1889.] 



On certain Ternary Alloys. 

 Valve Crucible. 



471 



Tall narrow Crucible: 8 hours. 



In these cases it was clear that the formation of eutectic alloys during 

 solidification was not the cause of the irregularities observed, whence 

 presumably the same is true for the less irregular results obtained 

 with the pipe-bowls. Ultimately we traced the cause to convection 

 currents set up through unequal heating of the walls of the contain- 

 ing vessels at different levels, and found that the imperfect separation 

 could be almost completely obviated by so heating the mass as to 

 avoid this inequality of temperature. This we ultimately effected by 

 employing crucibles very long in proportion to their diameter (large 

 test-tubes moulded on a core from a plastic mixture of fireclay and 

 syrupy silicate of soda, diluted with about three times its weight of 

 water), heated by immersion in a bath of molten lead some 8 or 9 

 inches deep, contained in an iron cylindrical vessel (the lower two- 

 thirds of a mercury bottle), surrounded by a concentric clay jacket 

 and heated by a number of bunsen burners playing into the annular 

 interspace. The molten metals being well intermixed in a crucible 

 (with a little potassium cyanide), the mixture was poured into a red- 

 hot clay test-tube, which was then quickly transferred to the hot lead- 

 bath, the mouth of the tube being covered with a heavy iron cap so as 

 to depress the test-tube into the lead to such a depth that the top of 

 the molten metal inside was some 2 inches below the surface of the lead 

 in the bath whilst the tube was kept vertical. Under these conditions, 

 practically complete separation was always brought about after six to 

 eight hours in the lead-bath. Usually several test-tubes were heated 

 simultaneously. After the required time had elapsed they were care- 

 fully lifted out without shaking, and set by to cool, still in a vertical 

 VOL. XLV. 2 i 



