26 Dr. Alder Wright and Mr. C. Thompson. [May 8, 



We have made a large number of experiments with various 

 mixtures of this kind. So far as our results are completed, they lead 

 to the following general conclusions : 



1. In all cases, the mixture of the three metals A, B, C, when 

 allowed to stand molten for a sufficient length of time at a tolerably 

 equable temperature, divides itself into two different ternary alloys 

 of unequal density, if the proportion of C present in the entire mass 

 falls below a certain limiting amount ; but, if the quantity of C 

 present is above this limit, no such separation takes place, only one 

 homogeneous alloy resulting. 



2. Under ordinary circumstances, the different alloys thus formed 

 are respectively a saturated solution of A in a mixture of B and C 

 (lighter alloy), and one of B in a mixture of A and C (heavier alloy); 

 the solubilities being such that, the greater the proportion of C 

 present, the more of A (or B) is dissolved. Certain metals, however, 

 appear to be capable of forming true chemical compounds in atomic 

 proportion, in which case the quantity of A (or B) dissolved does not 

 always vary directly with the amount of C present. 



3. The quantity of B dissolved by a given weight of A (or of A 

 dissolved by a given weight of B) in presence of a given weight of 

 C varies considerably with the nature of C. Moreover, although in 

 certain cases (e.g., the lead-zinc-tin alloys examined in Part I) a 

 considerable variation in temperature makes hardly any measurable 

 difference in the solubility, this is very far from being the general 

 rule ; the ordinary effect of increment in temperature is to increase 

 the solubility of A in BC, and of B in AC, in some cases to a very 

 considerable extent. 



4. The third metal C divides itself between the two alloys in a 

 fashion variable not only with the nature of A, B, and C, and with 

 the temperature, but also with the relative proportions subsisting 

 between A and B in the entire mass, and with the proportions of C 

 contained therein. If curves be drawn, as described in Part I, with 

 the percentages of C in one alloy as abscissae and the differences in 

 percentage between the two alloys as ordinates, two classes of curves 

 may be distinguished. In one, the percentage in the lighter alloy is 

 greater than that in the heavier ; calling the difference + , the curve 

 rises from the origin above the base line. In the other, the percentage 

 in the lighter alloy is less than that in the heavier one ; so that the 

 difference is now , and the curve falls from the origin below the base 

 line. With curves of the first kind, it generally happens that the 

 ordinate value increases gradually to a maximum and then 

 diminishes ; with some metals (e.g., silver-lead-zinc) the diminution 

 is only just perceptible ; with others (e.g., silver-bismuth-zinc) it is 

 more marked; whilst with some (e.g., tin-lead-zinc) it is carried so 

 far that at length the ordinate becomes 0, and subsequently in 



