374 Dr. C. R. Alder Wright. [Jan. 28, 



what must be the proportion between A and B, in order to yield with 

 a somewhat larger quantity of C an " ideal " alloy that will separate 

 into approximately equal quantities of the two ternary alloys formed 

 therefrom. Similarly, from these further results, the proportions 

 requisite for the accurate determination of other pairs of conjugate 

 points higher up still can be inferred; and so on. In this way the 

 critical curve is gradually traced out with a much smaller proportion 

 of wasted labour (owing to imperfect formation of truly conjugate 

 ternary alloys) than would otherwise be the case. This method of 

 procedure has accordingly been adopted in the investigation of 

 various critical curves of the kind which will be discussed in a future 

 paper. 



These considerations obviously suggest the possibility that, in 

 certain cases at any rate, the mean curve values deduced in the 

 earlier parts of this research may require some little degree of 

 revision, inasmuch as some of the pairs of alloys simultaneously 

 formed were derived from mixtures yielding one alloy in much 

 larger quantity than the other ; so that some small amount of experi- 

 mental error, due to imperfect formation of truly conjugate alloys, 

 might exist. Accordingly, a considerable number of the experiments 

 that might possibly be faulty from this cause have been repeated, 

 using proportions of A and B better suited to the end in view, viz., 

 formation of the two ternary alloys produced in not widely different 

 quantities relatively to one another. Further, a variety of additional 

 experiments have been made with the object of deducing the situa- 

 tions of pairs of conjugate points lying nearer to the limiting points 

 than those previously determined. These further experiments, how- 

 ever, have demonstrated the existence of two other sources of error, 

 not noticeable to so great an extent at the other parts of the critical 

 curve. 



In the first place, when an " ideal " alloy is used corresponding 

 with a pair of conjugate points lying near to the limiting point, the 

 two alloys formed differ far less widely in composition than is the 

 case with other pairs more removed from the limiting point ; conse- 

 quently, the densities of the two alloys do not greatly differ, which 

 circumstance appears greatly to impede their complete separation 

 from one another by gravitation whilst standing at rest molten. 

 Secondly, such "ideal" alloys appear to be extremely sensitive to 

 slight differences of temperature, at least as compared with mixtures 

 not so near the limiting point ; so that whilst a difference of 10, 20, 

 or even 50 C. makes but little difference in the compositions of the 

 two ternary alloys formed from a given " ideal " alloy not near the 

 limiting point, it produces a marked effect on an " ideal " alloy near 

 to that point; in some cases, indeed, a rise of a few degrees in 

 temperature will suffice to transform an " ideal " alloy separating into 



