260 N. L. Bowen — The Ttrnary System : 



times and which had quite different compositions. Neither 

 is it necessary, as it would be in the case of simple eutectic 

 relations, that the part lying towards the cold contact should 

 be the one giving evidence of having solidified first, for it 

 would be possible to have the temperature of final solidifica- 

 tion of such portions so lowered by the sinking out of crys- 

 tals that they might still be largely liquid when a lower and 

 hotter portion had already completely solidified. 



To the writer it seems that the settling out from a basic 

 magma of the more calcic plagioclases and the pyroxenes rich 

 in magnesia and iron is the dominant control 'in the differenti- 

 ation of the ordinary lime-alkali series of igneous rocks. 

 Possibly causes other than gravity, which may bring about a 

 similar relative movement of crystals and liquid, may be of 

 equal importance, an example being the intrusion or forcing 

 out of the liquid phase at a stage when most of the" magma 

 has crystallized. It would therefore be better to state that 

 fractional crystallization, as opposed to fractionation in the 

 liquid state, when aided by the processes mentioned, appears 

 to be the prime factor* in the differentiation of the series 

 mentioned. 



In making this result possible the occurrence of mix-crystals 

 is one of the fundamental factors. 



Dissociation in the Liquid Phase and its 

 Relation to Crystallization. 



There is one question on which the behavior of the artificial 

 melts throws some light and of which some mention might be 

 made, that is the question of the state of combination of the 

 various oxides in the liquid state. It has been maintained by 

 some that the oxides must exist uncombined ; at the time of 

 crystallization they combine and the combinations may r be dif- 

 ferent under different physical conditions. Unless this is true, 

 it is stated, the same magma could not under different condi- 

 tions give rise to different rocks. To this argument the 

 objection has been raised that there is no evidence that the 

 heat of combination is set free at the time of crystallization and 

 that the oxides must therefore be combined. From the view- 

 point of general chemistry, it might be stated that both views 

 are right and both wrong. We must consider that in any 

 melt there are certain amounts of the free oxides and of all 

 possible combinations of these oxides.f The necessity of such 



*Cf. Harker, A.: Fractional Crystallization the Prime Factor in the Dif- 

 ferentiation of Rock-Magmas. Communicated to 12th Int. Geol. Congress, 

 Toronto, 1913. 



f Cf. Harker, A.: The Natural History of Igneous Eocks, p. 165. 



