100 Mr. Herapath on True Temperature, and the [Aug. 



which corresponds to 120-2° Fahrenheit. This experiment, 

 therefore, differs from the theory -^th of a degree of Fahrenheit. 

 If Fahrenheit's thermometer really indicated the true increments 

 of tempefature, the result should have been the arithmetical 

 mean of 200-7 c and 45°, or 122-8°, which is 2-5° more than it 

 was observed. 



In another experiment, De Luc mixed equal weights of water 

 at 212° and 32°, and found the result 119°. 



212° of Fahr. in the tables, give T = 1172-6 

 And 32° giveT, = 1000 



2)2172-6 



And consequently r = 1086*3 



which makes the temperature of the mixture in Fahrenheit = 

 118-4°. This differs from the experiment -6 in defect, while the 

 •experiment is 3° below the arithmetical mean of 212° and 32°. 



Having contrived an apparatus for making the mixture so as 

 not to lose by radiation, or the unequal temperature of the mat- 

 ter of the vessel, 1 mixed about 30 ounces of mercury at 348-9° 

 with an equal weight at 31°, and found the mixture at 179-3° of 

 Fahrenheit. 



Here T = 1288-5 



And T, = 999-0 



2)2287-5 



Whence t = 1143-75 



which agrees with 179-9 of Fahrenheit. 



This experiment is -6° below the theory, and 10*7° below the 

 arithmetical mean. The other experiments that I made gave a 

 like coincidence with theory ; but as they have lately been read 

 before the Royal Society, it might be deemed improper at pre- 

 sent to quote any more of them here. 



Were the temperatures 3872° and 32° of Fahrenheit, 



Then 3872 would give 3000 



And 32 1000 



2)7o"oo 



2000 



which give 1472° for the mean temperature according to Fah- 

 renheit. The arithmetical mean of 3872 and 32 is 1952 ; that is, 

 480° above the computation. Wherefore could we make expe- 

 riments on these extreme ranges, our thermometers would differ 

 no less than 480°, on the supposition of the common theoretical 



