32 SCIENCE PROGRESS 



expected that a deformation such as that supposed to be pro- 

 duced by motion through the ether will be accompanied by a 

 corresponding stress in the material sufficient to cause sensible 

 double refraction ? If such double refraction exists, it should 

 be possible, by placing a piece of transparent material between 

 two nicols whose common axis is perpendicular to the ether 

 drift, and setting the analyser for complete extinction, to 

 observe a revival of the light on rotating the whole apparatus 

 through 90 . As a matter of fact, in the actual experiment the 

 calculated effect was too small to make this method possible, 

 and the effect was tested for by comparison with known 

 amounts of double refraction produced in a strip of glass by 

 small stresses. Such a horizontal strip of glass loaded at the 

 ends and supported in the middle, when placed between crossed 

 nicols, produces revival of the light except in the neighbourhood 

 of a horizontal band in the middle, called the "neutral axis," 

 because there is no stress there. Just above and below this 

 band, the strain is very slight indeed, and can be made of the 

 order of that expected on account of motion through the ether 

 by suitable adjustment of the stress. By including this appa- 

 ratus, together with the transparent material under examination, 

 between the two nicols, Lord Rayleigh was able to reduce the 

 test to one for an apparent displacement of the neutral band. 

 For the strain above the neutral band would be an extension, 

 while that below would be a compression ; the addition, there- 

 fore, of what was equivalent to a uniform strain over the whole 

 piece would alter the resultant position of no strain. Obser- 

 vations of the behaviour of this band upon rotation were carried 

 out with water, carbon-bisulphide and glass, in each case using 

 considerable lengths of the substances so as to multiply the 

 effect ; in no case, however, was there a movement so large as 

 that calculated. For the liquids, the experiment was of sufficient 

 delicacy to conclude that not 2 per cent, of the calculated 

 double refraction was observable. The experiments with glass, 

 owing to increased difficulties such as loss of light, were by 

 no means so conclusive as to a negative result. They were, 

 however, repeated subsequently with improved apparatus by 

 Prof. Brace, 1 with the result that it can be definitely stated 

 that, in the case of glass also, no double refraction of the order 

 expected can be detected. 



1 D. B. Brace, Phil. Mag. 6th Series, vol. vii. 1904. 



