Rankin and Wright — Tertiary System CaO-Al 2 0-Si0 2 . 35 



~&^. 7 shows that such is the case. There are 19 of these boundary 

 curves along which the temperature rises continuously ; that is, 

 there are 19 boundary curves which do not show a maximum. 



Approximate temperatures along boundary curves may be 

 deduced by extrapolation from the melting temperatures 

 determined for points within the fields but for more exact 

 temperatures heating curves and quenchings are required. For 

 the experimental determination of the temperature of any 

 point on a boundary curve it is not necessary to use the com- 

 position for that point ; any composition may be used which is 

 on a straight line passing through this point on the boundary 

 curve and the composition of either solid phase which occurs 

 on this boundary curve. The reason for this is obvious if we 

 consider the successive changes which take place when a ter- 

 nary mixture is heated. 



In heating curves of ternary mixtures the first break (heat 

 absorption) observed corresponds usually to the temperature of 

 a quintuple point. If the heating is continued, melting pro- 

 ceeds and equilibrium follows along a boundary curve until 

 one of the two solid phases in equilibrium with liquid and 

 vapor has completely melted ; at which point we find the 

 second break, which is the melting temperature corresponding 

 to that point on the boundary curve. Each point on the bound- 

 ary curves represents the composition of the liquid which is 

 in equilibrium at that point with the still solid portion of the 

 charge, but when we have progressed along the boundary curve 

 until it intersects the straight line drawn through the com- 

 position of the ternary mixure under investigation and the 

 composition of the primary phase of that mixture, then the 

 locus of the points representing the composition of the liquid 

 phase leaves the boundary curve and passes into the appropriate 

 field along the straight line. 



A more complete discussion of the changes which take place 

 during the heating of a ternary mixture will be given when 

 we come to consider ternary crystallization curves. For the 

 present it seemed to be sufficient to show : that the temper- 

 ature for any point on a boundary curve may be determined 

 from a number of different compositions ; that for any ternary 

 composition the point at which its heating curve will leave the 

 boundary curve is known ; that the second break in a heating 

 curve is usually the temperature of a point on the boundary 

 curve. For those ternary mixtures in which the second break 

 in the heating curve is not a boundary curve temperature, it is 

 necessary to determine by quenching the temperature at which 

 on heating the composition leaves the boundary curve. This is 

 done by determining the temperature above which one solid 

 crystalline phase is obtained imbedded in glass and below 



