550 
Proceedings of the Royal Society of Edinburgh. [Sess. 
a spectrograph of the usual type with quartz lenses, and it is with it that 
the results recorded in the present paper are obtained. 
The incandescent mantle is not suitable as a source of light in the ultra- 
violet. When we attempted to replace it with an electric spark we could 
not get the latter bright enough with the means at our disposal. The iron 
arc, though brighter, was not found steady enough. So we came to the 
conclusion that it was not advisable to make the comparison exposures one 
after the other. They must be made simultaneously. We had therefore to 
recast our whole photometric arrangement. The diagram shows how this 
vras done. 
The arrangement is easier to understand if we suppose the rays to be 
going in the reverse direction, from the camera to the slit. SS' is the slit 
of the collimator. PQRT is a quartz prism of special design in which PQ 
= PT and QR = TR. The sides PQ and PT are each 2 cm., angle P = 70°, 
c 
angle R = 75°, and the axis of the quartz is perpendicular to the plane of the 
diagram. B x and B 2 are two square plates of ground quartz with their 
planes perpendicular to the plane of the figure ; they can be moved back- 
wards and forwards in the line of the axis of the collimator. G is the iron 
arc ; it is shown vertical in the diagram, although in reality it was 
horizontal. Now consider all the rays that can be drawn from the collimator 
lens to the slit. They are divided into two beams by the sharp edge P of 
the prism, and every ray of each of these beams after refraction through the 
prism PQRT meets one of the ground quartz plates, no matter at what point 
of its range the latter happens to be. Hence, when the light travels in the 
proper direction, from the arc to the slit and thence to the camera, the 
brightness of either of the two spectra formed by the two halves of the slit 
does not vary with the distance of the quartz plate from the slit, but only 
with the state of illumination of the latter. Hence it varies as the inverse 
square of the distance of the quartz plate from the arc. 
The illumination of B 1 is proportional to 1 /d-f and that of B 2 to l/d 2 2 . 
The amount of light falling from B 2 on B 1 is inappreciable in comparison 
with the amount received by B x direct from the arc. If B x and B 2 are at 
