DIFFERENTIATION PHENOMENA OF THE PROSPECT INTRUSION. 119 



naphthalene and water, and sulphur and toluene. The 

 addition of a third constituent considerably complicates 

 the phenomena. There is not much experimental work on 

 the properties of mixtures of three liquids ; but apparently 

 the results vary according to the solubility relations of O 

 with A and B respectively; so that C may either raise or 

 lower the temperature of separation into two layers, pre- 

 vent it altogether, or produce a separation into three layers, 

 according to its nature. The problem is also complicated 

 by the fact that some pairs of liquids after becoming less 

 soluble in one another begin to become more soluble in one 

 another and sometimes miscible again in all proportions, as 

 cooling is continued, before or after the crystallisation of 

 one of the constituents has begun. 



In a rock magma there are usually ten or twelve different 

 silicates present, some of which crystallise together, besides 

 water and two or three oxides. Whilst this renders a 

 complete mathematical or experimental discovery of the 

 properties of such a magma virtually impossible, it does 

 not alter the nature of the laws to which they conform nor 

 preclude the possibility of liquation. We must remember, 

 however, that any one of these numerous constituents may 

 have the property of greatly altering, in accordance with 

 the proportion of it present, the conditions of temperature 

 and pressure at which a given magma liquates. It is 

 greatly to be desired that experiments should be under- 

 taken with mixtures of molten silicates under high pres- 

 sures and in the presence of water vapour. The chief 

 difficulty, no doubt, is the costliness of the apparatus that 

 would be necessary to maintain a very high pressure at a 

 high temperature. In default of experiments we must 

 rely on petrological evidence ; but this is not likely to be 

 abundant, because it would be only when the differentiation 



1 Bancroft, ibid., p. 238. 2 Ibid., p. 103. 



