I now come to the concluding experiment with these instruments, and, it appears to me most interesting and anomalous. The four tubes being placed in pounded ice, the columns stood at the points indicated in the last column of the Table; in this situation the tops of the tubes were broken off, so as to admit the free pressure of the air, and instantly the thermometers fell, in the order of their numbers, ·54, ·43, ·40, ·35 of a degree, now indicating on their scales +·30, +·40, +·50, +·35. The remarkable features shewn by this experiment are; first, that the two thermometers sealed with vacuum, and the two having air over their columns, should have risen nearly equally, when two had their bulbs pressed with the whole force of an atmosphere, while the other two had no pressure externally, farther than that caused from changes in the pressure of the atmosphere. Next, that on being opened, those with air over them should have started down nearly as much as those with a vacuum; and on all these appears a permanent change from three to five-tenths of a degree. I confess that I am very much at a loss to account for these singular changes; atmospheric pressure on the bulbs would account for the change in those sealed with a vacuum; for we can easily suppose that a permanent form had been taken from long exposure to that pressure by the glass forming the bulbs: besides this permanent form, there appears to have been a spring inwards, which instantly sprung out on removal of the pressure by the admission of air over the mercury; but the same reasoning will not apply to the thermometers having air over the mercury; and before I attempt to make any suggestions as to the cause of these changes, I propose to institute the following experiments. Having had three thermometers blown and filled with mercury, I shall make one with a perfect vacuum over the mercury, the next with air over it, and the third with air condensed over it; and, noting the changes that may go on in these, I hope to be able to assign a cause or causes for the change. It is argued by some continental writers on this subject that the reason why we do not perceive any change in the freezing point in spirit-thermometers is from the great expansion of spirit above mercury, volume for volume, thereby requiring a much smaller mass of fluid to give the same length of a degree: this I propose to test by making a thermometer with the same size of tube and bulb as those to be experimented on with mercury. In mentioning these experiments to Professor Forbes, he kindly put me in possession of some spirit-thermometers, one of these, made in 1837, having a very large bulb--this, with three others, shewed no change in the places of their freezing points.