EEY. T. E. EOBINSON ON SPECTEA OF ELECTEIC LIGHT. 
981 
No. 5 is not in any other H spectrum; No. 82 only in Te, A and As, H ; No. 16 only 
in PbO. None of these gas lines is peculiar to that one gas ; thus No. 16 is found in N 
with nine metals and in H with five ; this method therefore fails to give the true gas lines. 
Most of those that are common to two gases will be found in the spectra of sodium, 
potassium, calcium, and lead, all of which are ingredients of glass ; it is hence evident 
that even reciprocating discharges disintegrate the sui’face of glass. 
A plan of mine seemed more promising. With an apparatus which I have already 
described, I took the spectrum in mercury vapour with mercury electrodes ; then filling 
the tube with a gas, I took the new spectrum. I tried N, O, H, and CO. O failed from 
its extraordinary action on the mercury, which in a few seconds blackened the tube, so 
that not a ray from the flash was visible. I expected that the gas lines would be thus 
easily obtained, but was disappointed. 
Hg Hg has 48 lines 
Hg, N, C.P., has 36 lines 
Hg, N, E., has 23 lines 
Hg, H, C.P., has 33 lines 
Hg, CO, C.P., has 41 lines 
Hg, CO, E., has 18 lines 
N,C.P., 
Of these N, E.,. 
H, C.P., 
CO, C.P., 
CO, R, 
has 10 not in Hg Hg 
has 4 
has 7 
has 10 
has 8 
Of these last Nos. 2, 46, 59, and 78 are found in N only (that is with Hg electrodes, 
for with other metals they occur in all the gases). Nos. 13 and 32 belong to CO, the 
rest are common to the three gases. It deserves notice that many of the lines of Hg Hg 
are wanting in the gases, the lowest number being 13 in CO, C.P., the highest (29) in 
CO, E. This may partly arise from mercury and its vapour having difierent sets of lines, 
the latter of which are displaced by those of the gas, fewer in number Were this the 
sole cause, the same lines would be missing in each gas spectrum ; and the differ- 
ence of the numbers show's that the presence of a gas prevents the develop- 
ment of some mercurial lines. Corresponding facts occur in other spectra, those of H 
especially. 
The effect of a metal is most perceptible near the negative boundary of a spectrum, 
where some lines are brilliant for a short distance and then continue with much less 
brightness. The contrast is so great that, without measurement, one could scarcely 
beheve the line to be the same. That its existence in the faint state is due to that 
metal w'hich causes its partial splendour, is disproved by its occurring with other metals. 
In illustration I refer to the spectrum of bismuth (N) at the negative edge as compared 
to that of the centre (Table XII.), where it is seen that though lines (especially in the 
green) which are scarcely visible in the latter are intensely bright in the former, and 
faint single lines are seen as two, yet no decidedly new one appears. This is most 
remarkable with the volatile metals. Thus As, H, C.P., shows at the negative edge forty- 
four lines, of which ten are ^^s and eleven more are “ bright.” These retain their lustre 
for 10' or 11', and then (except the three hydrogen bands) can scarcely be traced across. 
6 s 2 
