84 Hough — Mechanical Equivalent of the 



ticularly with the air thermometer which does not admit of 

 very close reading." In fact, so uncertain are the corrections 

 for radiation and conduction, that Griffith* asserts as a the 

 general principle on which he proposed to work, that of elimi- 

 nating the effects of radiation, conduction, etc'., rather than 

 that of ascertaining the actual loss or gain due to such causes." 

 He eliminates these effects by maintaining the walls of the 

 chamber enclosing his calorimeter at a constant temperature 

 and gradually raising the temperature of the calorimeter from 

 some point below to some point above that of the jacket, such 

 that the gain and loss by the calorimeter are equal. This he 

 calls the null point and determines it experimentally. The 

 correction for convection by this method is doubtful. Row- 

 land, who also bunches the losses due to radiation, convection 

 and conduction, estimates the loss by convection to be more 

 than 75 per cent of the total losses from these causes. He 

 likewise corrects empirically. Obviously a better plan would 

 be to eliminate not only the effects but the cause of these 

 errors by maintaining the calorimeter, the jacket and the inter- 

 vening medium at the same constant temperature, if a method 

 admitting of such a process is possible. In fact, the principle 

 of elimination of source of error is fundamental to all physical 

 measurements since minimization and correction formulae can 

 never be more than a series of successive approximations. 



The grounds then for a new method of determining L 

 are : (a) the absence of any authoritative determination ; (b) 

 the absence of any absolute method ; (c) the inherent sources 

 of error in the present indirect methods. These are suffi- 

 cient but there are weightier considerations : i. e. the advan- 

 tages resulting from the use of L as the primary heat unit. 

 Much can be said in favor of L instead of C as the primary 

 heat unit, especially since the development of steam calori- 

 metry by Bunsen and Joly. 



The substance under calorimetric observation may be in a 

 thermo-dynamic or in a thermo-static condition. The tempera- 

 ture may be changing, or it may be constant. In the first case 

 the thermometer should be accurate, delicate and sensitive. 

 That is to say, that not only should all corrections to reduce its 

 readings to the air scale be definitely known, but it should 

 respond to small variations of temperature in a readable degree 

 and respond quickly. In this case the readings must be taken 

 rapidly and are necessarily limited in number. In the second 

 case the thermometer should be accurate and delicate but not 

 necessarily sensitive to a higher degree. The readings may be 

 taken more leisurely, with greater precision, and are only 



* Griffith, Phil. Trans., 1893. 



