ON THE MECHANICAL EQUIVALENT OF HEAT 465 



V. CONCLUDING REMARKS, AND CRITICISM OF RESULTS AND 



METHODS 



On looking over the last four columns of Table LIII, which gives 

 the results of the experiments as expressed in terms of the different 

 mercurial thermometers, we cannot but be impressed with the unsatis- 

 factory state of the science of thermometry at the present day, when 

 nearly all physicists accept the mercurial thermometer as the standard 

 between and 100. The wide discrepancy in the results of calori- 

 metric experiments requires no further explanation, especially when 

 physicists have taken no precaution with respect to the change of zero 

 after the heating of the thermometer. They show that thermometry 

 is an immensely difficult subject, and that the results of all physicists 

 who have not made a special study of their thermometers, and a com- 

 parison with the air thermometer, must be greatly in error, and should 

 be rejected in many cases. And this is specially the case where Geissler 

 thermometers have been used. 



The comparison of my own thermometers with the air thermometer is 

 undoubtedly by far the best so far made, and I have no improvements to 

 offer beyond those I have already mentioned in the ' Appendix to Ther- 

 mometry/ And I now believe that, with the improvement to the air 

 thermometer of an artificial atmosphere of constant pressure, we could 

 be reasonably certain of obtaining the temperature at any point up to 

 50 C. within 0-01 C. from the mean of two or three observations. 

 I believe that my own thermometers scarcely differ much more than 

 that from the absolute scale at any point up to 40 C., but they represent 

 the mean of eight observations. However, there is an uncertainty of 

 0-01 C. at the 20 point, owing to the uncertainty of the value of m. 

 But taking m= -00015, I hardly think that the point is uncertain to 

 more than that amount for the thermometers Nos. 6163, 6165, and 6166. 



As to the comparison of the other thermometers, it is evidently un- 

 satisfactory, as they do not read accurately enough. However, the fig- 

 ures given in Table LIII are probably very nearly correct. 



The study of the thermometers from the different makers introduces 

 the question whether there are any thermometers which stand below the 

 air thermometer between and 100. As far as I can find, nobody has 

 ever published a table showing such a result, although Bosscha infers that 

 thermometers of " Cristal de Choisy-le-Eoi " should stand below, and 

 his inference has been accepted by Eegnault. But it does not seem 

 to have been proved by direct experiment. My Baudin thermometers 

 seem to contain lead as far as one can tell from the blackening in a gas 

 30 



