EEV. T. E. EOBINSON ON SPECTEA OE ELECTEIC LIGHT. 
947 
developed in diiferent spectra: the second (not resolved in Duboscq) consists of six 
bright and sharp lines. Even this power fails to decompose the remarkable blue-green 
and violet bands which characterize the hydrogen spectra C.P. Perhaps a third prism 
might, but there is no room for it on the theodolite. 
A third class of lines is narrower than the slit, down to the finest hairbreadth ; they 
are mostly sharp and well defined, sometimes very bright. Their occurrence is not 
easily explained ; for in the ordinary conditions of these observations each ray illumi- 
nates the whole slit, and the image of the line due to it ought to be as broad as that slit. 
The only explanation which has yet been proposed (by Pluckee) is that they are caused 
by the overlapping of two images the distance of whose centres is less than the slit. 
It is, however, liable to two objections — that the brightness of such an overlap cannot 
exceed twice that of the ground on which it is seen, and that it would be often resolved 
by a prism of higher dispersion: the combination 2BS.C disperses seven times as much 
as the Merz, and ought surely to break up some of them. This is not, as far as I have 
examined, the case; and some of these lines are as intense as any in the spectrum. 
These narrow lines are sometimes very thickly crowded, as in the green and blue of iron 
spectra, and in those of carbonic oxide ; and possibly they may compose the bright 
ground when so close as to be unresolvable. There is a seeming tendency in them to 
be grouped in two, three, or higher numbers, which, however, might disappear with 
more powerful prisms ; but with those I use one can scarcely avoid thinking that there 
is some special connexion between the components of such groups ; such are enclosed in 
brackets. 
Some of the most brilliant are double, as the orange one y, supposed to correspond 
with D (though its components seem to me more separated than those of that line), the 
splendid yellow S, and the greens rj and &. 
At that boundary of the spectrum which corresponds to the negative electrode (and in 
a much less degree at the positive) extremely intense lines are seen, especially in the 
green, which however are short : bismuth, zinc, lead, and arsenic are the best examples 
of this. These are generally supposed to be metallic lines, and to proceed from the 
intense dispersion of metal near the electrodes, especially the negative. It has even 
been proposed to consider this not crossing the entire breadth of the spectrum as a test 
of metallic origin, and to regard the others as gaseous. I, however, find that these very 
lines can be traced entirely across though sometimes very faint, except when the entire 
part is covered with a sort of bright haze, through or on which nothing can be seen 
which is not very bright. It seems to me that the existence of a line and its brightness 
depend on different causes ; and I shall give instances when the same line assumes very 
different aspects as the media of discharge are changed. I may add that the sort of 
haze just mentioned does not occur at the negative electrode. 
The spark discharge without a jar is so much fainter that none but the brightest lines 
can be seen. If the surface of the jar be increased (for the converse reason) more lines 
are visible, but I think no new ones are produced. One or two examples will be given, 
MDCCCLXII. 6 0 
