IN RELATION TO QUANTITATIVE CHEMICAL ANALYSIS. 
59 
and are particularly strong when the carbons, wet or dry, are immersed in pure and 
dry nitrogen. When the negative pole is a metallic wire immersed in water, and the 
positive a point of carbon, the lines are strongly developed at the positive pole only. 
When the current is reversed the spectrum is not so strong. The bands not being 
traceable to carbon dioxide, to a hydrocarbon or to air, and being at the same time 
produced with great intensity when graphite is surrounded by pure and dry nitrogen, 
they must belong either to a modified spectrum of carbon or to a compound of carbon 
with nitrogen, such as cyanogen. In order to test the probability of these flutings 
belonging to the cyanogen spectrum, it was thought desirable to try the effect of 
cyanides in solution, using for the purpose metallic electrodes. A sample of perfectly 
pure cyanide of potassium was prepared by dissolving the commercial salt in alcohol 
and crystallising it therefrom. A saturated solution of the salt gave no sign of the 
bands when submitted to the spark passed between gold electrodes. A hot saturated 
solution of mercuric cyanide was similarly treated, the negative electrode being gold 
and the positive carbon, the gold being all but immersed in the solution. A strong 
mercury spectrum was obtained, but the flutings were absent except at that point 
where they might be expected under ordinary circumstances, namely, at the positive 
carbon electrode, and even here they appeared as faintly as if water and not a saline 
solution had been used. Cyanides therefore do not yield this spectrum. 
It is remarkable that certain solutions which do not contain nitrogen in any form 
favour the formation of these bands, as for instance chlorides generally and zinc 
chloride as a saturated solution particularly. In order to ascertain whether the 
strength of the spectrum is dependent on the proportion of saline matter present, 
I examined a series of solutions containing varying proportions of calcium chloride, 
from xfnjth to tooTTo^L with the result that the strength of the bands was found 
to increase with the strength of the solutions. 
There are two lines in this spectrum of graphite (Journal of the Chemical Society, 
vol. xli., p. 90) which apparently commence these flutings ; their wave-lengths are 
3875 ‘7 and 3870'7, and yet two others which are absent from the spectrum of carbon 
when taken in oxygen with wave-lengths 3585’5 and 3584'0. 
Professors Liveing and Dewar give the general appearance of the cyanogen 
spectrum as observed by them (Proc. Royal Soc., vol. xxxiv., p. 123), and the group 
of lines between K and L, extending from 3883 to about 3830, much resembles the 
first group mentioned above, while the second series near N, lying between 3580 and 
3590, approximate closely to the second pair of lines to which I refer; nevertheless I 
cannot attribute these lines to cyanogen because they are not obtainable from 
cyanides. 
That these lines are absent from the carbon spectrum when taken under certain 
conditions is not conclusive evidence that they are not carbon lines, because, as I have 
already shown, the lines 4266'3, 3919'5, 3881‘9, 3875 '7, 3870*7, &c., are occasionally 
absent, yet these unquestionably belong to the carbon spectrum. 
