EEV. T. E. EOBINSON ON SPECTEA OE ELECTEIC LIGHT. 
97S 
I failed totally in getting the O spectra ; for at the very first discharge the mercury was 
so rapidly oxidated that not a glimmer could pass the thick deposit on the tube. This 
singular energy is probably due to the presence of ozone in the electrolytic oxygen 
(though it was passed through a tube filled with silver-leaf). I shall repeat the expe- 
riment, passing it through a red-hot tube. This fact may explain the absorption of 
oxygen in Geissler tubes, which was observed by Plucker ; for Hg vapour must be 
present in them, as they are exhausted by a mercurial air pump. The deposit in CO 
was black below, grey above : the first was dissolved by nitric acid, the other not ; it 
was probably some form of carbon from the decomposition of the gas ; of that, how- 
ever, there was a large residue present. The spectrum of Hg vapour has a few very 
brilliant lines, but the rest are narrow and faint on a very dark ground. I had 
expected by comparing it with Pt, Hg vapour. Table II., to obtain at once the lines of 
platinum, which I supposed would be sujperadded to those of Hg; but, to my great 
surprise, that spectrum \i 2 is fewer lines (28, while the other has 48), and only three occur 
in it which the other wants. A similar deficiency occurs in the gas spectra of Table 
XXL: X, C.P., wants 22 mercurial lines; N, E., 25; H, C.P,, 20; CO, C.P., 12; and 
CO, E., 30. These observations, it must be remarked, are strictly comparable; they 
were aU made with the Duboscq prism, nearly at the same time, and with the apparatus 
in precisely the same condition. 
From this may be inferred, either that mercury gives different lines from its vapour, 
or that the gaseous media have lines which interfere with those of the electrodes so as to 
destroy each other. 
I must reserve for a second communication the complete discussion of these and the 
other spectra, as it would, I fear, swell this one beyond all reasonable limits. 
[P.S. (See page 954.) A more probable explanation of this peculiar form of the 
spectrum was suggested to me by Mr. Stokes, which is as follows : — 
The aperture of the collimator was 1 inch, and its focal length 9 ‘5 inches. The 
distance of the spark from the slit was probably less than 1’5 inch, say 1'25. Hence 
if we assume for simplicity that all the light emerging from the collimator entered the 
object-glass of the observing telescope, which was nearly true, each element of the sEt 
1*25 1 
would be capable of being illuminated by rays from a length Z=-^Xl=y:g inch of 
the spark, pro\ided the spark reached so far, the centre of that line being in a pro- 
longation of the line joining the centre of the object-glass with the element in question. 
If the length of the spark were very great compared with I, the central parts and those 
near the electrodes would illuminate widely separated elements of the slit, and the varia- 
tions of brightness in a vertical direction would nearly correspond with the real varia- 
tions of brightness of the different parts of the spark — though of course there would be 
a “ ragged edge ” above and below, where the spectrum was formed by partial pencils. 
If, on the other hand, the length of the spark were infinitely small compared with I, 
6 R 2 
