54 



THERMOMETER AND PYROMETER. 



ratio of expansion must affect thermo- 

 meters and pyrometers, graduated on 

 the principle of equal degrees between 

 two fixed points, is obvious ; and it 

 early became an object of solicitude. 

 Drs. Halley and Brooke Taylor devised 

 the method of investigation namely, 

 by mixing together equal weights of 

 water at different ascertained tempera- 

 tures, and finding how much the ther- 

 mometer, plunged in that mixture, dif- 

 fered from the mean temperature. 

 From not sufficiently attending to the 

 various requisites to obviate error, 

 they did not arrive at the true conclu- 

 sion. We owe to De Luc the more 

 successful investigation of this problem; 

 by a train of nice experiments, in which 

 he endeavoured to guard against the 

 sources of error arising from the cool- 

 ing effect of the vessel containing the 

 mixture, and of the escape of vapour, 

 he proved that the different therm o- 

 metric fluids do not expand in a uni- 

 form ratio to the quantities of caloric 

 applied; but follow an increasing rate 

 as thdr temperature is raised. Mer- 

 cury he found to be the most regular 

 in its expansions ; yet it also showed 

 very sensible deviations. "When equal 

 weights of water at 32 and 212 were 

 mixed, the mercurial thermometer did 

 not indicate the mean temperature 122, 

 but only 119; an oil thermometer, 

 in the same experiment, stood no higher 

 than 117, and one of spirit, of wine at 

 108; while, with a thermometer filled 

 with water, the temperature of the mix- 

 ture appeared only to be 75. His ex- 

 periments showed the great superiority 

 of the mercurial over the alcoholic ther- 

 mometer ; but this superiority, it pro- 

 bably owes, as Mr. John Dalton has 

 remarked, in a great measure to the dis- 

 tance of the ordinary range of tempera- 

 ture, from the freezing and boiling point 

 of mercury; for the experiments of 

 De Luc show that the irregularities of 

 all fluids are much augmented about 

 the points of their consolidation and 

 passing into vapour. 



From the usual method of graduating 

 the mercurial thermometer from two 

 fixed points only, the error from inequa- 

 lity of expansion will be greatest at the 

 mean between the two points; when, ac- 

 cording to De Luc, it amounts to about 3 

 Fahrenheit below the real temperature. 

 In the following table are given the 

 result of De Luc's researches, on the 

 two fluids chiefly used for thermometers 

 mercury and alcohol. In the first 



column are the indications of the mer- 

 curial thermometer, according to De 

 Luc's or Reaumur's scale; in the 

 second, the indications of the alcoholic 

 thermometer at the corresponding tem- 

 peratures ; and in the third, are, what 

 ought to be, the real temperatures as 

 discovered by experiment. 



Mercurial Ther. Alcohol Ther. Real Tern. 



There is reason to believe, however, 

 that De Luc states the irregularity of 

 the mercurial thermometer too high. 

 Dr. Crawford investigated this point 

 with great care, and concluded, that 

 when the difference of temperature of 

 the two portions of fluid did not exceed 

 100 F., the average deviations of the 

 mercurial thermometer were not above 

 0.25 of a degree. De Luc himself 

 allows, that the result of the mixtures 

 must be inaccurate, if the capacity of 

 the water operated on is changed during 

 the experiment. In mixing together 

 hot and cold water, the probability is, 

 that the diminished volume of the mix- 

 ture causes a diminution of capacity ; 

 and, consequently, an increase of tem- 

 perature, beyond what is due to the 

 heat of the two portions mixed together. 

 By exposing a mercurial thermometer 

 in a vessel, in which the included air 

 was exposed to the frigorific influence 

 of melting snow, and the heat of watery 

 vapour at 212, he found that it indi- 

 cated 121, or only a single degree less 

 than the arithmetical mean.* From 

 comparison with air thermometers, Gay 

 Lussac inferred that the mercurial ther- 

 mometer was equable in its expansions, 



* Experiments on Animal Heat. 



