1821.] Phanojneria of Heat, Gases, Gravitation, <S^T. 405 



this neighbourhood who has; nor is there at present any 

 Society in Bristol to which I could apply for assistance, though 

 from the spirited exertions now making, there is reason to 

 believe, we shall soon possess one that will be every thing we 

 could wish. 



For making an experiment ='^ of this kind, I know of no body 

 better than mercury, which is fluid at a very low, and boils at a 

 very high, temperature. If two large and equal portions of this 

 fluid, at the temperatures intended, were rapidly poured into a 

 thin smooth vessel, previously brought to the expected mean 

 temperature of the mixture, or a degree or two higher; and if 

 the temperature of the mixture was taken by two thermometers, 

 the one two or three degrees lower, and the other about as many 

 degrees higher than the computed mean, I think the obstacles 

 to success, pointed out by M. Biot, might be avoided; and this 

 great problem, with one or two repetitions of the experiment, 

 foe resolved to a considerable degree of accuracy. Could the 

 experiment be made in an atmosphere just above the tempera- 

 ture of the mixture, it would prevent any error arising from 

 a^adiation, &c. ; and if the vessel into which the portions of 

 mercury are poured be but little larger than sufticient to hold 

 them, and spherical, it would subject the experiment to the least 

 influence of the matter of the vessel. But if the experiment can- 

 not be made in such an atmosphere, it would be advisable to 

 cover the whole apparatus for a few minutes, until the tempera- 

 ture could be equally difl'used, with a polished metal cover, 

 which would tend to prevent any loss of temperature by the con- 

 tact of air, or by radiation. These and some other precautions, 

 which circumstances w^ould easily suggest, would, it appears to 

 me, ensure to the experiment all the success of which, perhaps, 

 it is capable. 



When the difference between the temperatures of the portions 

 of fluid mixed exceeds 180° of Fahrenheit, the arithmetical and 

 computed means differ from each other by a quantity which 

 increases much faster than the difference of the temperatures. 

 If equal portions be mixed at the temperatures of — 33° and 

 4- 300°, the value of F,, will come out 121-32°, while the arith- 

 metical mean is 133*5° ; that is, upwards of 12° in excess. 

 Within these limits, mercury and air go on neo-rly pari passu, the 

 volume of mercury at the higher temperature being only about 

 "2*34° greater than that of air. But our theory seems to be inde- 

 pendent of all considerations of proportional variations of volume; 

 the only precaution necessary appears to be to measure the 

 temperatures by the air thermometer instead of the common 

 mercurial. If mercury at — 7° and + 641° be mixed, the com- 

 puted mean temperature will be no less than 36° below the 



« I have since succeeded in some experiments of this kind in high temperatures with 

 mercury, and have found the mean error of my theory by seven experiments of my own, 

 and two of De Luc's, not more than about the one-tenth of ^ degree ; the mean firror of 

 the old theory being eight deguees. '"'O tJiUiJfl ibid isi^ munm i/ii^ ^ ^..u^ii 



