1892.] On certain Ternary Alloys. 383 



of the tie-lines to the one side or the other, since the lowest 

 possible tie unites the points obtained with a percentage of 

 solvent metal = nil, whilst the upper ties dwindle down to a point at 

 the limiting point, so that for each of these limiting ties the per- 

 centage of solvent metal is the same in both heavier and lighter 

 alloy formed ; but it is noticeable that only in one other case out of a 

 dozen ternary alloys now under examination is a second maximum 

 noticeable, i.e., only in this one other case (lead-aluminium-tin alloys) 

 are the upper and lower ties found to slope in opposite directions ; in 

 all other instances, whether the ties slope to the right or to the left, 

 the direction of the slope is the same throughout the whole extent of 

 the critical curve. 



It is further noticeable that both with lead- zinc- tin and lead- 

 aluminium-tin alloys the position of the first maximum is such that 

 it occurs when the ratio of lead to tin in the heavier alloy is sensibly 

 near to that indicated by the formula SnPb 3 . Thus in the case of the 

 lead-zinc-tin alloys above described 



Calculated. Found. 



Sn 118 = 15-9? 1476 = 15'41 



Pb 3 621 = 84-03 81-02 = 84'59 



739 = 100-00 95-78 = lOO'OO 



The parallel results obtained with lead-aluminium-tin alloys will be 

 described in a future paper ; it may be noticed, however, that with 

 neither series of alloys is any marked elevation or depression in the 

 outline of the critical curve noticeable at the part corresponding with 

 the compound SnPb 3 , unlike the curve obtained with lead-zinc-silver 

 and bismuth-sinc-silver alloys, where the formation of the definite 

 compounds AgZn 5 and Ag 4 Zn 5 leads to marked alterations of outline 

 (vide infra). 



It was found impracticable to obtain any accurate valuations of 

 tie-lines situated nearer to the limiting point than No. 14 ; several 

 attempts were made, but the results exhibited too great an amount 

 of discordance amongst themselves, and too wide departures from the 

 curve indicated by the above experiments, to enable them to be re- 

 garded as trustworthy ; the causes of this being, as above stated, the 

 slight difference in density between the two alloys formed, and the 

 relatively large effect of temperature variation at this part of the 

 curve as compared with other portions further removed from the 

 limiting point. In fig. 4 (representing the above table plotted on 

 the triangular system) L is the limiting point deduced by Stokes* 

 2nd Method from a carefully made large-scale plotting, giving as the 

 most probable values A + A' = 45 and B + B' = 85, whence 



2 D 2 



