984 
EEV. T. E. EOBINSON ON SPECTEA OE ELECTEIC LIGHT. 
evident both that the spectrum of air is not formed by the mere superposition of those 
ofN and O, and that certain electrodes can excite lines in a mixture of two gases which 
are not visible in either of those gases taken separately, or cannot excite them in the 
mixture, though they can in either or both of the components. This unexpected fact is 
illustrated by an attempt which I made to obtain the lines of platinum, by compaiing 
its spectrum in mercury vapour with that of mercury. I expected, as I had done in the 
cases already mentioned of gases with Hg electrodes, that I should get the spectrum of 
Pt+that of Hg ; but the result was otherwise. Hg Hg had forty-eight lines, Ft Hg only 
twenty-five, so that the presence of Pt instead of Hg as electrodes out twenty-three. 
It however brought out four new ones, of which No. 15 7 is found with every solid metal, 
Nos._33, 59 %!', and 76 | with most of them. There is here no superposition, but, instead, 
either an antagonism of platinum and mercury, or the curious fact that the fluid elec- 
trodes act differently from the solid. It would be interesting to compare the spectra of 
tin in these two states. 
The case is the same with a chemical compound as with a mixture. On the prin- 
ciple of superposition, the spectrum of CO should be merely the sum of those of C and 
O. To avoid all uncertainty about metallic lines, I used electrodes of graphite in the 
two gases. In each case nothing but C and O were present, and I expected to get the 
same spectrum in both. They were, however, very unlike. Of the lines which they 
had in common. 
No. 3 a is nc. in . 
No. 10 /3' is a * in 
No. 15 7 is a in 
No. 22 ^ is intense 
No. 26 2 is wvb. in 
No. 39 is faint in 
. O, but is a * in 
. O, but only c. in 
O, but only b, in 
in O, but only b. in 
CO 
CO 
CO 
CO 
O, but one of many very fine in CO 
O, but a * as bright as in . CO 
The K group is not nearly so bright in O as in CO. 
Besides these differences, O has ten wanting in CO, among which are No. 38 ??, very 
bright, and No. 44 (f, in all its brilliancy ; while CO has eight wanting in O, of which 
No. 75 is a-*; the difference being greater than I have sometimes found with much 
less chemical agreement. 
The only conclusion which such facts permit is, that the spectrum is a simple resultant 
of all the actions present, some of which may combine to produce an exalted effect, 
while others may be antagonistic in any degree. 
4. As to the manner in which electricity produces the rays of these lines, whether by 
merely heating the medium, or by some luminiferous action analogous to its heating, 
nothing is really established. Heat is known to produce some brilliant lines when metallic 
vapours are introduced into flames ; and it is possible that a temperature far above that 
of our hottest flame might bring out all those which I have enumerated, and the multi- 
tude of others which I did not attempt to measure. But is the temperature of the elec- 
