360 
Proceedings of the Poyal Society 
will show the reasons for selecting one as the most reliable. On 
pouring iron into a sand mould there is, at the moment of solidifica- 
tion some overflow ; but no matter how tightly the sand was rammed, 
or to what temperature the mould was heated, the overflow varied so 
very much as to show that, as a method of measuring the expansion, 
pouring the iron into a sand mould was quite useless. The experi- 
ments conducted in this way showed an expansion of from 08 to 
4 ’5 per cent., showing the method to be unreliable. 
We then tried pouring the metal into a hollow sphere of iron 
whose capacity has been accurately determined. The sphere, how- 
ever, seemed to yield in some parts, so that the overflow did not 
represent accurately the real expansion ; but by weighing the filled 
sphere and the overflow, after we had arrived at an approximate 
value for the liquid iron’s density, a more reliable estimate of the 
expansion was found. This method gave results varying from. 4 to 
5 per cent, of expansion, but the results were always low. The 
method which gave not only the most concordant results, but which 
would a priori be likely to yield the most accurate estimate of the 
expansion, was that of floating a solid sphere of iron in liquid iron 
of the same composition. The metal used was ordinary grey pig, 
and was contained in a large pot and brought to a temperature near 
its freezing point and spheres of metal dropped in. They were found 
to sink at once when dropped in cold, and they remained under the 
metal till they had acquired a temperature just approaching visible 
red ; but at that temperature they rose to the surface, and as they 
gained more and more heat from the liquid metal their line of flota- 
tion rose higher and higher. Sometimes, if dropped in suddenly, 
the spheres did not float until they had begun to melt, but this was 
owing to their having cemented themselves to the bottom of the pot. 
When dropped in cautiously, or suspended by a wire, they sank only 
for the space of 20 to 25 seconds, and rose to the surface when barely 
red hot. The spheres were allowed to remain in the liquid till they 
began to melt, and then withdrawn and cooled, when a well-defined 
mark of the line of flotation was seen round the sphere. The spheres 
and their flotation segments were measured by several methods — 1st, 
by callipers ; 2d, photographed, and the photograph measured by a 
dividing engine ; and 3d, by a telescope and micrometer. The last 
method yielding the most concordant results. The spheres from 
