the other vessel being absorbed, displaced, or in a great degree driven
away by the colder covering of linen which closely embraced it.
Led on by this conjecture, several experiments were made with
cylinders covered with one, two, or more coatings of glue, and of
copal varnish; and the results, in fact, turned out favourable to the
supposition, tlhe cylinder with one coat of glue losing 10° of its heat
in 43', and that with two coatings in about 38'. With the copal
also the cooling of the instrument became more and more rapid as
the thickness of the varnish was increased; till, however, eight suc-
cessive coatings having been applied, the cooling ngain became less
rapid, and it was found that there was a maximum of thickness which
produced the greatest effect. No probable reason is yet assigned for
this limitation.
The next object was to find out what effect colour would produce in the experiments, and accordingly the cylinder with eight coatings of varnish was painted black; it was also painted in the same manner after all the varnish had been washed away; and lastly, it was like- wise painted white: in each of these instances the cooling was ac- celerated by the paint, nearly in the same proportion as in the pre- ceding experiments.
A nicety occurred now in the conducting of the experiments, which was thought to deserve particular attention : though the apparatus for confining che heat at the two flat ends of the cylinder was the best that could be contrived, yet it is not at all unlikely that some would escape in those directions, and thus occasion some fallacy in the results.
In order to investigate this point, a given number was previously assumed as the measure of the whole quantity of heat emitted by the whole instrument, without terminal coverings, during a certain period. The surface of the whole of the cylinder was then accurately measured, and also that of its vertical sides; and thence was inferred the proportion of heat that passed off through the sides of the instru- ment, and what proportion must have escaped through its uncovered ends. With these data it is easy to infer, from an experiment with the ends covered, what proportion of the heat, lost in the cooling, had escaped through the flat terminal surfaces when covered. In this manner it has been ascertained that, assuming the total of the loss of heat emitted by a cylinder, in a given time, for instance 55 ½', to be 10,000, the quantity that escapes through the vertical sides will be 7,015, and that which penetrates through the terminal sides and coverings =2,985
Admitting these computations, it will now appear how an estimate may be made, what proportion of the heat lost in any otlher experi- ment, actually escaped through tlhe vertical sides of the instrument: and as the quantity of the heat emitted may well be represented by tlhe time of the emission, there can be no difficulty in substituting the velocity for the quantity; whence it is inferred, that in the ex- périment, for instance, when the sides of one of the cylinders were blackened, the velocity with which heat is given off from the naked
