Birefringent Action of Strained Glass. 339 



plate, the compression being really a small dilatation against 

 resistance, the glass is lengthened along a line of compression, 

 and the resultant effect for light along that line is a retardation. 

 There is one thing, however, that I have neglected here, 

 perhaps unwarrantably, and that is the variation of index 

 with temperature (apart from strain) in the heated plate. 



Proposition XII. 



For light passing along the line of strain, as for light per- 

 pendicular to that line, the velocity of transmission is diminished 

 by the strain in the case of compression, and increased by the 

 strain in the case of tension. 



17. In seeking for a rigorous proof of this proposition, I 

 have met with greater difficulties than in any other part of the 

 subject ; but now, after many trials, I find that the preceding 

 experiment with the heated plate, when properly modified, 

 serves the purpose perfectly. In that experiment, as described 

 in last article, the observed retardation is a complex effect, due 

 apparently, but in degrees not known, to these four immediate 

 causes — (1) strain along the rays, (2) strain across the rays, 

 (3) local change of temperature of the medium, (4) local 

 change of length of the plate — difference of actions on the 

 two pencils being understood for each of the four. 



In the new experiment, the cause (3) is practically elimi- 

 nated thus : the light passes near one edge of the plate, and 

 the heat is applied for several seconds along the opposite edge, 

 which is 2 and 1\ inches distant from the pencils. The effect 

 of (4) is eliminated by virtual immersion of the plate in 

 Canada balsam, this body having exactly the same index (1*58) 

 as the plate. The effect of (2) may be safely neglected. 



As in the former experiment, the plate is laid horizontally, 

 and the light passes along its length, one pencil almost grazing 

 the lateral surface. Close to the plate, in front of the two 

 ends, and in the course of the two pencils, there are two small 

 panes of thin parallel glass, which are supported in fixed 

 positions independently of the plate, so as to form virtually 

 the two ends of a constant and fixed cell ; and the interval 

 (about '7 millim. wide) between panes and plate are filled 

 with Canada balsam. When the pieces are placed with proper 

 care, I find that the fringes given in the refractor, through 

 the composite plate (flint, balsams, and panes), are more than 

 sufficiently good, being finer than those given formerly 

 through the single plate of flint. 



Things being thus prepared, a flat bar of iron, at a tempe- 



Z2 



