Plagioclase Feldspars. 585 



the beginning of melting of the mix crystals, for the presence 

 of but 2 per cent of anorthite, on account of the steepness of 

 the liqnidus, is sufficient to cause a melting interval of about 

 50°. The temperature found for carefully selected, finely 

 ground material was 1100° ±10°. This may be taken as prac- 

 tically the melting temperature of albite itself on account of 

 the flatness of the solid us at this point. In spite of the stub- 

 born nature of albite in the matter of affording complete melt- 

 ing no great difficulty was found in fixing a temperature of 

 beginning of melting of this material. The temperature 1100°, 

 moreover, lies on a rational extrapolation of the liqnidus, 

 although extrapolation alone could not fix the point closely on 

 account of the steep slope of the liqnidus. 



It will be noted that the curve of Day and Allen lies nowhere 

 very far below the liquidus.* At Ab 2 AHj the difference is 20°. 

 Beyond AbgAi^ no breaks were obtained by them and extra- 

 polation was correspondingly uncertain. 



The Composition of the Liquid Phase. 



The composition of the phases present at any temperature 

 could, of course, be predicted from the diagram as determined, 

 but it was desired to obtain independent confirmation of it 

 from actual determinations of optical properties. All hope of 

 success in the case of the solid phase was soon abandoned, for 

 the crystals, embedded in the glass, are always extremely small, 

 sometimes so thin, in fact, that their birefringence is scarcely 

 discernible. 



In the case of the liquid phase the outlook was more hope- 

 ful, since the determination of a single property, viz., the 

 refractive index of the glass, is sufficient to fix the composi- 

 tion. In most cases it was only when the glass was present in 

 considerable excess over the crystals that its index could be 

 determined, since it was only then that an edge of glass of 

 sufficient size for comparison with the immersion liquid could 

 be obtained. The difficulty was enhanced by the small size of 

 the crystals embedded in the glass, for, on this account, five or 

 six layers of crystals might be contained in the average grain 

 and in the projection viewed under the microscope their appar- 

 ent proportion was correspondingly multiplied. The difficulty 

 was avoided in part by choosing only the thinnest grains, but 

 this lessened somewhat the accuracy of the comparison with 

 the liquid. In a few favorable instances, however, definite 

 results were obtained. 



To determine the refractive index the quenched charge was 

 crushed and the fine powderf immersed in liquids of known 



* See revised temperatures, this Journal (4), xxxi, p. 347, 1911. 

 + Wright, F. E., The Methods of Petrographic -Microscopic Kesearch, 

 Carnegie Institution of Washington, Pub. No. 158, p. 87. 



