338 Dr. J. Kerr's Experiments on tJte 



heating is found, accordingly, to be compression along the 

 length of the plate, very intense at the heated surface, and 

 diminishing very rapidly inwards. The effect obtained in the 

 principal experiment with the heated plate admits therefore of 

 only . one explanation : — a ray along a line of compression is 

 retarded by the strain*. 



This is not such a simple result as it may at first ap- 

 pear to be : for the observed retardation is the sum of two 

 terms, corresponding toeSmand (m— l)Be of the retardation a 

 of art. 12, the first due to change of refringent power, and the 

 second due to change of thickness of the glass along the rays ; 

 and the experiment gives us no information as to the ratio of 

 these terms ; nor do we certainly know as yet even the sign 

 of the first term. In this connexion I must mention some 

 earlier experiments with a bent plate. 



16. The plate was taken from the same piece of light flint 

 as the heated plate of the last article, and was of the same 

 length, but narrower ; and it was strained in the same way as 

 the bent plate of art. 3. It was fixed in the screw-press with 

 its faces horizontal, so that the two pencils (vertical and very 

 narrow) of the refractor traversed it along its length, one 

 pencil very close to a lateral surface, and the other at or very 

 near the middle of the plate. The action of this bent plate 

 was not perfectly regular, the strains being complex, partly 

 along the pencils and partly across : but upon the whole, and 

 quite clearly, the effects were exactly contrary to those already 

 described as given by the heated plate, a ray along a line of 

 compression being relatively accelerated by the strain, and a 

 ray along a line of tension relatively retarded. 



To explain the contrariety of effect in the heated plate and 

 the bent, consider the two terms of the resultant effect, as 

 they are particularized in the end of last article. For light 

 along the line of strain, assume that the second term is abso- 

 lutely the greater of the twof. Then (1) in the bent plate, the 

 glass is evidently shortened by the strain along a line of com- 

 pression, and therefore both the corresponding second term 

 and the total effect are accelerations ; and (2) in the heated 



* A ray along a line of tension is of course accelerated by the strain. 

 This effect is obtained very distinctly when the plate is placed as in the 

 principal experiment ; and the lateral surface, instead of being heated, is 

 cooled by the rapid evaporation of a little ether. 



t The assumption is apparently justified by what we know of the 

 laws of deformation of glass under directional stress, taken in connexion 

 with prop. VIII., which asserts the equality of the two terms when the 

 light is perpendicular to the line of strain. 



