The broad general principle of the phenomenon here referred to is this:—that fusible substances are (with a few exceptions) specifically heavier while in their molten condition than in the solidified state, or in other words, that molten matter occupies less space, weight for weight, than the same matter after it has passed from the melted to the solid condition. It follows as an obvious corollary that such substances contract in bulk in fusing or melting, and expand in becoming solid. It is this expansion upon solidification that now concerns us. Water, as is well known, increases in density as it cools, till it reaches the temperature of 39° Fahrenheit, after which, upon a further decrease of temperature, its density begins to decrease, or in other words its bulk expands, and hence the well-known fact of ice floating in water, and the inconvenient fact of water-pipes bursting in a frost. This action in water is of the utmost importance in the grand economy of nature, and it has been accepted as a marvellous exception to the general law of substances increasing in density (or shrinking) as they decrease in temperature. Water is, however, by no means the exceptional substance that it has been so generally considered. It is a fact perfectly familiar to iron-founders, that when a mass of solid cast-iron is dropped into a pot of molten iron of identical quality, the solid is found to float persistently upon the molten metal—so persistently that when it is intentionally thrust to the bottom of the pot, it rises again the moment the submerging agency is withdrawn. As regards the amount of buoyancy we believe it may be stated in round numbers to be at least two or three per cent. It has been suggested by some who are familiar with this phenomenon that the solid mass may be kept up by a spurious buoyancy imparted to it by a film of adhering air, or that surface impurities upon the solid metal may tend to reduce the specific gravity of the mass and thereby prevent it sinking, and that the fact of floatation is not absolutely a proof of greater specific lightness. But in controversion of these suggestions, we can state as the result of experiment that pieces of cast-iron which have had their surface roughness entirely removed, leaving the bright metal exposed, still float on the molten metal, and further that when, under the influence of the great heat of the molten mass, the solid is gradually melted away, and consequently any possible surface impurities or adhering air must necessarily have been removed, the remaining portion continues to float to the last. The inevitable inference from this is that in the case of cast-iron the solid is specifically lighter than the molten, and, therefore, that in passing from the molten to the solid condition this substance undergoes expansion in bulk.