DR. A. SCHUSTER ON THE SPECTRA OF METALLOIDS. 
45 
I have eliminated the effects of the electrodes by using different metals, such as 
aluminium, platinum, silver, brass, and iridium. Only in case of aluminium did I 
observe lines which could be traced to the electrodes. When the vacuum is very good 
a spectrum is sometimes faintly seen, which can be easily obtained from aluminium, 
and which I believe to be due to the oxide of aluminium. The spectrum, however, is 
capricious in its appearance, and as aluminium electrodes were used in the great 
majority of experiments, it is quite possible that an effect of the electrodes of other 
metals exists, but escaped my notice. 
Any possible effect of the glass was eliminated by using tubes of different width. I 
have finally seen all the lines or bands which I ascribe to oxygen in a glass receiver six 
inches in diameter. The electrodes were about one inch apart, and no part of the 
spark was within two and a half inches from the glass. 
4. Measurements. 
It is important to vary both the dispersive and magnifying power of the instrument. 
Some of the phenomena are best seen under small dispersion, others under large 
dispersion. I have examined the various spectra of oxygen with flint-glass prisms of 
55° refracting angle, using almost any number from one up to seven. Whenever 
measurements are to be made it is best to use as large a dispersion as possible. Some 
of the measurements which I shall give are made with a dispersion of seven prisms. 
During the investigation, however, I procured from Mr. Hilger, in London, two very 
heavy flint-glass prisms (spec. grav. 5), of 62° each. These prisms contain much lead : 
they are yellow, and therefore can only be conveniently used to about the solar line G. 
The dispersive power of each is fully equal to that of two ordinary flint-glass prisms of 
the same refracting angle. Each prism under minimum deviation separated the D lines 
through an angular distance of 52 seconds. The focal length of the observing 
telescope was 17'4 inches. The micrometer screw of the eye-piece with which the 
measurements were made had 100 turns to the inch. The distance between the two 
D lines when both prisms were used was 0'88, taking as unit one turn of the screw. 
Most of the measurements were made with these prisms. 
During the spring of 1872 I published an investigation on the spectrum of nitrogen 
(Proc. Hoy. Soc., xx., 484), in which I showed that when nitrogen is heated up with 
sodium in a vacuum tube the nitrogen bands disappear, and a line spectrum appeared 
in my tubes, which, when measured with the instrumental means at my disposal, was 
found to agree fairly well with the line spectrum of nitrogen. Salet has since shown 
that the cause why the bands disappeared in my experiments must be looked for in 
the absorption of nitrogen by sodium, under the influence of the electric spark. The 
lines which I saw were really due to the metallic sodium. In order to avoid any 
similar error, I have carried my measurements to a degree of accuracy which renders 
any confusion between lines of different bodies quite impossible. 
