32 Mr. Carl Barus on the 



34. In case of an uncompressed circuit shunted by un- 

 avoidable leaks, let R b R 2 , R3 be the resistances of the cell, 

 the shunt, and the metallic circuit respectively. Similarly 

 let i u i 2 , i 3 be the three partial currents corresponding re- 

 spectively to R 1? R 2 , R3. Let SRi/Ri be the effect of pres- 

 sure on the resistance of the cell (resistancertube), and Se/e 

 be the simultaneous increment of electromotive force. Finally, 

 in case of the compressed circuit, let accentuated symbols 

 replace the symbols for the uncompressed cell. Then a some- 

 what tedious reduction leads to 



__ i n (l + 8,/e)-i a ' (1 + R 2 /R 1 )(l + R 3 /Ri)-1 f9 . 



6Kl/Kl -~ ~V R^-Cn-IJH,/^ ' ' (2) 



_ y (gR 1 /R 1 )(R 2 /R 1 -(n-l)R 3 /R 1 ) j 3 -j/ 



6e ' e ~ i z (l + ^/ROCl+Ks/JaO-l h ' * W 



where SR 1 /R 1 = nSR 2 /R 2 ,and SR 2 /R 2 is the pressure-coefficient 

 of the oil through which the leakage obviously takes place. 

 Regarding SRj/Ri in (2), if the resistance, R 3 , of the metallic 

 circuit is relatively small, or if R x be reasonably small relative 

 to the leak R 2 , the considerations of § 27 at once apply. The 

 case of Be/e 9 however, is serious. Supposing, again, that 

 R 3 /Ri and Ri/R 2 are small, the last equation reverts to 



-Ee/e= (*W/(l + SRiARi))A' 3 , 



and thus the changes of resistance due to pressure are inter- 

 preted as changes of electromotive force. Even, zero or 

 electrometric methods will only measure the difference of 

 potential of two points, of the leak-shunted circuit. Hence 

 the permanent pressure-coefficients (te/e) of Table II. cannot 

 but be regarded with suspicion, and they are merely an 

 essential justification of Table I. et seq. 



35. A final difficulty in case of short-circuiting is not to be 

 lost sight of. A leak usually includes an electrolyte between 

 two different metals, and thus introduces an electromotive 

 force in a way quite beyond computation. \ 



136.. The consideration of the data of Table II. would now 

 be in order : but for the reasons just stated I hold the work 

 scarcely ripe for discussion, and will therefore withhold re- 

 marks until I can devise some more nearly faultless method 

 of-measurement. — 



37. In order to exhibit the full character, of the individual 

 observations of Tables I. to VIII., I shall discuss a series of 

 data which I was first inclined to interpret as an electrical 

 manifestation . of volume " accommodation " (§ 17). In 

 figs. 7 to 9 the observed intensities of current in relative 



