300 Shepherd, Rcrnkin, Wright — Binary Systems of 



close together, like Al 2 Si0 5 -Al 2 3 , and the temperature dif- 

 ference is small, the exact location of the eutectic composition 

 is very difficult to establish. Unfortunately, the microscope 

 is also of little assistance in this case. 



All such observations must be made in an oxidizing or 

 neutral atmosphere ; a reducing atmosphere not only reduces 

 some of these oxides and silicates, but even where this is 

 not the case, the name playing about the furnace opening 

 renders the optical pyrometer useless for exact measurements. 

 A hydrogen atmosphere is perhaps the worst. At tempera- 

 tures above 1000°, it reduces silica or silicates, setting free 

 silicon which renders the platinum or iridium crucibles " hot 

 short" and ultimately destroys them. This reaction is 

 doubtless due to the silicon hydride, since it is by no means 

 necessary that the silica and platinum be in contact in order 

 that this destructive action occur. A platinum crucible in 

 which a charge of silica was heated to 1100° in hydrogen, 

 increased in weight by more than thirteen milligrams. On 

 analysis, almost the theoretical quantity of Si0 2 was found. 

 The crucible was highly crystallized and could be broken up 

 to a coarse powder in the fingers. Wires less highly charged 

 with silicon are very brittle when hot, even though not 

 appreciably so when cold. 



Carbon is known to react readily with lime above 1700° to 

 give calcium carbide, so that a carbon atmosphere above 1200° 

 introduces an undetermined error into the work. 



One of the series studied (lime-alumina) presents the 

 troublesome case of a compound unstable at the melting 



point. If the composition of 

 the compound be X (fig. 3), 

 then the first crystals to separate 

 on the freezing of this mixture 

 are, of course, B. At the tem- 

 perature £ 2 these crystals should 

 combine with component A to 

 form the compound AB. But 

 it often happens that the crys- 

 tals become coated over with 

 the compound AB and are 

 thereby removed from the action 

 of component A. The result is 

 that the charge freezes at t x to 

 a mixture of A, AB, and B, 

 which of course is unstable in 

 a two-component system. Theoretically, if the charge were 

 held for a long time at a temperature between t x and t 9 , dif- 

 fusion should cause all free B to disappear, so that the mass 



Fig. 3. 



