138 
The direction of revolution in the ellipse is easily seen to de¬ 
pend upon the sign of the quantity a b sin D. Now from (4) and (7), 
§ 1., it follows that 
(4) ab sin D = % 2 — a 2 2 ; 
hence, on the opposite sides of wave-length 
(5) 'Vt* — 9 
in the spectrum, the revolution will be in opposite directions. 
Equation (3) of this article seems to embody a remarkable re¬ 
sult. If we could evaluate the magnitude of the coefficient k we 
see from (3) that we would be enabled to calculate the ratio of 
the semi-axes S/a for light polarized elliptically under the influence 
of a magnetic field, the wave-length of light being supposed to be 
nearly the same as that in the middle of an absorption line 1 ). And 
conversely, from an exact determination of the value of S/a in this 
case we can expect to derive a considerable amount of guidance 
as to the nature and magnitude of that important constant, the 
dissipation coefficient k. 
§ 7. In deducing formula (ti) of § 5., we have supposed that 
the value of c f is small. This restricts us to magnetic fields which 
are, from an experimental point of view, weak or (at most) mode¬ 
rately strong. If the impressed magnetic field is of considerable 
intensity, formula (6) of § 5. cannot be applied; we have to revert 
in this case to the exact expressions (3) and (4) of § 5. 
To take an example: let us fix our attention on the point 
( 1 ) À = — g 
in the spectrum and let us examine the behaviour, at this point, of 
the substance considered, in a strong magnetic field We have then 
(2) F= 0. 
Hence if we put 
(3) n=xn/4nceN 
1 ) An effect of this kind was actually observed by Macalnso and Corbino; 
however, so far as the writer is aware, a quantitative study of the ellipticity 
produced has not hitherto been made, at least not in the case of gaseous bodies 
(see R. W. Wood. Phil. Mag. for February 1908, page 273). 
