It has long been known to experimentalists that, in thermometers constructed with the greatest care, a change takes place after a lapse of time in the standard points, as given by the melting of ice and boiling of water under a fixed pressure; on this account it has been recommended by most writers, where the employment of thermometers is treated of, that they should from time to time be compared one with another, and also at the freezing point. This change is a rising of the mercury in the tube, so that, after a length of time, the mercury will not sink to the point laid off in the construction of the instrument. To investigate to what cause this change was due, formed the object of my experiments: Was it a change in the glass of which the bulbs are formed, or in the mercury with which they are filled? I was aware that thermometers filled with alcohol were not subject to this change, which would lead to the inference, that the change was in the mercury and not the glass; but then, in the spirit-thermometer, air is left above the column of spirit, whereas, in those constructed with mercury, the air is expelled, and there is a vacuum above the column; consequently, the bulb is pressed together with the force of an atmosphere on all sides; might not this force, acting for a length of time, cause some small alteration in the arrangement of the particles forming the glass of the bulb? This is the explanation accepted by most of the Italian and French writers on this subject. Some suppose that the mercury may contain air and moisture within its particles; but such a hypothesis I think inadmissible, as in the case of a vacuum over the mercury, these particles would seek the void, and cause rather a depression than a rising of the freezing point. Mr Daniell, in his Essay on Climate, adopts the same view; and Sir John Herschel, in his article "Heat," in the Encyclopædia Metropolitana, says: "The freezing point upon the mercurial thermometer has been supposed to undergo some slight variation, so as to appear too low upon the scales of those instruments which have been long made; and it is said that, in such cases, the just indication was again recovered by breaking off the end of the stem, so as to admit atmospheric air." But, as I had observed that the change went on for a time only, after which it ceased, and that it affected thermometers sealed with air over the mercury, as well as those with a vacuum, I undertook the following experiments:-- In September 1848 I made four thermometers having long degrees,--such that 1/10° might be easily noted, constructed of the same draft of glass tube; two of these I placed in boiling water, and kept them at that temperature for a week: my object in this was to learn if any change in the form of the bulb would take place from this slow process of annealing, as glass is known to undergo some change from such exposure.