Measurement of High Temperatures. 47 
making use of the hypothesis that the intensity of the 
grrenen from the surface inward. Here as in the equations 
or ordinary reflected light, R, depends upon the quantity 
y.' Imeasured in the following way the transparency of the 
petroleum flame which I wished to use in the experiments 
about to be described. 
Suppose such a flame to be brought into position before the 
Spectrophotometer so that a certain pencil of rays would fall 
upon the slit. Were the flame perfectly transparent, a second 
eascano similar one and equally bright would, if placed be- 
‘ind the first flame, double the intensity of the above men- 
tioned pencil of rays. Were on the other hand the first flame 
perfectly opaque, all rays reaching it from the second flame 
would be cut off, and the light arriving at the slit would suffer 
no increase in brightness. 
The nearest approach to a second precisely similar flame is 
the real image of the first one. This image would be, when 
of the same size, of weaker intensity than the flame itself; but 
its other properties should, provided the mirror absorb all 
wave lengths of light in equal proportion, coincide perfectly 
with those of the flame. 
T illuminated both halves of the slit of the spectrophotometer 
with two common petroleum lamps with flat burners. Before 
the lower flame, was adjusted a system of cross-wires, of which 
the horizontal ones appeared as dark lines in the polarized 
Spectrum. By means of these it was easy to tell which portion 
of the flame came into the field of vision. Behind this lamp I 
placed a concave mirror, so that a real image of the flame was 
Cast upon the flame itself. This image was of the same size 
but weaker than the flame. Since the horizontal wires ap- 
peared in the spectrum of the image as well as in that of the 
flame, forming another set of black lines, it was possible to 
adjust the mirror so that corresponding portions of flame and 
