of Edinburgh, Session 1881 - 82 . 
55 3 
linear,* we would require a temperature of about 1700°, which would 
be a bright red heat at least, before the piece would have sufficient 
increase of bulk to enable it to float. Now, from various experiments 
made by the author, it appears that many of the pieces which float 
show little or no redness when taken out, hence some other action 
seemed necessary to account for the buoyant tendency. Believing, 
however, that a large part of the heat might disappear as work of 
expansion, the author has carried out recently a large number of 
experiments with pieces of cast iron of various sizes and forms. 
These were carefully callipered before immersion, and as soon as 
possible after being taken out of the liquid metal in which they 
had floated (a few seconds in general was all that was necessary for 
their reappearance), and it was found in every case that a large 
expansion had taken place, such expansion being about j-Jo-th P ar ^ 
at least of linear measurement. Here, therefore, we have a sufficient 
change of bulk to account for the floating phenomena. And this is 
the more remarkable, as although the change in dimension is equal 
to the shrinkage in cooling from the molten to the cooled solid 
state, yet the temperature is very much below the melting-point, 
as in some cases the pieces were barely red, and would hardly melt 
lead. 
5. Some of the experiments may be enumerated : — 
(1) A piece of a test-bar 2 inches broad by 1 inch thick, and filed 
down exactly to 12 inches long, was placed in a ladle, floated, and was 
taken out within half a minute; it just showed redness, and it was 
now found to be fully of an inch longer than at first. It was 
callipered from time to time as it cooled, showing a decrease in 
length. Finally cooled in water, and now being just warm to the 
touch, was found to be nearly down to 12 inches ; when quite cold 
came back to the 12 inches. 
(2) A cutting from an 11-inch pipe was carefully callipered when 
cold, and found to be exactly 11 inches inside measurement, and out- 
side 12-g- inches ; it was placed in a ladle holding 3 tons, disappeared 
for about twelve seconds, came up to surface and was immediately 
lifted out, was again callipered and found to measure inside 11J 
inches, outside 1 3 inches, and this after it had become quite dark ; 
* Prof. Rankine, Manual of Civil Engineering. 
