148 Br. E. H. Barton on the Attenuation of Electric Waves 



Various trial values were then given to a and plotted as 

 abscissae of a curve, the corresponding values of u being its 

 ordinates. The value of a satisfying the equation was thus 

 found graphically as that at which the curve cut the axis of 

 abscissae. 



Experiments. — In the experiments the bridge used as an 

 absorber was placed absolutely at the end of the line, as it is 

 only when in that position that it can absorb all. It consisted 

 of pencil marks on a ground glass disk as previously described, 

 and before use each time was carefully adjusted to the requisite 

 resistance for absorption as mentioned in the paper already 

 referred to. A doubt has been expressed as to the constancy 

 of the value obtained for a resistance of this kind by the 

 Post-Office box, some similar resistances having been said to 

 measure differently according to the voltage used in the test. 

 Probably this might be the case to a slight extent if extreme 

 variations were used, but with the pencil-mark resistances 

 employed in this work no change of value was observable, 

 whether a single Leclanche cell was used or a battery of 

 several storage-cells, so that the resistance was determined 

 with all the accuracy needed for the present purpose. 



The other bridge, intended to reflect undiminished the waves 

 incident upon it, was placed a little short of the end of the 

 line, and occupied alternately two positions a quarter of a 

 wave-length apart. This precaution was adopted lest, far as 

 the bridge was beyond the electrometer, the head of the wave- 

 train reflected from the bridge might reach the electrometer 

 before the tail had passed it. In this case a slight interference 

 would arise at the electrometer, and its throws would not be 

 of the values due to simple passages and re-passages of the 

 wave-train. By using the above two positions for the bridge 

 this source of error is obviated, for, if one gave additive inter- 

 ference at the electrometer, the other would give about equal 

 subtractive interference, and their mean would be the result 

 desired as though no interference had occurred. 



The length l 2 is of course measured from the electrometer 

 to the point midway between these two positions of the 

 bridge. 



Further, to eliminate errors due to the unavoidable irregu- 

 larity of the sparking in the primary oscillator, a number of 

 readings, about 20, with the absorber only at the end of the 

 line were alternated with a like number of readings with the 

 reflecting bridge interposed. The separate ratios of the 

 electrometer-throws obtained under the two conditions were 

 then taken, and the probable error of their arithmetic mean 

 calculated by the theory of probabilities. 



