IN CONNEXION WITH THE SPECTRUM OF THE SUN. 
567 
wood at the back, over the centre of a brass circular plate free to move on another 
lower plate levelled by screws. 
The observing telescope, after a few preliminary observations to adjust the grating 
for verticality, was replaced by the long camera before mentioned. 
The focus for any one order was determined in much the same way as in the case of 
the refraction photograph ; that is, it was first roughly obtained by viewing the spec¬ 
trum on the ground glass screen by means of a fixed positive eye piece, but the final 
adjustment was likewise determined by a series of trial plates, and, when satisfactory, 
the sliding end of the camera was clamped up tight in order to keep the adjustment 
as rigid as possible. 
The spectrum of the second order was first obtained. 
In this the distance between the H and K lines was nearly 8 millims. 
With the angle between the collimator and the camera which I used, the H and K 
lines of the third order are brought to focus with the D lines of the second order, so 
that practically the grating and instruments generally are in the first instance adjusted 
for D, with the exception that the violet light is focussed on to the slit. 
In the photograph thus obtained the H and K lines are about 15 millims. apart, 
and the perfection of the grating is demonstrated by the definition of the photographs, 
except in those parts of them where, in consequence of the presence of the glass plate, 
the purity of the spectrum is marred by interference effects.* 
As the exposure for the third order spectrum was much longer than that required 
by the train of prisms, the photographs used were taken by means of a siderostat, the 
use of which gives great constancy to the direction of the beam of light employed. 
At times, an opera glass was used for obtaining a parallel beam; at others, the centre 
of the beam (of 12 inches diameter) thrown by the siderostat was grasped by an 8-inch 
object glass, and focussed on the slit. 
B. Determination of Wave-lengths. 
The small original negative was enlarged by many stages to obviate all chances of 
distortion, until a glass positive was obtained of sufficient size to commence operations 
upon. This was 24 centims. in length. The wave-lengths of the extreme points 
having been taken from Cornu’s map, it was divided into ten parts carefully by com¬ 
passes. It was then found that the intermediate wave-lengths did not correspond 
with those of Cornu. I at first feared that some distortion had, after all, been intro¬ 
duced ; and to test this I employed a photograph on which wave-lengths 38 and 40 
were taken as extreme points, and the distance between them divided into twenty 
portions in like manner. I then found that, although the point 39 was very slightly 
changed, all the rest of Cornu’s measurements between 39-40 fitted the photograph 
very well. That is, I found that the ten points between 39, as thus determined, and 
* Mr. Rutherfurd, who lias carefully studied these phenomena, ascribes them to the so-called “ Talbot 
Bands.” 
