140 



REPORT — 1900. 



experiments to discriminate between a summit due to the existence of a 

 chemical compound and one due to the form of a mixed crystal curve. In 



Fig. 8. 



Fig. 9. 



Ruuxebooin . 



m 



m 



Cone 



Cone 



such cases as Kurnakov's NaHg.j and KHgo, our AugAl, and Roberts- 

 Austen's AuAlj, where the formula of the summit is an exact and simple 

 one, there can be little doubt. A microscopic examination of the summit 

 alloy might not help, but that of alloys at some distance on either side 

 of the summit ought to settle the matter. If they are homogeneous, give 

 a uniform ignition colour, and etch all over at the same rates, the summit 

 must be due to mixed crystals ; while if the alloys show primary ci-ystalli- 

 sation embedded in a mother substance, the primary crystals continually 

 decreasing in amount as we go down the curve, the summit is probably a 

 compound. This is the structure we found near the two summits of the 

 AuAl curve. 



But Roozeboom points out as the best method of attacking such 

 questions the two following series of experiments : (1) Determine not 

 only the freezing-point of each alloy, but also the temperature at which it 

 sets to a solid mass. From these data, both of the curves l and s 

 could be plotted. The setting-point would probably not be very sharply 

 marked, but a recording pyrometer would indicate its whereabouts by a 

 greater rapidity in cooling after the point was passed. Cooling-curves 

 such as those of Sir W. Roberts- Austen might be made to give the 

 information needed for plotting the s curve. We, in a very imperfect 

 way, sought for the setting-points in determining our AuAl curves, but 

 found instead the usual horizontal lines of second freezing-points. The 

 existence of these renders mixed crystals improbable. 



The other line of research, adopted by Reinders and Van Eyk and 

 Hissink, is to extract the first crystals that form and analyse them, as 

 •well as the mother liquor from which they were taken. If this were done 

 for alloys on either side of a summit due to a compound, we should find 

 the crystals having all the same composition — namely, that of the com- 

 pound ; while if the summit were due to the existence of mixed crystals, 

 the solids extracted from alloys of various compositions would differ widely. 

 There can be no doubt that this process, troublesome though it would be, 

 is the proper way to attack the interpretation of a complicated freezing- 

 point curve. 



A few cases out of many may be mentioned where mixed crystals are 

 probable in alloys : between CugSn and Cu4Sn, in lead-thallium alloys, 

 in bismuth-antimony, in gold-silver, in alloys containing zinc or cadmium 

 with either silver, copper, or gold. 



A careful study of some of these cases is probably the most pressing 



