THE MEASURES AND THE NATURE OF HEAT. 497 



nearly in the inverse proportion of the square of its distance from the 

 focus ; and the elevation of a common thermometer appears to be nearly 

 proportional to the heat which is thrown on it in this manner. 



The expansion of solids probably approaches the nearest to the steps of 

 the natural scale, although even in this there seems to be some inequality ; 

 but that of mercury is scarcely less regular, and a portion of mercury 

 inclosed in a bulb of glass, having a fine tube connected with it, forms a 

 thermometer the most convenient, and most probably the most accurate, of 

 any, for common use ; the degrees corresponding very nearly with those 

 of the natural scale, although, according to the most accurate experiments, 

 they appear to indicate, towards the middle of the common scale of 

 Fahrenheit, a temperature 2 or 3 degrees too low. There is an inequality 

 of the same kind, but still greater, in the degrees of the spirit thermometer ; 

 and this instrument has also the disadvantage of being liable to burst in a 

 heat below that of boiling water ; although it is well -calculated for the mea- 

 surement of very low temperatures, since pure alcohol has never yet been 

 frozen, while mercury has been reduced to a solid by the cold of Siberia 

 and of Hudson's Bay ; but both mercury and linseed oil support a heat of 

 between 5 and 600 without ebullition. For higher temperatures than 

 this, a thermometer has been made of semitransparent porcelain, containing 

 a fusible metal, which may be compared with the upper part of the mercu- 

 rial scale, and then continued further ; and the expansion of such of the 

 metals, as are difficult of fusion, affords another mode of determining the 

 highest degrees of heat. Mr. Wedgwood's thermometer* derives its proper- 

 ties from the contraction of a small brick of prepared clay, which contracts 

 the more, as the heat to which it is exposed is higher : it may be extremely 

 useful for identifying the degree of heat which is required for a particular 

 purpose : but for the comparison of temperatures by an extension of the 

 numerical scale, we have not sufficient evidence of its accuracy to allow us 

 to depend on its indications ; and it is scarcely credible that the operation 

 of furnaces, of any kind, can produce a heat of so many thousand degrees 

 of a natural scale, as Mr. Wedgwood's experiments have led him to sup- 

 pose ; nor is the supposition consistent with the observations of other 

 philosophers. 



Mercurial thermometers are in general hermetically sealed, the tube 

 being perfectly closed at the end, in order to exclude dust, and to prevent 

 the dissipation of the mercury. When a standard thermometer is to be 

 adjusted, its freezing point is readily fixed by immersing it wholly in 

 melting snow or pounded ice ; but for the boiling point, some further pre- 

 cautions are required ; the easiest method appears to be, to immerse its bulb 

 in an open vessel of boiling water, to cover it with several folds of cloth, 

 and to pour hot water continually over it ; for if it were immersed to a 

 considerable depth, the pressure would raise the temperature of the boiling 

 point, and if it were not covered, the mercury in the tube would be too 

 cold. Attention must also be paid to the state of the barometer ; it must 

 either stand at 80 inches, or the place of the boiling point must be raised, 

 when the barometer is lower than 30, and lowered when it is higher ; the 

 * Ph. Tr. 1782, p. 305 ; 1784, p. 358 ; 1786, p. 390. 

 2 K 



