the other showing its proper crystal outline in different 

 parts of the same thin slice. The order of crystallisation in 

 these binary rocks is thus seen to depend upon the relative 

 proportions of the constituents, as compared with certain 

 standard proportions. 



It will be observed, however, that the behaviour here 

 indicated is not what is seen in the crystallisation of a solution 

 of, e.g., sodium chloride and water. In the latter case only the 

 excess of one or other constituent first crystallises alone, and 

 then, when the solution has been reduced to the standard 

 (eutectic) proportions, the two constituents crystallise together 

 as a fine mixture of salt and ice. The apparent discrepancy 



