496 Professor Silvanus P. Thompson [May 17, 



be ready to build thereon the wider generalisation wbicli at present 

 seems to escape ns. 



So far we have been applying an optical torque to previously 

 polarised light, and producing a torsion of it. It remains for me yet 

 to describe the means by which, in the hands of Professor Abbe and 

 Professor Sohncke, it has been demonstrated that natural, non-polarised 

 light is actually rotated when subjected to an optical torque. 



The way of doing this is to make use of the principle of in- 

 terference. Here is the slit from which a narrow beam of light- waves 

 issues. At a point a little distance away is a Fresnel's biprism 

 which splits up the light (without polarising it) into two beams, just 

 as if we had two slits or sources of light. These two beams pass 

 along, and meet uj)on this distant screen, and give us — what ? A set 

 of interference fringes having a bright line down the middle, because 

 this part of the screen is exactly equidistant from the two sources of 

 light. 



But these dark interference fringes that lie right and left can only 

 exist because, in the first place the vibrations have travelled unequal 

 paths differing by an odd number of half wave-lengths ; and secondly, 

 because (owing to the method adopted of using two images of one slit) 

 the phases of the emitted waves from the two sources are identical. 



This being so, let us now introduce across the two interfering 

 beams of light a special biquartz, made of right-and-left handed 

 quartz of only 1 • 88 mm. thick. This will rotate — if it rotates 

 natural light at all — the yellow light in one beam 45^ to the right 

 and that of the other beam 45° to the left. The angles will be a 

 little more for green and blue, a little less for red and orange. Con- 

 sequently we shall not get quite a perfect result for all kinds of 

 colours. But for the main body of the light the result is this : that 

 because the two beams have had their respective vibrations turned so 

 that, whatever their primitive positions, they are now at right angles 

 to one another, they cannot interfere. In other words, if it be true 

 that the quartz rotates natural light, the interference bands will die 

 out. [Experiment shown.] 



Here I have the light passing through the biprism only, and 

 giving us this narrow series of interference bands. You must notice 

 carefully — with opera-glasses if you have them — the narrow bright 

 and dark stripes. Now I shift this little diapliragra so that the light 

 passes through the biquartz as well. Instead of sharp interference 

 bands we have merely a dull line of nebulous light. The disappear- 

 ance of the fringes proves that quartz does twist the non-previously 

 polarised waves of light. 



That the magnetic field can also exert a magnetic torque on non- 

 polarised light is readily proved, at least when one already has the 

 biquartz. Two strips of heavy glass of exactly equal length and 

 similar quality, such as those I hold in my hand, must be introduced 

 in the respective i:>aths of the two beams : and one at least of them 

 must be surrounded by a magnetising coil. The biquartz has wiped 



