CAPTAIN W. DE W. ABNEY ON THE SOLAR SPECTRUM. 
459 
spectrum of the third order is brighter than the second order on one side, and the 
second order brighter than the first order on the other side. This shows that the very 
greatest care must always be taken in deducing any value for radiation in any particular 
region unless the constants of the particular grating employed are known. Both of 
these last-mentioned gratings have been employed for the producing the photographs to 
which the map is referred. Besides these, Professor Howland sent me for this work 
two beautiful concave gratings of about 3600 lines to the inch, of 5 feet 7 inches and 
4 feet 8 inches radii respectively. These have been employed for the still lower 
regions of the spectrum. 
Taking a hint from Professor Rowland, I used the concave gratings with their 
axes perpendicular to the photographic plate, but I adopted a plan which is somewhat 
different to that proposed and utilised by Professor Rowland. The annexed figure 
will give an idea of a mounting for the 7 feet 6 inches concave grating. 
Fig. 1. 
A A is a plank some 8 feet 6 inches long; at one end of it, E, is placed the grating, 
with its axis lying along the centre of the plank ; G G is a camera, as shown, the 
plate being inserted at H at exactly the distance of the radius of the grating from its 
centre. At a point bisecting the radius of the grating a pivot is fixed in the plank A, 
round which the plank B can turn, and another pin or pivot is placed at D, also 
exactly at a distance of half the radius from the first pivot. Round D the plank C C 
can turn, which carries a tube, F, with the slit S attached. S is placed exactly over D. 
By this arrangement the slit S and the tube F can always be made to point towards E, 
no matter to what angle the plank B is turned away from A ; and as D travels on the 
circumference of a circle in which the plane of the plate and the grating lie, the 
diffraction spectra are always in focus on H, and the scale of the spectrum is invariable. 
This plan I adopted for portability’s sake, and also because my laboratory is somewhat 
confined in extent. At K is a slow motion screw, of which Fig. 2 is a representation 
on a larger scale, which turns the tube bearing the slit, and I found this arrangement 
better than turning the slit itself: a is a band encircling the brass tube, b is a screw 
turned by K, which screws into a swivel at c ; when K is turned the tube F turns in 
3 n 2 
