28 JOURNAL OF GEOLOGY— SUPPLEMENT 



residual liquid will evidently work in the same direction as the 

 two processes of zoning and sinking of crystals, but it differs in not 

 having the same continuous nature, being rather an incident in the 

 history of the magma, though, no doubt, repeated several times in 

 some cases. 



THE SYSTEM ANORTmTE-EORSTERITE-SILICA 



After discussing the crystallization of liquids of the foregoing 

 system involving an olivine, a series of pyroxenes and silica, a 

 system involving a feldspar, an olivine, a single pyroxene, and siHca 

 will now be considered. This is the system anorthite-forsterite- 

 silica recently investigated by Andersen.^ 



The equilibrium diagram is shown in Fig. 4. 



Crystallization of Hquid of composition x, a mixture of cHno- 

 enstatite and anorthite, may take place in any of the following 

 ways. With perfect equihbrium and sinking of crystals prevented, 

 forsterite separates first and the composition of the liquid changes 

 along the line xy. At the temperature y the pyroxene clino- 

 enstatite begins to separate and forsterite to redissolve, the Kquid 

 changing in composition along the boundary curve FM. When the 

 temperature M is reached, the last of the liquid (composition M) 

 and the last of the forsterite disappear and the whole consists of 

 the pyroxene, cHno-enstatite, and anorthite. The same final result 

 would be obtained if the liquid were cooled very rapidly, that is, if 

 considerable undercooling occurred. 



If, however, forsterite crystals settled from the upper to the 

 lower layers, the consequence would be that at the temperature M 

 the liquid would be used up in the lower layers before all the 

 forsterite was redissolved and the whole lower portion would be 

 crystalline, consisting of the olivine, forsterite, the pyroxene, clino- 

 enstatite, and anorthite. The earlier crystals of oHvine might 

 collect in a separate layer, peridotite-like in nature. In the upper 

 layerS; however, the httle forsterite remaining is redissolved while 

 some Hquid of composition M still remains and, after complete 

 solution of forsterite, the liquid changes in composition along MN, 

 clino-enstatite and anorthite separating until at the temperature N 



' Olaf Andersen, Am. Jotir. Sci. (4), XXXIX (1915), 407. 



