It is of course essential that the extent of the thermometer-scale should be great enough to comprehend all the temperatures at which the substances generally required to be examined exist in a state of fluidity ; and this extent may be obtained when mercury is employed. According to the experiments of Mr. Dalton, mercury does not boil till it has acquired a temperature equal to 660' of Fahreulicit's scale ; and it does not freeze till it is subject to a degree of cold expressed by 39 divisions below the zero of that scale, or below the freez ing-point of water. Pure alcohol, on the other hand, has never been frozen, though it has been exposed to a degree of cold exceeding that which i4 expressed by below the zero of Fahrenheit; though at that remarkably low temperature it became viscid. Hence a spirit-thermometer is to be preferred to one of mercury when it is intended to ascertain the temperature of the air in high northern or southern latitudes : but since the spirit boils in air with a degree of heat expressed by of Fahrenheit, it is unfit for many of the purposes for which a thermometer is required. For instruments capable of measuring very high temperatures, see PYROMETER.
In the construction of a thermometer, the air should be carefully expelled from the tube, and even from the mercury or spirit within it : the variations in the density of the atmosphere cannot, under proper precaution, affect the instrument, since the tube is hermetically sealed. This precaution consists in not graduating the tube for some months after it has been sealed, or until the glass has accommodated itself to the altered circumstances of a vacuum within and atmospheric pressure without. The most carefully constructed thermometers will shift their fixed points if graduated too soon after filling. It must also be observed that the indications of temperature are not precisely expressed in terms of the dilatation of the mercury or spirit only, but in terms of the excess of that dilatation above the dilatation of glass. The apparent dilatation of mercury in a glass tube is equal to of its volume, between the temperatures of freezing and boiling water ; and its true dilatation between the name limits is of its volume.
A perfect thermometer would be one in which the expansions of the fluid in the tube were exactly proportional to the increments of heat which it might receive from tho substance whose temperature is to be determined ; but it cannot be said that any of the fluids which as yet have been employed in the construction of thermometers strictly pos sess this property. Mercury is the fluid in which it exists in the greatest degree ; but from the experiments of Deluc it has been ascer tained that, between the points of freezing said boiling-water,.the tem perature indicated by the mercurial thermometer is lower than the true temperature, the greatest difference (which, however, is only equal to of 116aurnur's scale, or 3'15 Fahrenheit), being in the middle between those two points on the scale. From the same experiments it is also found that when thermometers are regulated so as to agree at the points of freezing and boiling-water, whether the liquid be oil, spirit, or water, the indications are always below those of mercury, the difference being the greatest at the middle between those points. With oil of olives the difference is of Waurnur's scale Fahr.) ; with highly-rectified alcohol, Reaumur Fehr.); with half alcohol and half water, VI 114aumur (1547 Fahr.); and with water, lidaumur (4347 Fahr.). It must be observed that great irregu larities take place in the expansion of all fluids when near their boiling state, and that mercury contracts very suddenly when at the point of its congelation. The deviations of the spirit-thermometer from the true indications of heat are kuown to be rather greater than those of the mercurial thermometer. It may be added, that the alcohol in a thermometer-tube loses, in time, part of its strength; and that, in con sequence, the degree of expansion by a given increment of heat is not the wane an when the instrument was made. The expansion of alcohol for temperatures greater than about 173' Fehr., at which the spirit boils, cannot be ascertained practically, because the spirit at that temperature passes into a state of vapour ; and the comparison between the mercurial and the spirit thermometer ought not to be carried higher than that temperature ; or the scales for mercury and spirit ought to be regulated so as to agree with one another at the freezing point of water and at the temperature of 173' Fehr. ; if this were attended to, the differences between the indications of the mercurial and spirit thermometers, above that point, would be less than they appear to be by the tables of Deluc.
Later measurements by Ilegnault of the total expansion of mercury for three progreosive intervals of Fehr., give the following results: Between 32' and it is I part in ; between and it is I in 54*6I ; and between 392' and 572' it Is I In 54'01. In the mercurial thermometer it may be assumed, without sensible error, that between and 212' equal inorements of heat raise the thermometer through an equal number of degrees. The increase In the capacity of the glass bulb (especially for crown glass) almost exactly compensates for the increasing rate of the expansion of mercury, although for temperatures above 212' the compensation is not so exact. It has been found, also, that the temperature of Fehr., as measured by an air thermometer, is by a mercurial thermometer, on account of the increasing dilatation of mercury, with an increased temperature. Mr. Dixon, in his 'Treatise ou Heat,' says : " Different glasses have different co-eflieients of expansion, and also vary in the law of their dilatation at high temperatures, and as the amount of absolute dilata tion of mercury is small, this variation in the expansion of the glass envelope produces irregularities of considerable magnitude in the apparent dilatation of mercury. As the real expansibility of air is much greater, its appareut expansion in glass is not affected to the same extent by these variations in the rate of expansion of the latter, and accordingly in an air-thermometer the rate of expansion of the glass may be considered as sensibly uniform. When corrected, there fore, for the expansion of its envelope, such an instrument forms the most perfect thermometer with which we are acquainted in the preseut state of ecience." An air-thermometer, corrected for the expansion of its envelope, being compared with a mercurial thermo meter made with the peculiar description of glass employed by M. Ilegnault, the agreement between the two instruments was perfect up to C.; whereas, in a mercurial thermometer made of ordi nary tube, compared with one of crystal glass, although they agreed from to C., yet at higher temperatures the discrepancies were : Water, like other substances, suffers a diminution of volume by the abstraction of heat, but when it is cooled to a temperature between and of Fahrenheit's scale, it seems to have attained the maximum of density; and if the process of cooling be continued, it then increases in volume till it is converted into ice. Therefore, if a thermometer were made with water, and the top of the column were at Fahr., it would be impossible to know whether the temperature were or the expansion being nearly equal at equal distances within above and below of the scale. The cause is uncertain, but it is probably owing to a partial crystallisation, which may begin to take place in water when at a temperature expressed by about above its freezing-point.
The mercurial and spirit thermometers formerly differed considerably at temperatures below that of freezing water. By observations made during Sir Edward Parry's second voyage, the differences between the indications of the spirit and mercurial thermometers varied from to Fahr. between the temperatures + and the alcohol being always too low. At very low temperatures alcohol thermometers, unless very carefully prepared, differ greatly among themselves. Dr. Kano in his Arctic voyage records temperatures as low as from to but he admits that "it was not uncommon for ther mometers which had given us correct and agreeing temperatures as low as 40' to show at differences of from fifteen to twenty degrees." Iu remarkable contrast with this are the results obtained by Sir E. Belcher iu his Arctic voyage with the thermometers furnished to him by Mr. Welsh of the Kew Observatory. While wintering in Northumberland Sound, N., W., be made a comparison of seven thermometers, marked 2, 6, 8, 20, 8, 4, 5, with a standard instru ment, with the following results. The temperatures were all natural ones: The lowest temperature observed was on 12th January, 1853, namely 62%. The indices of the minimum thermometers read next day at 8 a.m., gave " The night bright and calm; no sensation of cold. is the external exposed thermo meter, but was never read excepting by its index of that graduation." Parry's coldest waa for 151 hours. In Belcher's Expedition, the following temperatures were observed : and below, far 166 continuous hours.
30" Of II 89 PP // --53* Of 1, 88 ,, If