Now if we carefully consider this peculiar action and seek a cause for the phenomenon, we shall be led to the conclusion that it arises from the expansion of that portion of the molten mass which is in contact with or close proximity to the comparatively cool sides of the ladle, which sides act as the chief agent in dispersing the heat of the melted metal. The motion of the scoria betrays that of the fluid metal beneath, and careful observation will show that the motion in question is the result of an upward current of the metal around the circumference of the ladle, as indicated by the arrows A, B, C in the accompanying sectional drawing of the ladle (Fig. 2). The upward current of the metal can actually be seen when specially looked for, at the rim of the pot, where it is deflected into the convergent horizontal direction and where it presents an elevatory appearance as shown in the figure. It is difficult to assign to this effect any other cause than that of an expansion and consequent reduction of the specific gravity of the fluid metal in contact with or in close proximity to the cooler sides of the pot, as, according to the generally entertained idea that contraction universally accompanies cooling, it would be impossible for the cooler to float on the hotter metal, and the curious surface-currents above referred to would be in contrary direction to that which they invariably take, _i.e._, they would diverge from the centre instead of converging to it. The external arrows in the figure represent the radiation of the heat from the outer sides of the pot, which is the chief cause of cooling. [Illustration: Fig. 2.] Turning from cast-iron to other metals we find further manifestations of this expansive solidification. Bismuth is a notable example. In his lectures on Heat, Dr. Tyndall exhibited an experiment in which a stout iron bottle was filled with molten bismuth, and the stopper tightly closed. The whole was set aside to cool, and as the metal within approached consolidation the bottle was rent open by its expansion, just as would have been the case had the bottle been filled with water and exposed to freezing temperature. Mercury affords another example. Thermometers which have to be exposed to Arctic temperatures are generally filled with spirit instead of quicksilver, because the latter has been found to burst the bulbs when the cold reached the congealing point of the metal, the bursting being a consequence of the expansion which accompanies the act of congelation. Silver also expands in passing from the fluid to the solid state, for we are informed by a practical refiner that solid floats on molten silver as ice floats on water; it also, as likewise do gold and copper, exhibits surface converging currents in the melting-pot like those depicted above for molten iron.