the cylinder. Six of these instruments, with proper stands, and
auxiliary implements of obvious construction, were prepared for the
sake of comparative experiments.
A previous trial was made with two of the cylinders, the vertical polished sides of the one being naked, and those of the other covered with one thickness of line white Irish linen, strained over the metallic surface. Here it was found, contrary to expectation, that in a cer- tain space of time the covered cylinder had lost considerably more heat than the naked one.
In reflecting on this experiment it occurred to the author, that in order to insure the accuracy of the comparison between experiments made at different times and at dilferent places, it would be necessary to fix on some particular interval of the scale of the thermometer above the temperature of the air by which the instrument is sur- rounded. He therefore determined that all experiments should begin at the temperature of 50°, and end at 40° above that of the sur- rounding atmosphere, an interval of 10° appearing to him sufficient for the purpose of his investigation. Finding also that most experi- ments would take up several hours, dm-ing which he could often not be present to observe the thermometer at the different points which ought to be ascertained, and observing that the rate of cooling of hot bodies afforded a pretty regular progression, he determined to in- vestigate this rate more minutely, with a view to obtain the means of introducing such interpolations as would complete the serics of observations. Accordingly, on a given line, on which were set ofl" the times of cooling, he applied ordinates representing the different temperatures corresponding to those times; and having joined the opposite terminations of these ordinates, he had the satisfaction to find that this latter connecting line was in fact the logarithmic curve, by means of which he would be enabled to supply by computation any intermediate points which happened to have been neglected during the observation. The problem according to which these in— terpolations are to be computed, is given at full length.
These previous precautions and expedients having been fully stated, the author proceeds next to the enumeration of his long series of experiments, the first of which is merely the comparison, which has already been mentioned above, between the naked and the covered cylinders. The result was, that the former was 55' in cooling 10°, while the latter cooled through the same interval in 36½’; whence it appears that clothing does in some instances expedite the passage of heat out of a hot body instead of confining it. The only mode in which it is thought that this unexpected result can be accounted for, is by admitting that, as air is known to adhere with considerable obstinacy to the surfaces of certain solid bodies, the particles of air which were in immediate contact with the surface of the naked cylinder were so attached to the metal as to adhere to it with considerable force; and as confined air is known to be a very warm covering, it seems probable that the retardation of the cooling in this vessel was owing to that invisible covering, the air in contact with
