of the Electric and Lumimferous Medium. 



279* 



Another field in which the influence of a gyratory character in the- 

 molecule might be expected to be prominent is that of optics, more- 

 particularly the influence of matter as it appears in refraction aud 

 reflection. As an introduction to this subject, Lorentz's law of the 

 relation between refraction and density is worked out ; the argu- 

 ment is purely statical and independent of the constitution of the 

 molecule, and closely follows a cognate investigation of Clausius- 

 as I afterwards discovered ; it is however retained, as it appears to 

 be exempt, within its proper scope, from the objections which are 

 valid against other modes of demonstration of that law that have 

 been proposed. When we pass on to disc ass dispersion, the forced 

 vibrations of the molecules come in : and these will be of different 

 type according as the molecule is taken to be a system vibrating 

 about a position of rest as has hitherto been tacitly done in optical 

 theory, or a system vibrating about a state of steady motion as it is 

 here required to be. The main result, for a medium devoid of non- 

 selective opacity such as would arise from conduction, is that the 

 Lorentz refraction equivalent ((/a 2 — l)/(^ 2 + 2)/> (not ytt 2 — 1 as in the 

 usual dispersion theories) is an additive physical constant, equal for 

 each simple medium to 2gr r /(p r 8 — p 2 ), where p r is a natural vibration- 

 frequency for the molecule and g r is a related constant. The only 

 simplification that comes in when the gyratory quality is absent is 

 that then g r is necessarily positive : in the present case the reasons 

 for taking it to be positive are not so conclusive. If in any term g r 

 were negative the character of the anomalous dispersion near the 

 corresponding absorption band would be the opposite to that indi- 

 cated by Kundt's law, which has hitherto always been observed to 

 hold good. 



The only way that is a priori unexceptionable for determining the 

 complex index of refraction of a strongly absorbing medium such as 

 a metal is Kundt's method of deviation by thin prisms : this gives 

 only the real part of the index, but it is shown that by taking* 

 advantage of oblique as well as of normal incidence approximate 

 values might be obtained for the other part as well. The fair agree- 

 ment of Kundt's values with those derived from experiments on 

 polarisation by reflection is however a confirmation that surface 

 films are not seriously operative in the latter method, which has 

 yielded both parts of the index. If there were no non-selective 

 absorption, the curve representing the real part of the index would 

 rise to infinity near each absorption band, then fall straight down to 

 the axis, coincide with the axis for an interval, and finally again rise 

 above it. But when there is also general absorption the curve will 

 turn back before reaching infinity, and it will not descend as far as 

 the axis, while the sharp corners will be eased off. These charac- 

 teristics are precisely those of anomalous dispersion, for example of 



