Rankin and Wright — Ternary System CaO-Alfi^-SiO^. 21 



ary curves. In the case of ternary systems at ordinary temper- 

 atures, this is usually done either by adding the phase under 

 investigation and finding out whether or not it dissolves in the 

 saturated solution or by allowing the saturated solution to crys- 

 tallize slightly so that the first crystals to form may be iden- 

 tified.* To apply this method at high temperatures a small 

 amount of the material of the desired composition is placed in 

 a furnace and held at constant temperature until on quenching 

 (sudden cooling by dropping into a bath of mercury or water), 

 only one kind of crystal is present, and the rest of the solution 

 is cooled to a glass before it has time to crystallize. The exact 

 temperature for any given mixture must be determined by a 

 series of trials with sufficient variation of the conditions of ex- 

 periment to rule out effects due to slowness of reaction or of 

 crystallization. Temperatures will thus be found where the 

 charge is wholly glass, and also where the product obtained by 

 quenching is wholly crystalline. The method also allows the 

 crystals time to grow to measurable size, thus assisting the 

 microscopic examination. In other words, the method devel- 

 ops the primary phase (BodenJcorper) embedded in the glass 

 (solution) and allows a rapid and very satisfactory mapping of 

 the fields of stability for each phase. The method is impracti- 

 cable when the temperature rises too high for the platinum 

 furnace (1600°) ; the iridium furnace (1600°-2100°) does not 

 serve well for it. .Naturally the method does not work so well 

 where two kinds of crystals are present. To get the most sat- 

 isfactory results, i. e., fair-sized crystals of the primary phase, 

 the charge should be held at a temperature just below that of 

 complete fusion for the composition in question. 



The material to be investigated by each of the three methods 

 is prepared as follows : The finely powdered oxidesf CaO (as 

 CaC0 3 ), A1 2 3 , Si0 2 are mixed in the desired proportions, placed 

 in a platinum crucible, and fused in a Fletcher gas furnace. 

 The charge is then removed from the crucible ; broken up in 

 a steel mortar ; ground to a fine powder in a mechanical agate 

 mortar and fused again. The process of fusion, with fine 

 grinding between fusions, is repeated at least three times to 

 obtain a product chemically homogeneous. In this process 

 there are two operations, during which there is danger of losing 

 a portion of the charge and also of introducing foreign mate- 

 rial, (1) during the breaking up of the charge in the steel 

 mortar, and (2) the subsequent grinding in the agate mortar. 

 If, however, great care is exercised charges can be prepared in 

 this way, without the addition of over 0*2 per cent foreign 

 material or a change of over 0*2 per cent from the original per- 

 centage of the oxides. We have found that charges of twenty 



* Cf. Eoozeboom, loc. cit. 



f Careful tests of the purity of the materials used showed that they con- 

 tained no appreciable amount of impurity. 



