330 DR. F. DixEr ON" THE [vol. Ixxviii^ 



' If then we confine ourselves to the chief minerals, and if we provisionally 

 look upon the two pyroxenic minerals as one unit, we find here the same three- 

 stages of crystallization as in the ordinary ternary systems consisting o£ 

 independent components.' 



In extension of this idea, the following series might also be 

 expected to occur, since magnetite crystallized earlier than felspar, 

 when the ratio of the magnetite-component to the felspar- 

 component exceeded a certain quantity: — 



B. (1) Magnetite. 



(2) Magnetite and felspar (simultaneously). 



(3) Magnetite, felspar, and pyroxene (simultaneously). 



Both series A and B actually occur in the Sierra-Leone norite,. 

 even to the development of phenocrj^sts of labradorite in the 

 less coarse varieties. Moreover, with a slighth" increased propor- 

 tion of pyroxene, the following moditication of A is sometimes, 

 produced : — 



C. (1) relspa,r. 



(2) Felspar and pyroxene. 



(3) Felspar, pyroxene, and magnetite. 



(The minerals of stage 2 and of stage 3, respectively, in this, 

 and the following systems ciystallize simvUtaneously.) 



There is also evidence in the formation of other binary and 

 ternary intergrowths, graphic or otherwise, of the existence in the 

 magma of a number of additional ternary systems, due to the 

 inclusion of olivine as a fourth independent component; they^ 

 include : — 



D. (1) Felspar. 



(2) Felspar and olivine. 



(3) Felspar, olivine, and magnetite. 

 And E. (1) Pyroxene. 



(2) Pyroxene and olivine. 



(3) Pyroxene, olivine, and magnetite. 



The development of the above ternary systems in the porite- 

 does not imply great variations in the composition of the magma 

 as originally intruded, nor general variations in the resultant rock.. 

 There is evidence from the experiments of A. L. Da}^ & E. S. 

 Shepherd,^ that magmas differing slightly in composition will, at 

 different temperatures, give rise to a limited number of eutectics- 

 which differ considerably one from the other in composition and iu 

 the character of the minerals formed. The experiments referred 

 to were made upon the lime-silica series of minerals, for which 

 three eutectic points were found ; but it seems very pi'obable that 

 the results would apply in some measure at least to the crystal- 

 lization of more complex magmas such as occur in nature. Local 

 variations in composition within the magma were probably set 

 up in the course of crystallization and cooling, and they were 

 doubtless complicated also by more or less superfusion and possibl}^ 

 viscosity." 



^ Amer. Journ. Sci. ser. 4, vol, xxii (1906) pp. 265 et seqq. 



* J. V. Elsden, ' Principles of Chemical Geology' 1910, p. 151. 



