1895. ] NEW YORK ACADEMY OF SCIENCES. 169 
portional to sin? w and cos? w, the change in intensity of one 
of them will be very rapid when w is in the neighborhood either 
of zero or of ninety degrees. When, however, w is near 45°, 
the change in intensity corresponding to a small change of angle 
is comparatively small. Hence, when the polarization is large, 
and w consequently either large or small, the position of equality 
can be determined with considerably greater accuracy than when 
the polarization is weak and w in the vincinity of 45°. 
The results obtained for large angles may therefore be con- 
sidered more trustworthy than “the results for small angles. 
VIII. 
MEASUREMENT OF THE AIR EFFECT. 
In the qualitative experiments previously described it was 
ascertained that the amount of polarization was at least not 
greatly affected by the contact of the air with the heated sur- 
face. Before proceeding to careful quanitative measurements it 
was necessary to determine whether or not its effect upon the 
phenomenon is altogether negligible. This could be easily done 
by means of the polarimeter. 
The sealed glass tube containing the platinum strip was again 
connected with the air- -pump, and the air exhausted until the 
pressure was about 4 mm. The current was turned on, the 
polarimeter arranged so as to receive the light emitted from the 
glowing surface at an angle of about 80°, and the Nicol turned 
until the images were brought into equality. The stop cock was 
then suddenly turned and the air admitted. No change what- 
ever could be perceived in the equality of the images. The ex- 
periment was repeated a number of times and in a variety of 
ways, but always with the same result. The conclusion was, 
that, if the air has any effect whatever upon the proportion of 
polarization in the beam, that effect is so slight as to be ALTO- 
GETHER NEGLIGIBLE; a result exceedingly fortunate for the pur- 
poses of this investigation, since, had it been necessary to work 
upon substances in a vacuum, the following experiments would 
have been much more difficult, if not altogether impossible. 
rx 
EXPERIMENTS UPON URANIUM GLASS. 
The chief difficulties which beset the investigation of Polari- 
zation by emission are, Ist, the difficulty of obtaining a per- 
fectly definite and regular incandescent surface with which to 
work; and 2nd, the difficulty of ascertaining with certainty the 
