17* Dr. Wright and Messrs. Thompson and Leon. [Feb. 



the other ; there can only be an interchange of constituents across 

 surface of separation. 



The critical curve may be otherwise defined as the curve expressing 

 the saturation of the solvent C with a mixture in given variable pro- 

 portion of the remaining substances A, B. That it is really such, 

 little consideration suffices to show. The determination according 

 of the critical curve furnishes us with definite information, even 

 though we do not go into the ulterior question of the condition of 

 conjugation. 



Perhaps the attainment of true conjugation might involve more 

 stirring than wonld be practically feasible with molten metallic mix- 

 tures. The most hopeful way would seem to be to fuse the 

 mass at a higher temperature than that intended for the experi- 

 ment, stirring it well, and then let down the temperature to that 

 intended, stirring all the time, and avoiding too rapid a fall of 

 temperature. 



If truly conjugate alloys were obtained, and portions of each we 

 taken and fused together at the temperature at which the allc 

 were made, the compositions ought to be the same as before. But 

 the alloys were merely associated, then, even if the stirring in the 

 second part of the experiment were sufficient to ensure conjugation, 

 the compositions wcnld not be the same as the original, nor wonld 

 they be independent of the proportion of the two alloys which the 

 operator took for fusing together. 



The triangular method of representation described by Sir G. 

 Stokes in the above note obviously possesses several advantages, ii 

 much as it represents in one diagram simultaneously a number of 

 results which the ordinary curves drawn with abscissae and ordinates 

 can only partially indicate, consequently necessitating several 

 different curves being drawn in order to represent graphically the 

 entire set of results ; thus the two branches of the "critical curv*-,' 

 obtained by directly plotting the figures yielded on analysis of the 

 lowest and uppermost portions respectively of the compound ingot 

 formed (in the case of a mixture separating into two different ternary 

 alloys), represent the two solubility curves (e.g., of zinc in lead-tin 

 and of lead in zinc-tin), whilst the " ties " or " tie-lines" indicate, 

 according as they slope to one side or the other, the relative propor- 

 tions of the " solvent " (e.g., tin) in the heavier and lighter alloys ; 

 so that, when (as in the case of mixtures of lead, zinc, and tin) with 

 certain proportions of "solvent " the heavier alloy, and with other pro- 

 portions the lighter one, contains the larger percentage, this variation 

 is at once indicated to the eye by the change in direction of slope of the 

 tie-lines (compare fig. 3). Further, when once the critical curve for 





