1890.] On certain Ternary Alloys. 27 



sign, i.e., the curve first rises above the base line to a maximum, and 

 then sinks again, and crosses the base line, falling below it. 

 Similarly, with curves of the second kind, it sometimes happens that 

 the ordinate value reaches a negative maximum, and then lessens 

 again, so that the curve again approaches the base line ; as yet, 

 however, we have not met with a case where the curve actually 

 crosses the line giving a + ordinate value. 



Mixtures of Lead, Zinc, and Tin at Higher Temperatures. 



The experiments described in Part I indicated that little if any 

 difference is produced in the composition of the two alloys into 

 which a given mass of these three metals separates by varying the 

 temperature at which the fused mass is maintained between 565 

 and 689, or between about 650 and 750; sensibly the same 

 solubility curves for zinc in lead-tin, and for lead in zinc-tin, resulting 

 in all cases. We find, however, that if a higher temperature be 

 employed, 750 850 and upwards, a measurable increment in 

 solubility is produced. At this more elevated temperature, the 

 volatility of zinc is considerably enhanced, so that the ratio between 

 the lead and zinc present in the compound ingot finally obtained 

 differs more from that in the mass of metals originally weighed up 

 than was the case in the experiments described in Part I ; in those 

 experiments, the average loss by oxidation and volatilisation per 

 100 parts of original metals jointly was nearly 4 parts, of which 

 scarcely anything was due to loss of tin, about one-third to loss of 

 lead, and some two-thirds to loss of zinc. In the experiments 

 described below, 100 parts of original metals lost on an average 

 about 10 or 12 parts altogether, almost the entire increment in the 

 loss being due to enhanced volatilisation of zinc, this larger amount 

 being due not only to the higher temperature, but also to the longer 

 time of fusion (some twenty-four hours instead of eight). To reduce 

 this loss by volatilisation to a minimum, the molten metals, when 

 poured into the red-hot narrow clay test-tubes, were covered with 

 layers of fused cyanide of potassium some 12 or 15 millimetres thick. 



The melting arrangements employed throughout in the experi- 

 ments described in this paper were substantially those detailed in 

 Part I : viz., the weighed metals were fused with cyanide of 

 potassium in a clean crucible, well stirred together for some time, 

 and poured into a red-hot clay test-tube, which was then kept hot 

 for some hours by immersion in a bath of molten lead, fused in an 

 iron cylindrical vessel surrounded by a clay jacket, several large 

 Bnnsen flames playing into the interspace being the source of heat. 

 The temperature was ascertained from time to time by means of the 

 platinum specific heat pyrometer, as described in Part I. It was 



