( 352 ) 
45° with a vertical line; in the second case they are either horizontal 
The method of procedure is very simple. The nearly ideal Fresnel 
rhomb transforms light of completely circularly polarized doublets 
into two perpendicular, linear vibrations. By means of a Nicol either 
the one or the other of the lines can be completely quenched, provided 
there is no trace of ordinary light in the double lines. 
The observations with the lines investigated, indeed show that 
such must be the case. The value of the result depends upon the 
Sensitivity of the method. 
The observations consisted in finding the two positions, on either 
side of the position of extinction, at which the light was just recog¬ 
nizable with certainty. 
If « be the angle the Nicol is then rotated from the zero position, 
Isin * a ^presents the brightness of the emergent light, I being the 
intensity of the linear vibration. 
We can be sure that the quantity of ordinary light, emitted by 
one of the components, must be below I sin* a. 
8. Results. The observations consist in determining the value of 
« or the different spectral lines. I will arrange the results according 
to the different types of subdivisions, observed normally to the field. 
Inplet, mercury, 1 = 5791, doublet parallel to the field, a = T 
sm a 0.0144. The observation is somewhat hindered by satellites 
of the principal lines. 
triplet, 1 = 5771, doublet in direction of field. 
n extremely strong fields every component is resolved into three 
lines. This strength of field was not reached. 
« = 5° sin 2 « = 0.0076. 
Quartet , sodium, X = 5898, doublet in direction of field. 
« = 5° sin 2 « — 0.0076 
Sexiet, sodium, a = 5890, quartet in direction of field. 
« —6° sin*a = 0.0108 
Nonet , mercury, 1 = 5461, strong green mercury line, sextet in 
direction of field. J 
« = 5° sin*a= 0.0076. 
- Sextet, thallium, * = 5351, quartet in direction of field, 
« = 8° sin s « = 0.0196 
robably a lower limit can be obtained, when a more satisfactory 
thallium tube is made use of. 
