196 C. Barus — Energy Potentialized in Permanent 



4. A few words on the errors involved are essential. Com- 

 parison of the values of PdL/L and Pdl/l shows that the 

 work lost upon kinks and flaws together with that spent upon 

 the frame-work is not seriously large so far as the present pur- 

 poses are concerned. This is also true of the energy elastically 

 potentialized, as may be found by direct tests. The satisfac- 

 tory measurement of the thermal datum t, however, is much 

 more difficult. Apparently the graduation of the thermo- 

 couple is simple : for it is merely necessary to make preliminary 

 observation of the throw of the needle of a ballistic galvan- 

 ometer produced by given increments of temperature. The 

 use of such graduation is, however, only permissible if the 

 temperature of the wire remains constant during the period of 

 oscillation of the needle. These conditions are never rigorously 

 given ; whereas even in case of jacketed wires it is a question 

 whether they are sufficiently given. The temperature of the 

 wire increases very rapidly to a maximum, and then decreases 

 by radiation, etc., reaching the original thermal value in a few 

 minutes. Again the cooling effect of the metallic wires of the 

 thermo-couple cannot easily be allowed for, neither can it be 

 considered negligible even in case of Alimentary wires. Finally 

 the error of heterogeneity is of serious consequence ; for the 

 thermoelectric measurement is virtually a thermal exploration 

 of the metal lying very near the point-junction. Hence since 

 the wire near such a point may be imperfect by reason of flaws 

 or composition so that more or less work is done here than at 

 other parts of the wire, it follows that the temperature thermo- 

 electrically obtained is not a mean datum for the wire taken as 

 a whole. 



Unfortunately the combined effect of the errors stated will 

 usually be the cause of too small a value of t. Some assurance 

 of the approximate truth of the results in Table 1 may, how- 

 ever, be obtained by observing that the experiments made are 

 to some extent differential in kind. For instance, caet. par., 

 about as much heat is evolved in the copper wire for an expen- 

 diture of only f the work applied to the brass wire. Sections 

 and thermo-couple are here the same and similarly adjusted. 



The importance of the thermal datum is such, however, that 

 special corroborative measurements are essential. To obtain 

 these, the above method was modified in such a way that the 

 wire was stretched successively in equal amounts. One end of 

 it was fixed and the other fastened on the circumference of an 

 iron drum of small radius, i o=0'9 cm . By revolving the latter 

 the wire is stretched and the friction of the axle sufficient to 

 keep it so. I chose successive angles of revolution, tt, by 

 which extensions <5Z/Z= "054 each, were easily produced, and 

 could usually be repeated 4 or 5 times. 



