160 BRIDGMAN. 



D.C. was reached that the specimen burned out. It was usual to make 

 readings at five to seven different D.C. intensities. 



The difference between D.C. and A.C. settings for each constant 

 D.C. were now plotted against the reciprocal of the frequency, and 

 the results extrapolated to zero (that is, infinite frequency). Of 

 course the chief point of the experiment is this extrapolation. It is 

 in the first place to be remarked that the extrapolation is not a wide 

 one, the observed frequencies covering a range of ten fold, so that the 

 distance of extrapolation is only one tenth of the distance covered 

 by the readings. It is in the second place to be noticed that we have 

 theoretical evidence as to the form of the curve in the neighborhood of 

 zero. The part of the plotted difference which involves the heating 

 eft'ect, and which we are trying to eliminate varies linearly as the 

 reciprocal of the frequency. The remainder, which is the departure 

 from Ohm's law, is constant, independent of the frequency. The 

 curves therefore extrapolate as straight lines, their curvature disap- 

 pearing as they approach the axis. Given this condition, the extra- 

 polation can be made with very little uncertainty. The curve is 

 already so nearly straight at the last two or three frequencies that in 

 most cases a straight line could be drawn through these points with- 

 out error. However, this is not quite satisfactory. In the early part 

 of the work I made the extrapolation by plotting the results on a large 

 scale, and drawing free hand the curve which seemed to best satisfy 

 the conditions. Later, however, in order to make the process of 

 extrapolation more mechanical and less subject to error by personal 

 bias, I adopted the following mechanical device. A steel spring was 

 made to conform as closely as possible to the observed points by 

 imposing on it three restrictions. It was made to go through, or 

 near, the observed point at the lowest frequency, another point near 

 the middle of the frequency range, and it was made to go through 

 such a point on the axis that it coincided as closely as possible with the 

 points at the upper end of the frequency range. This process will be 

 more fully described in the next paragraph. 



Three typical sets of readings, giving differences between A.C. and 

 D.C. settings in terms of cm. of bridge wire plotted against reciprocal 

 of frequency, are shown in Figures 7, 8, and 9. These figures are for 

 the two thicknesses of gold and the one of silver. It will be seen that 

 there can be little doubt about the extrapolation. The fact that the 

 extrapolated curve does not pass through the origin, but strikes the 

 axis at a point higher up, constitutes the evidence for the departure 

 from Ohm's law. The mechanical extrapolation was made as follow^s. 



