552 
Proceedings of the Eoyal Society 
surface interferes with their movement, larger pieces, especially 
those of a hall shape having a maximum of weight with a minimum 
of surface, being to he preferred. 
The conclusions drawn by the author from his experiments were, 
therefore, that when the cold solid comes in contact with the 
intensely heated molten metal, sufficient expansion is induced in the 
solid metal as to give it the required buoyancy to rise to the surface, 
and that this may occur by part of the surrounding heat being 
converted into mechanical work, causing a sudden increase in bulk, 
and partly by a dilatation due to rise of temperature of the solid. 
Indeed, it also appears likely that the high temperature surrounding 
the immersed solid (over 2000° F.) may suddenly cause a change in 
the condition of the particles of the solid which have most probably 
taken a strained condition during cooling. 
That some such condition exists in castings appears probable, as 
in some cases annealing has to be resorted to, to prevent liability 
to fracture. And from experiments made by the author on the 
deflection and set of cast iron test-bars,* it would appear that 
when such bars are at first loaded and allowed to deflect, a certain 
amount of set is observed. On further application of the same, or 
even of greater loads, this set decreases, until ultimately no set is 
apparent, showing that in all probability the work done by the load 
upon the bar, has up to this point been one of freeing the particles 
from a local strained condition due to cooling. The common 
foundry practice of hammering curved castings on the concave side 
so as to straighten them, also points to relief of strain in the particles 
by the vibration set up. Such a strained condition may very well 
exist even in so simple and regular a form of casting as a test-bar, as 
the outer surfaces, cooling more rapidly than the inner, must 
cause compression of the latter. 
4. If we take, as in common foundry practice, -^th part (or of 
an inch to the foot) as the change in linear dimensions from the 
molten to the finally cooled solid condition, we may calculate from 
coefficients of expansion the temperature to which the cold solid 
would require to be raised to enable it to float on the molten metal. 
Thus, if we take the rate of expansion per 180° as -g-Joth part 
* Trans. Inst. Engineers and Shipbuilders in Scotland , vols. xix. and xxi. 
Minutes of Proc. Inst. C.E., vol. lviii. part iv. 
