polarisation of light by reflexion from transparent bodies. 129 
correctly the index of refraction for the mean refrangible ray, 
and the still greater difficulty of determining the angle at 
which the intensity of the evanescent pencil is a minimum. 
The total amount of the errors in seventeen observations is 
259 minutes, which gives an average error of 15' for each 
observation. In general the observed angles are less than 
the calculated angles, the number of negative being to the 
number of positive differences as 174/ is to 85'. 
This circumstance arises from two separate causes, which 
ought to be carefully kept in view in all experiments on the 
polarising angles of bodies. 
1. In order to illustrate the first of these causes, let us take 
the case of Zircon, in which the intensity of the evanescent 
pencil is a minimum at 63° of incidence. At 64? the intensity 
of the pencil which vanishes at 63° is much greater than that 
of the pencil at 62° on account of its falling more obliquely 
upon the reflecting surface, and consequently the intensity 
of that pencil varies more rapidly in passing from 64° to 63° 
than from 62° to 63°. Hence, in determining the point of 
minimum intensity, it is more likely, from the way in which 
the observation must be made, that the observed angle will 
fall below than above the real polarising angle. 
2. As the differently coloured rays have different angles 
of polarisation, and as the most luminous rays of the spectrum 
have less refractive power than the mean refrangible rays, the 
observed polarising ought always to be less than the polaris- 
ing angle for the mean rays. Hence, all the observed angles 
in the preceding Table ought to be increased by a certain 
quantity, or, what is the same thing, the index of refraction 
for the most luminous rays ought to be employed instead 
MDCCCXV. S 
