557 
1909-10.] The Efficiency of Metallic Filament Lamps. 
wave-lengths by means of the spectroscope and thermopile or bolometer and 
plotting it as a function of the wave-length. Then if an ordinate he erected 
at 760 jul/ul, the radiant efficiency is the area of the curve on the side of this 
ordinate towards the shorter wave-lengths divided by the area of the whole 
curve. G. W. Stewart * has determined the radiant efficiency of the acety- 
lene flame in this way. The other method, which has been introduced by 
K. Angstrom,]* is not such a simple one. It has been applied with modifi- 
cations to the case of the carbon glow-lamp by C. E. Mendenhall, J who 
obtained 2 8 as the radiant efficiency of the latter at its normal brilliancy. 
This, I think, is the only previous determination of the radiant efficiency 
of a glow-lamp to which exception cannot be taken on the point of 
method. 
The method I have applied is, as far as I am aware, an original one, 
though it almost follows from the article of Nichols and Coblentz. It seems 
preferable to both the above methods. The radiation was first measured 
in the usual way by a thermopile and galvanometer with and without a 
1 cm. thick water filter, and the percentage of total radiation transmitted 
through the filter determined. In this case the radiation from the whole 
lamp was received by the thermopile. A straight part of the filament of 
average brightness was then focussed by an achromatic glass lens on the 
slit of a spectrometer with glass prism which was furnished with a Rubens 
linear thermopile in place of the cross wires. The same water filter was 
placed in front of the slit and an energy curve of the filament taken through 
the filter. For the region of the spectrum for which water transmits, the 
absorption of light in the glass and the loss of light by reflection at the 
glass surfaces may be taken as independent of the colour. If the prismatic 
energy curve thus determined be taken and an ordinate set up at the point 
760 iu / ul , the area on the side of shorter wave-lengths divided by the whole 
area gives the fraction of the radiation transmitted by the filter, which is 
light. It is only necessary now to find what fraction of the incident 
light is transmitted by the water cell. This was done by measuring 
the transmission coefficient of the latter when filled with water with a 
spectrophotometer for four different points in the spectrum. The value 
found was 0‘84. It is somewhat low, as the glass sides of the cell absorbed 
rather more than usual ; it did not not vary with the colour. 
* “The Spectral Energy Curve of the Acetylene Flame,” Physical Review , xvi., 1903, 
p. 123. 
t “Energy in the Visible Spectrum of the Hefner Standard,” Physical Review , xvii., 
1903, p. 302. 
f “On the Luminous Efficiency of the Carbon Filament,” Physical Review, xx., 1905, 
p. 160. 
