[ 262 ]

than in their fluid state, and consequently that the meltedmetals contract in becoming solid. In making the lastof these experiments some caution is necessary, particu-larly with regard to lead and tin : for if these metals aremade very hot, the pieces thrown in will melt so fast,that it cannot be judged whether they sink or swim : ifnot heated sufficiently, they prove so thick, that the ex-cess of gravity of the solid piece will not be able to over-come the tenacity of the fluid, whence the piece will besustained on the surface, more especially if it is small,and if the {kin which forms on the fluid has not been pre-viously taken off.

On trying the same experiments with iron, the eventwas different. Ingots of iron appeared, he fays, not hol-low on the surface, but sensibly elevated or convex. Ha-ving brought some iron into thin fusion, and carefullycleared the surface from gross matter, he threw in solidpieces of cast iron of different kinds : they all swum, likeice on water, and when pushed down beneath the sur-face, they immediately rose again ; a proof that the solidpieces were lighter or more expanded than the fluid, andconsequently that melted iron expands or increases itsbulk in becoming solid. To this property he ascribes theneatness with which cast iron receives impressions from.moulds, a neatness wondered at by those who work onlyin copper or other'metals: he fays he has seen a hundredand a hundred times, that though iron was poured quitsthick into the mould, it nevertheless took the figure well,its expansion forcing it into the smallest cavities : that hehas often seen the workmen surprized to find, that it waswith the utmost difficulty they could unscrew the mouldsin which iron had been cast, while nothing of this kindis ever observed to happen in the casting of other metals :that works of lead, copper, gold, and silver, are always

found