Nov. I, 1888] 



NATURE 



II 



\ elocity of wave transmission must undergo a sudden 

 and immense decrease — a circumstance always causing a 

 ^reat amount of reflection, just as when sound tries to 

 !^s from any one medium to a much denser one. 

 I)Ut the opacity of iron and other magnetic substances 

 lybe explained by the mere fact of their conducting 

 wer, just like other metals, and no noteworthy etfect of 

 eir large value of \i need be detectable. 

 If a non-conducting highly magnetic substance could 

 • found, it would probably reflect a great deal of light 

 : its surface, though it would not dissipate that which 

 entered it. Such a substance would be most interest- 

 ing to submit to experiment, but perhaps its existence 

 jjresupposes a combination of impossible properties. 



As to the phenomenon detected by Hall, it appears in- 

 timately associated with that of Faraday, and it will be 

 most simple to omit all reference to it for the present. 



A general idea of what is happening in the Faraday 

 and Kerr phenomena can be given thus. A simple vibra- 

 tion, like a pendulum-swing, or any other oscillation in one 

 plane, can be re-:olved into two others in an infinite variety 

 T ways ; just as one force can be resolved into any number 

 pairs of equivalent forces. The two most useful modes 

 , analyzing a simple vibration into a pair of constituents 

 are these : (i) two equal components, likewise plane 

 vibrations, each inclined at 45^ to the original one, as 

 when P Q is resolved into A B and C D (Fig. 49) ; and (2) two 

 equal circular or rotatory oscillations in opposite directions. 



as when P Q is resolved into P M Q and P N Q (Fig. 50). The 

 first method of resolution is useful in explaining Kerr's 

 effect, the second in explaining Faraday's. 



Cf the two component vibrations, A B and C D, into which 



V (J can be supposed analyzed, let some cause, no matter 

 what, make one gain upon the other, so that in travelling 

 along a line perpendicular to the paper one goes a little the 

 ■quicker : the effect at once is to change the character of 

 the vibration into which they will recompound. After the 

 gain, they no longer reproduce the original simple vibration 



V (i, : they give rise to elliptic, or it may be to circular, 

 vibrations ; this last, if the retardation is equal to a 

 quarter period. 



These are matters fully treated in any elementary 

 treatise on polarized light, and they are quite easily 

 illustrated by means of a simple pendulum. One may 

 assume them known. 



-Similarly with the second system of analyzing the 

 vibration into two opposing circular ones. If the com- 

 ponents travel through any interposed medium at the 

 same rate, they will, on emergence, reproduce the original 

 vibration in its original position ; but if one travels 

 quicker than the other they recombinc into a vibration of 

 ttie same character as at first, but turned through a certain 

 angle. Thus anything which retards one of the rectatii^ular 

 components behind the other changes the character of the 

 vibration from plane into elliptical ; while anything which 

 retards one of the circular components behind the other 

 leaves the character of the vibration unaltered, but rotates 

 it through a certain angle. 



So far one has said nothing but the simplest mechanics. 

 The next point to consider is what determines the rate at 

 which light travels through any substance .'' This we have 



discussed at length, and shown to be ,,]T*y Anything 



which increases either the electric or the magnetic per- 

 meability of the medium decreases the velocity of light. 

 Now, when a medium is already subject to a violent strain 

 in any one direction it is possibly less susceptible to 

 further strain in that direction and responds less readily. 

 Not necessarily so at all : such an effect would only be 

 produced when the strain was excessive, when the medium 

 was beginning to be overdone, and when its properties 

 began thereby to be slightly modified. There are 

 reasons for believing the specific inductive capacity of 

 most media to be very constant ; of some media, perhaps, 

 precisely constant ; but if there were any limit beyond 

 which the strain could not pass it is probable that on 

 nearing that limit the specific inductive capacity would 

 be altered — possibly increased, possibly diminished — one 

 could hardly say which. Quincke has investigated this 

 matter, and has shown that the value of K is affected by 

 great electric strain. 



Suppose now that a dielectric is subject to a violent 

 electrical stress, so that its properties along the lines of 

 force become slightly different from its properties at right 

 angles to those lines. The value of K will not be quite 

 the same along the lines of strain as across them, and 

 accordingly the rectangular component of a vibration 

 resolved along the lines of force will travel rather quicker 

 or rather slower than the component at right angles, 

 because the velocity of transmission depends upon K 

 as already explained : such a medium at once acquires 

 the necessary doubly-refractive character, and will show 

 Kerr's effect. 



Similarly with magnetization. It is well known that for 

 many media /u is not constant. Take iron, for instance. 

 For very small magnetizing forces the susceptibility is 

 moderate, and increases as they increase ; at a certain 

 magnetization it reaches a maximum, and then steadily 

 decreases. But not only is it thus very inconstant, its 

 ascending and descending values are not the same. To 

 forces tending to magnetize it more, the susceptibility has 

 one value ; to forces tending to demagnetize it, it has 

 another and in general smaller value. This property has 

 been specially studied by Ewing, and has been called by 

 him " hysteresis." Slightly susceptible substances cannot 

 be magnetized to anything like the same extent, and 

 hence the property in them has been less noticed, perhaps 

 not noticed at all. Nevertheless it must exist in every 

 substance which exhibits a trace of permanent magnetism, 

 and every substance I have tried appears to show some 

 such trace (see Nature, vol. xxxiii. p. 484). 



An already strongly magnetized medium will be rather 

 differently susceptible to additional magnetizing forces in 

 the same direction than to those in a contrary direction. 

 Nothing more is wanted to explain Faraday's effect. The 

 vibration being resolved into two opposite circular com- 

 ponents, one of them must agree in direction with the 

 magnetism already in the medium and try to magnetize 

 it for the instant infinitesimally more; the other component 

 will for the instant infinitesimally tend to demagnetize it. 

 The value of /x offering itself to the two components will 

 be different, hence they will go at different rates, and the 

 plane of vibration will be rotated. 



rhe direction of rotation will depend on whether the 

 value of ii. IS greater for small relaxations or for small 

 intensifications of magnetizing force ; and diamagnetic 

 substances may be expected to be opposite in this respect 

 to paramagnetic ones. Any substance for which /* is 

 absolutely constant, whatever the strength of magnetic 

 polarization to which it is submitted, can hardly be 

 expected to exhibit any hysteresis ; the ascending and 

 descending curves of magnetization will coincide, being 

 both straight lines, and such a substance will show no 

 Faraday effect. Similarly, any substance for which K is 

 absolutely constant, whatever the electric polarization to 



