222 C. Barus — Effect of Pressure on Conductivity, etc. 



portional to pressure ; and by deduction, that the immediate 

 electrical effect of rise of temperature, 3R' '/ R — dR/R, is a 

 decrement of specific resistance both in the case of the metal 

 (Hg) and of the electrolyte (ZnS0 4 +Aq.). This points out an 

 inherent similarity between the metallic and the electrolytic 

 conduction, in this instance. 



5. In J. J. Thomson's expression for specific resistance 

 R=<?=(2tc fi/K) (q/mx). suppose to fix the ideas, that /3, K 

 and q are constant ; whereas nl, the number of molecules split- 

 ting up per unit of volume per unit of time, and x the distance 

 passed over by the partial molecule moving at a mean velocity 

 c during the interval of freedom t, are regarded variable. 

 Clearly x can not be independent of in. Taking active mole- 

 cules alone into consideration, supposing them to be symmetri- 

 cally distributed and to move parallel to each other, x= Vl/mt. 

 It follows that R=(2tt fiq/K) x 3 /c. This is in accord with the 

 above data. Reduction of volume, —dv/v, isothermally by 

 pressure, diminishes x only. Reduction of volume isopiesti- 

 cally by cooling, diminishes both x and c. Hence the greater 

 diminution of R in the former instance (pressure). Finally, 

 by partial differentiation under the given conditions (dR/dm) 

 = — (In ftq/3 K) Vt/m*. From this it may be conjectured 

 (conjectured because t and m are not independent of each 

 other), that the effect on R of an additional number of mole- 

 cules splitting up, decreases rapidly with the total number, m, 

 splitting up ; i. e. that the numeric of the immediate electrical 

 effect of temperature, SR'/R' — dR/R, is smaller for the metal 

 than for the electrolyte. This also is in accord with the above 

 data. 



6. For solids I have only found available data in the case of 

 copper. According to Chwolson,* — 3R/R =1-3x10- 6 dP. 

 From Everett's tables -dv/v=-6xl0~ 6 dP. Hence dR/R= 

 2dv/v. On the other hand dfi'/JR'=-Q0± d0, and dv/v= 

 52X10' 6 o<9, whence dR'/R'=W dv/v. Hence dR/R-dR'/R' 

 is negative in case of the solid metal. Comparing with §5 it 

 appears that dR/dm probably passes through zero into a nega- 

 tive region, in proportion as the number of paths which the 

 current can take is indefinitely increased. 



*Chwolson: Carl's Rep., xiv, p. 26, 1878. In case of mercury, the only- 

 kindred results I found are due to Lenz (Stuttgart, 1882). But they are unfortu- 

 nately inaccessible. 



