[ 264 ]

soon aster the metal has been poured in. It is supposed,that as cast iron begins to set on the surface almost imme-diately on its being poured into the mould, so as to be-come covered with a solid shell, the cavities must proceedfrom the fluid part within shrinking while it becomessolid. Several experienced artists have also assured me thatthe melted iron will shrink from the hoop.

I shall always take the earliest opportunity of acknow-ledging mistakes, and my obligations to those who shallpoint them out. With regard to the method of makingthe rolls, it may be observed, that the disputed propertyof iron does not affect the essential part of the proposal,which may therefore still deserve attention, whether ironhas or has not that property : for though the particularway, recommended for joining the parts, should not an-swer, the artist cannot be at a loss to And means of sup-plying the defect, or of fixing an iron axis in a steel hoop.

To satisfy myself in regard to the fact, and to disco-ver whence any fallacy might have arisen in the conse-quence of an experiment which appeared so decisive, Imade some further trials.

A rod of iron being placed upright in the middle ofa steel ring, I melted some cast iron, and poured it intothe space between them. When cold, the cast iron firmlyembraced the rod, but parted without difficulty fromthe ring, though it had received very neat impressions fromsome marks on its surface.

I melted a quantity of the iron in a large crucible, andthoroughly cleared it from the gross matter on the surface.When in perfect fusion, I threw into it a solid piece ofthe same kind of iron previously heated : the piece droptto the bottom, but immediately rose up again to the top,as wood does in water : being pushed down with an ironrod, it rose again, and continued to float till it melted

and