[ 266 ]

hundred and ten ; though it had filled the mould in itsfluid Hate, having taken an impression from both theends. The real contraction must have been somewhatgreater than this ; because the mould must necessarilyhave acquired a considerable heat at the time of the iron’sfixing, and consequently in cooling shrunk along with it.

I have mentioned above that I had seen instances, inwhich melted iron applied itself firmly to unmelted piecesby which it was confined. The foregoing observationsoccasioned me to recollect the circumstances in whichthose instances had happened, and indeed naturally pointthem out. Cast iron shrinks from an iron or steel ring,which it filled and distended at the time of its fixing :but if this ring be previously made very hot, it might bepresumed that its shrinking would keep pace with thatof the cast iron, so that the latter would still continue tofill it.

Accordingly I heated the ring to a strong red or ratherwhite heat ; and placing it on a bed of sand, poured intoit the melted iron : when cold, the cast iron filled thering, and was firmly applied to it, so as to be in no dan-ger of being separated or moved by any force that therolls for flatting gold or silver wire are designed to un-dergo ; though the juncture was not, perhaps, sufficientlystrong for resisting so great force as other rolls must ne-cessarily bear in the flatting of larger metalline masses.

This last experiment is entirely agreeable to, and seemsto confirm, the foregoing. For though the ring or mouldbe supposed heated even to the degree in which cast ironsets i yet, if the melted iron shrunk in setting, it wouldhave become less than the mould, and continued so in thesubsequent period of the cooling.

V. Of