THE SPECTRUM OF NOVA PERSEI AND THE STRUCTURE OF ITS BANDS. 2538 
The collimator lens, though a fine object-glass, F/9, did not prove to be achro- 
matised for photographic rays; the focal curve of the spectrograph runs along zero from 
D to H,, then gradually rises, and at wave-length 3700 the ordinate is 2°8 cm. I had 
therefore to incline the photographic plate up to 30°. 
When the telescope is turned through twenty-four hours of hour-angle, the image in 
the focal plane of the spectrograph oscillates in simple harmonic motion, with an 
amplitude of 0°8 mm., along a line which is slightly inclined to the spectral lines. 
1 compensated this deficiency of the mounting by making the plate-holder adjustable in 
the direction of the spectral lines, and at right angles to them, the motion being effected 
by two micrometer screws of 0°5 mm. pitch. ‘The plate-holder takes the plate, 11 by 
4 em., in its upper half, which hes central in the camera, and in the space below it 
carries an eye-piece with stout cross wires. As fiducial line for keeping the plate 
stationary with regard to the spectrum, | employed the magnesium line 4481, which is 
almost covered by the stout wire in the eye-piece. A split spring, which presses against 
the jaws of the slit, cuts out a short line from the upper portion of the slit 1°5 cm. 
above the optical axis, and the magnesium terminals are placed close to it, inside a short 
glass tube, to guard the slit against tarnishing. By these means I am able to keep a 
line of the spectrum always on the same place of the photographic plate, and to replace 
the plate after days in its old position. The differential change of dispersion due to 
changes of temperature is, of course, not taken into account. During an exposure of 
the plate I moved the plate-holder every time 0°01 revolution of the micrometer screws, 
at intervals given by a table, and checked the position once an hour direct on the 
magnesium line. To illustrate the efficiency of this method, I mention that on photo- 
plate No. 23, comparison lines at a distance of 0°04 mm. appear separated, though they 
were exposed on twelve different occasions, five seconds each time, on two days, and at 
different hour-angles. 
At the time the new star was announced, wave-length 5200 t.m. was in the centre 
of the field of the camera, and 4000 at the end of the plate. No change was made in 
the position of the camera until the beginning of October 1901, when wave-length 4170 
was placed.in the centre. | 
The distance of the hydrogen lines H, and H, is 20 mm. on the plate, and one tenth- 
metre or Angstrom unit is represented on the plate by 0°1 mm. at 4=3500, 0°05 mm. 
go — 4300, and 0:025 mm. at A= 5200 t.m. 
3. The Measurements and their Reduction to Wave-Lengths.—The plates taken in 
March and April 1901, and again those after January 1902, were difficult to measure,— 
the former, owing to the gradual change of intensity of the spectrum, which presented 
few definite points to set on ; the latter, owing to the faintness of the spectrum, some 
parts of which could barely be distinguished from the accidental markings on the film. 
I finally adopted the rule to measure every point to which the eye was drawn, except 
those which I thought to be defects in the film. With respect to these, I became more 
careful as the work advanced; and it is possible that the earlier plates may contain 
