Notices respecting New Books, 59 



Dulong and Petit had already made this comparison, and calculated 

 a table which permitted the necessary transformation ; but this table 

 is inexact, even for the particular mercurial thermometer which 

 they employed, because their experiments were calculated with much 

 too high a coefficient of dilatation — the coefficient 0'375 of Gay- 

 Lussac. 



M. Regnault, after he had made a great number of experiments 

 on this subject, perceived that the different mercurial thermometers 

 were not comparable, either because they had not been constructed 

 of the same kind of glass, or because they had been blown in a dif- 

 ferent manner. At the same time, he was desirous of ascertaining 

 if mercurial thermometers, constructed of the same kind of glass, 

 would, although blown in a different manner, proceed sufficiently in 

 agreement to allow of their being regarded as comparable. If this 

 circumstance were realized, it would suffice to make, once for all, 

 the comparison of one of these thermometers with the air-thermo- 

 meter, and to admit the same table of correction for all similar in- 

 struments. To settle this question, M. Regnault executed a long 

 series of experiments with the aim of comparing with the air-ther- 

 mometer, not only mercurial thermometers formed of an identical 

 quality of glass worked in a diffei'ent way, but also those made of 

 the different kinds of glass which are met with commercially in 

 France, and which are employed in physical apparatus. The mer- 

 curial thermometers employed in these experiments were overflow- 

 thermometers {thermometres a deversement) ; they are more easily 

 constructed than the ordinary thermometers with graduated stem, 

 and present the great advantage, that it is always easy to keep the 

 whole of the column of mercury in the bath. In order to construct 

 overflow-thermometers, the mercury must be boiled in the instru- 

 ments several successive times and with great care, after which they 

 are left to cool with the recurved point of the capillary tubes kept 

 in a bath of mercury previously heated. Then the reservoirs and 

 capillary tubes are enveloped in melting ice, the open point remain- 

 ing plunged in the basin of mercury, and it is easily seen that the 

 thermometer has acquired the temperature of 0° when the column 

 of mercury remains stationary at the extremity of the capillary tube. 

 The mercury which escapes in consequence of the elevation of tem- 

 perature is collected in a little empty capsule, and weighed with 

 great care. To obtain the weight of the mercury which filled the 

 thermometer at 0°, the weight of the mercury which has escaped 

 through the elevation of the temperature above 0° is added to that 

 of the weight of the thermometer itself, subtracting from the whole 

 the weight of the empty apparatus. Then, knowing the weight of 

 the mercury at 0°, and weighing with care the quantity which 

 escapes in proportion as the temperature is elevated, it is easy to 

 conclude from it the temperature itself, and to compare it to that 

 which would be given by a thermometer constructed of the same 

 glass but with a graduated stem. M. Regnault successively analysed 

 with care the indications of the thermometer with graduated stem 

 and the overflow-thermometer ; he shows that by taking into ac- 



