96 Proceediu(js of the Royal Societij of Victoria. 



been able to construct a model for this case, as I do not possess 

 a piezometer. 



For a gas, equation (3) becomes 

 d <l> = a . p dv. 

 whence (by the gaseous laws) we obtain 



Hence entropy change in this case can be measured on a 

 pressure gauge provided with a logarithmic scale, being propor- 

 tional to the difference of the napierian logarithms of the initial 

 and final pressures. 



Here p^ is one atmosphere, or 1-016 x 10'' dynes, and v,, is the 

 specific volume of the gas under consideration. The latter 



quantity for air is 773-4: cm., so t^^-^" = 2 -88 x 1 0'' 



The diff^erence of the napierian logarithms is independent of 

 the value of the unit of pressure. It is convenient, however, to 

 arrange the zero reading to correspond to a pressure of one 

 atmosphere. Hence, if we graduate our scale so that 2-88 

 xlogg/ — where/ is estimated in atmospheres — corresponds to 

 the division marked 1, the readings will give directly the entropy 

 changes for air, reckoned in millions of mechanical units ; to 

 obtain the same quantity for any other gas we have only to 

 divide the readings by the specific gravity of the gas referred 

 to air. 



The model shown (see also Fig. 2), is constructed in accordance 

 with these principles ; its graduations will measure entropy 

 changes corresponding to an isothermal variation of pressure 

 between ^^^th of an atmosphere and 2 atmospheres. It is simply 

 an old pressure gauge — recording between these limits — which 

 has been supplied with a new scale. If the results be required 

 in thermal units, all readings must be divided by 42 ; the 

 quotients give the entropies, measured from the arbitrary zero 

 of the scale, directly in thermal units. 



Had the molecular volume been taken for v„ instead of the 

 specific volume, the instrument would have been direct-reading 

 for all ffases. 



