84 Mr. E. II. Barton. Electrical Interference Phenomena 



\! = 9 m., the value of the frequency, namely, /27r = 33| millions 

 per second, nearly. 



From these data the curve T, fig. 4, is plotted, the lengths of the 

 abnormal part being abscissae and the corresponding fractions of in- 

 cident wave energy transmitted being ordinates. 



It is thus noticeable that the theoretical curve, though of simi- 

 lar form, lies wholly below the experimental one. If, however, 

 we apply to the latter, a correction based upon Equation (4) and 

 the results of Experiment IV, we obtain curves which lie wholly 

 below the theoretical one. One such is shown in curve C, fig. 4, in 

 which (T is taken equal to O0029. 



Thus, while the comparison of theory arid experiment has not 

 resulted in the establishment of an exact agreement between them, it 

 is yet so far satisfactory to notice* that the three curves are precisely 

 similar in form, and that the theoretical one lies entirely between the 

 two experimental ones, with and without correction. It is also note- 

 worthy that Mr. Yule, in his recent workf on "The Passage of 

 Electrical Wave-trains through Layers of Electrolyte," found a dis- 

 crepancy of the same sign and of like amount between the experi- 

 mental curve and the theoretical one. 



The dotted lines above and below the main curve in fig. 4 indicate 

 the limits between which the latter lies. These limiting lines would 

 be horizontal but for the primary damping of the waves emitted by 

 the oscillator. 



We thus see that the curve in the neighbourhood of Xj/4 affords 

 the best means of determining b when it is nearly unity. For 

 the troughs of the curve are broad, and therefore the curve is at that 

 part for some distance nearly parallel to the axis of the abscissae. Thus 

 a great deviation from the true length XJ4 would make but a small 

 one in the corresponding ordinate. Secondly, for large values of 

 6, this ordinate is but very slightly affected by the primary damping. 

 Hence a great error in the measurement of a makes but. a slight one 

 in the determination of b from this part of the curve. Thus, taking 

 7 1 = 0'5, and accepting the experimental curve, we obtain b = O71 

 and r = 6, or taking the experimental curve, with correction, as 

 shown by the lower line in fig. 4, we get b = 0'8 and r = 9. And 

 it was between these limits 6 and 9 that r was calculated to lie. 



In Experiment FT. Since the same wires were used for the 

 abnormal part but closer together, namely, at a distance d' = 0'68 

 cm. instead of d = 8 cm., we have for the ratio of capacities 

 _ 1/4 log (d'/w) _ log (8/0-05) 



1/4 log (d/w) ~ log (0-68/0-05) ~~ 2 ' 



* Professor Hertz considered the agreement between experiment and theory as 

 close BB was to be expected. 

 f ' Phil. Mag./ pp. 643544, Dec. 1893. 



