30 



Mr. Carl Barus on the 



in glass, may be looked to as responding to strain of the kind 

 suggested. 



30. In the next six paragraphs it will be convenient to dis- 

 cuss some of the errors relating to both the measurements of 

 resistance and electromotive force. 



The effect of pressure on the polarization occurring in case 

 of the above cell and of exceedingly weak currents is some- 

 what difficult to handle. Warburg * and others have shown 

 that the reaction is essentially Na/glass/Hg changing to 

 Na/Na 2 Si0 3 glass/NaHg, or a mere transfer of sodium through 

 the walls of the glass, by aid of sodium silicate as a con- 

 veyancer. I shall therefore have to rely on the data of 

 Table II., in which, if a marked effect of pressure on polariza- 

 tion had been present, it would have been detected. It will 

 be remembered that I commenced with high initial pressures 

 (p =50 to 200 atm.) to obviate incidental disturbances. 



31. The effect of pressure on the oil in the piezometer was 

 specially investigated in each case. Cf. Tables III. to VIII., 

 § 38. 



32. A source of error is introduced by the temporary 

 shifting of the isothermal planes of the piezometer, during 

 compression, in virtue of the compressibility of the oil con- 

 tained in the tube. Adiabatic thermal changes during com- 

 pression or removal of pressure w r ould produce effects of a 

 sign opposite to those of the Tables. 



To remove both of these it was often my habit to wait 

 3 to 5 minutes, supposing the original thermal state was then 

 nearly enough regained. Longer waiting was not safe because 

 of fluctuations of the astatic system of the galvanometer. 



Fig. 6. 



Special tests were made as follows : — In fig. 6 let P Q be 

 the piezometer closed at Q, and receiving pressure in the 

 direction of the arrow. Let A B be the position of the 

 glass resistance-tube. Finally, let the hollow cylindrical 

 vapour-bath surrounding the piezometer stretch respectively 

 from b to d, or from a to c, or from c to e. Parts lying 

 within the vapour-bath have constant temperature, which 

 falls off rapidly from both ends of the bath towards the ends 

 of the piezometer. When pressure is applied the vertical 

 isothermal planes of the compressible oil are all moved from 



* Warburg, Wied. Ann. xxi. p. 622 (1884) ; xxxv. p. 455 (1888). 



