82 Mr. E. H. Barton. Electrical Interference Phenomena 



with and without the abnormal part were absolutely equal. It 

 became expedient, therefore, to use an electrometer which gave de- 

 flexions only when reflected waves were present. Such an instru- 

 ment was, fortunately, ready to hand in the differential electrometer 

 used and described by Dr. von Greitler,* and kindly left by him for 

 my use. 



In this form of electrometer two needles, rigidly connected with 

 each other and hanging upon the same quartz fibre, are employed. 

 One of these experiences a right-handed torque, owing to attach- 

 ments at one pair of points on the secondary, whenever these points 

 have a potential difference of either sign, while the second needle, 

 under like conditions, experiences a left-handed torque, owing to 

 attachments at a second pair of points on the secondary, and distant 

 a quarter of a wave's length from the first pair. Thus, when the 

 instrument is properly adjusted, the mere passing of a wave-train 

 leaves the needles undisturbed. 



If, however, reflexion is by any means produced near the electro- 

 meter, so that it is in a region of standing waves, it may give a 

 throw. 



The arrangement of apparatus adopted may be seen from inspec- 

 tion of fig. 1, which was modified by the removal of the single-needle 

 electrometer from EE', and the insertion of the differential electro- 

 meter at FF' HH'. The first needle was influenced by attachments 

 to F and F', and the second by attachments to H and H'. The main 

 wires between the first and second pairs of points were led in a loop, 

 so that they first passed the electrometer at FF', and a second time afc 

 HH'. Thus, all the attachments between the main wires and the 

 electrometer were quite short. 



The wires in the normal part of the secondary were O116 cm. 

 diameter. The abnormal part consisted also of copper wires, and 

 were in this experiment 0*019 cm. diameter, and 2'25 m. long. 



These were tried at two distances apart, namely, 1 cm. and 1*5 cm. 

 The electrometer throws in these cases were 1'8 and -fO'6 scale 

 divisions respectively. Whence, by interpolation, we have T37 cm. 

 nearly, as the distance apart at which these wires would yield no 

 reflexion. The electrometer throw with a bridge atBB' was twenty- 

 four scale divisions. 



Experiment IX. A Second Example of no Eeflexion from an Abnormal 



Part. 



In this experiment everything was the same as in the preceding 

 one, except that the abnormal part consisted of wires thicker and 

 wider apart than the rest of the line, instead of being thinner and 

 * Wiedemann's ' Annalen,' vol. 49, pp. 188189, 1893. 



