between Optical and Electrical Qualities of Metals. 177 
conclusive because they are absolute values, not containing 
any arbitrary factor. 
The coefficients w) and @ of equation (5) are directly 
obtained by ordinary electrical measurement. In the region 
of lower temperatures (down to about 250°) where the in- 
fluence of the square term #é*? is only small, the observed 
emission values agree perfectly with those computed from the 
electrical conductivity. But in order to obtain a sinular 
agreement between observed and computed emission values 
in the region of high temperatures, the adoption of a square 
term with the coefficient @ is absolutely necessary. 
The coefficient 8=0°0000033 has a positive sign in equa- 
tion (5), thereby indicating a more rapid increase of the 
resistance at high temperatures. This is in contradiction 
with measurements of Messrs. Benoit*, L. Callendart, and 
Holborn & Wient, who have all observed a slower change 
of resistance at high temperatures, which corresponds to a 
small negative value of 8. 
Accordingly, we must suppose that the increase of resistance 
at higher temperatures—as computed from our observations 
on emission—is only apparent, and that other facts influence 
the change of the emission of platinum in that region. 
Particularly a change (roughening) of the surfaces at high 
temperatures is not improbable, and that would account for 
a remarkable rising of the emissive power. Lastly, it is not 
improbable that the observed deviations are connected with 
perhaps the insufficient homogeneity of the “ residual rays.” 
Summary of the obtained Results. 
1. The reflecting-power of the investigated metals from 
A= 0°65 to A= 14y is given in Table I.; the emission-power 
for X=25'5y and 170° C. is to be found in Table IV. 
2. For long waves the intensity entering into the metals 
(100—R) is in inverse proportion to the square root of the 
electrical conductivity «, and to the square root of A, the 
wave-length of the incident radiation. This law, derivable 
from Maxwell’s theory, holds good the more strictly, the 
longer the waves are. This is proved in Table VII., which 
gives the observed and computed values of the constant 
Vr 
for four different wave-lengths of the infra-red spectrum 
* R. Benoit, Compt. Rend. \xxvi. p. 342 (1873). 
tL. Callendar, Phil. Mag. [5] xlvii. p. 191 (1899). 
z L. Holborn & W. Wi ien, W ied. Ann. lvi. p. 860 (1895). 
Phil. Mag. 8. 6. Vol. 7. No. 38. Feb. 1904. N 
C, =(100—R) Ve and C= 
