438 Andersen — System An orthite- For sterite- Silica. 



both cases, the further crystallization, of course, proceeds 

 in the quaternary system. It is obvious that the lines B K 

 and C K do not coincide with any crystallization curves in 

 the ternary system (like, for instance, G- N or N M) but simply 

 represent boundary curves between the spinel field and the 

 adjoining fields. 



Mixtures with excess of MgO. — It may be well in this con- 

 nection to add a brief record of some experimental work on 

 mixtures containing more MgO than is necessary to bind all 

 free silica in the compound Mg 2 Si0 4 (forsterite), that is, mix- 

 tures in the section of the system anorthite-MgO-Si0 2 

 which lies between the line anorthite-forsterite and the line 

 anorthite-MgO. Thermal work with the platinum resistance 

 furnace was possible only on mixtures near the anorthite cor- 

 ner of the diagram and in fact most of these mixtures belonged 

 to the spinel field. 



In the system anorthite-MgO spinel separates as the 

 primary phase in mixtures containing more than 7 per cent 

 MgO. At the liquidus point for the mixture An 93, MgO 7. 

 anorthite and spinel are able to coexist in equilibrium with 

 the melt. The temperature of this point is about 1440°. 

 This gives a very steep liquidus curve for anorthite, the tem- 

 perature gradient averaging about 16° for one per cent MgO. 

 A few points on the liquidus curve for spinel were also 

 determined. The curve rises quickly with increasing amount 

 of MgO. At a composition An 80, MgO 20 the temperature 

 of the liquidus point was higher than 1630°, a mixture 

 quenched from this temperature containing glass and spinel. 



The liquidus surface of the spinel field in the section 

 anorthite-forsterite-MgO of the system rises steeply from the 

 line anorthite-forsterite and soon reaches temperatures inac- 

 cessible in the platinum resistance furnace. A mixture An 60, 

 Fo 21, MgO 19 quenched from 1620° gave a product consist- 

 ing of spinel and glass. 



Various pure glasses of mixtures within the spinel field were 

 crystallized at temperatures between 800° and 1000°. Spinel 

 was not found in any of the resulting products. This shows 

 that the spinel field is a reaction field or, in other words, that 

 the spinel found in the mixtures considered is redissolved in 

 the usual course of cooling. Besides this nothing definitely 

 can be said about the crystallization curves of mixtures within 

 any part of the spinel field. 



The ternary field. 



All mixtures outside the triangle A K D (fig. 9) behave as 

 part of a true ternary system. These mixtures will here be 



