362 
Proceedings of the Royal Society 
Calculated by the two methods — 
(3r 2 + 7i 2 ) , 5236/^, and 
(3c? - 2^)*52367i 2 , 
where r — radius of segment 
h = height „ 
d = diameter of sphere. 
The first method always yields the highest results owing to the 
diameter of the segment appearing larger than it really is ; this being 
caused by the ridge of metal and scum which marks its flotation line. 
The difference between the two is not very great, but we take the 
value given by the last formula as the correct one. The amount by 
this method is 5‘62 per cent, of expansion. Further experiments 
were tried by heating balls of iron to various temperatures and 
immersing them in the liquid iron to find at what temperature they 
ceased to sink ; but this method fails for two reasons, — 1st, the iron 
freezes a quantity of metal on its exterior, thus increasing its volume ; 
and 2d, it gains heat so rapidly that before equilibrium is established 
its temperature has risen several hundreds of degrees. 
The expansion obtained by the above method is rather under the 
truth, because, although the sphere is just at its melting-point, the 
liquid iron is of necessity considerably above it, so that it is not at 
its maximum density, which appeared to be very little if any above 
the melting point. 
We find, then, that liquid cast iron expands at least 5*62 per cent, 
of its volume on freezing. 
2. Researches on Contact Electricity. By C. Gr. Knott, Sc.D. 
Communicated by Professor Tait. 
{Abstract.) 
In these experiments the general method pursued was by direct 
contact and separation of two circular plane metal disks, the lower 
one of which was insulated and connected to one pair of quadrants 
of a Thomson quadrant electrometer. The upper disk or plate of 
this condenser arrangement pressed during contact on the lower by 
its own weight, and was in connection with the other pair of 
quadrants and with the earth. The lower plate formed the upper 
surface of a cylindrical flask, whose temperature was determined by 
