The Ternary System Mg O-Al, O t -Si % . 319 



the crystals have one dimension or more below micro- 

 scopic size, the aggregate will appear homogeneous. 

 The substance present in largest amount may crystallize 

 in a meshwork which limits the size of the particles of the 

 other substances. In such a texture diffusion would be 

 so hindered that a very long heating might be required 

 to allow the excess constituent to grow to appreci- 

 able size. 



Aggregates in which the ternary compound was the 

 chief constituent were prepared by heating for periods 

 of 15 minutes to 5 hours at 900° to 1400°. While the 

 glass was heating the /*, or unstable, form of the ter- 

 nary compound began crystallizing at 900 c or less as fine 

 fibers which radiated from points on the surface of the 

 grains. At temperatures of 925° to 1150° this form 

 inverted to the -form. The inversion in different por- 

 tions of the same charge, and in different aggregates 

 from glasses of the same composition, had an extreme 

 temperature range of about 200°. This range is mostly 

 accounted for in a lag in the starting of the inversion. 

 When once started in a grain the inversion proceeded so 

 rapidly that only in a few instances was a grain found, 

 after quenching, to contain both forms. It is significant 

 that the apparent homogeneity of aggregates containing 

 the low temperature, or /"-form, covered a greater range 

 of composition than that of the high temperature form. 

 Finer grained and therefore more homogeneous-appear- 

 ing aggregates would presumably be formed at the lower 

 temperatures. The refractive indices of the aggregates 

 decrease practically linearly with increase of silica. 

 The average refractive index of clear aggregates con- 

 taining the ^-form ranges from 1-535 at 68% Si0 2 to 

 1-560 at 40% Si0 2 ; and of aggregates of the a-form, 

 1-515 at 64% Si0 2 to 1-540 at 44% Si0 2 . 



The form appearing at the lower temperatures is called 

 unstable because repeated attempts to produce it by heat- 

 ing the stable form at temperatures below those at which 

 the stable form could be produced from the unstable met 

 with no success. Thus there appears to be no true 

 inversion point in this case, but the temperatures at 

 which the rate of inversion becomes notable may be dif- 

 ferent for aggregates differing in silica content. An 

 apparent rise of this temperature amounting to about 

 100° for an increase of 30 per cent of silica was observed. 



