TEE REACTION PRINCIPLE IN PETROGENESIS 183 



ordinary way giving a mixture of the pyroxene MgSiOj and silica. 

 We thus obtain a mass consisting of olivine, p3n"oxene, and silica. 



Again it is found that the existence of the reaction relation 

 introduces a flexibility such that different products are obtained 

 with var3dng opportunity for reaction. Under certain conditions 

 no free silica is formed; under others a httle may be formed; and 

 under still others a considerable amount may appear. The com- 

 pound Mg2Si04 and MgSiOj may be called a reaction pair. By 

 this is meant that crystals of the first compound react with the 

 liquid to produce the second during the normal course of crystal- 

 lization. 



A reaction relation of this latter type may exist between three 

 or more compounds and the compounds, arranged in proper order, 

 may then be said to constitute a discontinuous reaction series. 

 An example of this kind of series is given by the system HaO-KaSiOj- 

 SiOa as worked out by Morey and Fenner.^ The equihbrium 

 diagram is given in Figure 4. A liquid of composition A begins to 

 crystallize with separation of KzSiOj. This is joined later by KaSizOj 

 and the two crystalKze side by side, the hquid changing in composi- 

 tion along the boundary curve between their fields. At Qi the liquid 

 reacts with K2Si03 converting it into KjSiOj • ^ H2O and if the oppor- 

 tunity for reaction is perfect the whole mass soHdifies. On the other 

 hand, if the opportunity for reaction is limited, some hquid Qi is left in 

 excess and it proceeds to crystalHze along Q1-Q2 with separation of 

 KzSiaOs and KaSiOj • | H2O. At Q2 the crystals of K2Si20s react with 

 the liquid to produce K2Si20s-H20 and soHdification is complete if 

 there is complete freedom of reaction. Here again there may be 

 some Kquid (Q2) left over if reaction is for any reason limited and 

 it will proceed to crystalHze along Q2-Q3 with separation of 

 K2Si20s • H2O and K2Si03 • IH2O. At Q3 the crystals of K2Si03 • ^B.J^> 

 are converted into K2Si03-H20. In this case there is always some 

 liquid in excess of that required for the reaction even with perfect 

 opportunity for reaction, but there is more of it when the reaction 

 is limited. The Hquid then crystalHzes along the boundary between 

 the fields of K2Si03-H02 and K2Si20s-H20, with separation of those 

 compounds, until the isotherm of 200° is reached, and if there is no 



' Journal American Chemical Society, Vol. XXXIX (1917), pp. 1173-1229. 



