THE ULTRA-VIOLET SPECTRA OE THE ELEMENTS. 
189 
positions would be likely to introduce serious errors, we have been obliged to 
be content with the photograph taken on one side being a little, though only a 
little, less sharp in definition than that taken on the other side. The grating was 
used with its plane perpendicular to the axis of the collimator, and it was brought 
into that position in the following way. The telescope and collimator having their 
axes directed as nearly as possible on to the centre of the circle, the telescope was 
placed opposite the collimator and the cross wires brought on to the image of the slit, 
and the reading of the circle taken. The grating was then placed in position and 
adjusted for level until the spectra occupied the middle of the field of view in all 
positions of the telescope. It was then adjusted in azimuth until the images of the 
D lines, from the light of a sodium flame in front of the slit, in the spectra of the 
second order, on the two sides of the normal were at equal angular distances from the 
axis of the collimator, as determined by the reading taken when the telescope was 
opposite the collimator. The grating was then clamped in that position. This 
adjustment had to be made by hand and was liable to disturbance in the clamping, 
so that it was afterwards found that the plane of the grating was not quite perpen¬ 
dicular to the axis of the collimator ; but as the errors arising from this in the 
measures on the two sides nearly compensate one another, the final error in the wave¬ 
length from this cause is very small indeed. 
For measuring the photographs a micrometer was constructed for us by Hilger. 
This is attached to the stage of a microscope and carries a small frame in which the 
photographic plate is held by springs. The micrometer-screw has 100 turns to the 
inch, and by the drum-head y^th of a turn or taooo^ 1 °f an inch can be read. A 
1 inch object-glass to the microscope was used, and measures were made by moving the 
plate until the lines of the photograph were successively bisected by a spider line in 
the eye-piece. The reading of the micrometer gave the distances between the lines. 
The source of light employed was, in the first instance, the arc from a De Meritens 
magneto-electric machine, in a crucible of magnesia into which iron wire was intro¬ 
duced. But from the overlapping of the spectra of different orders, and the large 
amount of light emitted by the arc, we found that the plates were so clouded in many 
places that the lines could not be well seen, and we abandoned the arc for the spark 
between iron electrodes. This was produced by a large induction coil, worked by 
5 Grove’s cells, and having a large Leyden jar connected with the secondary wire. No 
inconvenience arose from the overlapping of the different orders when the spark was 
used, because the parts of the spectra of higher and lower orders which overlapped the 
part of the spectrum of the fourth order to be measured were always considerably out of 
focus, the object glasses of telescope and collimator being uncorrected, and so the light 
of the lines in the overlapping spectra was diffused and produced only a faint clouding 
of the plate, which in no way interfered with the measure of the lines of the fourth 
order. 
