.AJAGNETO-OPTIC PHENOAIENA OP IPUN, NICKEL, AND COBALT. 
103 
6 changes sign for that value of i which makes the vector angle of B cos i/A equal to 
an odd number of right angles. Let us assume that x must lie between 0° and 180°, 
leaving the sign of Cq to be determined afterwards. To obtain the vector angle for 
any given angle of incidence we must calculate the vector angles of the various 
complex factors which occur in numerator and denominator of the fraction in 
equation (38*). This involves troublesome arithmetical work ; but it is preferable to 
the approximation on the supposition that R is large, used by J. J. Thomson 
(• Recent Researches,’ p. 498) in a similar investigation, as that method introduces 
an error of quite a large number of degrees. 
Using the constants for yellow light, I get the following values :— 
Angle of 
incidence. 
Vector angle of 
iW. 
Vector angle of 
+ cos i. 
Vector angle of 
— cos i 
75° 
-55° 15' 
-52° 20' 
117° 17' 
78° 54' 
-55° 18' 
-53° 6' 
100° 23' 
80° 
-55° 19' 
-53° 19' 
95° 2' 
whence are derived the following :— 
Angle of 
Vector angle of 
incidence. 
(B cos I’/A). 
Kerr. 
75° 
;c + 187° 38' 
Righi . 
78° 54' 
.c + 205° 21' 
SiSSINGH. 
80° 
a; + 210° 56' 
So that if 6 changes sign when i — 75°, 
If when i = 78° 54', then 
If Avhen i = 80°, then 
a: = 82° 22'. 
a: = 64° 39'. 
a; = 59° 4'. 
And probably if 6 changed sign when i = 78°, the corresponding value of x 
would be about 69°. 
I he uncertainty as to the exact value of the angle of incidence for which 6 
vanishes, and the large dilierence, caused by a small error of observation, in the 
resulting value of x, render this experiment unsuitable as a means of arriving at the 
exact value of x. It will, however, be useful in testing a value of x determined in 
some other way. 
