126 
MR. W. GEOFFREY DUFFIELD ON THE 
As there was a certain amount of risk of the window being blown out, it was usual 
to operate the feeding of the arc from a little distance by ropes passing round the 
pulleys on the covers ; save, however, for some slight chipping of the inner surface, 
due to straining, the window has remained intact. 
3. The Illumination of the Slit, 
During the present research, in which the large dispersion of the 21-feet 6-inch 
Rowland grating was employed, it was found that extremely long exposures were 
required, and a number of unsuccessful attempts were made before a photograph of 
the spectrum under pressure w r as obtained. It became, therefore, most important to 
obtain the maximum illumination of the grating from the light passing through the 
window of the pressure cylinder. 
The system of lenses employed is shown in the diagram : lenses K and L are at 
their focal distances from the arc and slit respectively. 
sue 
Grating 
Fig. 9. 
The cone of light from the lens L passes through the slit, and a little more than 
just fills the grating ; this is not the usual arrangement, but the conditions differ 
from those contemplated by Schuster, # who finds that the conditions for maximum 
uniform illumination are fulfilled when the angle subtended by the lens is four times 
that subtended by the grating at the slit. In the present instance the original cone 
of light is limited by the window in the cylinder, and the most efficient use is made 
of it by refracting it into the slit by a lens which subtends at the slit the same solid 
angle as the grating. The central maximum of the diffraction pattern for each ray 
then falls upon the grating, ensuring a sufficiently (but not perfectly) even illumina¬ 
tion of its surface, and the loss of very little light. 
Dealing with the concave grating, the size of the image of the arc on the slit is of 
some influence on the intensity of the spectral lines, because the astigmatic properties 
of the ruled surface serve to partially integrate the effects from different parts of the 
slit. In a prism spectroscope this is not the case, and the size of the slit does not 
affect the intensity of the lines. The lens K was therefore introduced in order that 
the image of the arc should just fill the slit; in this way the intensity of the 
illumination of the slit was not altered, since the angle subtended by the lens L at 
the slit remained constant, but the total quantity of light received over the area of 
the slit was then a maximum. 
* A. Schuster, ‘ Astrophysical Journal/ XXL, p. 209, 1905. 
