August 2 2, 1889] 



NATURE 



399 



difference which might have existed, but unfortunately it 

 took some little time to alter the wall. 



In order to obviate this difficulty the following device 

 was resorted to with the object of showing that there was 

 a difference in the behaviour of the wall when 10 centi- 

 metres thick to its behaviour when 20 centimetres thick. 

 (For at the time I did not see that the experiment was 

 inconclusive, the effects observed being the same whether 

 the back reflected at all or not). A small sheet of zinc 

 was placed at the back of the wall, and the effect on the 

 sparking observed while an attendant suddenly removed 

 or again feplaced the zinc. It was supposed that when 

 the wall was 20 centimetres thick and there was sparking, 

 that on suddenly placing the zinc on the back the spark- 

 ing would increase, owing to the phase of the reflection 

 from the back being half a period different from that 

 of the reflection from the zinc ; but when the wall was 

 10 centimetres thick that the presence of the zinc would 

 diminish the sparking. 



It was with no little surprise that the reverse was ob- 

 served. That is to say, placing a sheet of zinc about '^'^ 

 centimetres square on the back of the wall actually aided 

 the reflection from the back so as to diminish the spark- 

 ingwiththe 20 centimetre wall, but increasing it with the 10 

 centimetre wall. This observation made it look as if it 

 must be on the first reflection from the paraffin — that is to 

 say, on passing from a rare to a dense medium — that the 

 " change of phase " occurs, and not at the back, — at a re- 

 flection from a dense to a rare medium, as is ordinarily 

 supposed. For Hertz's experiment of loops and nodes 

 showed that there was no change of phase on metallic 

 reflection — that is, of the magnetic displacement. There 

 is a change of phase of the electric displacement. It is 

 important to bear in mind that the electric loop and the 

 magnetic node occurred at the same place, and of course 

 so too the electric node and the magnetic loop. 



In order to investigate this, attempts were made to 

 obtain Hertz's loops and nodes off a paraffin wall as reflec- 

 tor, but no reflection could be discovered, the intensity of 

 the vertically reflected rays being insufficient. However, 

 by inclining the incident radiation to an angle of 57°, the 

 intensity of the reflection was found to be amply suf- 

 ficient. With a circular resonator, which is for these 



Fig. 2. 



waves about 10 centimetres in diameter, sparks were 

 obtained close to the reflector, the circle being held at 

 right angles to the wall so as to be equally inclined to both 

 direct and reflected radiation, and this was confirmed by 

 a straight resonator giving none there. At 30 centimetres 

 from the wall ^ there was interference with the circle, and 

 vigorous sparking with the straight resonator. This being 

 about the right distance for the loop to be from the 

 reflector at an incident angle of s;"", 



- = / -{- /' = p sec /(i -f cos 2/) = T.p cos i. 



Thus there is no doubt that it is on the second reflection 

 that the change of phase occurs. 



Here, then, was a difficulty — the small sheet of zinc at the 



' It would occur at about 17 centimetres on vertical reflection. This ex- 

 periment was also tried with a metallic reflect jr. 



back of the paraffin undoubtedly reflected with a change 

 of phase, while, according to the Hertzian experiment, 

 metallic reflection is unaccompanied by change of phase. 

 On mentioning this to Prof. Fitzgerald, he pointed out to 

 me its complete agreement with wave theory. For by 

 considering the secondary waves produced by dividing up 

 a primary wave with reference to any point into half- 

 period zones, it can be seen that the effect of the primary 

 is equivalent to half of that arising from the central circle, 

 and in consequence is half a period behind the phase 

 which would be at the point if an infinitesimal portion of 

 the centre alone acted. For the effect of each ring can be 

 considered as destroyed by half the effect of its two 



neighbours, and thus half the effect of the central circle is 

 left uncompensated. But the distance of the edge of this 

 circle is half a wave-length further from the point than its 

 centre is, so that the resultant phase at the point will be 

 behind that due to the centre, but in front of that due to 

 the edge, which effect would be half a period behind that 

 arising from the centre. Taking the mean between them, 

 the resultant phase then at the point is a quarter'^ of a 

 period behind what it would be if the centre alone acted. 

 Thus it was that the reflection from the small sheet of 

 zinc differed from what 1 had expected it to be. 



Experiment showing Phase of Secondary Waves. —To 

 experimentally test this, the small sheet of zinc was used 

 as reflector in the Hertzian experiment of loops and nodes. 

 Employing the circular resonator, the position of inter- 

 ference was found to have shifted out from 17 to over 24 

 centimetres, which nearly corresponds to an acceleration of 

 phase of a quarter of a period, the wave going in all nearly 

 a quarter of a wave-length further, and nevertheless being 

 still only half a period behind the phase on starting. The 

 furthest out the loop could be is 25"5 centimetres : to 

 obtain this would require an indefinitely small reflector. 

 Of course, when the resonator was close in to the sheet, 

 no change of phase was found to occur, the sheet being 

 then practically infinite. 



Another interesting observation was made. A long 

 sheet of zinc 30 centimetres wide was found to act simi- 

 larly 10 the sheet 30 centimetres square, provided it was 

 placed with its breadth parallel to the electric displace- 

 ment. When thus placed at 24 centimetres from the 

 circular "resonator," there was interference, but on 

 rotating the reflector so as its length was parallel to 

 the electric displacement, sparking occurred, and now 

 the "resonator" had to be brought back to 17 centimetres 

 in order to again obtain interference. This experiment 

 is interesting in connection with the electro-magnetic way 

 of looking at the acceleration of phase as being due to 

 the accumulations of electricity on the edges of the re- 

 flector, which is the same as the reason why it is necessary 

 to use long cylindrical mirrors, as was pointed out by Prof. 

 Hertz in a letter last February to Prof. Fitzgerald. This 

 experiment is really the same as Stokes's experimentum 

 crucis, as Prof. Fitzgerald points out. 



If, instead of using the whole primary wave in the 



' That it aided t'le back raiher than the front was probably due to their 

 pliase not being an exact period tr half period diflerent from each other. 



